Conference Proceedings 2016

June 21 to 24, 2016
Boulder - Colorado
Safe Engineering Services & Technologies, Ltd.
(450) 622-5000 Toll free 1-800-668-3737
www.sestech.com
Copyright © 2016 SES ltd. All rights reserved. No part of any of the content of this document may be reproduced,
distributed, modified, or made available in any form by any photographic, electronic, digital, photostat, microfilm, or
other means, or incorporated into or used in any information storage and retrieval system, electronic or mechanical,
without the prior written permission of the copyright owner (SES) or the authors of the pages of the document to be
reproduced, distributed, modified, made available, incorporated into or used in any information storage and retrieval
system.
PRESIDENT’S ADDRESS
I am pleased to welcome you to the 26th annual CDEGS Users’ Group Conference hosted by SR3
Engineering LLC, in Boulder, Colorado. This year marks the CDEGS Users’ Group Conference
silver anniversary. For 25 years now, this annual tradition has proven to yield lively and fruitful
discussions related to technical projects and the often novel strategies used to tackle them, as well
as on the new, planned, and desired features of SES’s software packages. Over the last quarter
century, the community of CDEGS users and contributors has grown tremendously. The current
state of the software and the promise it holds for the future, is a direct result of their cumulative
contributions, and we proudly dedicate these proceedings to those who are not with us this week
in Boulder, but are always with us in spirit. Finally, SES is eager to see the impact of the mountain
air on the valiant efforts of the volleyball players on Tuesday and Thursday.
For the third year running, SES is unveiling a set of new modules and modernized interfaces to
enhance the user experience and improve your productivity. We present beta versions of the new
integrated SESResap and FFTSES computation modules that can now run directly from within
the module environment and where both the input data and the computation results can be
visualized in this same environment. Complementing the new soil measurement interpretation
environment, we present this year the new Soil Model Editor, which will simplify and streamline
the specification of soil models. A new SES Object Library is introduced, which includes not only
a comprehensive conductor database, but also several power cables, and comes equipped with
handy search and filter capabilities, while the new SESPlotViewer module ushers in the new
generation of plotting capability via the SES Plotting Engine, which will eventually be integrated
into all SES Software that needs to plot data. Even the main CDEGS interface is revisited in this
year’s release.
While we forge ahead with new interfaces and modules, we continue to improve and enhance our
long-serving software package features and computation capabilities as well. For example,
SESCAD, now offers direct conductors import from the SES conductor database, provides a
complete, autonomous interface to enter data for the MALT, MALZ and HIFREQ modules. Right-
of-Way and RowCAD now handle a direct kml file import, streamline the Total Interference
computation under steady state conditions, and include new and improved maximum soil
breakdown voltage and various fault arcing and flashover distance options. These, and other
already well-known components of SES’s software suite, continue to grow and improve.
We are grateful to SR3 Engineering for hosting this conference, and to Mr. Giancarlo Leone
and Mr. Doug Gilroy, this year’s Users’ Group chairman and vice-chairman, to Ms. Nicole
Gruber and Mr. Vinod Simha, this year’s secretary-treasurer and assistant secretary-
treasurer, for their hard work.
In addition to the organizers, I would also like to thank all participants for their contributions
through their questions and comments during the presentations, workshops and Q&A sessions,
or even between sessions in informal conversation. An additional thanks to those who took time
out of their busy schedules to prepare presentations and articles, in order to share their experience
and acquired wisdom. It is this exchange of knowledge and ideas that makes the conference so
valuable.
Copyright © 2016 SES ltd. All rights reserved. Page i

Whether in designing a new module, a better interface, adding functionalities to our software, or
providing you with support, SES puts your needs and opinions first. SES’s continuous expansion
has led us to now occupy a second office building, allowing us to serve you better through new
software and stronger support capabilities. Your questions, comments and requests, either made
as wish list items at this conference, or during the course of your work and consultations with our
support staff all year round, are not only welcome, but are of capital importance. Please do not
hesitate to share them with SES!
Finally, let me underline SES’s enduring gratitude to YOU, our legion of loyal users, who have
renewed technical support from SES, year after year, thus keeping SES in touch with your needs
and financially sustaining the R&D that goes into delivering to you the world’s most advanced
power system grounding and electromagnetic compatibility software, along with technical
support from internationally recognized experts, whose priority it is to help you when you need
them.
Farid Paul Dawalibi
Group photo of the first Users’ Group Meeting in 1991.
Page ii Copyright © 2016 SES ltd. All rights reserved.

CHAIRMAN’S WELCOME
26th Annual CDEGS Users’ Group Conference
Boulder, Colorado, USA

June 21 to 24, 2016
Dear CDEGS Users:

Welcome to the 2016 CDEGS Users’ Group Meeting!
This year’s meeting is set against Boulder, Colorado’s signature Rocky Mountain backdrop. The
meeting will be held at the Millenium Harvest House Hotel, and hosted by SR3 Engineering, LLC.
This year’s meeting promises to be filled with an abundance of technical content provided by users
and SES. The presentations often include users’ case studies, technical papers, and theoretical
applications of electromagnetic analysis. Additionally, SES provides a forum for users to request
features in the software, and presents new developments that are currently in the works. In my
opinion, the Users’ Group Meeting offers unparalleled value in terms of the content and
opportunity for interaction with the developers of the software and expertise in the industry.
As for the venue, the city of Boulder offers scenic views and close access to the Rocky Mountains,
as well as numerous activities and sights to see around the city. For the outdoorsy types, there are
numerous beautiful hiking and biking trails; for those more suited to the city, the unique, brick-
paved Pearl Street pedestrian mall offers shopping, food, drinks, and the best people-watching
around. If you’re into tea, the nearby Celestial Seasonings factory has a great tour, which includes
an unforgettable stop in the sinus-clearing “Mint Room”.
I want to take this opportunity to thank Farid and the entire SES staff for their support and
involvement in this meeting - without them, the Users’ Group Meeting would lose a significant
amount of technical contribution, as well as logistical support involved in the months preceding
the meeting.
I also want to thank the executive committee for their help in organizing the conference: Vice-
Chair Doug Gilroy from Bechtel, Secretary-Treasurer Nicole Gruber from Stanley
Consultants, and Vice-Secretary-Treasurer Vinod Simha from AEP. I want to
particularly thank Nicole for taking over the Secretary-Treasurer position at the last minute, and
without having had a year of preparation in the “Vice” position.
Thank you for your involvement in this wonderful organization. It is an honor to serve as the Chair
and host for this year’s meeting.
Sincerely,
Giancarlo Leone
2016 CDEGS Users’ Group Chair
Copyright © 2016 SES ltd. All rights reserved. Page iii

Table of Contents
Error! Cannot open file referenced on page 1
President’s Address.......................................................................................................................... i
Chairman’s Welcome ..................................................................................................................... iii
Part I: User Contributions ........................................................................................................ 1
1 Influence of ground consolidation on local soil resistivity, earth impedance and safety
potentials for a new transmission substation .......................................................................... 1-1
1 Introduction.................................................................................................................... 1-1
2 Background ..................................................................................................................... 1-1
3 Soil Characteristics and Ground Consolidation ............................................................ 1-2
4 Influence on the Earth Impedance and Electrode System ............................................ 1-4
5 Effect of the Man-Made Ground on the Surface Potentials ...........................................1-7
6 Conclusions................................................................................................................... 1-11
7 Acknowledgments......................................................................................................... 1-11
8 References..................................................................................................................... 1-11
2 Fault Conditions at Substations without a Ground Connection to the Main Grid: Earth Fault
Current Calculation and Ground Potential Rise (GPR) .......................................................... 2-1
1 Introduction................................................................................................................... 2-1
2 Methodology and Assumptions ..................................................................................... 2-2
3 Mitigating Excessive GPR .............................................................................................. 2-7
4 Conclusions.................................................................................................................... 2-9
5 References...................................................................................................................... 2-9
3 Evaluation of non-traditional transmission line shielding configurations for reducing
magnetically induced voltage on parallel metallic objects in joint use corridors ................... 3-1
1 Introduction................................................................................................................... 3-1
2 Analyzed Cases .............................................................................................................. 3-2
3 Results ........................................................................................................................... 3-8
4 Conclusion ................................................................................................................... 3-12
5 Acknowledgement ....................................................................................................... 3-13
6 References.................................................................................................................... 3-13
Part II: SES Contributions ........................................................................................................ 1
4 Inclusion of information regarding deep layer soil from remote long traverses to improve
deep soil resistivity estimates ................................................................................................. 4-1
1 Introduction................................................................................................................... 4-1
2 Methodology .................................................................................................................. 4-2
3 Case Study...................................................................................................................... 4-4
4 Conclusion ................................................................................................................... 4-15
5 References.................................................................................................................... 4-15
5 Can a Grounding Grid Impedance Be Capacitive at Power Frequencies? .............................. 5-1
1 Introduction................................................................................................................... 5-1
2 Computer Simulations ................................................................................................... 5-1
3 Results and Discussion .................................................................................................. 5-2
4 Validation ...................................................................................................................... 5-9
5 Conclusion ..................................................................................................................... 5-9
6 References.................................................................................................................... 5-10
6 Analysis of an Underground Urban Substation Grounding System ....................................... 6-1
1 System under Study ....................................................................................................... 6-1
2 Important Considerations Made in the Analysis........................................................... 6-4
3 Conclusions.................................................................................................................. 6-13
7 Studying Safety in Underground Distribution System Vaults with HIFREQ ..........................7-1
1 Introduction....................................................................................................................7-1
2 Modeling of the System ................................................................................................. 7-2
3 Case Study...................................................................................................................... 7-8
4 Conclusion .................................................................................................................... 7-15
5 References..................................................................................................................... 7-15
8 Calculation of Magnetic Field from Solenoids and Reactors of Different Shapes .................. 8-1
1 Introduction................................................................................................................... 8-1
2 Reactor Models Created in HIFREQ ............................................................................. 8-2
3 Calculated Magnetic Flux Density ................................................................................. 8-4
4 Conclusion ................................................................................................................... 8-19
9 Modelling Network of Metallic Plates and Cylindrical Conductors in HIFREQ .................... 9-1
1 Introduction................................................................................................................... 9-1
2 Formulation of New Features ........................................................................................ 9-2
3 Numerical Results ......................................................................................................... 9-5
4 Conclusions and Future Developments ....................................................................... 9-18
5 References.................................................................................................................... 9-19
10 Modeling of Lightning Channel using HIFREQ and FFTSES ..............................................10-1
1 Introduction..................................................................................................................10-1
2 The Antenna Model of Lightning ................................................................................ 10-2
3 FFTSES Input Data for the Transient Current at the Base of the Channel ................. 10-3
4 Modelling of Lightning Channel in HIFREQ .............................................................. 10-4
5 Numerical results......................................................................................................... 10-8
6 Conclusion ................................................................................................................. 10-23
7 References.................................................................................................................. 10-24
11 Advancements in the Treatment of Nonlinear Devices in the Frequency Domain .............. 11-1
1 Introduction.................................................................................................................. 11-1
2 Mathematical Model ..................................................................................................... 11-2
3 Application to Surge Arrester Lightning Performance Analysis .................................. 11-4
4 Comparison between time and frequency domain approaches ................................... 11-8
5 Conclusion .................................................................................................................. 11-15
6 References................................................................................................................... 11-15
Part III: New Features ............................................................................................................... 1
12 Improvements in SES Software ............................................................................................ 12-1
1 Introduction.................................................................................................................. 12-1
2 New User Interface Programs ...................................................................................... 12-1
3 HIFREQ: Junctions between Metallic Plates and Wires ..............................................12-7
4 ROWCAD/Right-of-Way: Total Interference under Steady-State, Arcing Distance, KML
files, and much more… ................................................................................................ 12-8
5 CorrCAD: Coating Effective Resistance, Changes of Cross-Section, and Improved
Display ........................................................................................................................12-10
6 SESShield-3D: Accuracy Level and Component Identification ................................. 12-11
7 SESThreshold: Undo/Redo and Copy Settings between Touch and Step and between
Zones........................................................................................................................... 12-12
8 SESCAD: Complete Editing Capabilities and Access to Conductor Database ........... 12-13
9 SESeBundle: Hollow Equivalent Conductor .............................................................. 12-14
10 SESEnviroPlus: User-Defined Methods for Corona Performance ............................. 12-15
11 Concurrent Runs with a Multi-User Network License ............................................... 12-16
13 A New Interface for CDEGS .................................................................................................. 13-1
1 Introduction.................................................................................................................. 13-1
2 Application Tour ........................................................................................................... 13-1
3 Conclusion .................................................................................................................. 13-11
14 Improvements in SESCAD .................................................................................................... 14-1
1 Introduction.................................................................................................................. 14-1
2 SESCAD as a Complete Editor - Advanced Options ..................................................... 14-1
3 Outer Radius on Conductor Type and Importing Conductor Types from SES Conductor
Database ...................................................................................................................... 14-3
4 Delete Unused Characteristics..................................................................................... 14-8
5 Network Gaps .............................................................................................................. 14-9
6 Cables in Object Database .......................................................................................... 14-11
7 Selection Handles ....................................................................................................... 14-11
8 Specification of Energization on Plates in HIFREQ ................................................... 14-13
9 Outward Normal and Edge Continuity for Solid Objects Made of Plates .................. 14-15
10 Importing Default Types when Importing Files ......................................................... 14-17
11 New Soil Model Editor Component ............................................................................ 14-17
12 Conclusion and Future Developments .......................................................................14-18
15 SESResap: A New Soil Resistivity Measurement Editor...................................................... 15-1
1 Introduction.................................................................................................................. 15-1
2 Application Tour ........................................................................................................... 15-1
3 Other Features ............................................................................................................ 15-11
4 Example ...................................................................................................................... 15-13
5 Conclusion .................................................................................................................. 15-13
6 Acknowledgements ..................................................................................................... 15-14
7 References................................................................................................................... 15-14
16 SoilModelEditor: New SES Software Tool for Soil Model Design ........................................ 16-1
1 Introduction.................................................................................................................. 16-1
2 SoilModelEditor as a Standalone Application .............................................................. 16-1
3 SoilModelEditor as a Component................................................................................ 16-4
4 Modifications to Soil Type Specification ..................................................................... 16-5
5 Conclusion and Future Development .......................................................................... 16-7
17 New Features and Enhancements for the calculation of safety limits in SESThreshold ...... 17-1
1 Introduction.................................................................................................................. 17-1
2 New Features and Enhancements for SESThreshold...................................................17-3
3 New Features and Enhancements for Zone Editor ..................................................... 17-9
4 New Features and Enhancements for both SESThreshold and Zone Editor ............. 17-12
5 Conclusion .................................................................................................................. 17-13
6 References................................................................................................................... 17-13
18 Introducing the New SES Object Library.............................................................................. 18-1
1 Introduction.................................................................................................................. 18-1
2 Conductor Database ..................................................................................................... 18-1
3 Cable Database ............................................................................................................ 18-3
4 The Graphical User Interface ...................................................................................... 18-4
5 Using SES Object Library .......................................................................................... 18-10
6 Conclusions and Future Developments ......................................................................18-16
19 New Features in SESeBundle: Finding an Equivalent Hollow Conductor Representing a
Bundle of Conductors............................................................................................................. 19-1
1 Introduction.................................................................................................................. 19-1
2 Methodology ................................................................................................................ 19-2
3 The User Interface ....................................................................................................... 19-2
4 Examples and Validations ........................................................................................... 19-3
5 Limitations................................................................................................................... 19-8
6 Future Work ................................................................................................................ 19-9
7 Conclusion ................................................................................................................... 19-9
8 References.................................................................................................................... 19-9
20 New Features and Improvements in RowCAD .................................................................... 20-1
1 Introduction................................................................................................................. 20-1
2 New Features ............................................................................................................... 20-1
3 Improvements ............................................................................................................. 20-7
4 Conclusion ...................................................................................................................20-9
5 Acknowledgments........................................................................................................20-9
6 References....................................................................................................................20-9
21 New Features in Right-of-Way Pro ....................................................................................... 21-1
1 Introduction.................................................................................................................. 21-1
2 Integrated TransposIT Software Tool .......................................................................... 21-1
3 Dummy Pipe-wall Conductor ...................................................................................... 21-6
4 Import ROWCAD MALZ File Entity into Total Interference .......................................21-7
5 Integrated Steady State Total Interference Accounting for Inductive and Conductive
Components ................................................................................................................. 21-9
6 Global Coordinates Shift in Total Interference Module ............................................. 21-11
7 Evaluating the Vulnerability of Arcing to Pipelines under Fault Conditions in Right-Of-
Way ............................................................................................................................. 21-11
8 References................................................................................................................... 21-16
22 New Advances and Features in CorrCAD Software Package ............................................... 22-1
1 Introduction................................................................................................................. 22-1
2 Coating Resistance Effects – Discrete and Distributed Holidays ............................... 22-2
3 Polynomial Functions for Polarization Curves ............................................................ 22-5
4 Multiple Cross-sections ............................................................................................. 22-10
5 Customizations and Options in the Polarization Plot ................................................ 22-11
6 Background Items in the Working Space Area.......................................................... 22-12
23 A New Interface for FFTSES ................................................................................................ 23-1
1 Introduction................................................................................................................. 23-1
2 Summary of the New Features .................................................................................... 23-1
3 Program Functionalities and Interfaces ...................................................................... 23-3
4 A Simple Example....................................................................................................... 23-11
5 Conclusion ................................................................................................................. 23-18
6 References.................................................................................................................. 23-18
24 Improvements and New Features in SESShield-3D ............................................................ 24-1
1 Introduction................................................................................................................. 24-1
2 User Interface Changes................................................................................................ 24-2
3 Creating, Editing and Selecting Objects ...................................................................... 24-4
4 Calculation and Results ............................................................................................... 24-8
5 SESShield-3D Related Documentation ..................................................................... 24-10
6 Conclusion and Future Developments ....................................................................... 24-11
7 References.................................................................................................................. 24-12
25 User-Defined Calculation Methods for Corona Performance in SESEnviroPlus ................ 25-1
1 Introduction................................................................................................................. 25-1
2 General Formulation ................................................................................................... 25-2
3 Application................................................................................................................... 25-4
4 Conclusion ................................................................................................................... 25-7
5 References.................................................................................................................... 25-8
Part IV: Recently Published Technical Articles ................................................................... 1
Part I: User Contributions
UGM 2016 – BOULDER, COLORADO
1 INFLUENCE OF GROUND CONSOLIDATION ON LOCAL

SOIL RESISTIVITY, EARTH IMPEDANCE AND SAFETY
POTENTIALS FOR A NEW TRANSMISSION SUBSTATION
Denis Baudin
EdifERA
Ty Canol, Clun Avenue
Pontyclun, Rhondda Cynon Taf, Wales, UK
Email: denis.baudin@edifera.com, Web Site: www.edifera.com
Abstract
Difficult ground conditions can be challenging not only for civil work but also when ensuring a safe environment
inside and outside HV installations. This article covers different scenarios that were investigated at a new build
transmission substation and presents some effects of the piling and backfilling on the earth values and surface
potentials using CDEGS MALZ [1].
1 Introduction
A new 400 kV substation is required for network reinforcement in the UK and to provide a direct
132 kV connection for the development of a large on-shore wind farm.
The location of the substation has advantages of being in an industrial estate in a fairly
uninhabited area and can be fed off an existing 400 kV transmission overhead line. The geology
of the surrounding land is not particularly favorable to the project, with a wide area of soil
resistivity being fairly high due to the rocky ground conditions. The substation itself is partly set
on a peat bog of some scientific interest, while a Marsh Fritillary colony had to be relocated.
This paper looks at the influence that the ground structure and consolidation work can have onto
an earthing system and ground potentials.
2 Background
2.1 Project Requirements
An earthing system is required to ensure the equipotentiality of the substation equipment and all
metallic structures within the site. It must provide safe potentials within the site for operators. It
must not create unsafe conditions for the public outside the site by exporting high potential
through the ground or via buried services onto third party electrodes.
2.2 Challenges
While being built on a greenfield site, further inspection revealed a number of constraints awaited
the civil work. A Marsh Fritillary colony, a butterfly threatened not only in the UK but across
Europe and therefore the object of conservation effort, must be relocated from the bottom end of
Copyright © 2016 SES ltd. All rights reserved. Page 1-1
PART I: USER CONTRIBUTIONS
the site. The land hosts a very particular type of peat bog wetland that also needs to be protected.
And finally, yet not so valuable to environmentalist, a wild landfill site with potential harmful
waste that is better left untouched.
The relatively high fault current and ground resistivity mean that a large system must be installed
to achieve an earth impedance low enough to keep the Earth Potential Rise within safe limits. The
limited land, available being in the middle of the industrial estate, is surrounded by the local
distribution network and the operator required reassurance that the new substation will not
impact on its installation and customers.
3 Soil Characteristics and Ground Consolidation
3.1 Natural Soil Conditions
To obtain the electrical characteristics of the site and to facilitate a preliminary study, soil
resistivity soundings using the Wenner method were taken pre-construction directly on the site
with further measurements taken in a slightly remote area where a satellite earth nest is to be
installed to lower the overall earth impedance of the system. The soil model derived for the larger
area is shown in Table 1 representing a three layer high-low-high structure with the low resistivity
layer easily reachable by means of vertical electrode.
Table 1: Initial Soil Model
Layer Resistivity (Ωm) Thickness (m)
1 143 2
2 39 14.1
3 147 ∞
3.2 Engineered Soil Characteristics
The initial proposal to ensure the ground could bear the weight of the equipment was to use piled
foundations. The client was made aware that the proposed 15 m to 18 m long piling option would
benefit the earthing installation if connected to the earthing system. Not only would the piles
equalise the potential across the site, their lengths would ensure maximum contact within the
middle layer meaning they would make full use of the lower resistivity ground. The preliminary
models were set-up in CDEGS MALZ to estimate the achievable earth value and following many
simulations, an agreed target earth impedance was set that would ensure the potentials at the site
and in surrounding areas would not create safety issues.
Following additional geotechnical and civil investigations, it was found that the use of piles for
the equipment alone would leave the ground unstable if substantial consolidation work was not
taking place. This was going to be achieved using a method known as Geo-synthetic Reinforced
Pile Supported (GRPS) where a mesh is installed on top of piles and then covered with backfill
material. Discussion took place to try and retain access to those piles but it was dismissed due to
time constraints. The targeted impedance however would still need to be achieved but without the
piles. To compensate for the loss of the piles, earth rods would be installed up to 8 m depth. A
Page 1-2 Copyright © 2016 SES ltd. All rights reserved.

reasonable number of rods would be needed to achieve the equivalent contribution of the original
piles, which were twice their lengths.
Figure 1 shows the original earth mat with piles in grey and the proposed replacement with rods
in black. The amount of electrode has been greatly reduced nevertheless computations show that
the piles under the geo-synthetic mesh shown in Figure 2, together with the proposed rods have
ensured no significant change to the earth impedance, Table 2.
Figure 1: Foundation Piles Model vs Rods Model.
Figure 2: Consolidated Ground and Soil Volume.
Table 2: Achieved Earth Resistance Comparison
Model Earth Resistance (Ω)
With 13.5 m Piles 0.368
With 8 m Rods 0.384
Further soil resistivity measurements were taken on the backfill material to try to anticipate the
effect of the man-made layer that had been used to consolidate the ground. However, the results
obtained could not be interpreted, as the coupling between the piles caused erroneous data.
Several simulations were carried out to investigate the effect of a high resistivity imported ballast
material in the area of the proposed substation. This used the advanced simulation facility in the
CDEGS software where a user-defined volume of soil may be simulated, immersed in a
horizontally-stratified background soil model.
A volume of soil 3 m thick and with a resistivity of 450 Ωm was incorporated to the original soil
model. An equivalent model for this area would be as shown in Table 3.

Table 3: Equivalent Soil Model
Layer Resistivity (Ωm) Thickness (m)
1 450 3
2 39 13.1
3 147 ∞
After several computations using MALZ and the volume of soil, it was found that the ‘effective’
resistivity of the infill material was reduced thanks to the presence of the numerous vertical steel
piles installed beneath that area.
4 Influence on the Earth Impedance and Electrode System
4.1 Overall System Earth Impedance
The model was developed in three stages. The substation earth mat, the overall earth electrode
associated with the 400 kV system and finally, the local network and tower lines was all accounted
for.
Two scenarios were investigated, one with the 400 kV system and local network bonded together
and a second one where the two systems would remain separate. It was established that the local
distribution network and the proposed HV installation had to be kept separate as the resulting
surface potential contours for the combined installations would involve an unsustainable amount
of mitigation.
To ensure safe earth potentials, the maximum overall earth impedance to be achieved for the
400 kV system should not exceed 0.16 Ω.
The electrode for each stage is detailed in the following sections.
4.2 Substation Earth Mat
The earth mat, shown in Figure 3, consists of potential grading provided around all equipment
and building; with equipotentiality being ensured by double connections to each item. The
substation earth mat resistance and piles in isolation was calculated as 0.384 Ω.

Figure 3: Substation Earth Mat.
4.3 Overall 400kV Earthing System
To reduce the earth impedance further, additional earth electrode had to be installed in remote
land to increase the area occupied by the earthing system. A perimeter electrode is shown on
Figure 4 running outside the future extension boundary.
Figure 4: Overall Earthing System.
A satellite mesh with vertical earth rods is proposed on the other side of the peat bog, which is
causing environmental concerns. To fully benefit from this parallel earth contribution, three
earthwires must be installed to interconnect the two electrode systems. These will need to cross
to the other side of the peat bog; however installation at standard depth has been prohibited. The
links will need to be routed using directional drilling underneath the peat at depth of 14 m. The
earthing system also connects to the 400 kV tower line which provides a parallel contribution (i.e.
tower line chain impedance). The overall earth impedance achieved is 0.147 Ω.
4.4 Overall Earth Electrode
When set into the surrounding local network, Figure 5, the overall earth impedance seen from the
400 kV substation reduced further, thanks to the coupling through the ground. As such, a new
value of 0.143 Ω was calculated.
Figure 5: Overall Electrode and DNO Network.
4.5 Summary of Earth Impedance Results
A summary of the earth impedance achieved for each model is given in Table 4.
Table 4: Equivalent Soil Model
Model Earth Impedance (Ω)
Substation Earth Mat in Isolation (Fig.3) | 0.384 |
Overall Earthing System (Fig.4) | 0.147 |
Overall Electrode and DNO Network (Fig.5) | 0.143 |
A post-construction impedance measurement was conducted and revealed the accuracy of the
model and CDEGS MALZ. The impedance, measured from the 400 kV compound, was 0.154 Ω
corresponding to an error margin of less than 7 %. The low error margin may be attributed to
disparity in the soil structure over the whole area, construction deviations, a shorter measurement
traverse than recommended and instrument error. Overall, a very satisfactory result, see Table 5.
Table 5: Measured Earth Impedance
Value Required to Ensure Safe Post-Construction Earth

Potentials (Ω) Impedance (Ω)
0.16 | 0.154 |

5 Effect of the Man-Made Ground on the Surface Potentials
5.1 Investigations into the Earth Mat in Isolation
As the simulations, to compute the results of the overall model, were taking significant time at the
early design stages, some sensitivity studies were carried out with a smaller model. For the Earth
Mat in isolation, particular attention was taken to the touch voltages within the site and the
surface potentials just outside.
Table 6 summaries the results of the Earth Mat in isolation for the different types of grounds for
and compares its effects on the potentials.
Table 6: Comparison of Potentials based on Soil Characteristics For the Earth Mat in Isolation
Surface Potentials
Ground Conditions Touch Voltages Comparison Comments
within the Earth Mat
23.86 % (552 V) 46.3 % (1072 V)
Natural Soil Reference Used for reference
Fig.6 Fig.7
21.95 % (508 V) 46.31 % (1072 V) The effect of the piles is

Natural Soil with Piles Better
Fig.8 Fig.9 negligible
30.86 % (714 V) 44.28 % (1025 V) Smaller contours due to

Backfill Material Worst
Fig.10 Fig.11 ground resistivity
28.34 % (656 V) 44.36 % (1027 V) The effect of the piles is

Backfill Material and Piles Higher
Fig.12 Fig.13 negligible
For the Earth Mat in isolation, adding the piles to the model was found to have no effect on the
surface potentials. The touch voltages within the substation area were improved. However, adding
a layer of higher resistivity backfill material made the touch voltages worse. When the two were
combined, the result was still worse than having the natural soil without backfill.
Natural Soil
[ID:Small - Natural Soil @ f=50.0000 Hz ]
LEGEND
LEGEND [ID:Small - Natural Soil @ f=50.0000 Hz ]
240 Maximum Value : 59.449 250
MAXIMUM VALUE : 99.95692
MINIMUM VALUE : 40.33281

Minimum Value : 0.138
Level 11 ( 95.0000 )
59.45 2 2
11 11 Level 10 ( 90.0000 )
200
190 53.52 3 11 11 11
11
3
11 Level 9 ( 85.0000 )
10
47.59 11 11 4
11 Level 8 ( 80.0000 )
4 11
11 11
41.66 150 11
Y AXIS (METERS)
10
11
11 10 Level 7 ( 75.0000 )
140 11
Y AXIS (METERS)
11
35.72 10
11 11
11 1111 11 11 11 10 9
11 1111 11 Level 6 ( 70.0000 )
11
10
111111
29.79
100 1111 11
10
Level 5 ( 65.0000 )
8
23.86 10 910
8 7 10
9
6
Level 4 ( 60.0000 )
90 17.93
5
4
3
2 2 1
Level 3 ( 55.0000 )
50
12.00
Level 2 ( 50.0000 )
6.07
Level 1 ( 45.0000 )
40 0
-520 -470 -420 -370 -550 -500 -450 -400 -350
X AXIS (METERS) X AXIS (METERS)
Touch Voltage (% Ref. GPR) [Wors] Potential Profile (% reference GPR)
Figure 6: Touch Voltages. Figure 7: Surface Potentials.

Natural Soil with Piles

[ID:Overall - with Volum @ f=50.0000 Hz ] LEGEND
LEGEND [ID:Overall - with Volum @ f=50.0000 Hz ]
240 Maximum Value : 43.903 250 MINIMUM VALUE : 40.34883
Minimum Value : 0.628E-02
Level 11 ( 95.0000 )
2
43.90 2
11 11 Level 10 ( 90.0000 )
200
190 39.51 3 11
11 11
11
3
Level 9 ( 85.0000 )
11
10
11 11
35.12 4
11 Level 8 ( 80.0000 )
4 11
11 11
30.73 150 11 10
Y AXIS (METERS)
11
11 10 Level 7 ( 75.0000 )
140 11
Y AXIS (METERS)
11 11
26.34 10
1111 1111
11
11 11 11 10
11 9 Level 6 ( 70.0000 )
11 11
11
10
111111
21.95
100 1111
11
10
Level 5 ( 65.0000 )
8
10 910
17.56 8 10
67
9
5 Level 4 ( 60.0000 )
90 13.18
4
3
2 2 1
Level 3 ( 55.0000 )
50
8.79
Level 2 ( 50.0000 )
4.40
Level 1 ( 45.0000 )
40 0
-520 -470 -420 -370 -550 -500 -450 -400 -350
Backfill Material
[ID:Overall - with Volum @ f=50.0000 Hz ]
LEGEND LEGEND
240 Maximum Value : 61.720
250 MINIMUM VALUE : 38.07707
Level 13 ( 100.000 )
2 2
61.72
12 Level 12 ( 95.0000 )
1212
190 55.55 200 3

12
12 12 12
12
3
Level 11 ( 90.0000 )
1112 12
12
12 12 13 11 Level 10 ( 85.0000 )
49.38 10
12 12 11 4
1112 12 12 9
4 1112 12 10
10 12
12 1112 12 12 11 Level 9 ( 80.0000 )
11 12
43.20 150 12 9
Y AXIS (METERS)
10
12 12
12 121112 10
12 12 1112 10
11
Level 8 ( 75.0000 )
140 11 12 12 11 11
Y AXIS (METERS)
12 12 9
37.03 11
12 12 11
12 12 11
11 10 12 12 1212 11 12
11 12 11 1012 9 Level 7 ( 70.0000 )
11 10
11 1111
9
30.86 11
11 11
12 10 10 11 Level 6 ( 65.0000 )
100 11 11 11
12
10
11
24.69 10
119108 11
8 710
9 Level 5 ( 60.0000 )
5 6
90 18.52
4
Level 4 ( 55.0000 )
2 3 2 1
50 Level 3 ( 50.0000 )
12.35
Level 2 ( 45.0000 )
6.17
40 Level 1 ( 40.0000 )
0
-520 -470 -420 -370
-550 -500 -450 -400 -350
Backfill Material and Piles

LEGEND LEGEND
250 MINIMUM VALUE : 38.15905
Level 13 ( 100.000 )
2 2
47.22
12 Level 12 ( 95.0000 )
1212
190 42.50 200 3

12
12 1212
12
3
Level 11 ( 90.0000 )
1112 12
12
12 12 13 11 Level 10 ( 85.0000 )
37.78 10
12 12 11 4
1112 12 12 9
4 1112 12 10
10 12
12 1112 12 12 Level 9 ( 80.0000 )
11 12
33.06 150
Y AXIS (METERS)
12 12 11 9
10 12
12 12 12 1011
12 12 11 Level 8 ( 75.0000 )
140 11 12 12 11 12 1011
Y AXIS (METERS)
12
28.34 12
11
12 12 11
12 12
9
11
11 10 11 12 12 1212 11 12 12 10 9 Level 7 ( 70.0000 )
10
11 11 11 12 1111
12
23.62 9
11
11
10 10
11
11 Level 6 ( 65.0000 )
100 12
11 11 12
11
10
11
18.89 10
119108 11
87910 Level 5 ( 60.0000 )
5 6
90 14.17 4 Level 4 ( 55.0000 )
2 3 2 1
9.45
50 Level 3 ( 50.0000 )
Level 2 ( 45.0000 )
4.73
40 Level 1 ( 40.0000 )
0
-520 -470 -420 -370
-550 -500 -450 -400 -350

5.2 Effects on the Overall Earthing System including the Separate Local
Network
When calculating the potentials for the overall earth model Figure 5, it was necessary to find out
the effect on the transfer voltage between the 400 kV system and the local network, additionally
to the touch voltages and surface potentials. No plots are provided to illustrate these calculations.
Table 7 summaries the results of the overall model for the different types of grounds and compares
its effect on the potentials.
Table 7: Comparison of Potentials based on Soil Characteristics For the Overall Earthing System
Ground Conditions Touch Voltages Comparison Transferred Voltage Comments
12.86 % (298 V) 40.22 % (931 V)

Natural Soil Reference Used for reference
Fig.14 No plot
8.46 % (196V) 40.72 % (943 V)

Natural Soil with Piles Better Slight increase
Fig.16 No plot
12.85 % (297) 40.38 % (935 V)

Backfill Material On Par Smallest Increase
Fig.18 No plot
12.80 % (296 V) 40.57 % (939 V) The backfill slightly

Backfill Material and Piles On Par
Fig.20 No plot reduces the effect of piles
For the larger model, adding the piles had no significance on the surface potential, and improved
touch voltages similarly to the smaller Earth Mat studied in Section 5.1. The backfill material
though had not much outcome itself apart from neutralising the benefit of the piles.
Natural Soil
LEGEND [ID:Rhigos Vattenfall an @ f=50.0000 Hz ]
[ID:Rhigos Vattenfall an @ f=50.0000 Hz ] LEGEND
Maximum Value : 42.441 MAXIMUMVALUE : 98.39339
250 Minimum Value : 0.189
MINIMUMVALUE : 53.63290
9 8 Level 13 ( 96.6667 )
42.44 9
10 10
1 12 13 11
12 12 12
Level 12 ( 93.3333 )
200 200 2 12 12 12 12
12 12
12
38.22 9
12 13
3 12
12 13 1313 13 13 Level 11 ( 90.0000 )
13 13 13 13 1312
13
4 13 13 13 13
13 13 13
33.99 1313 13 13 12
13
1313 13
5 13 1313 13 12 Level 10 ( 86.6667 )
13 13
150 6 1313 13
13
7 12 13 13 1313 13 12 10
13 13 12
29.77 8 13 13 13 Level 9 ( 83.3333 )
11 12
13
13 13
13 12 13 13
1112 13 1313
13 13
13 131313
100 13 12 Level 8 ( 80.0000 )
Y AXIS (METERS)
25.54 1211
12 13 13
13
100
Y AXIS (METERS)
12 Level 7 ( 76.6667 )
12
21.31 12
12 Level 6 ( 73.3333 )
12
12 13 12 10
1211 1211
50 17.09 1211
12
12 12
12
12
12
11 Level 5 ( 70.0000 )
1212 12 12
12
0 1
11 12 12
1212 12 12
12
12.86 11
12 Level 4 ( 66.6667 )
1 2 3 4 5 6 7 8 10 9118 7 6
0 8.64
Level 3 ( 63.3333 )
Level 2 ( 60.0000 )
4.41
Level 1 ( 56.6667 )
-50 -100
-540 -490 -440 -390
-550 -500 -450 -400 -350

Natural Soil with Piles

LEGEND [ID:Rhigos Vattenfall an @ f=50.0000 Hz ]
[ID:Rhigos Vattenfall an @ f=50.0000 Hz ] LEGEND
250 Minimum Value : 0.122E-01
8 Level 13 ( 96.6667 )
42.25 10 11 9
1 11 11 10
13 1313
2
200 200 13 131313 13
Level 12 ( 93.3333 )
38.03 13 13 13 13
3 13 13
13
12 13 Level 11 ( 90.0000 )
4 13 13 1313
12 1313 13 13
131313 13 1313
33.80 5 13
13 10 Level 10 ( 86.6667 )
13
150 6
7
13
13 12
8 1313 13 11
29.58 11 13 13 Level 9 ( 83.3333 )
13 13
13
12 13 13
13 13
100 Level 8 ( 80.0000 )
Y AXIS (METERS)
11 13 13 13
25.36 1212
13
100
Y AXIS (METERS)
12 Level 7 ( 76.6667 )
12
21.13 12
12 12 Level 6 ( 73.3333 )
12 12
12 13 12 10
1211 12 11
50 16.91 1211
1212
12
12
12
11 Level 5 ( 70.0000 )
1212 12 12
12
1112 12
0 1 1212 12 12
12
12.68 11
12 Level 4 ( 66.6667 )
1 2 3 4 5 6 7 8 10 9118 7 6
0 8.46
Level 3 ( 63.3333 )
Level 2 ( 60.0000 )
4.24
Level 1 ( 56.6667 )
-50 -100
-540 -490 -440 -390 -550 -500 -450 -400 -350
Backfill Material
[ID:Overall - with Volum @ f=50.0000 Hz ] LEGEND LEGEND
MAXIMUMVALUE : 99.21738
8 Level 13 ( 96.6667 )
10 11 9
13 10
42.40 1 13 12
12
190 200 2 12121312 9 Level 12 ( 93.3333 )
12 12
38.18 12 13
3 11
12
11 13 13 13 13 Level 11 ( 90.0000 )
12 12 13 13 13 13 12
13
4
12 13 12131313
11 1313 13 12
1313 12
11 13 13 13 13 13
33.96 5 1313 13
12
13 13 1313 13 12 Level 10 ( 86.6667 )
140 6 13 1313 13
13 13
7 1312 13 13 13 1313 13 10
131313 13 1213
8 1313 11 Level 9 ( 83.3333 )
29.74 11 12
13
13 13 12 13
13 131313
1112 1312 13 1313 13
Y AXIS (METERS)
13 131313
100 13
13 12 13
Level 8 ( 80.0000 )
1211
12
90 25.52 13
Y AXIS (METERS)
12 Level 7 ( 76.6667 )
21.29 12
12 12 Level 6 ( 73.3333 )
12 12
12 13 11 12 10
40 17.07
1211
1211 12
12
12
12
11 Level 5 ( 70.0000 )
1212 12
1212 12
12
1112 12
0 1 1212 12 12
12
12.85 11
12 Level 4 ( 66.6667 )
1 2 3 4 5 6 7 8 10
9118 7 6
-10 8.63
Level 3 ( 63.3333 )
Level 2 ( 60.0000 )
4.41
-60 Level 1 ( 56.6667 )
-100
-560 -510 -460 -410 -360 -550 -500 -450 -400 -350

Backfill Material with Piles
LEGEND [ID:Overall - with Volum @ f=50.0000 Hz ]
[ID:Overall - with Volum @ f=50.0000 Hz ] LEGEND
250 Minimum Value : 0.136E-01
10 10 8 Level 13 ( 96.6667 )
42.63 11 9
11 11
1 13
131313
13 13 13
13 13131313
13
13 1213 13 13 Level 12 ( 93.3333 )
200 200 2
12 1213 1313 13 13
13
38.37 13 13 13131313 12
3 12 13
11
12 13 13 13
12 13 13 13 13 Level 11 ( 90.0000 )
13 12
12 13 131313 13
4 11 13 13
1313 13 13 13 13 12
34.11 13
13 13 13
5 12 13 Level 10 ( 86.6667 )
1312 13 131312
10
150 6
7
1313
13
13121313 13 13 13 13 13
8 1313 1213 11
29.85 11 13 1313 Level 9 ( 83.3333 )
13 131212 13
13 12 131313
12 13
13 13 13 13
13 131313
100 11 13 12 Level 8 ( 80.0000 )
Y AXIS (METERS)
1212 13 1313
25.58 13
Y AXIS (METERS)
100 12 13 Level 7 ( 76.6667 )

12
12
21.32 13
12 12 Level 6 ( 73.3333 )
12 12
12 13
13 12 10
1211 12
12
50 17.06 12 12 12 11
Level 5 ( 70.0000 )
12 13 12
13
11
0 1 1212 12
12
12
12
11
12 Level 4 ( 66.6667 )
12.80 2 3 4 5 6 7 8 10
9118 7
1 6
Level 3 ( 63.3333 )
0 8.54
Level 2 ( 60.0000 )
4.28
Level 1 ( 56.6667 )
-50 -100
-540 -490 -440 -390 -550 -500 -450 -400 -350

6 Conclusions
Ground conditions can have a significant influence on the earthing system required at a
transmission or distribution substation and other installations. The particular land chosen for the
installation of the proposed site that was studied, concealed a number of challenges that civil work
had to overcome. This in turn, had created several design complications for the earthing system.
The first part of this document, detailed some of the work that was necessary to achieve a
reasonable overall earth impedance and an Earth Potential Rise (EPR) value that could be
managed, in order to provide a safe environment for operators and members of the public.
Secondly, it focused on the direct influence of the ground consolidation on the surface potentials
within the site and the transfer of potential onto nearby separate electrodes, by studying the effect
of the long piles installation and import of high resistivity soil.
The study showed that the size of the area occupied by the consolidated ground compared to the
earthing installation affected the results.
Generally, the piles were found to have a negligible influence on the surface potential contours,
while having a positive effect on the touch voltages.
Finally, it should be noted that the results of this study are not meant to be used as a guidance.
Every site is different and there are too many factors affecting the current flows and potentials,
therefore it is recommended to conduct a case by case study where ever the results are critical.
7 Acknowledgments
The author would like to thank his colleagues Mayur Gaglani and Robert Knott for their
assistance.
8 References
[1] "CDEGS MALZ", Safe Engineering Services, Frequency Domain Analysis of Conductor Networks,
Version Numbers: Major 15, Minor 2, Copyright(C) 1978 - 2015 by SES ltd.

2 FAULT CONDITIONS AT SUBSTATIONS WITHOUT A

GROUND CONNECTION TO THE MAIN GRID:
EARTH FAULT CURRENT CALCULATION AND
GROUND POTENTIAL RISE (GPR)
Ehsan Azordegan
Senior Engineer, Kinectrics Inc., Toronto, CANADA

Ehsan.Azordegan@Kinectrics.com
Abstract
The high voltage (above 44 kV) transmission system in Canada is mainly a four wire system, taking advantage of
the skywires (also known as overhead ground wires) for lightning protection and grounding coordination.
Although according to the Canadian Electrical Code (Rule 36-308) a permanent bond between the grounding
system of HV substations and the transmission line skywires (and distribution neutrals) is required, some
utilities that are exempted from the CSA code do not allow privately built substations (such as wind and solar
developments) to take advantage of their skywires (due to liability reasons) and force them to design a grounding
system that could withstand the most stringent fault conditions without the benefit of connecting to the overhead
ground wires (standalone conditions).
When the substation grounding system is not bonded to the overhead ground wires, at the event of a fault, the
entire fault current must return to the remote sources through the soil. In other words, not only there is no
conductive return path for the fault current back to the source, but also there will be no inductive reduction of the
fault current (shielding effect) due to the proximity of the faulted phase and the skywires. Therefore, designing a
standalone substation grounding system in areas with poor soil conditions and fault current magnitudes as high
as 10 kA ~ 20 kA can be extremely challenging. In such design cases, it would be beneficial to take into account the
effect of standalone grid resistance on the magnitude of fault current while exploring other solutions to lower the
standalone grid resistance to achieve GPR, step, and touch voltage coordination.
This paper presents a case study to elaborate on the challenges involved in designing a standalone grounding
system of a wind farm in northern Ontario (without benefiting from the skywire connection and poor soil
conditions), detailed fault current analysis considering the effect of standalone grid resistance on the fault
magnitude, studying various mitigation solutions for lowering the grid resistance, and finally verifying the
modelling results (obtained by the SES grounding software) against the current injection test results (obtained
during Kinectrics commissioning grounding tests at site).
1 Introduction
The grounding system of a windfarm in northern Ontario was originally designed using a
relatively optimistic soil model. During the commissioning phase of the project, the
interconnected ground grid impedance of the farm was measured to be higher than the design
value. According to the commissioning ground grid test results, a line to ground fault at the
240 kV bus of the wind farm transformer station could potentially result in ground potential rise
values as high as 12 kV. It should be noted that the grounding system of the farm was isolated
from the skywires of the 240 kV transmissions lines that the farm output generation was
connected to.

In order to achieve coordination, a number of mitigation solutions were studied and the most
feasible and practical solution was implemented. This paper presents the testing methodology, a
brief summary of studied mitigation solutions, and the performance of the upgraded ground
grid after installing the mitigation solutions. Current injection test results before and after the
upgrades are also reported to support the modelling.
2 Methodology and Assumptions
2.1 Soil Model
In order to offer mitigation solutions that can practically resolve the grounding coordination
issues, a realistic soil model must be obtained. The original design of this farm was based on a
two layer soil model presented in Table 1.
Table 1: Soil model that was used for the original design
Resistivity Thickness
Layer
(Ω-m) (m)
Top Layer 0.5 80
Central Layer 8,000 3.5
Bottom Layer 5,000 ∞
The soil model shown in Table 1 looked quite conservative at first glance, however, it was later
found out that soil surveys leading to this soil model were relatively short surveys (the longest
Wenner spacing between the probes was 20 m). Therefore, Kinectrics performed two additional
soil surveys in the area; one with Wenner spacings up to 70 m (total traverse of 210 m), and one
with Wenner spacings up to 200 m (total traverse of 600 m). Using the long soil surveys and the
measurements available from the original design, the soil model presented in Table 2 was
selected for modelling the grounding system of the farm. The curve fitting for this model is
shown in Figure 1.
Table 2: Soil model concluded from kinectrics tests on site
Resistivity Thickness
Layer
(Ω-m) (m)
Top Layer 1,284.7 0.55
Central Layer 16,000 8.18
Bottom Layer 1,650 ∞

Figure 1: Concluded soil model for the grounding studies.
2.2 Current Injection Test (Original Design)
To evaluate the performance of the installed grounding system of the main transformer station
(original design) at the farm, a current injection test was carried out. As stated earlier, the
grounding system of the substation was interconnected to the grounding system of all turbines
but it was isolated from the skywires of the 240 kV transmission lines. The test arrangement is
shown in Figure 2, where the current injection probe and all potential readings (P) are labeled.
Figure 2: Current Injection test arrangement at the farm.

The entire test arrangement along with the grounding system of the farm and the HV
transmission lines were modelled in the Hifreq module of CDEGS. Using the model, the
readings were adjusted for proximity between the current injection probe and the potential
probes as well as the coupling due to parallelism to the skywires on the transmission lines. The
adjusted measurements at all nine locations (P locations in Figure 2) were then averaged to
reduce the measurement error. Table 3 compares the average of the nine adjusted readings to
the modelled interconnected impedance obtained from the Hifreq model.
Table 3: Measured Vs Modelled Ground Impedance
Impedance at 60 Hz
Ground Impedance as Measured 1.9898.8° Ω
Ground Impedance as Modelled 1.96711.5° Ω
2.3 Fault Current Analysis
The Thevenin equivalent circuit for a fault seen from the point of common connection (PCC) was
provided by the owner of the 240 kV transmission lines and is shown in Table 4. The per unit
values have a 100 MVA base power and 220 kV base line to line voltage.
Table 4: Per unit Thevenin equivalent seen from a fault at PCC
R X X/R
Positive Sequence Impedance 0.00391 0.04368 11.171
Zero Sequence Impedance 0.00686 0.03982 5.807
According to the Thevenin impedances presented in Table 4, a single line to ground fault at PCC
has a fault magnitude of 6.98 kA. The power flow model of the farm also calculated the fault
contribution from the turbines to be 146 A. Therefore, during a bolted fault at the HV side of
substation, a total of 7.13 kA current will be present at the faulted 240 kV bus.
In order to calculate the portion of the fault current that will be returning through the soil back
to the remote sources (causing the ground potential rise), three factors must be taken into
account. First, the transformer winding configuration which can circulate some of the zero
sequence fault current within the transformer. Secondly, the effect of ground grid impedance on
the fault magnitude must be studied. Finally, the fault arriving from the turbines will have an
inductive shielding factor due to the underbuilt neutral, reducing the total fault getting injected
to the grid. These considerations will be discussed in the following sections to determine the
total injected fault current into the ground during a single line to ground fault.

2.3.1 Power Transformer Zero Sequence Model
The 240 kV / 34.5 kV – 4.16 kV wye grounded, wye grounded transformer with a buried corner
grounded delta tertiary (Ynynd1) has a zero sequence network model similar to the network
shown in Figure 3:
Zp Zs
Zt
Figure 3: T–model zero sequence network.
Although the zero sequence T-model of the transformer was not available, the equivalent zero
sequence impedance seen from the high side, while the other windings are grounded, can be
found in the test report to be 229.8 Ω. Using the Thevenin equivalent zero sequence network
presented in Table 4, the zero sequence network during a single line to ground fault at the HV
bus can be modelled with two parallel impedances; the zero sequence impedance seen from the
HV side of the transformer (229.8 Ω), and the zero sequence equivalent impedance of the
240 kV system (19.86 Ω). Due to the transformer winding configuration, the zero sequence
impedance during the fault slightly decreases (about 8%), resulting in a higher fault current.
However, a portion of this fault will not be injected into the grounding system and circulates
back to the transformer.
2.3.2 Effect of the Ground Grid Impedance on the Fault Magnitude
The skywires on 240 kV circuit are not bonded to the station ground of the farm. Therefore, a
fault at the HV bus of the station is not exactly a bolted fault. To properly account for the ground
impedance of the grounding system of the transformer station on the fault level, a HIFREQ
model of the farm was developed (snap shot of the model near the main substation is shown in
Figure 4), accounting for the 240 kV transmission lines as well as the transformer at the station.
A truncated 2 km section of the transmission line is included in the model and phase conductors
are energized by three voltage sources. Source impedances for these voltage sources are
calculated in a way to generate the fault current of 6.98 kA at the last span, just outside the main
substation. The output of the HIFREQ model shows that considering the transformer zero
sequence impedance and the ground isolation between the 240 kV circuits and farm ground
grid, a total fault current contribution from the 240 kV lines will be 6 221 A. The transformer
model presented in Figure 4 was built using three single phase transformers. The parameters of
the single phase transformers were extracted from the factory test report of the main
transformer.

Figure 4: Snapshot of the Hifreq model of the Farm near PCC.
2.3.3 Fault Current Contributions from the Farm
The power flow model of the farm reported a contribution of 146 A from the turbines at the
event of a fault at the HV side of the transformer at the main substation. However, due to the
underbuilt neutral on the 34.5 kV collection lines, a portion of the fault will be inductively
coupled into the neutral, not causing any ground potential rise at the station. To calculate the
percentage of the fault that will inductively return through the neutral (also known as shielding
factor), the electrical property of the neutral conductor as well as the geometry of the phase
conductors on the tower must be known. The 2/0 ACSR neutral has Rac of 0.426 Ω/km and a
gmr of 1.55 mm. Conservatively, assuming that the distance between the faulted phase and the
neutral is 5 m and the bottom soil resistivity is 16 000 Ω-m, using the Carson’s earth return
formula, the portion of the fault that will inductively return to the turbines will be 42.4%.
Therefore, from the total of 146 A, only 84 A will be injected to the grounding system.
2.3.4 Summary of Fault Current Calculations
Summarizing the fault analysis described in the previous sections, the total fault current that
will be injected into the interconnected grounding system at the event of a fault at the HV side of
the transformer will be 6 305 A. This current returns to the remote sources through the soil,
causing the GPR at the station. This current can be used in the simplified MALZ model to
energize the model, reducing the required computation time for studying different scenarios.

2.4 Ground Potential Rise (GPR)
At the event of a single line to ground fault at the 240 kV bus at the transformer station, the total
zero sequence fault current that will be injected into the interconnected grounding system
excluding the transformer circulating current and the inductive component of fault current on
the collection line neutral will be 6 305 A, as calculated in the previous section. Considering the
measured interconnected ground grid impedance presented in Table 3, a ground potential rise
due to this fault current will be 12 540 V during summer conditions.
3 Mitigating Excessive GPR
3.1 Investigated Solutions
Since the GPR was significantly higher than the limit of 5,000 V specified in the Ontario
Electrical Safety Code [1], it was required to study a number of mitigation solutions to lower the
GPR. The studied mitigation solutions ranged from expanding the ground grid and installing
ground wells to installing counterpoise conductors and even seeking permission from the
owners of the HV transmission lines to connect to the skywires.
After discussions with ESA (Electrical Safety Authority), a new GPR limit of 10 kV was agreed
upon, mainly due to the known severe soil conditions of the area. However, all the step and
touch voltages inside and outside the substation should coordinate with the allowable limits
specified in IEEE 80 [2].
Since the main contribution of the overall interconnected ground grid comes from the
interconnected turbine grids (by the neutral conductor on the 35 kV collector lines and also the
underground counterpoise conductors buried under the 35 kV collector lines), expanding the
ground grid at the main substation couldn’t lower the grid impedance to the acceptable levels.
Even the current split measurements, performed as part of the current injection tests, had
already shown that close to 96 % of the injected fault current leaves the ground grid of the main
transformer station through the connection of the local ground to the turbines.
Installing additional ground wells couldn’t also lower the ground grid impedance enough to
meet the required target value. The ground wells could improve the standalone ground
resistance (effectiveness varies by the size, number, and length of the wells) of the substation but
similar to the option of expanding the grid, the effectiveness on the overall interconnected
ground grid impedance was not significant.
After investigating the effectiveness of many mitigation solutions, it was concluded that the best
solution for lowering the interconnected ground impedance is to install counterpoise conductors
at the main substation. Possible installation routes were identified (total of three routes; A, B,
and C) and studies were carried out to identify the most practical and effective installation along
these routes.
The results of some of the studied scenarios are presented in Table 5. Considering the difficulties
involved in each installation and taking into account the errors associated with the modelling,
Solution ID 7 was selected for the final installation.

Table 5: Studying different counterpoise installation scenarios
Solution Installation Description Ground Grid GPR (kV)

ID Impedance (Ω)
1 Original Design 1.99 12,540
2 400 m @route B 1.90 11,955
3 300 m @route C 1.92 12,090
4 2.2 km @ route A 1.53 9,665
5 2.2 km @ route A +453 m @route B 1.48 9,305

+300 m @ route C

+300 m @ route C
+300 m @ route C
10 2.2 km @ route A +300 m @ route C 1.50 9,305
3.2 Current Injection Tests (After Installing Solution ID 7)
After the counterpoise installation was complete, the current injection tests were repeated. The
new counterpoise installation is shown by yellow lines in Figure 5. The ten potential readings at
P locations shown in Figure 5 were adjusted for the coupling (inductive and conductive) due to
proximity between the probes and parallel connection to the isolated skywires on the 240 kV
lines (using the Hifreq model of the farm). Table 6 compares the average of ten adjusted
measurements with the modelled value obtained from the MALZ model.
Table 6: Measured Ground Impedance & GPR after the counterpoise installation
Interconnected Grid GPR at the event of a

Impedance at 60 Hz SLG fault
Measured 1.3268.8° Ω 8,360

Figure 5: Current Injection test arrangement at the farm after the counterpoise installation.
It should be noted that as part of mitigating the excessive touch potentials on the station fence, a
gradient conductor was installed 2 m outside the fence. This installation had negligible impact
on the overall interconnected grid impedance.
4 Conclusions
A practical case study for lowering the interconnected ground grid impedance of a wind farm
along with the ground grid test results are presented in this paper. Due to inadequate soil
surveys at the design stage of the project, a relatively optimistic soil model was used for the
ground grid modelling of the farm which eventually resulted in a ground potential rise (GPR)
value of 12 kV. After investigating a number of mitigation solutions to bring the GPR under the
limit of 10 kV, it was concluded to install 2.6 km of counterpoise conductors at two routes.
Current injection tests before and after the counterpoise installations were performed to validate
the modelling results. Also, the importance of detailed fault current analysis was explained and
presented as part of the GPR calculations.
5 References
[1] Electrical Safety Authority (ESA), Ontario Electrical Safety Code. 2012.
[2] IEEE Power and Energy Society, IEEE Guide for Safety in AC Substation Grounding,
February. 2013.

3 EVALUATION OF NON-TRADITIONAL TRANSMISSION

LINE SHIELDING CONFIGURATIONS FOR REDUCING
MAGNETICALLY INDUCED VOLTAGE ON PARALLEL
METALLIC OBJECTS IN JOINT USE CORRIDORS
Josh Brown and David Lewis
POWER Engineers, Inc.

9320 SW Barbur Blvd
Portland, OR 97219
Emails: Josh.brown@powereng.com, David.lewis@powereng.com Web Site: www.powerengineers.com
Abstract
This paper is to explore the effects of shield wire and aerial counterpoise conductor configurations on
magnetically induced voltages for parallel metallic objects. Magnetic induction can present special challenges to
electric transmission line development in joint use corridors, especially when occupied by pipelines or railroads.
A sampling of typical voltage levels and transmission line configurations were evaluated. The results presented in
this paper will allow the reader to quickly understand the impacts of shielding type and provide a reference for
considering the inclusion of an aerial counterpoise conductor for their transmission line project which shares a
corridor with a parallel metallic object.
1 Introduction
Metallic objects parallel to electric transmission lines are subject to magnetically induced
voltages. Typically the objects of greatest concern for electromagnetic interference in
transmission line planning are metallic pipelines and railroads. One method of mitigation, as
suggested by the EPRI, to reduce the induced voltage on a rail system is the use of aerial
counterpoise on the transmission line [1].
Designing an aerial counterpoise system requires an understanding of the interactions of the
magnetic fields between the phase conductors, shield wires, and the object under analysis. Some
shield wires’ configurations can even increase the magnetic coupling between a transmission
line and parallel object, depending on the phasing.
The purpose of this paper is to explore the effects of shield wire and aerial counterpoise
conductor configurations on magnetically induced voltages for parallel metallic objects. The
results presented will allow the reader to quickly understand the impacts of shielding type and
whether to consider the inclusion of an aerial counterpoise conductor for their transmission line
project.
This analysis was performed using the HIFREQ module of CDEGS. Four different tower
configurations and four different shielding conditions were evaluated. The configurations also
included typical phase spacing for 115 kV and 345 kV transmission circuits.

2 Analyzed Cases
2.1 Transmission Tower Configurations
The effects of transmission line shielding are greatly affected by the base configuration of the
transmission line tower. Multiple tower configurations were evaluated as a part of this analysis.
These configurations produce varied magnetic fields as a result of the phase conductor
geometry. Each configuration is listed below, followed by an example figure.
Single-Vertical Configuration: All phases stacked vertically on one side of the tower, closest
to the parallel metallic objects. The base model placed a shield wire and Optical Ground Wire
(OPGW) overhead. When aerial counterpoise is modeled, either ACSR or OPGW is placed ten
feet below the lowest phase conductor closest to the parallel object. (Figure 1).
Figure 1: Single-Vertical Configuration.
Double-Vertical Configuration: Similar to the Single-Vertical Configuration, however the

two circuits are cross phased for magnetic field cancellation. The base model placed a shield
wire and OPGW overhead. When aerial counterpoise is modeled, either ACSR or OPGW is
placed ten feet below the lowest phase conductor closest to the parallel object. (Figure 2).

Figure 2: Double Vertical Configuration.
H-Frame Configuration: All phases at a uniform elevation with the SW and OPGW placed
atop each pole in the base model. For this configuration, results differ significantly depending
on the OPGW being placed on the pole closest to the object of interest or atop the farthest pole.
The effects of these differences are detailed in the results section. When aerial counterpoise is
modeled, either ACSR or OPGW is placed ten feet below the lowest phase conductor closest to
the parallel object. (Figure 3).

Figure 3: H-Frame Configuration.
Delta Configuration: Two phases stacked vertically on one side of the tower, closest to the
parallel metallic objects, with the third phase positioned on the opposite side of the tower. The
delta configuration only has one overhead wire, which is modeled as OPGW in the base model.
When aerial counterpoise is modeled, either ACSR or OPGW is placed ten feet below the lowest
phase conductor closest to the parallel object. (Figure 4).

Figure 4: Delta Configuration.

As shown in Figure 1 through Figure 4, each tower configuration was adjusted to have phase
spacing typical of either a 115 kV or 345 kV tower. All cases were analyzed with 1,000 amperes
through each phase conductor, and voltage corresponding to the tower phase spacing.
2.2 Shield Wire and Aerial Counterpoise Configurations
Transmission lines typically have shield wires overhead to provide lightning protection,
communications paths, and fault current return paths. These shield wires are often steel, but
may consist of aluminum or other conductive metals. OPGW in particular provides for both
lightning protection and communications paths. For this analysis, the base configuration of the
transmission line consists of one steel shield wire and one OPGW placed overhead, with the
exception of the delta tower.
Parallel metallic objects can have voltages induced on them due to magnetic fields from
transmission lines. One way to mitigate for magnetically induced voltages away from the
transmission line is to place aerial counterpoise between the phase conductors and the object of
interest. The aerial counterpoise has a current induced on it that opposes the net magnetic field
of the transmission line. Placing aerial counterpoise in this way reduces the magnetic fields for
an object of interest.
Multiple options exist for the placement of this aerial counterpoise conductor. The three most
common locations are: near the centerline of the transmission circuit, underneath the phase
conductor closest to the parallel object, or on an additional set of structures between the
transmission line and the parallel object. This analysis concentrated on the underbuild
conductor being placed ten feet under the lowest phase conductor. Placing an aerial
counterpoise near the centerline of the transmission line typically produces less benefit

compared to the selected solution because it is not between the transmission line and object of
interest. Building additional structures for improved aerial counterpoise can be an option but is
substantially more costly and time consuming then an underbuilt conductor.
As previously discussed, transmission lines often have a steel shield wire and an OPGW placed
overhead. OPGW is primarily aluminum, which is more conductive than the steel shield wire
allowing more current to be induced. Therefore, a third configuration investigated in this
analysis moved the OPGW from an overhead position to an underbuild counterpoise position.
As shown in Figure 5, when the OPGW is used as an aerial counterpoise, the overhead shield is
replaced with a steel shield wire.
Figure 5: OPGW Underbuild Examples.
2.3 Modeling Details
The transmission line models extend for approximately 5.5 miles in the modeled space. The
conductors were modeled without sag. The parallel metallic objects are modeled with a one mile
length, and centrally located with respect to the transmission line path. Details of the physical
and electrical characteristics for this analysis are described below.

Figure 6: Modeled Shared Corridor.
Phase conductors were modeled as a single conductor. The type of metal and bundling is
insignificant to the magnetic effects on the parallel metallic objects. Each phase conductor is
energized with 1,000 amps of steady state current with 120 degrees between phases. A one volt
GPR-potential energization was placed at the end of each phase conductor to allow a place for
the current to flow outside of the model.
Tower structures were modeled as solid steel, relative resistivity of 12 and permeability of 150,
with a radius of 0.25’. Each tower is placed ten feet in the soil to approximate the typical
structure grounding. A 100 ohm-m uniform soil model was used for this analysis. Structure
spans were modeled at 800 feet for the 345 kV lines and 600 feet for the 115 kV lines.
The OPGW, shield wire, and ACSR underbuild conductors are modeled with a radius of 0.0156
feet (3/8” diameter). The OPGW and ACSR are modeled as aluminum conductors, while the
shield wire is modeled with electrical characteristics consistent with steel. Each shield wire and
underbuild conductor is electrically connected to every tower structure.
The parallel metallic objects are modeled as steel pipe and are perfectly insulated. By modeling
with perfect insulation, the voltage on the metallic object is only developed by magnetic effects
from the modeled transmission line. The pipelines have a radius of 0.25 feet with a hollow inner
section of 0.20 feet. These pipelines are buried to a depth of three feet and located various
distances away from the modeled centerline.
Both pipelines and railroads have leakage paths to the adjacent soil. For a pipeline this is
typically expressed in a resistance per unit area [2]. For railroads this is defined as the resistance
between two rails 1,000 feet in length. The calculation and modeling of these effects has been
discussed at previous CDEGS Users Group Meetings [3]. Utilizing perfect insulation allows for
comparison of worst-case results.

3 Results
All results for this analysis were produced utilizing the HIFREQ module of CDEGS. The impact
of each of the configuration was analyzed for parallel objects at the following distances (in feet)
from the centerline of the transmission line: 50, 100, 150, 200, 300, 400, 500, 600, 800, and
1000.
Magnetic induction produces a voltage on the objects with a positive value at one end and a
negative value at the other end. Since the objects are exactly parallel the magnitudes are equal,
results are reported as the magnitude of the ground potential rise at one end of the object. All of
the results in Sections 3.1 through 3.3 are for a one mile parallel object.
The results are broken down by configuration type as the effects for the cases analyzed are
predominately impacted by the phasing configuration.
3.1 Vertical Configurations
In the single circuit vertical phasing configuration the underbuilt ACSR conductor provides the
greatest reduction in GPR along the parallel object for both the 115 kV and 345 kV transmission
lines. The single circuit results can be seen in Figure 7 below. The underbuilt OPGW option
provides very little benefit, and could produce higher voltages as seen in the objects farther from
the line in the 345 kV model.
Figure 7: Single Circuit Vertical Configuration Results.
An underbuilt conductor actually amplifies the magnetic induction for a cross phased double
circuit line. Figure 8 shows the increased induced voltages versus a standard shielding
configuration. This happens because the induced current in the aerial counterpoise reduces the
balance provided by the cross phased transmission line. One important consideration is that
single circuit operation of the transmission line may need to be considered for any actual AC
interference analysis.

Figure 8: Double Vertical Configuration Results.
3.2 H-Frame Configurations
The H-frame configuration is unique in both the separation of conductors and total separation
of the shield conductors. For this configuration, when the OPGW is overhead it is located atop
the pole closest to the object of interest. Placement of an OPGW below the phase conductor
closest to the object produces the largest reduction in magnetic coupling. The ACSR aerial
counterpoise can actually increases the GPR on the parallel object, because the interaction
between the ACSR underbuild and overhead OPGW conductor. As a result, the voltage increases
on objects farther from the transmission line, but still reduces voltages for objects closer to the
line. The results for this configuration are in Figure 9.
Figure 9: H-Frame Configuration Results.
Another option for the H-frame configuration is to place the overhead OPGW atop the pole
farthest from the parallel metallic object. Without any aerial counterpoise this increases the
magnetic coupling comparing the base results above, Figure 9, to those below, Figure 10. This
configuration option places the overhead OPGW such that it contributes to a larger net magnetic
field on the object of interest. The OPGW underbuild is identical to the previous case as it is
replaced with a steel shield wire. In this configuration the ACSR underbuild is an improvement
compared to a configuration without underbuild.

Figure 10: H-Frame Results Alternate OPGW Location.
3.3 Delta Configuration
The delta configuration has less magnetic coupling to the object of interest with either of the
aerial counterpoise options. Which option provides the greatest benefit is dependent on the
distance of the object from the transmission line. For objects close to the transmission line the
OPGW underbuild provides the greatest reduction in voltage, while objects farther away benefit
more from ACSR aerial counterpoise. Figure 11 contains the results for the delta configurations.
Figure 11: Delta Configuration Results.
3.4 Summary of Results
The results for the configuration types are similar even at different voltage levels. There are
differences between the voltage levels but the trends remain very similar. The 115 kV and 345 kV
results have been tabulated at select distances representative of the curves shown earlier. Table 1
and Table 2 show these results for the 115 kV and 345 kV results respectively. All of the results in
the tables are reported as a percentage of induced voltage compared to the base configuration.
Any cases that exceed the base are in red text.

Table 1: Tabulated Results for Selected Distances 115 kV
Percentage of Induced Voltage Compared to

Aerial Base Configuration
Configuration Counterpoise
Type 50 100 200 500 800
feet feet feet feet feet
ACSR 54% 44% 36% 34% 34%

Vertical
OGPW 43% 56% 79% 95% 99%
ACSR 96% 116% 123% 122% 122%

Double Vertical
OGPW 90% 125% 143% 146% 146%
ACSR 89% 95% 109% 128% 133%

H-Frame
OGPW 67% 64% 67% 84% 90%
H-Frame ACSR 64% 61% 55% 49% 48%

Alternate
OPGW OPGW 42% 39% 39% 52% 59%
ACSR 54% 44% 36% 34% 34%

Delta
OGPW 43% 56% 79% 95% 99%

Table 2: Tabulated Results for Selected Distances 345 kV
Percentage of Induced Voltage Compared to

Aerial Base Configuration
Configuration Counterpoise
Type 50 100 200 500 800
feet feet feet feet feet
ACSR 64% 62% 62% 67% 70%

Vertical
OGPW 56% 64% 98% 138% 151%
ACSR 74% 100% 130% 137% 136%

Double Vertical
OGPW 72% 115% 173% 199% 200%
ACSR 77% 80% 85% 110% 126%

H-Frame
OGPW 61% 60% 58% 82% 101%
H-Frame ACSR 60% 59% 51% 41% 39%

Alternate
OPGW OPGW 45% 44% 38% 48% 59%
ACSR 59% 47% 31% 23% 32%

Delta
OGPW 40% 20% 18% 65% 90%
4 Conclusion
The location and type of shield conductors as well as aerial counterpoise has a large impact on
the magnetic coupling to parallel metallic objects in a shared corridor. The optimal solution is
largely dependent on the transmission line configuration and the separation distance between
the transmission line and object of interest. Installation of aerial counterpoise on a transmission
line is much cheaper when included in the original design, in part due to structure loading
considerations. Whether or not it needs to be installed, including provisions for the installation
of an aerial counterpoise conductor maybe a good investment of time and resources for
corridors shared with railroads or pipelines.
While the results presented in this paper cover a limited set of the possible transmission line
configurations, the goal is to provide a reference on the order of magnitude that different
shielding configurations provide for magnetic coupling. Table 3 below shows the preferred
shielding design for each transmission line configuration based on the configurations examined
in this paper.

Table 3: Preferred Shielding Configuration
Transmission Line Optimal Design for Optimal Design for

Configuration Close Objects Distant Objects
Vertical Single Circuit ACSR/OPGW UnderBuild ACSR UnderBuild
Vertical Double Circuit None* None*
H-Frame OPGW UnderBuild None
H-Frame Alternate OPGW OPGW Underbuild ACSR Underbuild
Delta OPGW UnderBuild ACSR UnderBuild
*Double Vertical Configurations with rolled phasing do not benefit from aerial counterpoise, however there
may be issues in single circuit operation of line without an underbuilt conductor.
Installing an optimized shielding or aerial counterpoise configuration does not eliminate the
need for detailed analysis of a joint use corridor. Depending on the type of object the acceptable
voltage limits on the object may vary widely. NACE and EPRI provide documentation on the
acceptable limits and analyses that must be performed for pipelines and railroads respectively
[1][4].
5 Acknowledgement
Authors thank colleague Rob Schaerer for his review of this paper.
6 References
[1] EPRI, "Power System and Railroad Electromagnetic Compatibility Handbook," EPRI, 2006.
[2] Doug Gilroy, "Pipeline Coating Resistance and Coating Stress Voltage Basics," in CDEGS Users'
Conference Proceedings, Rock Creek Resort, Montana, USA, 2009.
[3] Luis Valcarcel, Syvie Lefebvre and Farid Paul Dawalibi, "A Simplified Method for Computing
Equivalent Parameters for Rail Ballast Resistance in Inductive and Conductive Models," in CDEGS
Users' Conference Proceedings, Montreal, Canada, 2010.
[4] NACE, "Mitigation of Alternating Current and Lightning Effects on Metallic Structures and Corrosion
Control Systems," NACE, 2014

Part II: SES Contributions
4 INCLUSION OF INFORMATION REGARDING DEEP LAYER

SOIL FROM REMOTE LONG TRAVERSES TO IMPROVE
DEEP SOIL RESISTIVITY ESTIMATES
Majid Siahrang
Safe Engineering Services & technologies ltd.

Email: info@sestech.com, Web Site: www.sestech.com
Abstract
It sometimes happens that, due to limitations imposed by the available space at the site of a grounding system,
performing soil resistivity measurements to large probe spacings to extract adequate information regarding the
resistivity of the soil at deep layers is not possible at the location of the installation. However, it may still be possible
to carry out soil resistivity measurements to larger probe spacings at locations which are some distance away from
the site. When the measured data corresponding to the common probe spacings along the local and remote
measurement traverses are relatively consistent, the raw measurement data obtained from both traverses can be
directly combined and processed for the soil model interpretation. However, when the measured data along local
and remote traverses are highly inconsistent at common probe spacings, the measurement results obtained from
the remote traverse should be processed independently and the interpreted deep layer soil structure should be used
as supplementary data for the interpretation of the soil structure at the site location.
1 Introduction
Having adequate information about the soil structure is highly important in the design and
analysis of grounding systems. Ground resistance, touch and step voltages, and ground potential
rise (GPR) very much depend on the electrical resistivity of the soil, not only at grounding grid
depth but also at depths on the same order as the maximum dimension of the grounding system
or deeper [1].
For grounding design and analysis studies, the soil is typically modeled as a horizontally layered,
isotropic electrical medium. A multilayer soil structure model allows for the consideration of
variations in moisture content, concentration of mineral salts, grain size, compactness,
temperature, and other factors affecting resistivity, as a function of depth [2]–[4].
Data obtained from soil resistivity measurements are used to estimate the electrical resistivity of
the soil as a function of depth. Four-probe measurement techniques applied along line traverses
are usually the method of choice for soil resistivity measurements. With these methods, four
probes are arranged along a straight line (the measurement traverse) so that while a current is
forced to circulate through the soil between two of the measurement probes (the current probes),
the potential difference between the two other probes (the potential probes) is measured. As the
spacing between the probes is increased from small to large values, the ratio of the measured
potential difference between the potential probes to the current circulating between the current
probes is recorded as the so-called “apparent resistance”. This set of the apparent resistance
versus probe spacing data should be analyzed to find an equivalent horizontally layered soil
structure that is consistent with the measurement data [5].

PART II: SES CONTRIBUTIONS
The data obtained at each probe spacing depends on the resistivity values of a range of soil depths,
with the weight of deeper layers increasing with probe spacing. Data from larger probe spacings,
therefore, contain more information regarding the resistivity of deeper soil. To obtain reasonable
accuracy in grounding system design and performance prediction, it is necessary to carry out soil
resistivity measurements along traverses whose lengths are on the order of several times the
maximum dimension of the grounding system under study [1].
In many cases the available land at the site under study may not allow soil resistivity to be carried
out measurements with adequately large probe spacings, to reveal information regarding deep
soil characteristics. Under such circumstances, it may still be possible to measure soil resistivity
along longer traverses, at locations which are some distance away from the site. If the long
resistivity measurement traverses are not located too far from the station site, then under the
assumption that the deep layer soil characteristics are the same at the site location and at the
remote measurement locations, the data obtained from the more distant soil resistivity
measurements can be used in the determination of the deep soil resistivity at the station site.
When the data measured at the short probe spacings are relatively consistent at the local and
remote measurement locations, the measured data at large spacings obtained from the longer
traverses can be directly included with the available data from the local traverses. Then the
combined measurement data can be used as input data points for the soil structure interpretation
process [6].
On the other hand, when the soil resistivity values measured along traverses are not consistent at
shorter probe spacings, the measurement results obtained from the long traverse should be
processed independently and the characteristics of the interpreted deep layer soil structure should
be used as supplementary data for the interpretation of the soil structure at the site location.
In this paper, through two case examples, it is shown that for cases where the shallow soil
characteristics are quite different at the location of short and long traverses, then assuming that
the variation of the deep soil characteristics between the measurement locations is not important,
the data obtained from the long traverses can be used to improve the estimation of the soil
structure at the location of the local short traverses.
2 Methodology
As observing sharp changes in the topology of land surface and top layer soil structure from one
location to another nearby location is not uncommon, encountering significant differences in the
electrical resistivity of soil at shallow depths can be expected within a given area. On the other
hand, the soil structure tends to be more stable and vary more smoothly as we move downward
from the earth surface. Consequently, for relatively nearby locations, the electrical resistivity of
the soil at deep layers is expected to be on the same order of magnitude.
Even when the soil structures at two nearby locations are not quite the same at greater depths, as
the soil resistivity data measured at large probe spacings represent an average surface response
of the deep soil structure over a wide area (as compared with the data measured at small probe
spacings, which are highly localized), these data can still be relied upon to be representative of the
surface response of the deep soil at both locations and therefore suitable for use in the modeling
of a grounding grid.

Although the weight of the soil resistivity of deep layers becomes more important in measured
apparent resistivities as probe spacings increase, the effect of the shallow depth soil is always
present in the measured data. That is to say, even when deep soil structures are identical at two
locations, significant differences between the shallow depth soil structures can result in
inconsistent measured data at these locations not only at small probe spacings, but also at large
probe spacings.
Since the measured soil resistivity at a large spacing is influenced by the soil resistivity at
shallower depths, the direct addition of the measurement data obtained from the longer traverses
at one location to the measurement results obtained from the short traverses at another location
more or less nearby makes sense if the measurement readings for the common probe spacings are
consistent between the two sets of traverses. If the measured data at smaller probe spacings along
the two sets of traverses are significantly inconsistent, then the measured apparent resistivity
obtained from a remote long traverse should be interpreted separately and the estimated
information regarding the deep layer soil structure (i.e. the resistivity and thickness of equivalent
soil layers) should be considered as supplementary conditions in the inversion process of data
obtained from the measurement carried out along the local short traverses. The procedure is
explained in the flowchart shown in Figure 1.
Start
Remote Local
measurement measurement
data data
Soil structure
interpretation
Soil structure
interpretation
Deep layer
soil
structural
data
Soil structure based on local measurement data

supplemented with data obtained from remote long
traverses
Figure 1: Flowchart of procedure to include data obtained from a remote long traverse in soil structure
interpretation process.

3 Case Study
In this section, we apply the methodology introduced in the previous section to improve the soil
structure estimate at the location of a small grounding grid whose geometrical and physical
parameters are shown in Figure 2.
Let us assume that the dimension of the available (and accessible) land at the installation location
is just about the size of the grounding grid, i.e. 40 m by 40 m. However, it is assumed that not too
far from the grid, there is accessible space which makes it possible to perform soil resistivity
measurements up to larger probe spacings.
In the following two examples, we assume that the soil resistivity measurements are carried out
according to the Wenner four-probe method [5] along two traverses, i.e. one at the installation
location (before installing the grid) which will be referred to as “local traverse” and the other with
larger probe spacings at the location of the aforementioned available space which will be referred
to as “remote traverse”. Note that although only one local and one remote traverses are used in
this example for illustrative purposes, several traverses are normally used for an actual grounding
study.
Figure 2: Grounding system design.
3.1 Case Example I
As the first case example, let us suppose that the actual soil structure at the installation location
is a four-layer soil, with the characteristics shown in Table 1. This table also includes the soil
structure at the location of the remote measurement traverse. According to the data shown in
Table 1, the soil structure up to a depth of 8.5 m is quite different at the location of the local and
remote resistivity measurement traverses. On the other hand, as shown in Table 1, the deep layer
soil structures at those two locations are identical.

Table 1: Actual soil structures at soil resistivity measurement locations
Table 2 presents the apparent resistivity values that would be measured along the local and
remote traverses (prior to construction of the installation). The maximum electrode spacing is
limited to 15 m in the local measurement, while for the remote measurement, as the available
space allows for it, the maximum electrode spacing is 150 m. The data presented in Table 2, are
actually synthetic data obtained from simulation of the soil resistivity measurements, using the
RESAP module of the SES software (by locking the thickness and the resistivity of the soil layers),
based on horizontally layered soil structures presented in Table 1.
Table 2: Soil resistivity measurement data
The measured apparent resistivities along the local and remote traverses are plotted in Fig 2. As
shown in Table 2 and Figure 3, the large difference between the top soil layers at the location of
the measurement traverses is reflected in the measured apparent resistivities, so that there is a
large discrepancy between the measured data along the traverses at common short probe
spacings.

Figure 3: Measured apparent resistivity along local and remote traverse as a function of probe pacing.
The data listed in Table 2 are processed with the RESAP module of SES software to obtain
equivalent horizontally layered soil models that fit the measurement results for the local and
remote traverses separately. Figure 4 and Figure 5 show the results of the soil resistivity
interpretation process, respectively, for the local and remote locations.
The green curve shown in Figures 4 and 5 represent the apparent soil resistivity values that would
have been measured for the interpreted soil structure. As can be observed, the computed curve of
the apparent soil resistivity matches the measured data quite well. This is an indication that the
data inversion algorithm (i.e. RESAP) has done a good job in finding equivalent soil structures
based on the measurement data.
Comparing the interpreted soil structure corresponding to the local soil measurements (see
Figure 4) with the data presented in Table 1, it is observed that the characteristics of the two
bottom soil layers are not detected in the deduced soil structure. This arises from the lack of data
corresponding to sufficiently large probe spacings along the local traverse. On the other hand,
thanks to the available data corresponding to larger probe spacings, the interpreted soil structure
for the remote location is quite consistent with the actual structure at this location.

Figure 4: Interpreted soil structure based on local traverse only.
Figure 5: Interpreted soil structure based on remote traverse only.

To improve the soil structure estimate at the location of the grounding grid, we apply the
procedure shown in Figure 1 to use the data regarding the deep layer soil structures as imposed
conditions for reinterpreting the measurement data obtained from the local measurement
traverse. As shown in Figure 6, the resistivity values of the last two deepest layers of the
interpreted soil structure at the remote location are set as constraints for the resistivity values of
the two deepest soil layers at the location of the local traverse. The thickness of the bottom layer
is set to infinity, while the thickness of the penultimate layer is left to be determined by the data
inversion algorithm (i.e. RESAP).
Figure 6: Imposing data regarding the characteristics of bottom layers as constraints in interpreting soil
data obtained from local traverse.
Figure 7 represents the results of the soil data interpretation based on data obtained from the
local traverse, supplemented by the soil data interpreted from remote traverse. As can be
observed, the reinterpreted soil structure is in very good agreement with the actual soil structure,
at the location of the local traverse (i.e. the location of the grounding grid).

Figure 7: Interpreted soil structure, based on local traverse, supplemented by soil data interpreted from
remote traverse.
Table 3 presents the computed values for the ground resistance of the grid (introduced in Figure
2), considering the actual soil structure, along with the three interpreted soil structures presented
in Figures 4, 5 and 7. The listed ground resistances are computed using the MALZ module of the
CDEGS software package.
Table 3: Computed grid resistance
Comparing the computed values for the ground resistance corresponding to the three interpreted
soil models with the ground resistance computed when considering the actual soil structure (at
the installation location), we can see that the computed ground resistance when the soil model is
based on the data obtained from the local measurement data, supplemented by the structural soil
data regarding the deeper soil layers is in an acceptable agreement with the real ground resistance
of the grid. As the actual soil structure and the reinterpreted soil structure are quite consistent, it

is expected that the performance of the grounding grid considering this soil model will be
realistically representative of the real performance of the grounding system under study.
3.2 Case Example II
The second case example presented in this article is quite similar to the first one, the only
difference, being that the electrical resistivity of the two topmost soil layers at the local and remote
locations are interchanged, as can be seen in Table 4.
Table 4: Actual soil structures at soil resistivity measurement locations
Table 5 presents the apparent resistivity values that would be measured along the local and remote
traverses (prior to construction of the installation), according to the Wenner four-probe method.
As before, the maximum electrode spacing is limited to 15 m for the local measurements, while
for the remote measurements the maximum electrode spacing is 150 m. Again, it is worth
mentioning that the data presented in Table 2 are actually synthetic data obtained from
simulation of soil resistivity measurements, assuming the homogeneous, horizontally layered soil
structures presented in Table 4.
As shown in Table 5 and Figure 8, the large difference between the top layers of the soil at the two
measurement locations is reflected in the measured apparent resistivities, so that there is a large
discrepancy between the measured data along the traverses at common, short probe spacings.

Table 5: Soil resistivity measurement data
The measured apparent resistivities along the local and remote traverses are plotted in Figure 8.
Figure 8: Measured apparent resistivity along local and remote traverse as a function of probe spacing.

The data listed in Table 5 are processed with the RESAP module to obtain equivalent horizontally
layered soil models that fit the measurement results for each of the local and remote traverses
separately. Figure 9 and Figure 10 show the results of the soil resistivity interpretation process,
respectively, for the local and remote locations.
Figure 9: Interpreted soil structure based on remote traverse only.
The computed curve of the apparent soil resistivity (i.e. the green curve) matches the measured
data (blue dots) quite well. This is an indication that the data inversion algorithm (i.e. RESAP)
has done a good job in finding equivalent soil structures, based on the measurement data.
Comparison of the interpreted soil structure corresponding to the local soil measurements (i.e.
shown in Figure 10) with the data presented in Table 4 shows that the characteristics of the two
bottom soil layers are not visible in the deduced soil model. This arises from the lack of data
corresponding to adequately large probe spacings along the local traverse. On the other hand,
thanks to the available data corresponding to larger probe spacings, the interpreted soil model for
the remote location is quite consistent with the actual structure at this location.

Figure 10: Interpreted soil structure based on local traverse only.
To improve the soil structure estimate at the location of the grounding grid installation, we apply
the procedure shown in Figure 1 to use the data regarding the deep soil as imposed conditions for
reinterpreting the measurement data obtained from the local measurement traverse. As shown in
Figure 11, to this end, the resistivity values of the two deepest layers of the interpreted soil
structure at the remote location are set as constraints for the resistivity of the two deepest soil
layers at the location of the local traverse. The thickness of the bottom layer is set to infinity, while
the thickness of the penultimate layer is left to be determined by the data inversion algorithm (i.e.
RESAP).
Figure 11: Imposing data regarding the characteristics of bottom layers as constraints in interpreting
soil data obtained from local traverse.

Figure 12 presents the results of soil data interpretation based on data obtained from the local
traverse supplemented by soil data interpreted from the remote traverse, following the
methodology described in Figure 11. The reinterpreted soil structure is in a very good agreement
with the actual soil structure at the location of the local traverse (i.e. the location of the grounding
grid).
Figure 12: Interpreted soil structure, based on local traverse, supplemented by soil data interpreted
from remote traverse.
Table 6: Computed grid resistance
Table 6 presents the computed values for the ground resistance of the grid (introduced in Figure
2), considering the actual soil structure, along with the three interpreted soil structures presented
in Figures 9, 10 and 12.
Comparing the computed values of the ground resistance corresponding to the three interpreted
soil models with the ground resistance computed with the actual soil structure at the installation
location (see Table 5), we can see that the computed ground resistance considering the soil model
which is based on the data obtained from the local measurement data, supplemented by the
structural soil data regarding the bottom soil layers, is in an acceptable agreement with the real
ground resistance of the grid. As the actual soil structure and the reinterpreted soil structure are

quite consistent, it is expected that the performance of the grounding grid considering this soil
model be representative of the real performance of the grounding system under study.
4 Conclusion
This article describes a procedure that makes it possible to include information obtained from soil
resistivity measurements carried out at a more distant location to improve the soil structure
estimate at the location of a grounding grid installation, where limitations imposed by the
available space do not make it possible to perform on-site soil resistivity measurements with
adequately large probe spacings. In particular, it has been shown that when the data measured at
smaller probe spacings, at the local and more distant measurement locations, are highly
inconsistent, to improve the soil structure estimation at the location of the grid the measured
apparent resistivity obtained from the remote long traverse should be interpreted separately and
the estimated information regarding the deep layer soil structure (i.e. the resistivity and thickness
of equivalent soil layers) should be considered as supplementary conditions in the inversion
process of data obtained from the measurements carried out along the local short traverses.
To examine the performance of the discussed methodology, we applied it to two case examples in
which the data measured along a remote soil resistivity measurement traverse were used to
improve the estimate of the soil structure at the site of a grounding grid installation.
5 References
[1] R. D. Southey and F. P. Dawalibi, “Improving the Reliability of Power Systems with More
Accurate Grounding System Resistance Estimates,” in Proc. Int. Conf. Power Syst. Technol.,
Oct. 2002, vol. 1, pp. 98–105.
[2] F. P. Dawalibi, J. Ma, and R. D. Southey, “Behaviour of Grounding Systems in Multilayer Soils:
A Parametric Study,” IEEE Trans. Power Del., vol. 9,no. 1, pp. 334–342, Jan. 1994.
[3] H. Lee et al., “Efficient Grounding Designs in Layered Soils,” IEEE Trans. Power Del., vol. 13,
no. 3, pp. 745–751, Jul. 1998.
[4] F. P. Dawalibi and N. Barbeito, “Measurement and Computation of the Performance of
Grounding Systems Buried in Multilayer Soils,” IEEE Trans. Power Del., vol. 6, no. 4, pp.
1483–1490, Oct. 1991.
[5] R. D. Southey , M. Siahrang, S. Fortin and F. P. Dawalibi, “Using Fall-of-potential
Measurements to Improve Deep Soil Resistivity Estimates” IEEE Trans. Industry App., vol.
51, no. 6, pp. 5023–5029, Nov. 2015.
[6] F. P. Dawalibi, S. Tee, N. Mitskevitch and J. Ma, “Shallow Surface Soil Modeling Associated
with Local and Seasonal Resistivity Variations” The 5th International Conference on Power
Transmission & Technology, Beijing, China, Oct. 2005.

5 CAN A GROUNDING GRID IMPEDANCE BE CAPACITIVE

AT POWER FREQUENCIES?
Robert D. Southey, Marjan Mehrabi
Safe Engineering Services and technologies ltd.

Email: Info@sestech.com, Web Site: www.sestech.com
Abstract
Previous work performed by SES has found that the impedance of a conductor located in a horizontally layered soil
can decrease with increasing frequency, if the conductor is located in a layer which has a lower resistivity than the
layers below. The present paper further investigates this phenomenon, based on a large grounding system modeled
by HIFREQ and MALZ in a number of two-layer soils, and proposes an explanation for the observed behavior of the
grounding system. It is shown that in extreme cases, the HIFREQ module is required to obtain realistic results,
which can only be obtained by accounting for inductive effects between conductors and induced currents in the
earth.
1 Introduction
How could a grounding grid impedance possibly be capacitive? The circuit involved, from the
current injection point into the grounding grid to the ultimate drain at remote earth, appears to
be nothing but inductive and resistive in nature, with the grounding conductors through which
the current initially flows being primarily inductive and the earth being more resistive than
anything else. For those wags, who will point out that at sufficiently high frequencies, the
permittivity of the soil will introduce a capacitive component, it should be pointed out that at
60 Hz, the complex conductivity of the soil, given by the formula σ + 2пfε, is overwhelmingly
resistive, due to the very small value of epsilon (8.854 x 10-12), even for very high soil resistivity
values. The authors invite the reader to try to come up with a system, using electromagnetic field
theory, transmission line equations or circuit theory, whose input impedance is capacitive, when
made up of only resistive and inductive elements. Failure is the likely result, until the reader,
applying Maxwell’s equations, realizes that the paradigm of a point source of current being
injected into a grounding grid, out of nowhere, then flowing onward to remote earth, is somewhat
unphysical. This is the beginning of much frustration, but also the inkling of a solution.
But before we spoil the fun, let us show you some rather surprising results obtained from HIFREQ
simulations of a large grounding system in a top layer of low resistivity soil, underlain by very high
resistivity bedrock.
2 Computer Simulations
As shown in Figure 1, the grounding grid is roughly 500 m x 500 m, with 45 m x 45 m meshes and
18 mm x 20 m ground rods at all nodes. It is made up of 4/0 copper, and is buried 0.5 m deep in
a two-layer soil whose bottom layer resistivity is 15,000 ohm-m, a value which can be encountered
in the Canadian Shield, and whose top layer characteristics are varied for maximum effect. A

nominal current is injected into the center of the grounding grid. This is done for two frequencies:
0 Hz and 60 Hz. The experiment is carried out with HIFREQ, then repeated with MALZ.
Figure 1: Grounding Grid Studied.
3 Results and Discussion

Tables 1 and 2 summarize the results of the study, which focuses on the computed grid impedance,
defined as the ground potential rise at the point of current injection into the grid divided by the
injection current. In Table 1, the top layer thickness is set to 250 m and the resistivity of this layer
varied from 5 ohm-m to 80 ohm-m, in order to find a scenario that maximizes the capacitive effect
and the difference in impedance between 0 Hz and 60 Hz. In Table 2, the top layer resistivity is
set to 5 ohm-m and the top layer thickness varied from 5 m to 200 m, again with the express aim
of impressing the reader with dramatic differences in 0 Hz and 60 Hz impedance and as capacitive
an impedance as possible.
The winner in both categories is a 5 ohm-m soil top layer: when its thickness is 10 m, it results in
a grid impedance with an angle of -25 degrees. On the other hand, when its thickness increases to
60 m, its 60 Hz impedance is a full 4.9 times lower than its 0 Hz impedance and a similar factor
with respect to the 60 Hz impedance computed by MALZ.

Table 1: Top Layer Thickness is 250 m and Bottom layer Resistivity is 15,000 Ω.m for all cases.
Grid Impedance Results in Grid Impedance Results in Impedance Ratio

Top Layer HIFREQ (Ω) MALZ (Ω) (HIFREQ 0 Hz/
Resistivity
60 Hz 0 Hz 60 Hz 0 Hz HIFREQ 60Hz)
5 Ω.m 0.01555 ∠ 54 0.02698 ∠ 0 0.03522 ∠ 24 0.02821 ∠ 0 1.74
10 Ω.m 0.01723 ∠ 46 0.05009 ∠ 0 0.05617 ∠ 16 0.05009 ∠ 0 2.91
20 Ω.m 0.02258 ∠ 18 0.09182 ∠ 0 0.09553 ∠ 10 0.09171 ∠ 0 4.1
25 Ω.m 0.02774 ∠ 7 0.11124 ∠ 0 0.11425 ∠ 8 0.11124 ∠ 0 4.01
30 Ω.m 0.03417 ∠ 0.4 0.12996 ∠ 0 0.13245 ∠ 7 0.12928 ∠ 0 3.8
40 Ω.m 0.04951∠ -8 0.16581 ∠ 0 0.16757 ∠ 6 0.16476 ∠ 0 3.35
50 Ω.m 0.0669 ∠ -12 0.2 ∠ 0 0.20121 ∠ 5 0.19864 ∠ 0 2.99
60 Ω.m 0.0856 ∠ -15 0.23288 ∠ 0 0.23361 ∠ 4 0.23151 ∠ 0 2.72
70 Ω.m 0.10522 ∠ -16 0.26469 ∠ 0 0.26516 ∠ 4 0.26288 ∠ 0 2.52
80 Ω.m 0.1255 ∠ -17 0.29559 ∠ 0 0.29579 ∠ 3 0.2936 ∠ 0 2.35
Table 2: Top Layer Resistivity is 5 Ω.m and Bottom layer Resistivity is 15,000 Ω.m for all cases.
Grid Impedance Results in Grid Impedance Results in Impedance Ratio

Top Layer HIFREQ (Ω) MALZ (Ω) (HIFREQ 0 Hz/
Thickness
60 Hz 0 Hz 60 Hz 0 Hz HIFREQ 60Hz)
5m 0.41265 ∠ -20 0.6457 ∠ 0 0.64658 ∠ 1 0.64428 ∠ 0 1.56
10 m 0.1769 ∠ -25 0.3759 ∠ 0 0.38014 ∠ 2 0.3773 ∠ 0 2.13
20 m 0.06861 ∠ -24 0.2138 ∠ 0 0.22087 ∠ 4 0.21708 ∠ 0 3.11
30 m 0.03846 ∠ -17 0.15228 ∠ 0 0.16077 ∠ 6 0.15634 ∠ 0 3.96
50 m 0.02027 ∠ 1 0.09863 ∠ 0 0.10825 ∠ 8 0.10304 ∠ 0 4.87
60 m 0.01722 ∠ 10 0.0845 ∠ 0 0.0942 ∠ 11.4 0.0888 ∠ 0 4.91
80 m 0.01462 ∠ 25 0.06639 ∠ 0 0.07592 ∠ 13.7 0.07014 ∠ 0 4.54
100 m 0.01384 ∠ 35 0.05533 ∠ 0 0.06445 ∠ 13 0.05846 ∠ 0 3.99
150 m 0.01404 ∠ 49 0.04003 ∠ 0 0.04848 ∠ 17 0.04211 ∠ 0 2.85
200 m 0.01487 ∠ 53 0.03198 ∠ 0 0.04022 ∠ 21 0.03352 ∠ 0 2.15
How is this possible? Should not the grid conductor inductive impedance be greater at 60 Hz than
at 0 Hz, thus acting as an inductive choke and reducing the effective electrical extent of the
grounding grid, thereby increasing its total impedance, not to mention its inductive reactance?
That is what the authors believed and had the HIFREQ developers scrambling to find the
pernicious bug responsible for this untoward behavior, to no avail. “Wave effects,” muttered one
eminent researcher. “Wave effects” became the new mantra… leading to an attempt to explain the
observed behavior by going back to basics and applying Maxwell’s equations to highly simplified
systems, involving infinite media and spherical electrodes.
The only problem with Maxwell’s equations is that they only work properly with physical systems,
which respect such fundamentals as conservation of charge, a condition not respected when we
inject current from an imaginary point source into a grounding grid.
On the other hand, if we start to think about a truly physical system, we realize that the current
flowing into the grounding grid must be coming from somewhere. Where? Typically over a
transmission line, carrying current over the earth for many kilometers from a remote source. And
what effect might this transmission line have on the grid impedance? Aha! Now we are close to an
answer. Consider the simplified circuit illustrated in Figure 2, which shows the overhead line,
assumed for simplicity to have a single phase and no shield wires, a resistance to remote earth,
representing a path from the grounding grid to a multitude of points a great distance away in all
directions, whence the current finds its way back to the remote source, and last but not least, a
resistance representing the volume of earth running under the transmission line from the faulted
grounding grid back to the remote source. What is the nature of this volume of earth? It is
primarily resistive and is inductively coupled to the phase conductor running above it.
Figure 2: Simplified Circuit Illustrating Action of Overhead Line on Current Flow to Remote Earth.
In order to illustrate the effect of the soil volume under the faulted phase, a simple HIFREQ
simulation was performed. A 30 km long conductor representing one phase of an overhead
transmission line was modeled in HIFREQ, 10 m above ground, parallel to the earth surface, as
shown in Figure 3. The soil model was specified as a two-layer soil with 5 Ω.m top layer resistivity
and 50 m thickness, over an infinite 15,000 Ω.m resistive layer. A nominal current of 10 kA flows
through the overhead conductor. Observation points were modeled, on a vertical plane, midway
along the line and orthogonal to it.
Figures 4 and 5 show the induced earth current density flowing in parallel to the line, in the low
resistivity top layer of the ground, throughout the 50 m tall by 1000 m wide surface of observation
points shown in Figure 3. A rough estimate of the total current flowing through this surface would
be 50 m x 1000 m x 0.18 A/m2 = 9 kA, if one accepts that 0.18 A/m2 represents the average current
density through the surface. This represents a large proportion of the 10 kA current flowing in the
line and can therefore be expected to have a significant impact on the GPR of a grounding grid to
which this line is eventually faulted. Figure 6 indicates the phase angle of the current flowing in
earth, which ranges from -110 to -137 degrees, which is characteristic of an induced voltage in a
circuit that is more resistive than inductive.

Figure 3: Simplified System Modeled with HIFREQ to Illustrate Action of Overhead Line on Current Flow
in Earth beneath Line.

Figure 4: Current Density in Top Layer of Soil versus Position. Overhead line is located at X=500 m. Each
curve represents a different depth, ranging from 0.1 m to 50 m.

Figure 5: Current Density Magnitude in Cross Section of Top Soil Layer, Midway along Overhead
Conductor Length.
Figure 6: Current Density Angle in Cross Section of Top Soil Layer, Midway along Overhead Conductor
Length.

If we analyze the circuit illustrated in Figure 2, we immediately realize two things:

1. The inductive coupling of the overhead line to the earth beneath it will tend to draw
current out of the grounding grid, just as it tends to do with shield wires, thus tending to
reduce the GPR of the grounding grid for a given applied injection current (or increase the
current flowing into the grounding grid for a given applied GPR) and thus reduce its
apparent impedance. This induced current may not be highly apparent under normal
circumstances, but when the top layer has a very low resistivity and sufficient thickness, it
can have a noticeable effect, even at 60 Hz. This is all the more true when the bottom soil
has a very large resistivity, which impedes the flow of current from the grounding grid to
remote earth, thus making the induced current in the top layer all the more important
compared with the trickle flowing through the bedrock.
2. The induced current in the earth will be roughly 90 degrees out of phase with the fault
current in the phase conductor, since the induced emf in the earth is 90 degrees out of
phase with the source current in the phase wire, whereas the earth impedance is close to
being purely resistive: the current flow will be the induced emf divided by the earth
impedance, yielding something like a 90-degree angle. A hand calculation or an FCDIST
run will demonstrate that the phase angle of the induced earth current flowing away from
the grounding grid leads the phase angle of the current flowing from the phase wire into
the grounding grid, resulting in an apparent capacitance. Place this in parallel with the
grid impedance to remote earth and you obtain a net capacitive impedance.
So, we have explained the more dramatic phenomena observed in Tables 1 and 2. However, the
grounding system is not capacitive for all cases run, nor is the ratio of 60Hz grid impedance to
0 Hz grid impedance necessarily smaller for lower top layer soil resistivity values or for greater
thickness of the lower resistivity top layer. As far as the low resistivity top layer thickness is
concerned (refer to Table 2), it can be argued that when it is too small, the resistance of this narrow
layer is simply too high to allow induced currents to be large enough to make a great difference,
so the 0 Hz and 60 Hz grid impedances have similar magnitudes. On the other hand, when the
top layer thickness is very large, the grid impedance to remote earth is relatively small to begin
with, so the effect of induced current in the earth is less noticeable… not to mention the fact that
the induced current in the earth decreases as a function of increasing depth, so increasing the
thickness of the top layer does not necessarily increased induced currents in the earth. As a result,
the grid impedance begins to increase with increases of the top layer thickness, when this top layer
becomes very thick.
As for the phase angle of the grid impedance, which becomes more and more inductive as the top
layer thickness increases, it is surely again a question of the induced current in the earth becoming
a smaller and smaller component of the grid impedance observed, to the point of becoming almost
negligible. When this happens, what we see is what we have become conditioned to expect: a large
inductive component, due to current flow through the inductive grid conductors, with a
progressively decreasing contribution to the grid impedance from the thicker and thicker top layer
of very low resistivity earth in which the grid is immersed.
A similar argument can be made with respect to the evolution of grid impedances in Table 1 as a
function of top layer resistivity. Noting that the top layer thickness is quite large in all cases, i.e.,
250 m, it appears that as the resistivity of that layer increases, although the induced earth current
surely becomes progressively smaller, it does not decrease linearly with increasing top soil layer

resistivity and therefore becomes a more important component of the grid impedance, as the
resistance of the path from the grid to the bedrock becomes greater and greater at a linear rate.
This would explain the progressively increasing capacitance of the grid impedance as the top layer
resistivity increases. It would appear, however, that the best cancellation occurs at an
intermediate top layer resistivity, when the grid conductor inductance is still significant compared
with the resistance of the earth between the grid and remote earth, thereby allowing the capacitive
angle of the induced current in the earth to have an optimal effect.
4 Validation
As can be seen in Tables 1 and 2, the HIFREQ simulations were duplicated in MALZ, for
validation. As can be seen in these tables, both HIFREQ and MALZ yield virtually identical results
at 0 Hz, based on their different computation methodologies to calculate earth effects on
interactions between grid conductors. On the other hand, when the grounding system is energized
at 60 Hz, the MALZ and HIFREQ results are quite different, which is to be expected, due to the
modeling assumptions made by each software module.
HIFREQ not only accounts for induced voltages in the earth, as discussed above, it also accounts
for inductive coupling between the grid conductors and indeed, between any overhead conductors
and buried grid conductors, using a full-wave solution of Maxwell’s equations. As a result,
HIFREQ results are much more realistic. On the other hand, MALZ accounts for through-earth
conductor interactions from a quasi-static point of view and accounts for conductor self-
impedances when computing the distribution of current throughout the grounding grid and the
resulting grid performance. However, MALZ does not address inductive coupling between the
conductors modeled. Although this difference is not great in typical grounding applications, it can
make a big difference in unusual studies such as the one described here, in which the grounding
system is extensive, the top soil layer is fairly thick and the bottom layer resistivity is many orders
of magnitude greater than that of the top layer.
This is not the first time that such grid performance has been revealed by HIFREQ. In a previous
study [1], it was found that as the frequency increases for a large grounding grid, the impedance
may decrease.
5 Conclusion
The present paper demonstrates that grounding grid impedance can decrease by a factor of 4.5 or
more, when the energization frequency is increased from 0 Hz to 60 Hz, and that its impedance
can become markedly capacitive at power frequency. This behaviour, predicted by the HIFREQ
module, is a result of inductive coupling between grid conductors and of currents induced in the
earth by overhead conductors. To become observable, this phenomenon requires that a large
grounding system be considered in a low resistivity top soil layer, underlain by a very high
resistivity bottom layer. A comparison of MALZ and HIFREQ shows that the HIFREQ module is
required to obtain realistic results in such extreme cases.

6 References
[1] S. Fortin, Y. Yang, J. Ma and F. P. Dawalibi, "Frequency Behavior of Impedance of Large
Grounding System in Horizontal Multilayer Soils," The International Conference on Electrical
Engineering 2007 (ICEE), Hong Kong, July 8 -12, 2007.

6 ANALYSIS OF AN UNDERGROUND URBAN SUBSTATION

GROUNDING SYSTEM
Yexu Li(1), Haijun Zhou(2), Wei Liang(2), Longyang Yan(2) and Farid P. Dawalibi(1)
(1) Safe Engineering Services & Technologies ltd. (2) Beijing Safe Engineering Services & Technologies
Email: info@sestech.com, Web Site: www.sestech.com Email: info@seschina.cn , Web Site: www.seschina.cn
Abstract
With the rapid growth of China national economy there is a need to minimize the required substation area and
enhance the aesthetic look of substations in populated areas. Consequently, underground substations are used,
mainly in urban cities nowadays. Therefore, designers responsible for substation grounding face additional
challenges unique to computer modelling in urban underground substations in order to meet safety standards. This
paper discusses the necessary considerations peculiar to urban underground substation studies. By comparing the
analysis with a classical suburban grounding study, this paper shows that in order to provide an adequate
grounding system that meets specific criteria for personnel safety inside and outside the substation and remain
below maximum thresholds for maintaining the integrity of equipment during a fault condition, it is important to
include in the computer model complex soil structures consisting of layered soils along with finite soil volumes
such as air pockets, transmission and distribution cables, as well as modelling accurately neighboring metallic
infrastructures.
1 System under Study

The substation that is the subject of this study occupies an area of approximately 148.2 m by
68.4 m. The substation is a 500 kV high voltage and is entirely underground with a grounding
system located at a depth of 25.6 m. The metallic steel rebars of an indoor large building are
interconnected to the station grounding system. More than 600 steel piles are used to support the
substation. Figure 1 and Figure 2 show computer generated images of the underground
substation. Figure 3 shows the entire network under study. Figure 4 shows a perspective view of
the grounding system under study. Figure 5 shows a three-dimensional view of a portion of the
600 steel piles required to support the underground buildings.
Figure 1: Shanghai 500 kV underground substation (considered in this study).

Figure 2: Left - Chengdu 500 kV underground substation. Right – 110 kV underground substation.
Figure 3: Entire network under study.

Outdoor Grounding：0.8m
GIS Grounding
1st Floor：8.9 m
2nd Floor：16.3m
3rd Floor：21.3m
Main Grounding
Grounding Rods
Grid：25.6m
Figure 4: Perspective view of the system under study.
Steel piles
connecting to the
substation building
Steel piles
Figure 5: Steel piles.

2 Important Considerations Made in the Analysis
2.1 Air Pockets in the Soil Model
It is well known that grounding system performance and safety are closely related to soil
characteristics. The performance of the grounding grid is heavily dependent on the soil structure.
For regular aboveground substations, modelling horizontally layered soil structures may be
sufficient to complete the study of the grounding system. However, for urban underground
grounding systems, the presence of numerous underground buildings and similar facilities such
as substation control rooms, GIS buildings, adjoining shopping malls and underground parking
spaces introduces air pockets that represent electrical discontinuities in the soil. These air pockets
may affect significantly the grounding performance of the system as well as the fault current
distribution in the soil. This situation may raise safety concerns within the underground facilities.
Consequently, modeling accurately these air pockets becomes a critical task in the analysis of such
underground substations.
On-site soil resistivity measurements were performed before the substation was built. The
measured soil resistivity data were analyzed using the RESAP computation module, and an
equivalent multilayer soil model was obtained. In order to account for the air pockets in the soil
of the substation, the hybrid soil model shown in Table 1 was used throughout the study. Figure 6
shows the computer model including the final hybrid soil model.
Table 1: Final hybrid soil model
Resistivity (ohm-m) Depth (m) Dimensions

Surrounding 800 0 ~ 1.3 Length (m) 163.5
Multi-layer Air Pocket
Soil
40 1.3 ~ 5.5 Width (m) 67.65
7 5.5 ~ 30 Top Depth (m) 5.6
60 30 ~ 50 Bottom Depth (m) 25
Figure 6: Grounding system with the final hybrid soil model.

A simple model was used to demonstrate the effect of the building air pocket on the grounding
resistance. Figure 7 is a sketch of the soil structure in the substation and underground commercial
buildings. The grounding resistance is R1 without the building air pocket while the grounding
resistance with the presence of the air pocket is R2. The ratio of the grounding resistance with
(R1) and without (R2) considering the air pocket is defined as A = R2/R1. The ratio of the
underground building air pocket size (Us) width to the grounding system size (Gs) is defined as
B = Us/Gs. The native soil is assumed to be uniform with a soil resistivity of 100 Ω-m.
From Figure 8, as expected, we can see that when the building air pocket size increases, the system
grounding resistance is also increasing. When the underground building width is about 3 times
that of the grounding system size, the grounding resistance value is approximately five times
larger than the resistance without the building air pocket! Furthermore, for a ratio B of about 8
or more, the resistance ratio A starts stabilizing and remains practically constant. This is because
when the size of the air pocket is large enough, the grounding system earth current sees the air
pocket as an extra soil layer of infinite extent and cannot take advantage of the areas beyond the
air pocket. Therefore, increasing the building (or air pocket size) beyond a certain critical size
compared to the grounding system size will have little impact on the grounding system resistance.
Consequently in general, when an urban substation grounding system performance is evaluated
without due consideration to the underground facilities such as buildings and commercial space
areas, the computed grounding system performance will be based on an optimistic evaluation of
reality that may lead possibly to underestimating hazardous conditions.
Figure 7: Sketch of the soil structure in the substation and underground buildings.

Ratio of Grounding Resistances
Ratio of Air Chamber Width to Grid Width

Width
Figure 8: The influence of different air pocket sizes on the system ground resistance.
2.2 Effects of Steel Piles on Grounding System Performance
The grounding system of the underground substation was computed as built. It is emphasized
that the steel piles must be considered in the grounding system analysis although a large portion
of the steel piles is immerged in air pockets covering public areas such as shopping malls. The
following three scenarios were studied to demonstrate the effects of the steel piles on the
grounding system performance. Figure 9 illustrates the fault currents flowing in the grid and in
the soil for all scenarios.
1) Scenario 1: Main grounding system without steel piles;
2) Scenario 2: Steel piles without main grounding system;
3) Scenario 3: Main grounding system with Steel piles.
Table 2 compares the substation grounding resistance with and without the steel piles that are
used for supporting the buildings. As can be seen, the presence of steel piles results in a decrease
of the grounding resistance from 0.083 ohm to 0.0714 ohm, a reduction of about 14%. Table 3
shows the grounding performance during a 220 kV fault with a maximum short-circuit current
level of about 50kA. The following observations can be drawn from this table.
1) The maximum GPR of the grounding system is 288.58 V and 333.26 V with and without
considering the steel piles, respectively. This represents a decrease of about 13%.
Removing the grounding system has a small effect on the ground potential rise (GPR).
Indeed, the GPR does not change significantly (the increase is about 0.5 %).
2) The maximum touch voltage is 97.12 V and 112.19 V with and without considering the steel
piles, respectively. This is a decrease of about 13.4 %. Remove the main ground network,
and the maximum touch voltage remains almost the same (the increase is about 0.03 %).
3) The maximum step voltage is 12.51 V and 17.96 V with and without considering the steel
piles, respectively. This represents a decrease of about 30 %. Removing the grounding
system does not change step voltages significantly (the increase is about by 3 %).

The study shows that the steel piles alone have a slightly better contribution to the grounding
system performance than the grounding system itself. However, when the grounding system and
steel piles are combined, the overall grounding network impedance is slightly better than any one
of its components alone.
Table 2: Grounding resistance with and without the steel piles
Scenario Grounding Impedance (ohm)
1 Main grounding system without steel piles 0.083∠2.940
2 Steel piles without main grounding system 0.0718∠2.840
3 Main grounding system with steel piles 0.0714∠2.850
Figure 9: The diagram of the fault currents flowing in & out of the grid.

Table 3: 220kV Single-phase-to-ground fault - Maximum GPR, Touch and Step Voltages
Maximum Touch Maximum Step

Maximum GPR (V)
Voltage (V) Voltage (V)
Scenario
Main Earth Earth
2nd floor 2nd floor 2nd floor
Grid surface surface
No steel piles, Main 316.46 333.26 105.42 112.19 17.96 17.62
1
grounding grid
Steel piles, no main 270.84 290.01 76.04 97.15 12.89 10.57

2
grounding grid
Steel piles, and main 269.41 288.58 75.19 97.17 12.51 10.30
3
grounding grid
2.3 Underground Cables in Fault Current Distribution
The objective of the fault current split calculations is to obtain the earth current (current
discharged by the grounding system to earth). Under most conditions, the total fault current
doesn’t discharge entirely in the substation grounding system. Part of the fault current, which
does not contribute to the average ground potential rise (GPR) of the grounding system, will
return to remote source terminals and to transformer neutrals through shield wires, neutral wires
or conductors of the grid. It is well known that the power cable sheaths can carry a large amount
of the fault current back to the remote sources due to the strong EMF effects between the cores
and sheaths. The urban underground substations are often connected to other metallic networks
via underground power cables. It is therefore important to determine how much of the fault
current returns to remote sources via cable sheaths and neutral wires of the transmission lines
and distribution lines connected to the substation.
Computer simulations have been performed using both Right-Of-Way (based on a circuit
approach) and MultiFields (based on a field approach) software packages. The networks are
shown in Figures 10 and 3, respectively, representing the scenario of a 220 kV single-line-to-
ground fault at the station. Table 4 provides the results of the fault current distribution calculation.
The earth currents discharged by the grounding systems were determined to be 9.4 % of the total
fault current.

Figure 10: Single-phase-to ground fault distribution network diagram.

Table 4: Fault current split calculation results
Earth-to-
Current Sheath Current Earth Current
Voltage Terminal Fault
Contribution
Level Number Current
（kA） Magnitude (kA) ∠ Degrees Magnitude (kA) ∠Degrees
Split (%)
500kV 1 8.36 8.160 -174.161 8.648 -73.726 10.3%
2 2.58 2.466 -176.131 0.205 -54.288 7.9%
3 1.99 1.837 -177.545 0.174 -26.962 8.7%
4 3.4 3.253 -174.621 0.345 -62.120 10.1%
5 4.77 4.511 -173.587 0.580 -60.314 12.2%

220kV
6 4.96 4.769 -173.413 0.591 -67.869 11.9%
7 3.94 3.636 176.845 0.369 32.885 9.4%
8 3.88 3.509 -178.017 0.392 -18.027 10.1%
9 5.80 5.488 -173.230 0.736 -61.568 12.7%
Remote Currents (kA) 39.68 37.579 4.6 3.74 -53.5 9.4%
Transformer (kA) 10.63 N/A
Total Faults (kA) 50.31 N/A
2.4 Bonding or not the Grounding System to City Buried Metallic

Infrastructures
In general, in an urban area, the low voltage distribution neutrals and the city metallic
infrastructure such as residential metallic water pipe networks are inevitably directly or indirectly
connected to the substation grid. Their omission in the computer model can cause negligible or
significant inaccuracies in the computed grounding performance depending on the topology of
the electrical network. This is because the power line fault currents can dissipate easily in the
metallic infrastructures, a situation that in most cases enhances the safety status inside the
substation. On the other hand, the transferred voltages to the city metallic infrastructure may
endanger the safety of people in the zones outside the substation. For example, when a fault occurs
on the network and metallic pipes are connected to the ground network, high potentials may
propagate along the pipes. If the local soil potential at distant locations are low then touch voltages
may be high. If a nearby pipe is not connected to the grounding network, the local soil potential
is typically high while the pipe GPR is low, resulting also in high touch voltages. In either case, the
designer should verify the safety of nearby residential and industrial pipes.
Realistic computer models including a skeleton of the extensive urban business and residential
metallic pipe network (see Figure 11) were modeled. In order to examine the effect of the pipe
network on the system performance, a study was conducted with and without the metallic pipes
included in the model. The results are presented in Table 5. As can be seen, the GPR, touch and
step voltages can be very different if water pipes are not considered. The maximum GPR with
water pipes is about 25 % less than when the water pipes are ignored. Table 6 summarizes the
results for both bonded and not bonded pipeline scenarios.
Figure 11: Complete computer model for safety evaluation at the substation.

Table 5. Effects of the City Buried Metallic Infrastructures

Maxim GPR (V)
Voltages (V) Voltages (V)
Scenario
The 2nd The 2nd Main The 2nd
Main Grid Main Grid
Floor Floor Grid Floor
No Pipes 467.09 642.64 154.77 203.21 19.67 23.27
With Pipes 371.69 514.84 107.46 149.70 13.43 23.00
Table 6. Bonding effects of the City Buried Metallic Infrastructures

Scenario
Voltages (V) Voltages (V)
Pipes connected to grid 37.74 3.49
Pipes disconnected to grid 28.86 7.01
Figure 12: Touch Voltages of the residential pipes: pipes bonded to the grounding system.

Figure 13: Touch Voltages of the residential pipes: pipes disbonded to the grounding system.
2.5 Extreme Contingency Failure
All drawings indicated that there are no exposed power lines or electric structures outside the
underground substation switchyards. Consequently outdoor faults are unlikely to occur
aboveground. However, in order to provide a more comprehensive assessment of the
underground level of substation safety status an extreme contingency failure analysis outside the
underground substation was modelled and studied.
Contingency for fault
Figure 14: Extreme Contingency Failure.

The computation results show that when an outdoor fault occurs, step voltages are satisfactory
and well below the safety limit based on the native soil anywhere within the entire underground
site whether indoor or outdoor. However, the maximum touch voltage exceeds the safety limit at
some locations based on native soil but will be safe with an 18 cm thick, 1000 ohm-m (when wet)
layer of gravel or equivalent. Concrete with reinforced rebar provides similar safe conditions.
3 Conclusions
The performance of an urban underground substation grounding system has been analyzed using
modern computer techniques. Special analysis considerations peculiar to urban underground
substation grounding systems have been discussed. It is necessary to model accurately the various
underground air pockets in native soil, ground conductors, massive reinforced steel rebar, steel
piles, power cables, residential neutrals, surrounding commercial building metallic
infrastructures and metallic pipes as well as correctly simulating fault currents including
circulating currents in order to determine grounding performance accurately and avoid
inappropriate overestimates or underestimates.
For an urban underground substation, bonding of an urban substation grounding system to the
city buried metallic infrastructure, in most cases, will enhance the safety status inside the
substation. However, it is important to ensure that the transferred voltages to the surrounding
city metallic infrastructures will not endanger the safety of people in the zones of significant
influence. The main grounding system does not have a major influence on the overall grounding
grid performance when massive steel piles are present. However, air pockets in buildings can
significantly alter the safety status. Finally, unlikely extreme contingency failure outside the
substation should be studied to ensure public safety especially in an urban populated area.

7 STUDYING SAFETY IN UNDERGROUND DISTRIBUTION

SYSTEM VAULTS WITH HIFREQ
Marc-André Joyal, Majid Siahrang, and Robert D. Southey
Safe Engineering Services & Technologies ltd.

Abstract
This paper presents a study of underground distribution system vaults with the MultiFields software package, more
particularly the HIFREQ module. We study the case of a worker splicing a cable on a de-energized circuit that runs
parallel to an energized one and shares the same vaults, in which all cable shields are tied to the same grounding
system. A fault is represented on the energized circuit and the resulting potentials transferred to the de-energized
cables via the grounding system and through inductive and capacitive coupling are computed. In order to assess
safety in the working vault, it is needed to compute the touch voltages to which the worker is subjected when the
cable is cut, for different contact scenarios. Since the vaults in these kinds of systems are typically made of
reinforced concrete, which must be modeled, simulations can last a long time, when the interaction between the
rebar cage and its environment needs to be accurately modeled. This paper proposes modeling simplifications in
order to reduce run time to a minimum, while maintaining reasonable accuracy. Limitations of these
simplifications are also discussed.
1 Introduction
In recent years, SES has been asked to study the effectiveness of various methods of temporary
protective grounding in underground distribution system. It is more difficult to protect workers
in such systems than in overhead systems due inaccessibility of the phase conductors at the work
or even near the work site, prior to and indeed during part of the work, for the application of
temporary grounds. Furthermore, the work site is more complex, consisting of cut single-phase
or three-phase cables, for which touch voltages must be computed across the cut and with respect
to the local ground, which can consist of a ground rod or the reinforced concrete skeleton of a
vault, if indeed it is bonded to the work site ground. Whether grounded or not, the interactions of
the vault rebar with the worker and with the rest of the system need to be represented, which can
result in time-consuming simulations, if the reinforcing steel is modeled in detail at all vaults.
This paper discusses the modeling of such system and proposes techniques to dramatically reduce
the computation time when modeling the reinforcing steel of vaults. The focus of the discussion
is the case of a worker splicing an underground cable that runs parallel to a cable experiencing a
fault. Starting from this original case, different scenarios are derived, considering the different
components that can be touched by the worker, the grounding configuration, and the soil model.
The paper is organized as follows. In Section 2, the HIFREQ model of a short distribution
underground system in which the vaults are fully modeled is presented. Simplifying assumptions
are then proposed, leading to a reduced model, and the results of both models are compared to
validate the method. Using these assumptions, realistic cases are studied in Section 3. In order to
generalize the proposed approach, the initially studied scenarios are built such that the phase
conductors of the circuit under repair are left floating. The effectiveness of grounding different

conductors is then studied, in an attempt to achieve an equipotential zone (EPZ) [1, 2]. The effect
of the soil structure is also considered with four different models.
Note that all the example files used to obtain the results presented in this paper are in the folder
UGC 2016\Example Files\Studying Safety in Underground Distribution System Vaults with
HIFREQ.
2 Modeling of the System

In underground distribution systems, vaults are positioned at intervals along the system duct
banks, in order to allow workers to access different portions of the circuits. In urban areas, many
circuits may share common vaults, in which case their shields could all be connected together and
to the vault grounding system. In such an environment, if somebody works on a de-energized
circuit that is close to other energized circuits, considerable potentials can be transferred to the
work site by conduction, via the interconnected cable shields, and due to inductive and capacitive
coupling.
To study such a situation, let us first model a simple system consisting of two circuits, two vaults
and one substation, as shown in Figure 1. This system is described in Section 2.1 and its HIFREQ
model is presented in Section 2.2. A simplified version of the model is proposed in Section 2.3 and
finally, Section 2.4 addresses the effect of including the worker in the model.
Figure 1: Illustration of the simple system modeled, with naming convention of worker contact points (A
to E).
2.1 Description of the System
To demonstrate the modeling method, the simplified system shown in Figure 1 is modeled. Its
characteristics are given below:
 The total length is 2,000’, with two cable vaults spaced 1,000’ apart.
 Each vault is made of steel (12.1 p.u. resistivity, 250 p.u. permeability, and 0.5” diameter)
rebar cages with dimensions of 16’ x 8’ x 8’, embedded in concrete. The top of the vault is
1’ below grade. The grounding system of each vault consists of a 10’ long ground rod (9.5’
in the soil below the vault, 0.5’ inside the vault). As is often the case in reality, the rebar is
not connected to the ground rod or anything else, for that matter. However, because it is

embedded in concrete, the rebar is considered to be in intimate contact, electrically

speaking, with the surrounding soil.
 The substation grounding grid dimensions are 150’ x 120’, each mesh being about 21’ x 20’.
The conductors are 4/0 stranded copper and the grid is 1.5’ deep.
 The circuits studied are part of 25 kV 3-phase system, with a maximum load current of
334 A per circuit;
 The maximum available current when a single-phase-to-ground fault occurs at the
substation is 19.4 kA with an X/R ratio of 15;
 A 100 Ω-m uniform soil model is studied initially.
During splicing of a cable under repair, the three-phases of the circuit to which it belongs are de-
energized at the substation and isolated at switches located elsewhere. The phase conductors of
this circuit can therefore be considered to be floating at both ends.
The scenario studied in this paper is the following: A single-phase-to-ground fault occurs on one
phase of the energized circuit at Vault #2. At that time, a worker in Vault #1 is repairing one phase
of the de-energized circuit. The cables are 500 MCM copper single-conductor with a copper
longitudinal corrugated tape (LCT) shield of 10 mils. The insulation rating is 15 kV. Additional
characteristics are as follows:
 Core: 789 mils diameter, copper conductor;
 Insulation: 175 mils thick, 3 p.u. relative permittivity;
 Shield: 10 mils thick, copper conductor;
 Jacket: 80 mils thick, 3 p.u. relative permittivity.
All told, the modeled system consists of six cables. The shield of each cable is grounded in every
manhole, except for that of the spliced cable, which is disconnected by the hapless worker in the
vault under repair.
2.2 HIFREQ Model of the Actual System
A detailed HIFREQ model has been built in order to model the system described in the previous
section and shown in Figure 1. This model takes advantage of the capability of HIFREQ to model
power cables in detail and to compute the potential accurately on each cable component. Also,
individual observation points are used on the vault floor in order to compute the potentials where
the worker is standing. Note that the actual interior medium of the vault is air, which cannot be
represented by HIFREQ, which models soil both inside and outside the vault. However, since the
rebar cage is metallic, small, and densely meshed, it is more or less equipotential; therefore,
almost no current can flow inside, whatever the medium is. As a result, the interior of the rebar
cage behaves as if it were an air volume, which is confirmed by a simulation using finite volumes
in MALZ. The reader can see the input files for more details on the models.
Figure 1 shows the worker contact points A to E. The floor potential (F) is not shown in the figure
and is obtained from individual observation points on the vault floor. For this study, we are
interested in the potentials of the following six components:
 A and B: phase conductors on both sides of the cut cable;

 C and D: shield conductors on both sides of the cut cable;

 E: grounding system of the vault;
 F: floor (or rebar) of the vault.
The system is energized as follows:
 Phases A, B, and C of the energized circuit are fed with three voltage sources of 14.4 kV
(25/√3);
 A single-line-to-ground fault at the substation was first simulated. To get a fault current
of 19.4 kA with X/R = 15, it was determined that the three source impedances ZS (see
Figure 1) should be ZS = 0.048 + j0.72 Ω (i.e., 0.048 Ω resistor in series with a 0.00191 H
inductor);
 To get a 334 A load current, the three load impedances are set to ZL = 48 Ω (see Figure 1).
The HIFREQ model has been run with a fault on one phase in Vault #2. The results showing the
potentials at locations A to F in Vault #1 are provided in the first column of Table 1. Note that the
simulation time of this reference case was about 11 hours.
Table 1: Potentials A to F Computed with Different Techniques
Potentials (V)
Locations Floor Potential
Whole System Floating Rod
Factor
A 531<128° 532<128° 532<128°
B 1,369<54° 1,370<54° 1,370<54°
C 531<128° 532<128° 532<128°
D 1,369<54° 1,370<54° 1,370<54°
E 800<73° 801<73° 801<73°
F 120<78° 112<73° 120<78°
2.3 Simplified HIFREQ Models
The very long simulation time when rebar cages are modeled is due to the great number of
conductor segments in the model. In this section, we propose two methods that lead to a
simplified model of the system in which all rebar cages are removed and replaced with equivalent
components.
We first present the simplest technique extensively used in similar cases in the past, whereby the
vault grounding system (i.e., the rebar cage and the 10’ rod) are replaced by a single equivalent
rod and the floor potentials are obtained from this equivalent rod GPR, multiplied by a Floor
Potential Factor (FPF). After showing the limitations of this method when the rebar cage is
floating, we propose an alternative method in which the rebar cage is modeled as a floating rod
parallel to the grounding rod of the vault.
2.3.1 Method of the Floor Potential Factor
In this first method, the vault grounding system is replaced by a single equivalent rod. The
procedure is as follows:

1. In a separate input file1, only one vault is modeled (i.e., the rod and the rebar cage, whether
the two are connected together or not) and the following values are obtained:
a. The vault ground impedance ZR (ZR = VR/IR), determined by injecting a current
(IR) into the vault’s ground rod and computing its GPR (VR). Note that even if the
vault rebar cage is not connected to the rod, it will decrease the effective ground
impedance of the ground rod.
b. The Floor Potential Factor (FPF = VF/VR), where VF is the floor potential (which is
about the same as the GPR of the rebar cage).
2. In another file, a single rod is modeled without the rebar cage and, by trial and error, its
required length is determined such that its ground impedance is the same as that of the
actual rod in the presence of the rebar cage (ZR).
Using this method to model the previous example, we obtained an equivalent rod length of 11.5’
and an FPF of 0.14. With the rebar cages replaced by these equivalent rods, the HIFREQ model
of the distribution system ran in less than one minute and lead to the results provided in the
second column of Table 1. As can be seen, the computed potentials are in very good agreement
with those obtained from the detailed model, except for the floor potentials that show a difference
of about 7%. This difference is believed to be due to potentials transferred through earth from the
substation grounding grid to the vault under study, an effect that is not taken into account with
the single equivalent rod, which considers only energization of the vault grounding system from
within. For better accuracy, the method in the following subsection is proposed.
2.3.2 Method of the Floating Rod
The principle of the proposed method is similar to that described above, in the sense that the vault
structural steel is modeled with a simpler structure. The difference is that a separate floating rod
is used to represent the vault’s rebar cage and while the vault’s dedicated ground rod is modeled
separately. The lengths of both rods and the distance between them are determined such that their
self and mutual impedances are the same as those of the actual rod and rebar cage. The detailed
design procedure is outlined next, using the same two-vault circuit example:
1) As described in the previous subsection, one vault (the rebar cage and its rod) is modeled in a
separate input file. The ground rod’s apparent impedance ZR (ZR = VR/IR) and its mutual
impedance with the rebar cage ZFR (ZFR = VF/IR) are determined by injecting a current (IR) into
the ground rod and its GPR (VR) and that of the rebar cage (VF) are computed. In this example,
we find ZR = 28.3 Ω and ZFR = 3.96 Ω.
2) In another input file, the same components are modeled, but the current is now injected into
the rebar cage (the floor), in order to determine its apparent impedance, ZF = VF/IF. As a
verification, the mutual impedance ZRF = VR/IF computed from this simulation must be equal
(or at least very close) to ZFR computed in the previous step, due to the Reciprocity Principle.
For this example, we find ZF = 5.61 Ω and ZRF = 3.95 Ω.
1The input file can be a MALZ or a HIFREQ file. Both give similar results, since only below ground structures are modeled and in
this case, conductive coupling usually dominates over inductive and capacitive coupling. However, the MALZ simulation is much
faster since the vault is made up of a large number of conductor segments.

Figure 2: HIFREQ model of the short system with the rebar cage replaced by a floating rod.
3) In two additional input files, Steps 1 and 2 are simulated, except with the rebar cage replaced
by a vertical conductor that is placed parallel to the vault ground rod. The goal here is to find
the length of both rods and the spacing between them such that we end up with the values of
ZR, ZF and ZRF determined in Steps 1 and 2. The proposed process is as follows:
a) The actual length of the vault ground rod alone can be used as an initial value in this
process for the vault ground rod. The initial length of the floating rod representing the
rebar can be obtained by estimating the length required to obtain a ZF value of 5.61 Ω (from
Step 2), leading to an initial length of 77’.
b) In a first input file, both rods are included and placed close to each other (e.g., spacing of
0.1’) and the vault ground rod is energized with a known current and the following
quantities are computed: ZR = VR/IR and ZFR = VF/IR.
c) In a second input file, the same arrangement as that of Step b) is used, but the energization
is applied to the floating rod, leading to ZF and ZRF.
The lengths of the rods and the separation distance are then modified until the values of ZR,
ZF and ZRF obtained are the same as those computed in Steps 1 and 2 with the actual rod and
rebar cage. This trial and error process can be somewhat time-consuming, depending on the
experience of the designer and the complexity of the soil structure, but can ultimately save a
lot of run time for an extensive system. Furthermore, this process need only be undertaken
once. A good way to proceed is to modify the files created in steps 3-b) and 3-c) together in
SESCAD and apply the modifications in both files at the same time to see the impact of the
modifications on the four impedance values (ZR, ZF, ZRF, and ZFR). The following information
may help with the design process:

 Increasing (decreasing) the rod length decreases (increases) its self-impedance and
increases (decreases) the mutual impedance between both rods;
 Increasing (decreasing) the spacing between the rods decreases (increases) the mutual
impedance between them and increases (decreases) their self-impedance.
At the end of this process, the ground and floating rod lengths are 7.25’ and 75’ and they are
spaced 0.2574’ apart. Note that the final lengths are on the same order as their initially
estimated values.
4) Finally, once the simplification of the system vault is achieved, every vault in the system is
replaced by the equivalent model, as shown in Figure 2.
The results obtained with this Floating Rod model are shown in the third column of Table 1. As
can be seen, all computed potentials are very close to those of the complete model. Furthermore,
this model took less than one minute to run.
2.4 Addition of the Worker in the System
In the previous example, the potentials were computed with the cable cut, but the worker was not
in contact with any conductor. However, when the worker closes the circuit between two
conductors or between the floor and a conductor, the potentials on both sides of the worker may
change, since the circuit is no longer the same. If the system impedances are low compared with
the worker’s body resistance, this change may be quite small. On the other hand, when system
impedances are large, as can be the case when evaluating contact with “floating” phase conductors
that are capacitively coupled to their grounded shields, the change can be quite large. In any case,
better accuracy is obtained when the worker’s body impedance is included in the circuit. This
means representing the body by a lumped impedance between the two conductors that are
touched. This can lead to multiple HIFREQ models, one for each pair of contact points that need
to be studied. This provides a second reason to use the floating rod, as it is needed to compute
touch potentials when the worker contact points consist of a conductor and the vault floor.
Table 2: Potentials A to F Computed when the Worker is Modeled between Points B and F
Potentials (V)
Locations
Whole System Floating Rod
A 531<128° 532<128°
B 164<98° 162<98°
C 531<128° 532<128°
D 1,369<54° 1,370<54°
E 800<73° 801<73°
F 121<77° 120<77°
As a simple example, we study the case where the worker is touching Points B and F. This scenario
is likely to occur if the phase conductor is touched by a worker who is also touching the vault floor

or wall. In this example, the worker is modeled with a 1,000 Ω resistor, in accordance with [3]2.
This scenario is first studied with the detailed model; then the results are compared with those
obtained with the Floating Rod method. The potentials obtained are listed in Table 2, which shows
that the differences between both methods are negligible. Also, a comparison between VB and VF
obtained in Table 2 and in Table 1 clearly shows the non-negligible effect of adding the worker in
the system, who dramatically lowers the potential of Contact Point B. In Section 3.4, more will be
said about the necessity of considering the worker, depending upon the system configuration.
3 Case Study
In this section, realistic underground distribution systems are studied in the two following contact
scenarios:
 A-B: The worker touches both phase conductors, A and B;
 B-F: The worker touches one phase (B) and is in contact with the floor (F).
An original system is first assumed with the same characteristics as those of the previous section,
except for the following parameters:
 The overall length is 20,000’;
 There are 20 vaults (vault spacing is 1,000’).
Then, different soil models and grounding configurations are simulated, such as the position of
the worker and of the faulted vault.
These cases are modeled with the Floating Rod model proposed in the previous section and the
results are compared to those obtained with a Hybrid model. In this latter system, the complete
rebar cage is modeled for the vault under study, but the other vaults are modeled using the
Floating Rod model. This Hybrid model is used as the reference, since it is impossible to model
accurately every vault of the system, due to the large amount of conductor segments that would
be needed.
In the following subsections, we test the effect on the Floating Rod method of varying the vault
spacing, changing the worker and fault locations, using different soil models, and grounding
different conductors. These results are intended to be used to assess the validity of the method
and find its limitations.
3.1 Variation of the Vault Spacing
In order to verify the validity of the simplified model, the first test we do is to vary the vault spacing
and compute the touch voltages for the two contact scenarios mentioned above, i.e., A-B and B-F.
They are first simulated for vault spacings of 1,000’ and 5,000’. Next, in order to assess the model
for shorter vault spacings, we model two systems with spacings of 100’ and 500’, respectively, but
we reduce the overall system length from 20,000’ to 2,000’, in order to reduce the computation
2Note that other references may suggest other values, as it is the case in [4], where a value of 500 Ω is suggested for the body
impedance.

time. This modification clearly has an impact on the potentials in the vault under study, because
there is a lesser total number of ground rods in the system. However, since the Floating Rod and
Hybrid models we are comparing always have the same lengths, it is not thought that the
conclusions of this study will be affected by this modification.
Results of the eight models are presented in Table 3. They show that although the Hybrid and
Floating Rod models lead to similar results, discrepancies between the two methods are
noticeable when the vault spacing is small. In this latter case, a maximum error of about 5.5 % in
the potential magnitude is seen. This seems to be due to conductive coupling through the soil
between the surrounding grounding systems and the vault under study. Recall that when
modeling the vault with the Floating Rod model, coupling between the vault ground rod and the
rebar cage equivalent floating rod is considered, but not the coupling with its surrounding
environment. Furthermore, in this case, the vault under study is very close to the substation, with
which the through-earth resistive coupling must be much stronger than with other vaults, due to
the large substation area. As a result, when the substation is submitted to a ground potential rise,
the potentials transferred to the rebar are not negligible and could be different, depending on
whether the floating rebars are explicitly modeled or with the floating rod model, even if both have
the same grounding impedances. This is because their shapes are different. This discrepancy
could possibly be rectified by shifting the positions of the two rods with respect to the substation
grounding grid.
To investigate coupling between the station grid and the vault, we first computed the mutual
impedance between the rebar cage and the substation grid with spacings of 100’ and 1,000’.
Impedances of 0.315 Ω and 0.0486 Ω were obtained, respectively, for the two cases. When the
detailed vault model was replaced by the equivalent Floating Rod model, values of 0.285 Ω and
0.0482 Ω were obtained for spacings of 100’ and 1,000’, respectively. These results clearly show
that when the vault is close to the substation, the potentials transferred from the substation are
highly sensitive to the distance of the modeled vault conductors from the substation.
Table 3: Potentials of Structures Touched by Worker for Different Vault Spacings
System Models / Vault Spacing

Contact Points 100’ 500’ 1,000’ 5,000’
A (V) 254<87° 622<67° 872<46° 2,326<47°
Hybrid
B (V) 254<87° 631<68° 893<49° 2,585<63°
Floating A (V) 254<88° 623<67° 872<46° 2,326<47°
Rod B (V) 255<88° 632<68° 893<49° 2,585<63°
B (V) 104<120° 132<109° 695<90° 2,157<103°
Hybrid
F (V) 90.3<113° 90.5<91° 66.1<81° 234<69°
Floating B (V) 98.7<120° 131<110° 695<90° 2,156<103°
Rod F (V) 85.3<112° 89.6<90° 65.6<81° 231<69°
3.2 Effect of the Worker Location
As in the previous subsection, work site potentials are computed for Contact Points A-B and B-F.
However, here, the vault spacing is kept constant, whereas the work site and faulted location are
varied, as follows:
 The worker is in Vault #1 and the fault is in Vault #2;

 The worker is in Vault #9 and the fault is in Vault #10;

 The worker is in Vault #19 and the fault is in Vault #20.
Results for all these cases are shown in Table 4. The first observation is that although the errors
between the simplified and hybrid models are quite low, these results clearly show that the highest
discrepancies (about 3%) between both models are seen when the vault under study is far from
the substation. This is seen more clearly for the B-F contact scenario, i.e., when the worker is in
contact with the floor.
Table 4: Potentials of the Structures Touched by the Worker in Different Vaults
Worker Location
System Model /
Vault 1 Vault 9 Vault 19
Contact Points
(near substation) (middle of system) (end of system)
A 872<46° 1,166<56° 984<108°
Hybrid
B 893<49° 1,436<69° 1,080<110°
Floating A 872<46° 1,167<56° 987<108°
Rod B 893<49° 1,437<69° 1,083<110°
B 695<90° 1,023<102° 356<82°
Hybrid
F 66.1<81° 253<57° 328<63°
Floating B 695<90° 1,022<102° 347<82°
Rod F 65.6<81° 247<57° 318<63°
3.3 Effect of the Soil Model
Another interesting parameter to study is the soil structure. In this subsection, the effect of the
following soil models are tested:
 100 Ω-m uniform soil;
 5 Ω-m uniform soil;
 Two-layer soil model: 100 Ω-m over 5 Ω-m;
 Two-layer soil model: 5 Ω-m over 25 Ω-m.
Due to the important effect of the soil on the grounding performance of a system, it is necessary
to properly adjust the grounding rod and equivalent floating rod lengths, as well as the spacing
between them. The final configuration of the Floating Rod systems, as a function of soil model,
are provided in Table 5. This table shows that the soil resistivity does not have an important
impact in the case of uniform soils, since the parameters are the same for the 5 Ω-m and 100 Ω-
m models. On the other hand, layered soils greatly affect these parameters. For instance, it is seen
that high over low resistivity soil models lead to longer ground rods and shorter floating rods
whereas low over high resistivity soils lead to the opposite.
This latter scenario is actually a limiting factor that must be taken into account. In such cases, it
may become difficult to model the vault with a simple floating rod since it would need to be very
long. As seen for the case of the 5 Ω-m over a 25 Ω-m soil model, this floating rod has a length of
235’. For larger ratios between the top and bottom layers, this value could become much larger.
On the other hand, although this case shows an apparent limit of this floating rod method, it would
be possible to remedy the problem by using horizontal conductors instead of vertical rods.

Table 5: Parameters of the Simplified Vaults in the Different Soil Models
Soil Models
System Parameters 100 Ω-m 5 Ω-m 100 Ω-m 5 Ω-m over
uniform uniform over 5 Ω-m 25 Ω-m
Ground Rod Length
7.25 7.25 8.4 5
(feet)
Floating Rod Length
75 75 14.5 235
(feet)
Rod Spacing
0.2574 0.2574 0.4074 0.1174
(feet)
Table 6: Potentials of the Structures Touched by the Worker for Different Soil Models
Soil Models
System Model 100 Ω-m 5 Ω-m 100 Ω-m 5 Ω-m over
uniform uniform over 5 Ω-m 25 Ω-m
A 872<46° 731<26° 972<42° 679<29°
Hybrid
B 893<49° 732<30° 987<45° 684<33°
A 872<46° 730<26° 972<42° 679<29°
Floating Rod
B 893<49° 732<30° 987<45° 684<33°
B 695<90° 555<72° 745<87° 532<73°
Hybrid
F 66.1<81° 45.9<35.3° 44.1<60.5° 61.0<49.4°
B 695<90° 554<72° 745<87° 534<73°
Floating Rod
F 65.6<81° 44.0<32.5° 44.6<60.4° 60.5<50.8°
Results of the potentials obtained with the Floating Rod and Hybrid methods for the four soil
models are provided in Table 6. They show good agreement between both methods. The largest
discrepancy is about 4.3% and is seen in the case of the 5 Ω-m uniform soil for the floor potential,
which is already very low compared with the other contact point, to begin with.
3.4 Effect of the Grounding Configuration
In the cases studied so far, rebar cages have always been disconnected from the vault grounding
system, which is the more general case. However, these rebar cages are often connected to the
vault ground, for enhanced worker safety. The objective is to build an equipotential zone (EPZ)
[1, 2] so that each conductor that the worker may touch is at about the same potential, resulting
in a reduced amount of current flowing through the worker’s body during a two-point contact.
There are different ways to build equipotential zones and these methods are out of the scope if
this article. However, the method studied herein implies both of the following steps:
 The rebar cage and the shields of the cut cable are connected to the grounding system (it
is understood that the shields of the other cables are also bonded to the grounding system)
in order to make sure that all conductors that are not intended to carry the current under
normal load situations are at the same potential;

 Bracket grounds are added at accessible locations on both sides of the worker’s vault, as
close as possible to this latter. At each bracket, the cores of each cable of the de-energized
circuit are bonded together and to the local grounding system, such that they are at the
same potential as that of the neutral wires (or shields) and the other grounded metallic
structures at the bracket location. As seen in Figure 3, a bracket is modeled as a simple
connection between the three phases of the de-energized system to ground.
In principle, if applied correctly, the above mentioned steps used together should provide a
working area that is more or less equipotential, the only remaining issue being contact with the
core of the cable under repair. In the following subsections, these steps are modeled separately in
order to see their individual effects on the model. Based on these results, simplifying assumptions
in the model will be derived.
Figure 3: The worker is splicing the cable in Vault #1 and brackets are on both sides of this vault, i.e., at
the Substation and at Vault #2.
3.4.1 Grounding of Rebar Cages and Cable Shields
In some distribution systems, rebar cages may be connected to the grounding system in every
vault, which significantly reduces their grounding impedance. In addition, if the shields of the cut
cable on both sides of the worker are connected to the grounding system, conductors that are not
intended to carry load currents should be at the same potential. This considerably reduces worker
exposure to potentially hazardous touch voltages.
Starting from the original case of this study, connection of the ground rod to the rebar cage in
each vault is first assumed. Then, in the vault where the cable is spliced, its shields on both sides
of the splice are connected to the ground rod. No bracket grounding is present.
Table 7: Potentials Computed when the Rebar Cage and the Shields are Grounded
Potential (V)
Location Hybrid Equivalent Rod Hybrid Equivalent Rod
(A-B) (A-B) (B-F) (B-F)
A 338<30° 337<30° 327<172° 327<172°
B 350<35° 349<35° 373<78° 372<78°
C 147<124° 147<124° 147<124° 147<124°
D 148<124° 148<124° 148<124° 148<124°
E-F 148<124° 148<124° 148<124° 148<124°
When the vault’s rebar cage is bonded to the vault’s ground rod, only one equivalent rod is needed.
The elimination of the floating rod greatly simplifies the modeling process, as there are fewer
parameters that need to be matched in coming up with an equivalent. Each vault has a ground
impedance of 5.508 Ω, which can be modeled as an equivalent 77.7 foot rod. Work site potentials
computed using the Hybrid and the Equivalent Rod models are listed in Table 7, for the A-B and
B-F contact scenarios. The results from the Hybrid and Equivalent Rod methods are in very good
agreement. Furthermore, it can be seen that structures not intended to carry current are at very
similar potentials, i.e., VC ≈ VD ≈ VE ≈ VF.
3.4.2 Bracket Grounding
Let us now examine the modeling of bracket grounding, as illustrated in Figure 3. Other than the
brackets, the original model is used, whereby the ground rod in each vault is not connected to the
vault rebar. The B-F contact scenario is modeled, the results of which are shown in the first two
columns of Table 8. In the two other columns of this table, the same case is studied, but the
potentials VA to VF are computed by removing the resistor modeling the worker, i.e., leaving an
air gap between the worker contact points.
Table 8: Potentials Computed when the Vault under Repair is between Two Bracket Grounds
Potential (V)
Location Hybrid Floating Rod Hybrid Floating Rod
(with worker) (with worker) (without worker) (without worker)
A 188<118° 189<118° 189<118° 189<118°
B 558<77° 558<77° 558<77° 558<77°
C 188<118° 188<118° 188<118° 188<118°
D 559<77° 559<77° 559<77° 559<77°
E 356<87° 356<87° 356<87° 356<87°
F 51.2<90.9° 50.8<90.8° 48.8<91.7° 48.3<91.6°
Interestingly, these results show that when bracket grounding is used, the impedance of the
worker does not significantly affect the potentials to which the worker is submitted when touching
the metallic conductors. This is an important conclusion that greatly simplifies the simulation of
these scenarios for the two following reasons:
 When using a lumped impedance to model the worker, if we need to compute potentials
for all possible scenarios worker contact scenarios, i.e., A-B, A-C, A-D, etc., a considerable
number of HIFREQ models must be created. However, when bracket grounding is used,
only one run is necessary to determine the six potentials applicable for all contact
scenarios.
 It is not always clear what the worker body impedance should be. As seen previously, a
value of 1,000 Ω is specified in [3] whereas a resistance of 500 Ω is suggested in [4]. When
modeling a case with bracket grounds, there is no need to bother with the worker body
impedance (as long as this value is much higher than the cable core resistance), until it is
time to interpret the results.
The reason why the worker body impedance can be ignored when bracket grounding is used
follows from the simplified circuit of the system shown in Figure 4. In this single-phase circuit,
without loss of generality, the case of a worker touching the core on both sides of the cut cable
(Contact Scenario A-B) is assumed. In this circuit, the core and shield impedances are shown as
lumped impedances. When no bracket grounding is used, as illustrated in Figure 4-a), the core of

the spliced cable is left open at both ends. However, since cables are made of at least two
conductors (core and shield) separated by a dielectric, they act as capacitors when they are left
floating (shown as C1 and C2 at the ends of the cable in Figure 4, although they are in fact
distributed throughout the length of the cable). Because the 60 Hz reactance of these capacitances
are quite large, ranging from roughly j 1,000 Ω/mile to j 10,000 Ω/mile [5], they can dramatically
limit the current flow through the body impedance and therefore the resulting body voltage,
depending upon the length of the cable and the body impedance value chosen. On the other hand,
when bracket grounding is employed, as illustrated in Figure 4-b, the cable capacitances are
shunted, leaving the cable core impedance, which is on the order of 1 Ω/mile, in series with the
1,000 Ω or 500 Ω body impedance. Clearly, the current flow through the body is now determined
only by the body impedance (and the voltages generated by the fault to begin with), which is so
high in relation to the remaining impedances in the system that it can be replaced by an air gap,
with little change in the resulting voltage across the body.
(a)
(b)
Figure 4: (a) Simplified model of a single-phase system with a cut floating core shunted by a body and (b)
modified circuit with brackets.
3.4.3 Brackets with Equipotential Zone
Finally, potentials VA to VF are computed for the Hybrid and Equivalent Rod cases with both
bracket grounding and bonding of all cable shields to the vault rebar and ground rod. Results are
summarized in Table 9, with the worker’s body removed from the system. As can be seen, the
agreement between both methods is very good. It is notable that despite the bracket grounding,
there is a significant voltage remaining between the cut cable core and the vault ground, as a result
of induced voltages in the de-energized cable between the bracket grounding points and the work
site and as a result of current flow and associated voltage drop in the cable shields between vaults.
As is to be expected, if we move both brackets very close to the worker, i.e., within the worker’s
vault, we find that all the conductors are all at the same potential, namely 148<124° V.

Table 9: Potentials A to F Computed for “Equipotential Vault”, with Bracket Grounding, when Worker is
Removed from System
Potentials (V)
Locations
Hybrid Equivalent Rod
A 244<-156° 244<-156°
B 204<78° 204<78°
C 103<151° 103<151°
D 103<151° 103<151°
E-F 103<151° 103<151°
4 Conclusion
Simplifying assumptions have been proposed in order to model underground distribution system
vaults with HIFREQ. Methods that makes it possible to substitute vault rebar cages with simple
equivalents have been proposed, which significantly reduce computation time. It is seen that when
the rebar cage is not grounded, more steps are needed to carry out the analysis, since an equivalent
system of two rods needs to be designed, having the same self-impedance and mutual impedance
properties as the vault ground rod and vault. It is also shown that different grounding and
temporary protection techniques currently used can significantly simplify the model. Among
others, grounding the rebar cage makes it possible to replace the two-rod system by a single
equivalent rod.
Different configurations have been tested to demonstrate the accuracy of the proposed modeling
methods. In light of the tested scenarios, it is seen that the greatest discrepancies between the
simplified and more detailed models occur when the leakage current flowing through the earth is
high and when the work site is very close to the substation. Note, however, that for the scenarios
studied, the greatest error was on the order of 5.5%, which is considered acceptable for most
applications.
Layered soils have also been examined: a low over high resistivity soil and a high over low
resistivity soil. The accuracy of the results was good in both cases, but it was seen that for the low
over high resistivity soil, it could be necessary to model horizontal conductors instead of vertical
rods.
Finally, the effectiveness of bonding the vault rebar to the cable shields and vault ground has been
shown, as has the effectiveness of bracket grounding. It is seen that bracket grounding may
nevertheless result in sizeable work site voltages.
5 References
[1] J. T. Bonner, B. Erga, W. W. Gibbs and V. M. Gregorius, "Test results of personal protective
grounding on distribution line wood pole construction," in IEEE Transactions on Power Delivery,
vol. 4, no. 1, pp. 810-817, Jan 1989.
[2] "Worker protection while working de-energized underground distribution systems," in IEEE
Transactions on Power Delivery, vol. 19, no. 1, pp. 298-302, Jan. 2004.

[3] "IEEE Approved Draft Guide for Safety in AC Substation Grounding," in IEEE P80/D10, April 2013,
vol., no., pp.1-236, May 15 2015.
[4] Occupational Safety & Health Administration [OSHA]. (n.d.). "Appendix C to § 1910.269 --
Protection From Hazardous Differences in Electric Potential, " (Standards-29 CFR). Retrieved from
https://www.osha.gov/pls/oshaweb/owadisp.show_document?p_table=STANDARDS&p_id=9871.
[5] H. N. Muller, Jr. and J. S. Williams, "Electrical Characteristics of Cables," in Electrical Transmission
and Distribution Reference Book, 4th ed. East Pittsburgh, PA, USA: Westinghouse, 1950, pp. 64-95.

8 CALCULATION OF MAGNETIC FIELD FROM SOLENOIDS

AND REACTORS OF DIFFERENT SHAPES
Peter Zhao, Simon Fortin and Farid P. Dawalibi

Email: info@sestech.com Web Site: www.sestech.com
Abstract
The magnetic field and flux density generated by air-core reactors with cross sections of different shapes are
studied. The calculated total magnetic field and flux density on a horizontal surface perpendicular to the axis of the
reactor at its mid-point shows interesting patterns, in the form of concentric rings bounded by the segments
forming the reactors. This kind of pattern is observed not only inside but also outside the reactors, for reactors with
symmetrical or asymmetrical cross sections. The variation of the levels of total magnetic flux density as a function
of the number of segments per turn used to model the reactor is also provided, both on the horizontal surface and
along a vertical surface passing through the axis of the reactor.
1 Introduction
Air-core reactors are often used in substations for current limitation purposes, voltage regulation
and reactive power compensation applications in power systems. By their very nature, air-core
reactors can generate powerful magnetic fields, which can be a source of concern in substations
due to possible induced currents in the substation fence or grounding grid. As a result, air-core
reactors regularly need to be included in computer models of substations built using the HIFREQ
module.
A common way to create a simple air-core reactor is by means of a solenoid (or coil) of wire wound
vertically around the core. For this simple geometry, the magnetic field anywhere inside a solenoid
carrying current 𝐼 is given approximately by 𝐻 = 𝑛𝐼, where 𝑛 is the number of turns of wire per
unit length in the solenoid. This formula holds exactly only when the solenoid is infinitely long.
In this same limit, the magnetic field outside the solenoid is zero. For realistic solenoids of finite
length, the magnetic field is not uniform inside the solenoid and is not zero outside of the solenoid;
a suitable HIFREQ model must be built to calculate the actual magnetic field.
In this article, the magnetic field and flux density inside and outside reactors of different shapes
are calculated using HIFREQ. The field is calculated along two planes of different orientations:
one is perpendicular to the reactor axis in the middle of the reactors, and the other is parallel and
passing through the reactor axis. In the simulations, the axis of the reactors is modeled in the
vertical direction.

2 Reactor Models Created in HIFREQ
Figure 1: Screen to define a reactor.
In SESCAD, a reactor can be easily modelled using the built-in element “Spiral”. By clicking
Insert | Arc/Circle/Spiral/Helix in the menus of SESCAD, the screen shown in Figure 1 will
pop-up and a reactor can be created after filling the data shown in that figure.
In the designed models, the characteristics of the reactors are as follows:
 Z Coordinate of Top of Spiral: -12 m.
 Z Coordinate of Bottom of Spiral: -1 m.
 Inner Radius: 1.5 m.
 Outer Radius: 1.5 m.
 Number of turns: 550.
 Segments per turn: ranges from 3 to 16 for reactors of different shapes.
(a) (b) (c)
Figure 2: Edit Transformations screen.

Table 1: Cross Sections of the Defined Reactors
Equilateral Triangle Square Pentagon Hexagon
Heptagon Octagon Nonagon Decagon
Hendecagon Dodecagon Tridecagon Tetradecagon
Pentadecagon Hexadecagon Parallelogram Rhombus
The radius of the conductors in the reactors is 0.006 m. Two short energization conductors with
the same radius are connected to the two ends of the reactors. A source of 715 A is impressed on
the short conductor at the lower side of reactor and another source of 250 kV is assigned to the
short conductor at the upper side. The computation frequency is set to 60 Hz. The X, Y and Z
components of the translation vector is (0, 0, 0) as shown in the Edit Transformation screen
(Figure 2). The data in Figure 2 (a) is used to generate the symmetrical reactors. The data in Figure
2 (b) and Figure 2 (c) are used to generate the unsymmetrical parallelogram and rhombus shaped

reactors (4 segments per turn), respectively. The cross sections of the generated reactors are
shown in Table 1.
(a) (b)
Figure 3: Horizontal and vertical observation surfaces for a reactor with hexadecagon cross section.
The horizontal observation surfaces are defined inside and outside each reactor, separately, with
the spacing between consecutive observation points being set to 0.01 m at the reactor center (Z =
-6.5 m). Inside the reactors, the observation points fill the cross section of the reactors, coming to
within 0.01 m of the wall of the reactors. Outside the reactors, the observation surface also starts
at 0.01 m from the reactor wall and extends 3 m away from the wall. The vertical observation
surface covers the following region: X from -2.05 m to 2.05 m; Y from 0 m to 0 m; Z from -0.05 m
to -13 m, with the spacing between consecutive observation points set to 0.01 m, which is used for
all reactors with the exception of a few points which were too close to the conductors used to
model the reactor, and which were removed. The vertical observation surface is passing through
the reactor axis. As an example, the defined horizontal observation surface and vertical
observation surface for a reactor with hexadecagon (16 segments per turn) cross section are shown
in Figure 3 (a) and (b), respectively.
3 Calculated Magnetic Flux Density
3.1 Total Magnetic Flux Density along the Horizontal Profiles inside the
Reactors
3.1.1 Reactors with Symmetrical Cross Sections
The total magnetic flux density along the horizontal observation surface inside the reactors with
symmetrical cross sections is plotted in Figure 3, Figure 4, and Table 2 for reactors with 3 to 16
segments per turn, respectively. The magnitude of the magnetic flux density is quite uniform
inside the reactors, except in the area close to the corners. When displaying a large number of
levels, however, a very interesting phenomenon can be observed in that the total magnetic flux
density forms multiple concentric rings bounded by the conductors forming the reactor. This
phenomenon is apparent in all studied symmetrical reactors, independently of the number of
segments per turn. The maximum magnetic flux density inside the reactor also decreases as the
number of segments per turn used to build the reactors increases. This is mainly due to the

increasing distance between the axis of the reactor and the conductors forming the reactor for the
cases with larger number of segments per turn.
Figure 4: Total magnetic flux density along a horizontal observation surface inside a reactor in the shape
of an equilateral triangle.
[ID:Square Reactor @ f=60.0000 Hz ]
1.5 SPOT LEVELS x 1.E+3

Maximum User Limit: 46.198
46.20 46.15
46.20 46.14
46.19 46.14
46.19 46.14
46.19 46.14
0.5 46.19 46.14

46.19 46.13
46.18 46.13
46.18 46.13
46.18 46.13
46.18 46.13
Y AXIS (METERS)
46.18 46.12
46.17 46.12
46.17 46.12
46.17 46.12
46.17 46.12
-0.5
46.17 46.11
46.16 46.11
46.16 46.11
46.16 46.11
46.16 46.11
46.16 46.10
46.15 46.10
46.15 46.10
46.15
46.15
-1.5
-1.5 -0.5 0.5 1.5
X AXIS (METERS)
Total Magnetic Induc. (MicroTesla)
Figure 5: Total magnetic flux density along a horizontal observation surface inside a reactor in the shape
of a square.
Table 2: Total Magnetic Flux Density along a Horizontal Observation Surface inside the Reactors
Levels of total magnetic flux density (mT) used for Levels of total magnetic flux density (mT) used for
the reactors with 5 or 6 segments per turn. the reactors with 7 to 16 segments per turn.
Pentagon Hexagon
Heptagon Octagon

Nonagon Decagon
Hendecagon Dodecagon
Tridecagon Tetradecagon

Pentadecagon Hexadecagon
3.1.2 Reactors with Unsymmetrical Cross Sections
In order to further confirm this phenomenon, reactors shaped in the form of a general
parallelogram and of a rhombus (as shown in Table 1) were studied. The total magnetic flux
density calculated along the horizontal observation profiles inside the reactors is shown in Figure
6 and Figure 7, respectively.
Figure 6: Total magnetic flux density along horizontal profiles inside a reactor in the shape of a
parallelogram.

Figure 7: Total magnetic flux density along horizontal profiles inside a reactor in the shape of a rhombus.
These figures show that the magnitude of the magnetic flux density along the horizontal
observation surface inside the reactors is also in the form of multiple concentric rings bounded by
the conductors forming the reactor, even though the cross section of the reactors is highly
unsymmetrical.
3.2 Total Magnetic Flux Density along Horizontal Profiles outside the
Reactors
The total magnetic flux density along the horizontal observation surface outside a rhombus
shaped reactor is shown in Figure 8. For other reactors, the plots of the total magnetic flux density
along the horizontal observation surface outside the reactors are listed in Table 3.

Figure 8: Total magnetic flux density along horizontal profiles outside a reactor in the shape of a
rhombus.
Table 3: Total Magnetic Flux Density along Horizontal Profiles outside the Reactors
Levels of total magnetic flux density (mT) used for

Parallelogram
the plots in this table.

Equilateral Triangle Square Pentagon
Hexagon Heptagon Octagon
Nonagon Decagon Hendecagon

Dodecagon Tridecagon Tetradecagon
Pentadecagon Hexadecagon
These figures show that the magnitude of the magnetic flux density along the horizontal
observation surface outside the reactors is much smaller than that inside the reactors, as shown
in Section 3.1. The total magnetic flux density also forms multiple concentric rings bounded by
the conductors forming the reactor, as was the case inside the reactors, and this no matter whether
the cross section of the reactors is symmetrical or not. For the symmetrical reactors, the level of
the total magnetic flux density increases as the number of segments per turn increases (keeping
the other reactor parameters constant) and approaches a stable value, unlike inside the reactors
where the field decreases with increasing number of segments per turn. This is also due to the
conductors moving outward when increasing the number of segments per turn.
3.3 Total Magnetic Flux Density along a Vertical Observation Surface

through Reactors
The total magnetic flux density along the vertical observation surface is shown in Figure 9 for a
reactor shaped in the form of a parallelogram and a rhombus, respectively. For the symmetrical
reactors, the results are shown in Table 4. The levels used to display the total magnetic flux density
are the same for all plots in this section; they are shown in Figure 9. The variation of the total
magnetic flux density along the vertical surface is much larger than for the horizontal observation
surfaces in Sections 3.1 and 3.2.

Parallelogram Rhombus
Figure 9: Total magnetic flux density along vertical profiles through reactors in the shape of a
parallelogram and a rhombus.
Table 4: Total Magnetic Flux density along Vertical Profiles through the Reactors
From left to right, the number of segments per turn increases from 3 to 9

From left to right, the number of segments per turn increases from 10 to 16
These plots show that the total magnetic flux density inside the reactors is much larger than that
outside the reactors except in the area close to the open ends.
3.4 Magnetic Field Plotted using SESSystemViewer
In the previous sections, the total magnetic flux density was plotted as color intensity (Spot 2D)
plots over two different surfaces. In this section, the total magnetic field is shown as 3D plots using
SESSystemViewer. Plots of the total magnetic field on a horizontal observation surface located
entirely inside the reactors, outside the reactors, or cutting across the reactors, are listed in Table
5 and Table 6 for the square-shaped reactor and hexagon-shaped reactor, respectively. On the left
side of the tables, the reactor and the magnetic field on the horizontal observation surface are
shown in the plots in a single view. On the right side, the conductors forming the reactor are
hidden in order to clearly show the results.
Table 5: Total Magnetic Field on a Horizontal Observation Surface for a Reactor with Square Cross
Section
Slanted view from top: conductors are shown; Slanted view from top: conductors are hidden;
observation surface cuts across the reactor. observation surface cuts across the reactor.

observation surface is inside the reactor only. observation surface is inside the reactor only.
Slanted view from bottom: conductors are shown; Slanted view from bottom: conductors are hidden;
observation surface is outside the reactor only. observation surface is outside the reactor only.

Table 6: Total Magnetic Field on a Horizontal Observation Surface for a Reactor with Hexagonal Cross
Section

observation surface is inside the reactor only. observation surface is inside the reactor only.

observatio surface is inside the reactor only. observation surface is inside the reactor only.

The results over the observation surface crossing the reactors clearly demonstrate that the total
magnetic field is greatly enhanced inside the reactor comparing to that outside the reactor.
Although the plots are all in 3D Perspective view, the total magnetic field outside of the reactor
appears to be drawn on a lower flat surface and that inside the reactor on a higher flat surface as
a result of this effect. However, when the total magnetic field is drawn only outside of the reactor,
it clearly shows the concentric decay pattern when going away from the wall of reactor.
4 Conclusion
The total magnetic flux density due to air-core reactors with cross sections of different shapes was
presented along a horizontal observation surface perpendicular to the reactor axis (and in the
middle of the reactor) and along a vertical observation surface passing through the reactor axis.
The calculations were performed with the HIFREQ module of MultiFields.
The magnetic flux density inside the reactors is quite uniform (except near the two open ends)
and much higher than that outside the reactors. The deviations from uniformity exhibit an
interesting pattern. Along a horizontal observation surface cutting through the middle of the
reactor, the magnetic flux density is in the form of concentric rings bounded by the segments
forming the reactors, not only inside but also outside the reactors, no matter whether the reactors
have symmetrical cross sections or not. Inside the reactors, the level of total magnetic flux density
increases when going from the axis to wall of the reactor due to the increased contributions from
the conductors forming the wall. Outside the reactors, the level of total magnetic flux density
decreases when going away from the reactor, for the same reason. For the symmetrical reactors,
the level of total magnetic flux density decreases inside the reactors with increasing number of
segments per turn, but changes in the opposite manner outside the reactors.
The variation of the levels of total magnetic flux density along a vertical observation surface
passing through the axis of the reactor is also illustrated.

9 MODELING NETWORK OF METALLIC PLATES AND

CYLINDRICAL CONDUCTORS IN HIFREQ
Ali Aghabarati, Rouzbeh Moini, Peter Zhao, Shabnam Ladan,

Simon Fortin and Farid P. Dawalibi

Abstract
This paper presents recent progress in HIFREQ for analyzing networks of metallic plates and cylindrical wire
conductors. In this new version of HIFREQ, direct energizations of metallic plates can be realized in term of current
injections into the plates. Also, junctions between plates and wires are now allowed. The latter permits the
treatment of arbitrarily oriented wire segments crossing or being attached on the surface, edges, and vertices of
metallic plates. Numerical computation results are presented at low and high frequencies for several metallic
structures with multiple wires attached to metallic plates. The results illustrate the accuracy of the method in terms
of current distribution, charge density, input impedance, and electromagnetic fields.
1 Introduction
In recent versions of SES software (15.1.4080 and later), the HIFREQ program was extended to
support the modeling of metallic plates. Metallic plates are often employed to form many kinds of
objects, such as transformer tanks, cars, electronic cabinets, etc., that can be present in
substations and other environments subject to electromagnetic interference. Models built with
plates are often more efficient and accurate than the wire-grid models that were traditionally used
to represent those objects. With this addition, HIFREQ is now capable to provide accurate
modeling for a wider class of problems related to grounding networks and electromagnetic
interference.
One of the difficulties involved in defining networks of plates and wires is the junctions that can
occur between plate surfaces and cylindrical wire conductors. Figure 1 shows a schematic view of
a general wire-plate network with existing junctions embedded in multilayer soil medium. In
previous versions of HIFREQ, such junctions were forbidden and would cause the program to
stop. However, numerous practical applications require junctions between wires and plates. For
example, connecting the metallic enclosure of various types of equipment found in substations to
the grounding grid requires such junctions. As another example, in the higher frequency range
the functionality of a category of electromagnetic radiating devices is based on thin wire
configurations connected to conducting surfaces and excited by potential difference at the
attachment points. In addition, the shape and electrical functionality of many electromagnetic
design problems is such that some parts should be treated as wires connected to the ground plane.
In this version of HIFREQ, junctions between wires and plates are now fully supported. Such
junctions allow transfer of current between wires and plates. This makes it possible to energize a
metallic plate by specifying a current (Lead), Voltage, or Potential (GPR) source on a wire and
connecting it to the plate. Wires can connect at any location on a plate, including the edge or the
corners. Wires crossing plates are automatically subdivided at the intersection point; several wires
can connect to the same plate or even to the same patch on a plate. Also, plates can now be directly
energized by specifying the amount of current entering a plate patch.
The remainder of this article is structured as follows. In Section 2, the formulation of the applied
boundary conditions to the configuration of a wire in contact with a conducting metallic plate is
presented. In Section 3, several numerical examples are given. The first example describes the use
of HIFREQ for the computation of the capacitance of a spherical capacitor at very low frequency.
Then, the input impedance (admittance) of a monopole antenna on a limited ground plane of
various shapes is presented. The electromagnetic interference's (EMI) analysis of various
practical problems involving wires in contact with metallic plates is also described.
Figure 1: The configuration of a wire-plate network with junctions embedded in a multilayer soil
medium.
2 Formulation of New Features

This section briefly describes the methodology behind the new features introduced in HIFREQ
since the last User’s Group Meeting.
2.1 Wire and Plate Junctions
The approach used to handle junctions between wires and plates in HIFREQ is as follows.
In order to determine the current distribution, plates are subdivided into smaller planar
quadrilateral elements called patches. To make sure that the current distribution varies smoothly
in the plate, plates are subdivided in such a way that any two patches either do not contact one
another or join perfectly along one edge.
Figure 2 shows a particular patch, with corners at 𝐴⃗, 𝐵

⃗⃗, 𝐶⃗ and 𝐷
⃗⃗. This patch has a junction with a
wire segment. The variation of the current flowing in each patch is assumed to be of bilinear form,
and is parameterized in terms of the current flowing into the patch across its four edges. With this
assumed form for the current distribution in the patch, the current flowing into the patch through
each edge of the patch is constant along that edge. These values of the normal edge currents for
each patch and the longitudinal and leakage currents for the wire segments constitute the

unknowns of the problem. By current conservation, the leakage current of a patch is the sum of
those four edge currents plus the current of all the wire segments attached to the patch.
Figure 2: Showing the reference direction for unknown wire segment current split into four edge
currents.
The unknown edge currents and line currents are determined by enforcing that the voltage drop
Δ𝑉 along the path connecting the center of the patch and the center of attached wire segment
satisfy Ohm’s Law:
𝛥𝑉 = 𝑍𝑃 𝐽 + 𝑍𝑊 𝐼 (1)
where 𝑍𝑃 represents the internal impedance of the patch (which is a function of its resistivity,
permeability, and thickness) and 𝑍𝑊 represents the internal impedance of the attached cylindrical
wire segment (which is a function of its resistivity, permeability, and radius of the segment). The
variable 𝐽 represents the total current density flowing across the patch while 𝐼 represents the
longitudinal current flowing along the wire segment.
2.2 Direct Energizations of Plates
Plates can also be directly energized, by specifying the current that enters the plate. More
precisely, the precise plate patch in which the current enters the plate must be specified, along
with the Energization Type Code of the energization applied to that patch.
As shown in Figure 3, the identity of the patch that is energized can be specified in one of two
ways:
a) By giving the patch number along the two main directions of the plate at the intersection
of which the patch is located.
b) By giving the position of a point in space, in which case the energization will be applied to
the plate patch which is located closest to that point.
Note that the second type of patch specification for energization purposes is less sensitive to the
detailed way in which plates are subdivided into patches.
Several energizations can be specified on the same plate. However, it is not allowed to specify
several energizations on the same patch of a plate. Direct energizations can also be specified on
patches where one or more wire segments are attached.

Figure 3: Illustration of Add / Edit Plate Energizations dialog for specifying direct energizations on
metallic plates in HIFREQ. (a)- By giving the patch number along the two main directions of the
energized plate. (b)- By giving the position of a point in space.
For each energization, the required data are as follows:

 Energization Type: Indicates the type of energization selected for the analysis. At this
moment, only Current (Lead) energizations are supported for metallic plates.
 Patch Selection: Indicates which method is selected to identify the energized patch.
When By Position is selected, the energized patch is identified by providing the X, Y, and
Z coordinates of a point located close to the patch. Therefore, the closest patch to this point
on the plate containing that patch is considered as being energized. When By Patch
Numbers is selected, the energized patch is identified by providing the explicit patch
numbers along the two principal directions of the plate.
 Patch Numbers: Specifies the subdivision numbers along the 'ab' and 'bc' directions that
identify the energized patch.
 Reference Point Coordinates: Specifies the X, Y, and Z coordinates (in meters or feet)
of the point used to identify the energized patch.
From a formulation point of view, the presence of a direct energization on a patch simply
generalizes the relation used to obtain the leakage current of a patch in terms of its edge currents
and of the longitudinal currents of the wires attached to the patch.
The energization of the network can also be performed using an externally applied electric or
magnetic field, in the form of a static field or of a plane-wave (see examples of this type in [1]).

3 Numerical Results
Several numerical results are now presented that demonstrate the accuracy and versatility of
HIFREQ for analyzing networks of wires and plates. The first example describes the use of
HIFREQ for the computation of the capacitance of a spherical capacitor at very low frequency.
Next, the input impedance (admittance) of a monopole antenna on a limited ground plane of
various shapes is computed as an indicator of the validity of the method at high frequencies.
Finally, various examples of electromagnetic interference (EMI) analysis involving wires in
contact with metallic plates are also presented.
3.1 Concentric Spherical Capacitor at Very Low Frequency
Figure 4-(a) depicts a plate model of a concentric spherical capacitor. The geometry is made from
two concentric metallic spheres with an inner radius of 𝑟1 = 0.5 𝑚 and an outer radius 𝑟2 = 1 𝑚.
Each spherical electrode is modeled by juxtaposition of 420 quadrilateral metallic plates tiling the
surface of the sphere. It should be noted that spherical objects made from metallic plates can be
readily created in SESCAD using the command Insert | Spherical Surface, as shown in Figure
4-(b).
Note that in addition to simple spheres, metallic plate representations of other common
geometrical object such as conical, cylindrical surfaces, disks, and bricks (or prisms) can also be
created in SESCAD by specifying a few parameters in the corresponding dialogs. For example,
The Create Conical Surface dialog box (Insert | Cylinder / Cone) allows creating a model for
the surface of a cylinder, a cone or a general portion of conical surface in terms of metallic plates
or conductors. Similarly, solid objects representing other types of common geometries can be
specified using the commands Insert | Disk / Ring, Insert | Spherical Surface, and Insert
| Brick / Prism.
The energization of the capacitor in the electrostatic regime is simply obtained by connecting the
two thin spherical shells using a wire connection. A Voltage energization of 1 volt (at 1 Hz) is
applied to the connecting wire. This results in induced positive charges (+𝑄) and negative charges
(−𝑄) on the inner and outer spherical electrodes of the concentric capacitor and in a radial electric
field between the metallic shells.

(a)
(b)
Figure 4: (a) - Spherical capacitor with a wire junction between the two electrodes and an applied
Voltage energization. (b)- The Create Spherical Surface dialog box that allows creating a model for
the surface of a sphere, a hemisphere or a general portion of spherical surface in terms of conductors or
metallic plates.
The surface of each individual quadrilateral plate on the spherical electrode is discretized using 4
patches. The capacitance of concentric spherical structure is then obtained numerically by
examining the value of the longitudinal current discharged from the interior electrode to the
exterior electrode through the attached connecting wire. The computed current corresponds to
the overall charge induced over the electrodes and can be obtained from the input impedance of
the system as follow:

𝐼
𝐼 = 𝒿𝜔𝑄 = 𝒿𝜔𝐶𝑉 ⇒ 𝐶 = (2)
𝒿𝜔𝑉
Using the computed value of the discharge current, the evaluated capacitance of the system at
1 Hz is:
𝐼 𝒿0.7108 × 10−9 (3)

𝐶= ⋍ = 0.1131 × 10−9 F = 113.1 [pF]
𝒿𝜔𝑉 𝒿 × 2𝜋 × 1 × 1
An analytical expression for the electric field between the two electrodes of a concentric spherical
capacitor can be found by applying Gauss’ Law. Accordingly, the voltage difference between the
spheres can be calculated by integrating the electric field along a radial line in order to obtain the
theoretical value for the capacitance as follow:
𝑄 4𝜋𝜀0 4𝜋 × 8.8541 × 10−12

𝐶= = = = 0.1111 × 10−9 = 111.1 [pF] (4)
𝑉 (1 − 1) (
1 1
− )
𝑟1 𝑟2 0.5 1
A very good agreement is observed between the theoretical value for the concentric spherical
capacitance at electrostatic regime and numerical value obtained by HIFREQ at very low
frequency. More importantly, the accuracy of the obtained result reveals the numerical stability
of the HIFREQ computation engine at very low frequency, recognized as the problem of low
frequency breakdown in EFIE [7]. The spatial distribution of the amplitude of electric field
between the spherical electrodes is demonstrated in Figure 5. The pattern is in agreement with
the following theoretical relation:
𝑄 𝐶𝑉 4𝜋𝜀0 𝑉 1 1
𝐸= = = = = 2 [V⁄m]
4𝜋𝜀0 𝑟 2 4𝜋𝜀0 𝑟 2 1 1 1 1 (5)
4𝜋𝜀0 𝑟 2 (𝑟 − 𝑟 ) 𝑟 2 ( − ) 𝑟
1 2 0.5 1
Figure 5: Radial variation of electric field intensity inside the concentric spherical capacitor.

3.2 Calculation of Input Impedance at High Frequency
In this section, the numerical computation results for the input impedance (or admittance) of two
benchmark examples at high frequency are presented. The geometries considered include a wire
attached to a smooth surface or to a vertex of a plate.
Figure 6-(a) shows the geometry of a monopole antenna and its grounding plane. The monopole
antenna consists of a single wire attached to the centre of a flat metallic plate. A Voltage source
located at the wire attachment point energizes the wire antenna. The length and radius of the wire
are 41.9 cm and 0.8 mm respectively, mounted perpendicularly at the center of a 91.4 cm square
plate. By using SESCAD, individual metallic plates can be easily inserted from the command
Insert | Metallic Plate. This launches the Edit Object screen (Figure 6-(b)) for creating and
editing metallic plates in the model.
(a)
(b)
Figure 6: (a) - A monopole wire attached at the centre of a flat plate and fed at the attachment point.
(b) - Edit Object dialog box for editing detailed plate subdivision in SESCAD.

The Edit Object screen also allows to specify all the data related to a plate such as: Plate Name,
Plate Shape (rectangle, parallelogram, or general quadrilateral), Thickness of the plate,
Relative Resistivity (with respect to annealed copper) of the metallic material of the plate,
Relative Permeability (with respect to vacuum) of the metallic material of the plate, the
Cartesian Coordinates of the four vertices (A, B, C, and D) of the plate, desired number of
Total Subdivisions (or surface patches) for the plate, and also the ratio of the size of consecutive
patches along the plate edges (Compression Factor).
The Advanced Subdivision feature, shown in Figure 6-(b), can also be used to define more
detailed plate subdivision settings to be applied for any specific plate. When detailed subdivision
settings along AB and BC sides are specified, the Total Subdivision specified in the Edit Object
screen is ignored.
In this section, the frequency of the Voltage source varies from 140 MHz to 200 MHz with a step
of 1 MHz. At each frequency, the input admittance of the wire-surface network is computed from
𝐼(𝜔)
the ratio of longitudinal current at the feed point to the supplied voltage (i.e. 𝑌(𝜔) = 𝑉(𝜔)). The
numerical computation results are compared with calculated and measured values by Newman-
Pozar [6], and the result reported by Hwu-Wilton [3].
Figure 7 depicts the variation of the input admittance with the frequency, obtained from HIFREQ.
The input admittances extracted from above mentioned references are also shown in this figure.
As can be observed, the agreement between the results obtained with HIFREQ and the reference
publications is very good.
Figure 7: Input admittance versus the frequency of the monopole fed at the attachment point.
The next example pertains to a quarter wave monopole mounted near a vertex of plate network
consisting of three square plate, at 300 MHZ. Figure 8-(a) illustrates the geometry of the problem,
where the monopole is attached near the corner edge formed by three 0.4 𝜆 square plates. The
radiating monopole (radius=0.0015 𝜆) is positioned normal to the top plate (plate number 1) and
is energized at the attachment point using a unit amplitude Voltage source. As shown in Figure

8(b), labeling of plates can become visible in SESCAD by invoking the View | Labeling | Show
Plate Numbers command.
(a)
(b)
Figure 8: (a) - A quarter wave monopole wire mounted near the corner formed by three square metallic
plates. (b) – SESCAD command for showing the labeling for plate numbers.

The variation of input impedance as a function of attachment position is illustrated in Figure 9.

Once again, the HIFREQ results are compared with calculated and measured results by Newman-
Pozar [6], and Hwu-Wilton [3]. The agreement between the HIFREQ results and the reference
method is again satisfactory.
Figure 9: Input impedance versus the attachment position of a quarter wave monopole mounted near
the corner made by three metallic plates.
3.3 Immunity Test for Automotive EMC
This section presents numerical results for some challenging electromagnetic compatibility
(EMC) problems.
The following numerical computation pertains to a practical test of automotive EMC, commonly
used for determining the immunity of passenger area to electromagnetic disturbance from off-
vehicle radiation sources. Electromagnetic interference tests are usually performed by
manufacturers to demonstrate the electromagnetic compatibility of their products according to
standards requested by related organizations such as CISPR, SAE and the ISO [8].
The off-vehicle sources of interference can be natural or man-made. They can be high voltage
power lines [1], electronics circuits with switching of large amount of currents, electromagnetic
pulses of indirect lightning [9], or other atmospheric sources of noise. For the first EMI analysis,
the electromagnetic fields generated by a biconical antenna are considered as the source of radio
frequency interference (RFI), and placed in close proximity of the metallic body of a vehicle as
shown in Figure 10. Biconical antennas are commonly used in automotive EMC because of their
capability of producing a desired field strength over the range of test frequencies [8][10].
As shown in Figure 10-(b), the antenna (modeled with 31 conductor segments) is located 3 m away
from the vehicle at a height of 2.4 m above ground. The soil resistivity is set to 100 Ω − 𝑚. The
metallic body of the vehicle (5 𝑚 × 2 𝑚 × 1.7 𝑚) is located 0.3 𝑚 above the ground. The immunity
of vehicle body against the incoming electromagnetic interferences is determined by examining
the field strength for the passenger area, with and without the presence of car. The test frequency
is set to 20 MHz. For this purpose, a wire model of biconical antenna is considered in HIFREQ.
The antenna is energized at its center point using a unit amplitude Voltage source (dimensions
taken from [9]). The geometry of the vehicle is modeled using 34 quadrilateral metallic plates and
6 wire conductors, which represent a metallic body with opening windows. As seen in Figure 10,
wire conductors are in contact with the edge of plates. The electromagnetic field are evaluated for
an observation surface which crosses the vehicle, as seen in Figure 10-(b). The electromagnetic
field amplitude on this profile is used to assess the field penetration level through the openings
(windows). The combined wire and plate HIFREQ model is solved at the desired frequency using
a total number of 1375 surface patches and 454 conductor segments.
(a)
(b)
Figure 10: (a) - Overview of the immunity test problem for Automotive EMC. (b) - Model of EMI problem
in SESCAD, along with defined surface and line profiles for electromagnetic field sampling.
To control the total number of patches over a portion of a network that includes plates, you can
select that group of plates and invoke the Edit | Edit Plate Attributes command. The attributes
(including the subdivision settings) for the selected group of plates can be modified. This dialog
box is also useful for specifying the thicknesses of selected plates, their relative permittivity and
relative permeability or the number of patches or subdivisions for the plates based on a total
number of patches on all plates, the total number of patches on each plate or the maximum patch
area allowed (see Figure 11).

Figure 11: Dialog box for editing attributes of plates in SESCAD.
Figure 12-(a) shows the variations of the electric field over the selected observation surface. The
results demonstrate how the metallic body of the vehicle interacts with the incident field of the
antenna and the variation of electric field intensity for the regions inside and outside the vehicle.
For observation points which are located inside the vehicle, the amplitude of the interfering field
is reduced noticeably compared to the exterior regions. In order to better quantify the level of
protection observed in this EMI study, the electric field intensity is examined in presence and
absence of the metallic body along a line profile that passes through the vehicle from the side
windows (see Figure 10-(b)). The computation results are shown in Figure 12-(b) for both
scenarios. Comparing the results, the electric field is seen to have consistently lower amplitude
inside the vehicle. For the point at the center of vehicle, the ratio of field amplitude with and
without the presence of vehicle is 6.77 × 10−4 /5.63 × 10−3 = 0.12, which indicates a Shielding
1
Effectiveness (SE) of around 20 log (0.12) = 18.4 dB.

(a)
(b)
Figure 12: Spatial distribution of the amplitude of the electric field around a vehicle modeled by plates and
wires and exposed to a radio frequency interference source in its proximity. (a)- Observation surface, (c)-
Line profile with and without the presence of vehicle body.

3.4 Lightning Effects on the Wires and Cables inside a Helicopter
The last test problem shows how HIFREQ in combination with the FFTSES module can be used
to examine the effects of lightning on cables and wires installed inside a simplified helicopter
model. The presence of induced interference on long cables and wires as a result of lightning is a
safety concern for the aviation industry [11][12]. Simulation results of this class of EMI problems
can demonstrate the benefit (or necessity) of using shield cables to minimize the effects of
lightning on critical onboard systems, despite the drawbacks of increased weight, routing
limitations or costly design techniques.
The geometry of the EMI study is shown in Figure 13, in which a combination of wires and plates
are used as a mock-up of the helicopter with a few opening windows. The dimensions of the
problem are 15.1 m (length), 10 m (width) and 3.3 m (height). The SESCAD model consists of 117
plates, 61 wire segments and a total number of 34 junctions between wires and plates, and
represents necessary components that can significantly interact with electromagnetic fields. The
figure also illustrates two long “single wire” and “shielded cable” running along similar routes
inside the fuselage. The diameter of the wire is 2 mm, and the shield cable is RG58 type (core
diameter of 0.91 mm, a layer of insulation material with resistivity of 1 × 1010 Ω. m and thickness
of 1.02 mm around the core, and a sheath with an outer diameter of 3.5 mm). In the model, it is
assumed that the lightning current penetrates through the tip of the blade used to represent the
main rotor, while the tail of the helicopter is assumed to be the lightning exit point. These points,
designated as the strike points, present a relatively higher probability to be hit by lightning
compared to other parts of the helicopter fuselage.
Figure 13: The SESCAD model using plates and wires for a mock-up helicopter subjected to lightning.

The entry and exit point for the direct lightning strike are as depicted in Figure 13 [10]. The strike
points are modeled in HIFREQ using the Add / Edit Plate Energizations dialog box to specify
direct energizations on metallic plates (see Figure 14). The energizations are specified by
providing valid Energization Type codes which are defined in the Define Energization
Types dialog as shown in Figure 14-(a). In this example, the coordinates of the strike points, are
used to specify the location of the two current sources for the analysis of the EMI problem. Note
the opposite polarity of these two current sources, which is used to define the lightning entry exit
points.
(a) (b)
(c)
Figure 14: (a)- HIFREQ System Energization dialog to define the energization source parameters for
plates. (b)-Edit Object dialog box for creating and editing plates attributes in SESCAD. (c)- Plate
Energizations dialog box in HIFREQ for specifying direct lightning energizations on a blade of the
principal rotor of the helicopter.
A typical lightning current waveform is considered, with a time variation represented as the sum
of two double exponential functions, as follows:
𝐼(𝑡) = 9000(𝑒 −40000𝑡 − 𝑒 −3000000𝑡 ) + 115(𝑒 −100000𝑡 − 𝑒 −22200𝑡 ) (6)
This transient waveform is shown in Figure 15. It is characterized by a rise time of 1.44 𝜇𝑠 and a
half-value time of 24 𝜇𝑠. The results are obtained at 512 frequency samples (determined by the
FFTSES Forward operation) and extracted by the database management feature of FFTSES for
the inverse FFT operation. Detailed information on the steps for conducting such a procedure are
described in a How-To manual [13].

In this study, we examined the level of induced current at the middle of a single wire and a shielded
cable, located in the proximity of a highly electrified fuselage leading the lightning current from
its entry point to its exit point.
Figure 15: Current waveform of the studied lightning signal.
The results for the induced transient current in the wire and cable are given in Figure 16 (a) and
(b) respectively. It can be seen that two sharp peaks occur in the middle of the single bare wire
with amplitudes around 38 mA and 20 mA, while for the shielded cable the intensity of the
induced interfering current is much lower and does not exceed 0. 016 mA at its peak. In this case,
the shielding effectiveness provided by the cable shield against the lightning discharge current is
around 67.5 dB. The distribution of the electromagnetic fields generated by the lightning impact
is shown in Figure 16-(c). As illustrated in that figure, the field strength for the regions located
inside and outside the helicopter varies significantly. For the exterior regions, electric fields of
V
strength 5 × 106 m can be observed, while thanks to the shielding due to the metallic body of the
helicopter, the level of electric field inside the helicopter is reduced to values that are smaller
V
than 100 in some regions.
m

(a) (b)
(c)
Figure 16: HIFREQ results for the helicopter model hit by lightning. (a) – Induced current observed in the
middle of single wire as a function of time. (b) - induced current observed in the middle of shielded cable as
a function of time. (c)- Computed amplitude of the electric field (V/m) over the observation surface around
the helicopter.
4 Conclusion and Future Developments

This paper presented recent developments in the HIFREQ program for analyzing various
problems involving networks of metallic plates and cylindrical conductors with arbitrarily
configured and arbitrarily connected wire and plates. Numerical results, including comparisons
with calculations and measurements found in the literature, showed that wire-to-plate junctions
can now be modeled accurately in HIFREQ, both at low and high frequencies. The computation
results also demonstrate the versatility, accuracy and efficiently of the latest version of the

HIFREQ program for solving practical EMI problems consist of combined networks of wire and
plates with complex configurations and multiple junctions between wires and plates.
The next year will see further developments of the plate feature in HIFREQ, notably in the
following areas:
 Multiple junctions between plates: Presently, at most two plates (and any number
of wires) can connect along the edge of a plate. This will be generalized to allow for any
number of plates sharing a common edge.
 Data Visualization and Processing for Currents and Potentials of Plates:
Currently, results related to plates, such as the current flowing into plate patches or the
potential of the patches, are not displayed in SESsystemViewer or when creating
Configuration plots in CDEGS – Examine. Also, plates are not taken into account
when calculating touch voltages. These features will all be implemented in the near future.
5 References
[1] A. Aghabarati, R. Moini, S. Fortin, P. Zhao and F. P. Dawalibi, “Modeling of Plate Structures In
HIFREQ,” UGM 2015 –SAN DIEGO, CALIFORNIA.
[2] Gibson, Walton C, "The method of moments in electromagnetics". CRC press, 2014.
[3] S. U. Hwu, D. R. Wilton, and S. M. Rao, "Electromagnetic scattering and radiation by arbitrary
conducting wire/surface configurations," IEEE Antennas and Propagation International Symposium,
AP-S. Digest, pp. 890-893. 1988.
[4] N. J. Champagne, "On attaching a wire to a triangulated surface." Antennas and Propagation
International Symposium, vol. 1, pp. 54-57, 2002.
[5] A. Selby and F. Dawalibi, “Determination of current distribution in energized conductors for the
computation of electromagnetic fields,” IEEE Transactions on Power Delivery, no. 2, pp. 1069-1078,
1994.
[6] E. H. Newman and D. M. Pozar, "Electromagnetic Modeling of Composite Wire and Surface
Geometries," IEEE Transactions on Antennas and Propagation, vol. 26, no. 6, pp. 784-789, 1978.
[7] J. F. Lee, R. J. Burkholder, "Loop star basis functions and a robust preconditioner for EFIE scattering
problems." IEEE Transactions on Antennas and Propagation, no. 8, pp. 1855-1863, 2003.
[8] V. Rodriguez and L.P. ETS-Lindgren, "Automotive Component EMC Testing: CISPR 25, ISO 11452-2
and Equivalent Standards." IEEE Electromagnetic Compatibility Magazine, no. 1, pp. 83-90, 2012.
[9] S. Ladan, A. Aghabarati, R. Moini, S. Fortin and F. P. Dawalibi, “Induced Disturbances by High Voltage
Transmission Lines on Nearby Stationary Vehicles”, Accepted for presentation at IEEE International
Symposium on Electromagnetic Compatibility (EMC 2016), Ottawa, Canada, July 25-29, 2016.
[10] B. A. Austin and P. F. Andre, "Characteristics of the wire biconical antenna used for EMC
measurements," IEEE Transactions on Electromagnetic Compatibility, no. 3, pp. 179-187, 1991.
[11] M. Apra, M.D. Amore, K. Gigliotti, M.S. Sarto and V. Volpi, "Lightning indirect effects certification of
a transport aircraft by numerical simulation," IEEE Transactions on Electromagnetic Compatibility,
no. 3, pp. 513-523, 2008.
[12] M. H Vogel, "Impact of lightning and high-intensity radiated fields on cables in aircraft," IEEE
Electromagnetic Compatibility Magazine, no. 2, 56-61, 2014.
[13] SES Technologies, "Lightning Transient Study of a Communication Tower," How To Engineering
Guide, SES Technologies, Quebec, Montreal, 2012.

10 MODELING OF LIGHTNING CHANNEL USING HIFREQ

AND FFTSES
Shabnam Ladan, Ali Aghabarati, Rouzbeh Moini, Simon Fortin and Farid P. Dawalibi

Abstract
In this paper, a wire model of a lightning channel based on monopole antenna above ground is implemented in
HIFREQ. In order to tune the propagation speed of the lightning current to a physical value, a distributed
impedance is applied to the monopole representing the lightning channel. The current at the channel base is
simulated by means of a Lead Energization, which injects current at the bottom of the monopole representing the
channel. The appropriate use of FFTSES together with HIFREQ yields an accurate computation of the
electromagnetic fields around the channel. The computation results show that the vertical component of the
electric field induced by the lightning channel is the main cause of coupling between the lightning channel and
nearby structures. The presence of an elevated strike object is also investigated. To this end, the antenna model of
the lightning channel is added to a wire mesh model of the structure hit by lightning at the strike point.
Comparisons between results obtained with the antenna model of lightning channel and those obtained with a
single Lead energization, demonstrate that the model consisting of a single Lead energization can underestimate
the coupling level between the struck object and the neighboring structures.
1 Introduction
The interactions of lightning-radiated electromagnetic fields with electric power lines and other
electrical circuits can cause malfunctions of critical installations or can even destroy such
installations. In order to design an adequate lightning protection system, a detailed
understanding of lightning and its electromagnetic fields is necessary. Lightning is a transient,
high-current electric discharge spanning distances in the kilometer range. Cloud-to-ground
lightning is less common than other kinds of lightning (i.e. cloud-to-cloud and inter-cloud
lightning), but it is more important for protection studies of electric and electronic apparatus used
in power systems, information technology systems, etc. The complete electric discharge, known
as flash, has a time duration of about 0.5 s and is made up of various components, including
stepped leader, return-strokes and dart leaders (Fig. 1-(a)). The part of a flash which is of greater
interest for protection purposes is the return-stroke phase. Indeed, the return-stroke current
produces most of the damage attributed to lightning. Therefore, mathematical models designed
to reproduce certain aspects of the physical process involved in the return-stroke phase are
utilized to predict the damaging effects of lightning [1].

Figure 1: (a) Downward negative cloud-to-ground lightning stroke (b) Return-stroke phase represented by
a monopole antenna energized at ground level by a current source (𝑐 is the speed of light in free space).
The expression ‘lightning return-stroke model’ is generally used to describe a specification of the
time and height-dependent current in the return-stroke channel (RSC) so as to make possible the
calculation of the resulting electromagnetic fields [2]. The aim of this article is to show how to use
the combined powers of the HIFREQ and FFTSES modules in order to implement an appropriate
model of RSC allowing for an accurate evaluation of the transient electromagnetic fields around
the channel.
2 The Antenna Model of Lightning

A suitable model for the lightning channel should be characterized by a minimum number of
adjustable parameters and be consistent with the measured characteristics of the return-stroke
such as the current at the base of the channel, the variation of the intensity of the light emitted by
the channel with height, the upward propagation speed of the luminosity front, and the
electromagnetic fields at various distances from the channel. Other physical parameters could
also be included in a realistic model for describing lightning phenomena. However, considering
these extra parameters would add to the complexity of the model, making it impractical for most
applications. Thus, despite the availability of many lightning models in the literature, the antenna
theory (AT) model of RSC [3] was selected for the purpose of analysis.
The AT Model follows rules based on various measurements presented by Rakov and Uman [1].
In this model, the lightning RSC is modelled by a thin wire antenna, excited by a current source.
A distributed impedance along the wire antenna is used to control the propagation speed of the
current wave along the channel [3] and the related electromagnetic fields.
The full wave capabilities of the HIFREQ module are well suited for the electromagnetic analysis
of a lightning channel based on the AT model. To this end, the lightning RSC is considered as a

monopole wire antenna [3] above a conducting ground. Fig. 1-(b) depicts a monopole antenna
representation of the return stroke phase. HIFREQ also allows for the analysis of complex
structures, including a network of wires and metallic plates in the presence of a multilayer soil, at
the same time as the lightning channel. An appropriate use of the FFTSES module also permits a
very accurate analysis of lightning in the time domain. The goal of this article is to provide
guidelines to help use the MultiFields or CDEGS software package modules for the
implementation of the AT model of lightning.
A description of the current wave shape at the base of the channel and of the input parameters
used in FFTSES to specify this current is the subject of the next section. The AT model of RSC
based on a monopole antenna above a uniform soil with distributed impedance is described in
Section 3. The last section is dedicated to numerical results obtained with the implementation of
the AT model in HIFREQ. This chapter presents a wide range of validation results for the lightning
channel model adopted in this paper. Finally, an example showing how to use this model for
towers hit by lightning is also presented.
3 FFTSES Input Data for the Transient Current at the Base of

the Channel
In the implementation of the return stroke model, the transient current at the base of the channel
is considered as energization of the model. Indeed, FFTSES allows the use of various transient
current wave shapes for the energization procedure. The sum of two Heidler functions [4] is
commonly used in the literature to express the current at the base of the channel and it has been
adopted for the lightning analysis in this study. Accordingly, the transient variation of the
energization is depicted as follows:
𝑡 𝑛1 𝑡
𝐴 (𝜏11 ) 𝑡 𝐵 ( )𝑛2 𝑡
− 𝜏21 −
𝑖(𝑡) = 𝑒 𝜏12 + 𝑒 𝜏22 (1)
𝜂1 1 + ( 𝑡 )𝑛1 𝜂2 1 + ( 𝑡 )𝑛2
𝜏11 𝜏21
1
𝜏 𝜏
−( 11 )(𝑛 12 )𝑛1 (2)
𝜂1 = 𝑒 𝜏12 1 𝜏11
1
𝜏 𝜏
−( 21 )(𝑛 22 )𝑛2 (3)
𝜂2 = 𝑒 𝜏22 2 𝜏21
The numerical values of two channel-base current waveforms corresponding, respectively, to
typical first and subsequent return strokes, are given in Table 1. These values are based on
observations of Berger et al. [5].
Table 1: Parameters of the two Heidler functions used to reproduce the channel-base current wave shape
𝐴 (kA) 𝜏11 (µs) 𝜏12 (µs) 𝑛1 𝐵 (kA) 𝜏21 (µs) 𝜏22 (µs) 𝑛2
First stroke 28 1.8 95 2 - - - -
Subsequent 10.7 0.25 2.5 2 6.5 2 230 2

stroke

The first return stroke channel-base current is characterized by a peak value of 30 kA and a
maximum steepness of 12 kA/µs, whereas the subsequent return stroke current has a peak value
of 12 kA and a maximum steepness of 40 kA/µs. Therefore, in the following, the analysis will be
restricted to a subsequent return stroke current because of its higher frequency content compared
to the current of a first return stroke.
Figure 2 depicts the characteristics of the channel-base current used in the FFTSES module. The
total time of the analysis is set to 𝑇𝑚𝑎𝑥 = 800 µs. This time window is subdivided into 213 = 8192
213
time samples, corresponding to 212 = 4096 frequencies up to 𝐹𝑚𝑎𝑥 = 𝑁𝑦𝑞𝑢𝑖𝑠𝑡 𝑟𝑎𝑡𝑒 = =
2𝑇𝑚𝑎𝑥
213 1
1600µs
= 5.120 MHz with frequency intervals of ∆𝑓 = 𝑇 = 1250 Hz. Consequently, the time step
𝑚𝑎𝑥
800µs 800µs
for transient analysis is given by ∆𝑡 = 2𝑛
= 213
= 0.097 µs. The transient wave-form of the
channel-base current on a 800 µs time scale and 1 µs time scale are also plotted in that figure. The
latter shows clearly that the rise time of the injected current is about 0.4 µs and is discretized by
0.4 µs
0.097 µs
≅ 4 points.
To ensure the convergence of the MoM in HIFREQ, the segment length should not exceed one
sixth of the minimum wavelength (corresponding to the maximum frequency), defined by 𝜆𝑚𝑖𝑛 =
𝐿𝑖𝑔ℎ𝑡 𝑠𝑝𝑒𝑒𝑑 𝑖𝑛 𝑓𝑟𝑒𝑒 𝑠𝑝𝑎𝑐𝑒 3×108
= ≅ 60 m. Thus, the maximum segment length permissible in
𝐹𝑚𝑎𝑥 5.12×106
𝜆𝑚𝑖𝑛
HIFREQ, is = 10 m. For the channel-base current chosen for this study, a lightning channel
6
1300 𝑚
with a height of 1300 m must be modeled by at least 130 = segments.
10 𝑚
4 Modelling of Lightning Channel in HIFREQ

The schematic of a lightning RSC located above ground is shown in Figure 1-(b). In this figure, the
RSC is represented by long conductive thin wire excited by a current source at its lower end and
at the ground surface. The transient behavior of the current source at the base of channel was fully
described in the previous section.
Modeling of the lightning return stroke typically involves a description of the time and height
variations of the current wave along the channel, which is needed for the calculation of resultant
electromagnetic fields. Outputs of an adequate model should be consistent with the observed
characteristics of the return stroke, namely propagation speed of the luminosity front (often used
as a proxy for the current wave front), and electromagnetic fields at different distances from the
channel.
The AT model represents the lightning channel as a lossy monopole antenna above a conducting
ground. The losses along the channel are introduced by distributed resistances on the monopole.
The value of resistance per unit length is selected to provide an agreement between model-
predicted and measured electric fields near the RSC channel. In this regard, the following internal
resistance is added to wire conductors used to model the RSC channel [3]:
𝑅𝑎𝑑𝑑 = 0.5 Ω/m = 500 Ω/km. (4)

a- 800 µs time scale.
b- 1 µs time scale.
Figure 2: Transient characteristics of the input current.

A set of distributed inductive loads is also applied to the simulated lightning channel, in order to
tune the propagation speed to a desired value. Note that this additional distributed inductance
along the channel has no physical meaning and is invoked only to reduce the speed of current
wave to a value lower than the speed of light.
An unloaded horizontal, perfectly conducting wire above perfectly conducting ground behaves as
a lossless uniform transmission line, and the propagation speed along the wire is constant and
1
equal to the speed of light 𝑐 = , where 𝐿0 and 𝐶0 are per unit length inductance and
√𝑐0 𝐿0
capacitance, respectively. The capacitance 𝐶0 , and inductance 𝐿0 , for a cylindrical metallic wire
of radius 𝑎 can be estimated using the following equations given by Kodali et al. [6]:
2𝜋𝜀0
𝐶0 (𝑧) = F/m (5)
2𝑧
ln ( )
𝑎
𝜇0 2𝑧
𝐿0 (𝑧) = ln ( ) H/m (6)
2𝜋 𝑎
where 𝑧 is the height above ground. According to the above equations the propagation speed of
the wave is:
1 1
𝑉0 = = = 𝑐 = 3 ∗ 108 m/s (7)
√𝐶0 (𝑧)𝐿0 (𝑧) √𝜀0 𝜇0
If we introduce an additional, height-variable distributed inductance along the antenna without
any resistive loading, the resultant height-variable propagation speed along the simulated
lightning channel will be given by [7]
1
𝑉(𝑧) = (8)
√𝐶0 (𝑧)(𝐿0 (𝑧) + 𝐿𝑎𝑑𝑑 (𝑧))
where 𝐿𝑎𝑑𝑑 (𝑧) is the additional height-dependent distributed inductance per unit length. As a
result, for a specified height-variable speed, 𝐿𝑎𝑑𝑑 (𝑧) is given by
1
𝐿𝑎𝑑𝑑 (𝑧) = − 𝐿0 (𝑧) (9)
𝑉 2 (𝑧)𝐶0 (𝑧)
For a channel with a height of 1300 m, and a radius of 0.01 m, the values of 𝐶0 and 𝐿0 at z = 1300 m,
are 4.5 pF/m, and 2.5 µH/m, respectively. Note that these values are applied to the entire channel.
This is a commonly used simplification [6] based on the fact that the dependence of 𝐶0 and 𝐿0 on
z is weak (logarithmic). To have a propagation speed near 1.3 × 108 𝑚/𝑠, the value of 𝐿𝑎𝑑𝑑 is
computed from (9) and is approximately equal to 9.98 µH/m. Consequently the following internal
reactance is added to wire conductors used to model the RSC channel:
𝜔𝐿𝑎𝑑𝑑 = 2𝜋 ∗ 60 Hz ∗ 9.98 µH/m = 4 Ω/km (10)
It should be noted that all the above computation were performed for a lossless channel. In the
case of lossy channel, the value obtained for additional inductance per unit length has to be
adjusted (by trial and error) to obtain a specified average speed.
The lightning channel is represented by a lossy vertical antenna above ground. The length of the
monopole is 1300 m while its radius is 0.01 m. As mentioned in the previous section, the

maximum segment length permissible for the channel is 10 m. Thus, 130 segments are used to
discretize the channel. According to equations (4) and (10), and as seen in Figure 3, a distributed
reactance of 4 Ω/km and a distributed resistance of 500 Ω/km are considered along the channel
(monopole antenna). As mentioned before, these values permits to tune the propagation speed
along the channel around the value of 1.3 × 108 m/s.
Figure 3: Distributed elements for RSC channel.
Figure 4: Energization for RSC channel.
The channel-base current is defined by means of a Lead Energization, which injects current at the
bottom of lightning channel. To this end a short conductor energized by a Lead with unit current,
is placed between the lower end of the monopole and the ground (Figure 4). In order to avoid
unrealistic values of the electric field near the channel at very low frequency, a short conductor
with a Free Energization is placed at the upper end of the monopole. The orientation of this
segment should be selected from the surrounding medium to the junction of the short conductor
with the monopole. The use of Free Energization informs HIFREQ that current flow into the
origin of the energized segment should be drawn from a fictitious adjacent network and not from
the surrounding medium [8]. The latter prevents formation of a large leakage current along the
channel at very low frequency. Note that there is no need to specify an energization value when
using a “Free energization (Figure 4). A detailed description of the model is depicted in Figure 5.
The computation of the induced currents and electromagnetic fields is performed in the frequency
domain using HIFREQ, for selected frequency components of the spectrum of the channel-base
current. The computations are performed at each frequency for a unit current.
The transient analysis of the structure is performed by the use of the FFTSES Fourier analysis
module. They are obtained through an Inverse Fast Fourier Transform applied to the product of
the frequency domain unit response by the spectrum of the lightning surge.

Figure 5: Lightning RSC model in HIFREQ.
5 Numerical results
5.1 Model validation
Figure 6 schematically shows a lightning channel above a lossy half-space, with resistivity 𝜌 and
relative permittivity 𝜀𝑟 . With reference to Figure 6, first the electromagnetic fields generated by a
lightning discharge over a perfectly conducting ground plane are evaluated. Then, the effects of a
soil of finite conductivity on the value of electromagnetic field are examined. To this aim, we
present the vertical and horizontal components of the electric field and the horizontal component
of the magnetic field at a distance 𝑟 from the channel at a depth of 𝑑𝑖 inside the ground or at height
𝑑𝑒 above the soil interface. In each case, the results are compared with those obtained by other
methods available in the literature.

Figure 6: Model configuration.
5.1.1 Electromagnetic fields above a perfectly conducting ground plane
The computation of the electromagnetic fields are performed at a height 𝑑𝑒 = 0.1 m above the soil
interface, at horizontal distances 𝑟 = 50 m, 𝑟 = 500 m and 𝑟 = 5000 m from the channel.
Figure 7 depicts the temporal distribution of electric and magnetic fields predicted by HIFREQ
and a transient method of moments (T-MoM) technique [3] using the same AT model for the
lightning channel. In HIFREQ, the value of the soil resistivity is set to 0.001 Ω. m in order to model
a perfectly conducting plane. Since the observation points are very close to the ground plane
interface, only the normal component to the interface of the electric field and the tangential
component to the interface of the magnetic field are computed.
As seen in this figure, except for small differences in the values of the field amplitude and rise
time, the result predicted by the two methods are similar. The differences between the field
amplitudes predicted by the two models may be related to the differences between the current
waveforms of the two models that are mainly due to different techniques implemented for the
solution of the equations in the time and frequency domains. Also, differences between the rise
times of the field waveforms are thought to be due to numerical instabilities in the T- MoM model.
It should be noted that the magnetic field at very close distances to the channel can be obtained
from Ampere’s law. Accordingly at a distance 𝑟 = 50 m from the channel, the wave shape of the
magnetic field can also be extracted from the current at the channel base, just by dividing the
latter by 2𝜋𝑟. The agreement between the values of computed magnetic field and those predicted
by Ampere's law demonstrates the validity of the adopted model.

r=50m r=50m
Horizontal magnetic field (A/m)

Vertical electric field (KV/m)
40 40
35 35
30 30
25 25
20 20
15 15
10 10
5 5
0 0
0 10 20 30 0 10 20 30
Time (Microsecond) Time (Microsecond)
r=500m r=500m

Vertical electric field (KV/m)
2000 3
2.5
1500
2
1000 1.5
1
500
0.5
0 0
0 5 10 15 20 0 5 10 15 20
r=5000m r=5000m
Vertical electric field (V/m)
80 0.2
60 0.15
40 0.1
20 0.05
0 0
0 10 20 30 40 0 10 20 30 40
-20
AT model in AT model in Ref [3]

HIFREQ
Figure 7: Lightning channel electromagnetic fields for a perfectly conducting ground plane 𝑑𝑒 = 0.1 m.

Note that HIFREQ uses Sommerfeld integrals for the computation of the electromagnetic fields,
while the T-MoM technique uses analytical expressions, describing the time-domain radiation of
a monopole antenna above a perfectly conducting ground plane. Thus, the T-MoM technique is
not capable of computing the electromagnetic fields for a soil with finite resistivity, while the
lightning channel model in HIFREQ can be extended to arbitrary soil configurations.
5.1.2 Underground electromagnetic fields
In this section, we present the electromagnetic fields due to lightning at a depth of 𝑑𝑖 = 1 m inside
the ground at two observation points located at horizontal distances 𝑟 of 50, and 300 far from the
lightning channel base. To this aim, we present the vertical and horizontal component of the
electric field (i.e. 𝐸𝑧 , 𝐸𝑟 ) and the horizontal component of the magnetic field (i.e. 𝐻𝑟 ) under the
ground surface.
As the transient MoM technique [3] is not able to take of the presence of a soil with finite resistivity
into account, a two dimensional finite element based technique (FEM) taken from [9] is
considered for purposes of the model validation. In the FEM study, the modified transmission
line with exponential decay (MTLE) is adopted for modeling the lightning return stroke channel
with a current height decay of 𝛼 = 2000 m and assuming a return stroke speed of 𝑣 = 1.3 ×
108 m/s [9]. According to the MTLE model, the current distribution along the channel is
expressed as
𝑧 𝑧
𝑖(𝑧, 𝑡) = 𝑖 (0, 𝑡 − ) 𝑒 −𝑣 (11)
𝑣
where 𝑖(𝑧, 𝑡) is the channel current at height 𝑧.
Two values of resistivity are chosen for the purpose of comparisons. The first value is selected
as 𝜌 = 100 Ω. m , representing a relatively low resistivity soil and the second value is selected
as 𝜌 = 1000 Ω. m, representing a higher resistivity soil. For both cases, the value of relative
permittivity 𝜀𝑟 of the soil is set to 10. Figures 8 and 9 show the electromagnetic fields at a depth
of 1 m from the soil interface for 𝜌 = 100 Ω. m and 𝜌 = 1000 Ω. m, respectively. Comparison of the
results obtained with HIFREQ and those obtained with the finite element technique taken from
[9] demonstrates a fairly good agreement between the two models.
The study of [9] is only limited to a two-dimensional finite element analysis across the soil-air
interface. For an accurate and realistic three-dimensional (3-D) finite element modeling, the open
region problems should be confined to a sufficiently large region of analysis that can also limit the
number of unknowns to a manageable size. In addition, appropriate artificial boundary conditions
should be applied to prevent the waves from being reflected by the boundaries. Considering the
fact that for such 3D open region studies FEM simulation would result in a large and inefficient
matrix system, the presented HIFREQ approach is computationally significantly efficient for
analyzing lightning radiated electromagnetic fields.

r=50m r=300m

35 5
30
4
25
20 3
15 2
10
1
5
0 0
0 2 4 6 8 10 0 2 4 6 8 10
r=50m r=300m
100 10
Horizontal electric field (V/m)

0 0
0 2 4 6 8 10 0 2 4 6 8 10
-10
-100
-20
-200
-30
-300
-40
-400 -50
-500 -60
r=50m r=300m
35 3.5
30
2.8
25
20 2.1
15 1.4
10
0.7
5
0 0
0 2 4 6 8 10
-5 0 2 4 6 8 10 -0.7
AT model in HIFREQ MTLE model in FEM [9]
Figure 8: Underground electromagnetic fields - 𝜌 = 100 Ω. 𝑚, 𝜀𝑟 = 10, 𝑑𝑖 = 1 m.

r=50m r=300m
40 5

35
4
30
25 3
20
2
15
10 1
5
0
0 0 2 4 6 8 10
0 2 4 6 8 10 -1
r=50m r=300m
200 50

0
-200 0 2 4 6 8 10 0
0 2 4 6 8 10
-400
-50
-600
-800
-100
-1000
-1200 -150
-1400
-1600 -200
r=50m r=300m
350 35
Vertical electric field (v/m)
300 30
250 25
200 20
150 15
100 10
50 5
0 0
-50 0 2 4 6 8 10 -5 0 2 4 6 8 10
AT model in HIFREQ MTLE model in FEM [9]
Figure 9: Underground electromagnetic fields - 𝜌 = 1000 Ω. m, 𝜀𝑟 = 10, 𝑑𝑖 = 1 m.

5.1.3 Lightning-induced voltages on overhead lines
Figure 10 shows the configuration of the lightning channel and its position relative to a single-
phase transmission line. The length of the transmission line, which is placed at a height of 10 m
above the ground, is assumed to be 1000 m and its characteristic impedance is 567 Ω. The
lightning channel is positioned at distance of 50 m from the center of the transmission line. A
homogeneous soil with resistivity of 0.001 Ω. m (essentially, a perfect ground plane) and a
dielectric constant of 1 is considered in the simulation procedure.
Figure 10: Lightning channel and the transmission line configuration.
For the evaluation of the lightning-induced voltage level at a specified location on the line, first, a
vertical conductor with a resistance of 1 MΩ, connecting that location to the ground, is considered.
Then, the induced voltage is obtained by multiplying the current crossing that conductor by the
resistance value.
Figure 11 shows the level of the induced voltages at three different points (𝑥0 = 0 m, 𝑥0 = −250 m
and 𝑥0 = −500 m) of the transmission line when 𝑍𝐿𝑜𝑎𝑑 = 567 Ω (characteristic impedance of the
line). To show the propagation effects, including reflections from both ends of the transmission
line, the level of the induced voltages for the case of the line terminated by 𝑍𝐿𝑜𝑎𝑑 = 56 Ω
(mismatched conditions) is also presented. In order to assess the validity of the lightning channel
model used in HIFREQ, the results are compared to those of [10], which are based on a
combination of the MTLE model of lightning channel and a field-to-transmission line coupling
model. The comparison demonstrate a very good agreement between the two models.

𝑍𝐿𝑜𝑎𝑑 = 𝑍𝑐 = 567 Ω 𝑍𝐿𝑜𝑎𝑑 = 56 Ω
x0 = 0m X0 = 0m
100 120
90
80
Induced voltage (KV)

60
60
30
40 0
-30 0 2 4 6 8
20
-60
0
-90
0 2 4 6 8
-20 -120
x0 = -250m X0 = -250m
100 120
90
80
60
60
30
40 0
20 -30 0 2 4 6 8
-60
0
-90
0 2 4 6 8
-20 -120
x0 = -500m X0 = -500m
100 30
80 20
60 10
40 0
0 2 4 6 8
20 -10
0 -20
0 2 4 6 8
-20 -30
AT model in HIFREQ MTLE model in transmission line theory [10]
Figure 11: The level of induced voltages at various locations of the transmission line.

5.2 Modelling lightning strikes to structures
The presence of an elevated strike object can also be included in the AT model of lightning [11].
As the AT model of the lightning channel is based on a network of thin wires, it can be used by
HIFREQ for the electromagnetic analysis of buildings, towers, etc., hit by lightning.
To this end, the wire model of the lightning channel described in previous chapter has to be added
to the wire mesh model of the structure hit by lightning at the strike point.
As an example, the telecommunication tower and grounding system described in [12] are
considered. A detailed description of the geometry of the tower and of its wire network model can
be found in [12]. The wire structure of Section 3 representing the lightning channel is also added
on top of the tower. First, the response of telecommunication tower when it is hit by lightning is
studied, then the neighbouring structures are added and considered in the analysis. Two different
neighbouring structures around the tower are examined. The first structure is the transmission
line described in the previous section (shown in Figure 14) while the second one is a flat
rectangular metallic plate (shown in Figure 16). In both cases, we compare the AT model of
lightning to the case where the lightning channel is not modelled and the lightning current is
injected directly at the top of the tower using a Lead energization, which was the approach used
in [12].
5.2.1 Isolated telecommunication tower hit by lightning
In this section, in order to assess the differences between the lightning channel and the Lead
energization procedures, the electromagnetic fields around the individual telecommunication
tower hit by lightning are investigated. The computations are performed at a horizontal distance
𝑟 = 60 m from the tower, and at a height 𝑑𝑒 = 0.5 m above the soil interface. Next, a comparison
is made between the two energization procedures in terms of body current and GPR for a person
touching the tower.
Figure 12 depicts the horizontal and vertical components of electric and magnetic fields for two
soil resistivity values, namely 100 Ω. m and 1000 Ω. m. It can be seen from the figure that the two
procedures for representing the lightning current predict the same level of magnetic fields and
horizontal electric fields, while the difference between the predicted vertical electric fields by the
two models is noticeable. Thus, the significant differences observed between horizontal electric
field levels predicted by the two models of lightning are not important for practical applications.
For a high resistivity soil, the level of the horizontal component of the electric field becomes
comparable to that of the vertical component. The two models of lightning predict similar
horizontal electric fields but the vertical electric field levels are different.
Note that since the main difference between the lightning channel and Lead energization
approaches is restricted to the vertical component of the electric field induced by the lightning
channel section, major differences between their inducing effects are expected only when the
coupling between the tower hit by lightning and surrounding structures is mainly due to the
vertical component of the electric field.

𝜌 = 100 Ω. m , 𝜀𝑟 = 10 𝜌 = 1000 Ω. m, 𝜀𝑟 = 10
1000 2500
Horizontal electric field

Horizontal electric field
800 2000
600 1500
(V/m)
(V/m)
400 1000
200 500
0 0
0 2 4 6 8 0 2 Time (Microsecond)
4 6 8
Time (Microsecond)
20000 20000
Vertical electric field
Vertical electric field

15000 15000
(V/m)
10000 10000
(V/m)
5000 5000
0 0
-5000 0 2 4 6 8 -5000 0 2 4 6 8
Time (Micosecond)
Time (Microsecond)
50 40
Horizontal magnetic field
Horizontal magnetic field
40 30
30
(A/m)
20
(A/m)
20
10 10
0 0
0 2 4 6 8 0 2 4 6 8
Lightning channel Lead energization
Figure 12: Electromagnetic fields around a tower hit by lightning.
Figure 13 depicts the body current for a person touching the tower hit by lightning and the GPR
of the tower at the point being touched. Two cases of soil with low resistivity value (𝜌 = 100 Ω. m)
and high resistivity value ( 𝜌 = 1000 Ω. m) are chosen for the analysis. As the conductive coupling
is dominant, the two energization procedures (namely, lightning channel and Lead energization)
provide similar results without regard to the value of soil resistivity.

𝜌 = 100 Ω. m , 𝜀𝑟 = 10 𝜌 = 1000 Ω. m, 𝜀𝑟 = 10
20 50
Body current (A)
Body current (A)

0 0
-20 0 2 4 6 8 0 2 4 6 8
-50
-40
-60 -100
-80 -150
100 500
0
GPR (KV)
GPR (KV)
0
0 2 4 6 8
-100 0 2 4 6 8
-500
-200
-300 -1000
Lightning channel Lead energization
Figure 13: Body current and GPR for various soil resistivity.
5.2.2 Transmission line near a telecommunication tower hit by lightning
This section examines the inducing effects of the tower hit by lightning on a nearby single-phase
transmission line. Figure 14 shows the configuration of the telecommunication tower and its
position relative to the transmission line. The value of the soil resistivity is set to 100 Ω. m and its
relative permittivity is set to 10.
Figure 15 shows the level of induced voltages at three different points (𝑥0 = 0 m, 𝑥0 = −250 m
and 𝑥0 = −500 m) of the transmission line when 𝑍𝐿𝑜𝑎𝑑 = 567 Ω (characteristic impedance of the
line), when the lightning current is modeled using the two energization procedures (lightning
channel and Lead energization).
Due to the absence of the channel, the induced voltage predicted by Lead energization procedure
is much lower than that predicted when modeling the lightning channel. Also, a comparison
between the induced voltages on the line for the case where only the lightning channel is modeled
(without the presence of the tower, Fig. 11) with those where the tower is present (Figure 14)
demonstrates that the presence of the tower adds a few resonances but does not affect drastically
the level of induced voltages. The resonances in question are mainly due to the multiple reflections
of the surge current at the extremities of the tower.

Figure 14: Transmission line near a telecommunication tower hit by lightning.
5.2.3 Metallic plates near a telecommunication tower hit by lightning
The study of electromagnetic interference (EMI) on solid structures located near towers hit by
lightning is of a great interest for the electromagnetic compatibility (EMC) community. These
solid structures could have diverse shapes, varying from a truck to a solar farm. In all cases, the
solid structures can be considered as a juxtaposition of metallic plates [13] and [14]. Indeed, the
basic shape of such structures is a rectangular plate. Consequently, in the following, the analysis
will be restricted to such a basic shape. Figure 16 shows the configuration of telecommunication
tower hit by lightning and its position relative to a metallic plate. The 5 × 5 m square plate is
located at a horizontal distance 𝑟 = 𝑦1 = 60 m from the tower and at a height 𝑑𝑒 = 𝑧1 = 0.6 m
above the soil interface. All analyses are performed for two configurations of the plate namely,
horizontal and vertical.

x0=0 m
200
Induced Voltage (kV)

150
100
Channel
50
Lead
0
0 2 4 6 8
-50
Time (Microsecond)
x0=-250m
100
Induced voltage (kV)
80
60
40
Channel
20
Lead
0
-20 0 2 4 6 8
-40
Time (Microsecond)
x0=-500m
80
60
Induced voltage (kV)
40
20
Channel
0
Lead
-20 0 2 4 6 8
-40
-60
Time (Microsecond)
Figure 15: The level of induced voltages at various locations along the transmission line.

Model 1: Lightning channel
Model 2: Lead energization
Energization
Figure 16: Structure represented by a metallic plate near a telecommunication tower hit by lightning.
The body current for a person touching the plate and the plate GPR at the touched point are used
as an evaluation of the difference between the lightning channel and the Lead energization
procedure. Figure 17 and Figure 18 depict the body current, and the GPR for the horizontal and
vertical configurations of the plate, respectively. All computations are performed for different soil
resistivity values varying from 0.001 Ω. m (perfect ground plane) to 1000 Ω. m, with the value of
the relative permittivity set to10. The following remarks are applicable:
1. The multiple reflections of the surge current at the tower extremities are the origin of the
oscillations in all plots.
2. As the effects of the vertical component of the electric field are dominating the coupling
between the tower and the metallic plate, the body current is more pronounced for the
case of the vertical plate configuration.

3. The level of body current predicted by the Lead energization procedure is smaller than that
for the lightning channel model.
4. At late times, the GPR levels of the plate are always larger for the lightning channel model
than those predicted by the Lead energization model.
Figure 17: Body current and plate GPR for various soil resistivity - Horizontal configuration.

Figure 18: Body current and plate GPR for various soil resistivity - Vertical configuration.
6 Conclusion
A technique for modeling a return stroke lightning channel using HIFREQ and FFTSES was
presented. This technique was applied to the study of the electromagnetic fields generated by
lightning, and to induction effects on a nearby transmission line and neighboring structures when
lightning hits the ground or a communication tower. The computation results were compared to
published work, with excellent agreement. The results obtained when modeling the return stroke
lightning channel were also compared to the case where the lightning current is specified by a
direct Lead energization at the point of strike. This comparison showed that the presence of the
lightning channel can have a significant effect on the results when induction effects are important,
but less so when the phenomena being studied are dominated by conductive effects.
7 References
[1] V. A. Rakov and M.A. Uman, “Lightning Physics and Effects,” Cambridge University Press, 2003.
[2] V. Cooray, “Lightning Electromagnetics,” IET Power and Energy series, 2012.
[3] R. Moini, B. Kordi, G.Z. Rafi and V.A. Rakov, “A New Lightning Return-Stroke Model Based on Antenna
Theory Model,” J. Geophys. Res., Vol. 105, 2000.
[4] F. Heidler, “Traveling Current Source Model for LEMP Calculation,” in Proc. 6th Int. Symp. Tech.
Exhibition on Electromagn. Compat., Zurich, Switzerland, 1985.
[5] K. Berger, R.B. Anderson and H. Kroninger, “Parameters of Lightning falshes,” Electra, no.1, 1975.
[6] V. Kodali, V.A. Rakov, M.A. Uman, K.J Rambo, G.H. Schentzer and J. Schoene, “Triggered-Lightning
Properties Inferred from Measured Currents and Very Closed Electric Fields,” Atmos. Res., Vol.76, 2005.
[7] S. Bonyadi-Ram, R.Moini, S.H.H. Sadeghi, and V.A. Rakov, “On representation of Lightning Return-
Stroke as a Lossy Monopole Antenna with Inductive Loading,” IEEE Trans. Electromagnet. Compat.,
Vol. 50, No.1, 2007.
[8] “CDEGS 15.4.8631”, Safe Engineering Services & Technologies ltd. Laval, Québec Canada, 2016
[Online]. Available: www.sestech.com.
[9] J. Paknahad, K. Sheshyekani, F. Rachidi, and M. Paolone, “Lightning Electromagnetic Fields Properties
and Their Induced Currents on Buried Cables. Part II: The Effect of a Horizontally Stratified Ground,”
IEEE Trans. Electromagnet. Compat., Vol. 56, No.5, 2014.
[10] C.A. Nucci, F. Rachidi, M.V. Ianoz and C. Mazzetti, “Lightning-Induced Voltages on Overhead Lines,”
IEEE Trans. Electromagnet. Compat., Vol. 35, No.1, 1993.
[11] B. Kordi, R.Moini, W. Janischewskyj, A.M. Hussein, V. Shostak and V.A. Rakov, “Application of the
Antenna Theory Model to a Tall Tower Struck by Lightning,” J. Geophys. Res., Vol. 108, 2003.
[12] “Lightning Transient Study of a Communication Tower,” How To….Engineering Guide, Safe
Engineering Services & Technologies ltd. Laval, Québec Canada, 2016.
[13] S. Ladan, A. Aghabarati, R. Moini, S. Fortin, and F.P. Dawalibi, “Induced Disturbances by High Voltage
Transmission Lines on Nearby Stationary Vehicles,” accepted to be presented in IEEE Int. Symp. on
Electromagnet. Compat., Ottawa, 2016.
[14] A. Aghabarati, R. Moini, S. Ladan, S. Fortin, and F.P. Dawalibi, “Electromagnetic Shielding Properties
of Spherical Polyhedral Structures Generated by Conducting Wires and Metallic Surfaces,” accepted to
be presented in IEEE Int. Symp. on Electromagnet. Compat., Ottawa, 2016.

11 ADVANCEMENTS IN THE TREATMENT OF NONLINEAR

DEVICES IN THE FREQUENCY DOMAIN
Stéphane Franiatte(1,2), Octavio Ramos(1), Marc-André Joyal(1), Simon Fortin(1),

Jean-Marc Lina(2), Alain April(2) and Farid P. Dawalibi(1)
(1) Safe Engineering Services & Technologies ltd. (2) École de Technologie Supérieure
Email: info@sestech.com, Web Site: www.sestech.com Web Site:www.etsmtl.ca
Abstract
Frequency Domain (FD) methods are a valuable complement to Time Domain (TD) methods for the analysis of
electromagnetic transients in power systems. However, it is still difficult to include non-linear elements in the FD.
This article presents some improvements on a technique developed several years ago for analyzing systems
including non-linear elements in the FD. The article focuses on the modeling of surge arresters on distribution lines
for the analysis of insulation failure due to lightning. This work includes improvements to the Simulated Annealing
metaheuristic (SA) technique presented last year (solutions are found using realistic MOV arresters parameters)
and an example is given where results obtained by the proposed method are compared to those obtained with the
widely accepted Time Domain tool ATP-EMTP.
1 Introduction
When studying the lightning performance of surge arresters using time domain (TD) solvers such
as ATP-EMTP, an equivalent electric circuit including the surge arresters is built to represent the
transmission line system under study. A lightning strike on the line is simulated and the circuit is
solved entirely in the time domain. Hence, the frequency dependence of the elements in the circuit
has to be obtained by other methods or even neglected in some cases. Approximations are usually
required in computing the equivalent impedances in the circuit.
Fortin et al. [1] proposed an alternative approach, studying the lightning performance of surge
arresters directly in the frequency domain (FD) and solving the complete network in the frequency
domain. Once the response (voltage and current) of the network including surge arresters is
obtained, the voltage and current through the surge arresters in the time domain are obtained
using the inverse Fourier transform.
The importance of developing Frequency Domain methods for the analysis of Electromagnetic
Transients resides in the fact that most power system elements are frequency dependent and these
can be modeled in a straightforward manner within FD methods. Since the basic principles of FD
methods are different from those of the Time Domain, FD methods are very useful in verifying
TD methodologies.
This frequency-based approach requires solving of a large number of coupled non-linear
equations. Last year, an automated method for finding solutions of the nonlinear system of
equations through stochastic global optimization was proposed [2]. The original problem as
stated in [1] (nonlinear equations system solving) was transformed into a global optimization one
by synthesizing objective functions whose global minima, if they exist, are also solutions to the
original system. The global optimization task was carried out by the stochastic method known as
Simulated Annealing (SA). This article describes progress in using this method. Tests of the
approach for realistic values of alpha are carried out, and the results are compared with those
obtained using a well-known Time Domain solver named ATP-EMTP.
2 Mathematical Model
2.1 Time Domain Compensation Method
The Time Domain compensation method is used in time domain solvers such as ATP-EMTP to
solve networks containing nonlinear elements. According to the Compensation Theorem, any
network element can be replaced by a current source whose current is equal to the current flowing
through the replaced element. This theorem is applied in [2] alongside an LU decomposition of
the network’s nodal admittance matrix to simulate the effects of nonlinear elements, without the
need of re-computing the admittance matrix. To take into account the effects of the nonlinear
element, the following equation system is solved first, for each time step:
(0)
𝑣𝑘𝑚 = 𝑣𝑘𝑚 + 𝑧𝑇 𝑖𝑘𝑚 (1)
𝑑𝑖𝑘𝑚 (2)
𝑣𝑘𝑚 = 𝑓(𝑖𝑘𝑚 , , 𝑡)
𝑑𝑡
(0)
where 𝑣𝑘𝑚 indicates the voltage drop between nodes 𝑘 and 𝑚 without the nonlinear element while
𝑧𝑇 is the Thevenin equivalent impedance (Figure 1).
Note: In this article the following conventions are followed in formulas; lowercase letters are used
to represent time domain variables, uppercase letters are used to represent frequency domain
variables, boldface letters are used to represent vectors or matrices and normal letters are used to
represent scalar values.
Equations (1) and (2) are solved simultaneously for 𝑖𝑘𝑚 , the current through the nonlinear branch.
The final voltage solution is found by superposing the voltage drop obtained when applying the
non-linear current source 𝑖𝑘𝑚 :
𝒗 = 𝒗(0) + 𝒛 ∙ 𝑖𝑘𝑚 (3)
As shown in [3], the compensation method can also be used with 𝑀 parallel nonlinear branches:
𝑀
𝒗=𝒗 (0)
+ ∑ 𝒛(𝑗) ∙ 𝑖 (𝑗) (4)
𝑗=1
Figure 1 Compensation method: left) linear network without the nonlinear element; right) the equivalent
circuit.

2.2 Frequency Domain Compensation Method
To fully take into account frequency dependence of the elements of the network, a new Frequency
Domain Compensation Method was proposed in [1]. The model for the network model is stated
directly in the frequency domain, and can be seen as the frequency counterpart to Eq. (4):
𝑀
(0)
𝑽𝑖 (𝜔) = 𝑽𝑖 (𝜔) + ∑ 𝒁(𝑖,𝑗) (𝜔) ∙ 𝑰(𝑗) (𝜔) (5)
𝑗=1
For a network containing 𝑀 nonlinear elements, the voltage across the 𝑖 𝑡ℎ nonlinear element is
(0)
obtained from this equation. 𝒁 and 𝑽𝑖 are frequency dependent constants. The current 𝑰(𝑗) is
obtained from the model H for the nonlinear element by applying a Fourier transform:
𝒊(𝑡) = 𝐻(𝒗(𝑡), 𝑡) ↔ 𝒗(𝑡) = 𝐻 −1 (𝒊(𝑡), 𝑡), 𝒊, 𝒗 ∈ ℝ𝑁 (6)
A self-consistency equation is then derived by combining the linear Eq. (5) with the nonlinear
element’s equation (6):
𝑀
(0)
𝒗𝑖 (𝑡) = ℱ −1
[𝑽𝑖 (𝜔) + ∑ 𝒁(𝑖,𝑗) (𝜔) ∙ ℱ[𝐻 (𝑗) (𝒗𝑗 (𝑡), 𝑡)]],
𝑗=1 (7)
𝒗 ∈ ℝ𝑁
Eq. (7) represents a self-consistency condition on the voltage across the nonlinear elements, and
can be solved iteratively. A suitable parameterization of 𝒗𝑖 as a function of time must first be
obtained, then the parameters can be adjusted to satisfy equation (7); this can be stated as a root-
finding problem:
𝑀
(0)
𝑓𝑖 (𝒗) = 𝒗∗,𝑖 − ℱ −1
[𝑽∗,𝑖 (𝜔) + ∑ 𝒁(𝑖,𝑗) (𝜔) ∙ ℱ[𝐻 (𝑗) (𝒗∗,𝑗 )]] = 0,
𝑗=1 (8)
𝒗 ∈ ℝ𝑁×𝑀
Considering the set of 𝑀 nonlinear elements in the network for which (8) needs to be solved, a
system of nonlinear equations is constituted:
𝑓1 (𝒗)
𝑓2 (𝒗)
𝐹(𝒗) = [ ] = 0, v∈ ℝ𝑁×𝑀 (9)
⋮
𝑓𝑀 (𝒗)
The nonlinear equations system solving task is transformed into an optimization problem by
synthesizing objective functions whose global minima, if they exist, are also solutions to the
original system. The Simulated Annealing metaheuristic method [4] is used to solve this system.

3 Application to Surge Arrester Lightning Performance

Analysis
3.1 System Description
The case presented here is the same as that presented in [1]. Figure 2 shows the simulation circuit.
The circuit under test is composed of 7 poles with an earth impedance of 24 ohms at power
frequency. The phase conductors are 4/0 ACSR with a radius of 7.15 mm. The neutral conductor
is a 1/0 ACSR conductor with a 5.05 mm radius. Surge arresters are located at poles #2 to #6.
Poles #1 and # 7 are terminated with the line surge impedance value of 555 ohm. As was done in
[1], we study the case in which lightning strikes Phase C at mid-span between Poles #4 and #5. It
is considered that phases C and A are more susceptible to lightning strikes due to their positions.
To reduce computation time, we have modeled only Phase C in the study, since the transient
behavior is dominated by this phase during the lightning strike.
Figure 2 Seven distribution poles modeled.
The nonlinear elements are represented as gapless metal oxide varistors (MOV). The MOV
nonlinear V-I relationship, corresponding to Eq. (6), is expressed by the following equation:
𝛼
𝒗(𝑡) (10)
𝒊(𝑡) = 𝐻(𝒗(𝑡), 𝑡) = 𝑖𝑠 ( ) , 𝒊, 𝒗 ∈ ℝ𝑁
𝑣𝑠
In Eq. (10), , sometimes referred to as squareness, can be different for different arresters. Typical
values for  lie between 2 and 75. The constant value 𝑣𝑠 represents the voltage threshold of the
device: the arrester begins to conduct significantly when the voltage across its terminals reaches
or approaches this value. The value 𝑖𝑠 is the amount of current flowing through the device at the
voltage threshold .Figure 3 displays the V-I characteristic of a typical MOV arrester for several 𝛼
values, with 𝑖𝑠 = 700 𝐴 and 𝑣𝑠 = 15 000 V.

Figure 3 Typical MOV arrester V-I characteristic for different 𝛼 values.
The lightning surge is modeled as an ideal current source characterized by a 1.2/20s wave; two
lightning models were studied, the double-exponential, Eq. (11), and the more widely used Heidler
model [5], equation (12):
𝑖(𝑡) = 𝑖𝑚 (𝑒 −𝑎𝑡 − 𝑒 −𝑏𝑡 ) (11)
𝑖(𝑡) = 𝐴 ∙ 𝐺(𝑡, 𝜏1𝐴 , 𝜏2𝐴 ) + 𝐵 ∙ 𝐺(𝑡, 𝜏1𝐵 , 𝜏2𝐵 )

1 (𝑡⁄𝜏1 )2 −𝑡
𝐺(𝑡, 𝜏1 , 𝜏2 ) = ∙ ∙ 𝑒 ⁄𝜏2 (12)
𝛾 1 + (𝑡⁄𝜏 )2
1
𝜏
−√2 1⁄𝜏2
𝛾=𝑒
Figure 4 Heidler lightning surge.

3.2 Frequency Domain Solver: HIFREQ and Simulated Annealing
The HIFREQ and FFTSES engineering modules of the CDEGS software package are used to
perform the computations. The HIFREQ module is used to compute the currents and voltages
throughout the conductor network in the frequency domain, while the FFTSES module is used to
convert the results into the time domain. The Simulated Annealing approach to solve nonlinear
Eq. (9) is programmed in Matlab.
Since last year [2], improvements have been made in the convergence capabilities of the
Simulated Annealing implementation. Solutions can be found for realistic 𝛼 values up to 45.
Figure 5 and 6 below display results for 𝛼 = 25 and 𝛼 = 45 , respectively. These results were
obtained using an input signal based on the Heidler model of Eq. 12, with 𝐴 = 25 kA.
5-A
Voltage across the
arresters.
It can be seen here that
voltages are properly
clamped around the value
of 15 kV.
5-B
Current through the
arresters.
Figure 5 Voltage and Current solutions, 𝛼 = 25.

6-A
Voltage across the
arresters.
6-B
Current through the
arresters.
Figure 6 Voltage and Current solutions, α = 45.
3.3 Time Domain Solver
In order to validate the proposed methodology, the widely accepted Time Domain solver ATP-
EMTP is used to reproduce the experiment. The ATP circuit model is presented in Figure 7. MOV
surge arresters are modeled as nonlinear resistances given that it was impossible to utilize the
available MOV models in ATP with the given nonlinear characteristics and time step.
Transmission lines were modeled using the frequency dependent Marti model with a frequency
range of 0.01 Hz up to 100 MHz with a selected modal conversion matrix at 10 MHz. The
simulation time step was selected to be 0.1 ns. Notice that the smallest travel time in the system
is of 0.145 s.

I
V
I
LCC LCC LCC LCC LCC LCC LCC
I
R(i)
R(i)
R(i)
R(i)
R(i)
0.087 km 0.087 km 0.087 km 0.043 km 0.043 km 0.087 km 0.087 km
Figure 7. ATP-EMTP simulation circuit.
Results obtained with this approach are compared with the ones obtained with the Frequency
Domain method in the next section.
4 Comparison between time and frequency domain

approaches
Figure 8 shows the results with both methodologies when the lighting surge is of a double
exponential form. The Frequency Domain results are plotted in black and the Time Domain
results are plotted in colors. All signals present a similar rise time and a similar oscillation
frequency of 1.7 MHz. This value corresponds to a resonance that is seen when reflections occur
at the arresters. However, the Time Domain results present less attenuation for current in the
arresters at poles 4 and 5 resulting in a difference of up to 3 kA.
Figure 8 Results for a double-exponential lightning surge, 25 kA input signal. Results obtained with the
Frequency Domain method are in black and results obtained with the Time Domain method are in color.
Figure 9 presents the current for the arrester at pole 5 which is similar to the current for the
arrester at pole 4. In this case, the results from the Frequency Domain method present more
attenuation than those obtained with the Time Domain method. Figure 10 presents a close up on
Figure 9 where it can be seen that the current difference between the two methodologies is around

3 kA. On the other hand, the rise time and the oscillation frequency of the two signals, (Time
Domain and Frequency Domain), are very similar. This resonance frequency of about 1.7 MHz
corresponds to a half wavelength separation between the arresters, causing reflected waves to be
sustained during the period of the surge.
Figure 11 to Figure 16 present comparisons for arresters at poles 2, 3 and 6. In general these figures
show that the Time Domain results for these arresters behave in the opposite fashion to those
obtained with the Frequency Domain method regarding the magnitude of oscillations, i.e., the
Time Domain signals present oscillations of smaller magnitude in the first microseconds and
oscillations of larger magnitude in the middle part of the time window, whereas the Frequency
Domain results present larger magnitudes at the beginning of the signal and attenuate as time
goes by.
Due to the fundamental differences between the Time Domain and Frequency Domain methods,
multiple causes could explain the discrepancies stated above. One of them is the ability of the
proposed Frequency Domain method to take the frequency dependence of the network into
account, which is something more difficult for Time Domain methods. For instance, the
impedance of the poles in the ATP software was set to a constant value of 24 Ω given that Time
Domain methods do not have dedicated models capable of representing grounding structures for
a wide range of frequencies. On the other hand, since the actual poles were modeled with HIFREQ
in the proposed method, their frequency dependence was fully taken into account. As an
illustrative example, Figures 17 and 18 show the resistance and the reactance of one pole with
respect to the frequency. It is seen that for very low frequencies, the pole impedance is about the
same as that used in ATP, but for higher frequencies, the differences are more pronounced. As a
matter of fact, an important frequency to look at is the system’s resonance frequency of 1.7 MHz.
At that frequency, the amplitudes of the oscillations are highly dependent on the reflection
coefficient at the arresters, which is determined from the pole impedance assuming that the
arresters impedance is negligible when the lightning strike occurs. By looking more closely at
these results, we find that at this resonance frequency, the pole resistance is 84.4 Ω whereas its
reactance is 261j Ω. Similar observations could be done with all components included in the
network, showing the potential lack of accuracy of Time Domain methods with respect to
Frequency Domain methods.

Figure 9. Current trough arrester 5.
Figure 10. Close up on Figure 9.




Figure 17 Resistance of the pole with respect to the frequency, computed with HIFREQ.
Figure 18 Reactance of the pole with respect to the frequency, computed with HIFREQ.

5 Conclusion
In this article a technique aiming to solve nonlinear elements in the Frequency Domain first
introduced in [1] has been improved. A difficulty encountered by Fortin et al. in their work to solve
networks containing nonlinear devices in the frequency domain was related to the method used
to solve the self-consistency equation (7). The optimization algorithm used to solve the system
often failed to converge for high values of the nonlinearity constant 𝛼, and required the user to
frequently update some parameters to eventually find a solution, even for lesser values of 𝛼. A
new approach based on the Simulated Annealing metaheuristic technique was presented in [2],
and improved since then: as demonstrated in this article, solutions were obtained for 𝛼 values as
high as 45 and no human intervention during the convergence process was required. The
proposed Frequency Domain method was implemented and compared to a widely accepted Time
Domain method based on travelling waves, i.e. ATP-EMTP. Comparisons between results
obtained with both approaches show that the results obtained with the proposed method match
the general behavior of those obtained with the Time Domain tool. The method presented here is
still a work in progress and it is expected to be refined in the years to come. This work represents
a first step towards the accurate solution of nonlinear elements in the Frequency Domain, a task
that has proven to be challenging.
6 References
[1] S. Fortin, W. Ruan, F. P. Dawalibi, and J. Ma, "Optimum and Economical Deployment
Method of Surge Arresters on Distribution Lines for Insulation Failure due to Lightning -
An Electromagnetic Field Computation Analysis," 2002.
[2] S. Franiatte, S. Fortin, J.-M. Lina, M.-A. Joyal, and F. P. Dawalibi, "Treatment of
Nonlinear Devices in the Frequency Domain," presented at the SES Users Group
Conference, San Diego, California, USA, 2015.
[3] H. W. Dommel, "Nonlinear and Time-Varying Elements in Digital Simulation of
Electromagnetic Transients," Power Apparatus and Systems, IEEE Transactions on, vol.
PAS-90, pp. 2561-2567, 1971.
[4] S. Kirkpatrick, C. D. Gelatt, and M. P. Vecchi, "Optimization by Simulated Annealing,"
Science, vol. 220, pp. 671-680, 1983.
[5] F. Heidler and J. Cvetic, "A Class of Analytical Functions to Study the Lightning Effects
Associated With the Current Front," European Transactions on Electrical Power, vol. 12,
pp. 141-150, 2002.

Part III: New Features
12 IMPROVEMENTS IN SES SOFTWARE
Simon Fortin
Safe Engineering Services and Technologies ltd.

Abstract
This article describes the improvements made to the SES software program since the 2015 CDEGS User’s Group
Meeting.
1 Introduction
This article briefly describes the main improvements that were made to SES software since the
2015 CDEGS User’s Group Conference.
2 New User Interface Programs

The development of the new user-interface programs for SES Software, undertaken a few years
ago, has continued at a good pace. This version of the software includes the first release of
several modules and components and also includes improvements to modules introduced for
the first time last year.
2.1 CDEGS
First and foremost, the CDEGS main interface that acts as a front-end for packages in the
MultiGround, MultiGroundZ, MultiFields, and CDEGS families has been updated to adopt the
new look and feel that is now standard for SES programs. This new version of CDEGS supports
all the features of the original version of CDEGS, and can be used interchangeably with that
program. It also adds some interesting enhancements, in particular with respect to the selection
of the Session Mode and with the management of the JobID list.
The article “A New Interface for CDEGS” describes this new program in greater detail.

PART III: NEW FEATURES
2.2 SESResap
This is the first (Beta) release of the SESResap module. This module acts as a complete
replacement for CDEGS – Specify for RESAP. It can load and save all existing RESAP files.
Cases can also be run from within the program and both the input data and the computation
results can be visualized from this same environment.
The new program also includes enhancements compared to CDEGS – Specify:
 Clearer schematics with electrode labelling that follows popular conventions from
instrument manufacturers,
 Measurements data grid with optional columns that help verify the input data,
 Possibility of adding comments for any data point,
 Interactive plot of the data,
 Clearer specification of analysis parameters,
 Useful field data sheet that helps achieve quality measurements.
The article “SESResap: A New Soil Resistivity Measurement Editor” gives more details.

2.3 SESFFT
This is also the first Beta release of the SESFFT module, the new user-interface module for the
FFTSES engineering program. As is the case for SESResap, this new program offers full support
of existing FFTSES files and can be used instead of CDEGS – Specify for FFTSES. As for all
new SES Software user-interface programs, SESFFT offers a fully integrated work environment
for FFTSES. It allows you to edit and visualize input data, run the computations (in both
Forward and Inverse mode) and visualize the computation results. It also includes tools to
help build the FFTSES Input Databases starting from HIFREQ or MALZ computation
databases.
The new program also includes several new features and improvements compared to CDEGS –
Specify, including the direct visualization of the frequency spectrum of the input signal in
Forward mode, the specification of window or regularization functions to smooth out time-
domain signals and attenuate undesired high-frequency components, and others.

Consult the article “A New Interface for FFTSES” for more details.
2.4 Soil Model Editor
Another addition to SES Software is the new Soil Model Editor module. This module, which
can be used as a stand-alone program but is also used as a component in several other SES
Software programs, allows editing the various soil structures that can be used in SES Software.
It can currently be used to edit soil structure data for the MALT, MALZ, HIFREQ, and TRALIN
programs but will eventually be capable of performing this task for all SES Software programs
requiring soil structure data, making it easier to share soil structure data between those
programs. It is currently integrated into the SESTRALIN and SESCAD programs.
See “SoilModelEditor: New SES Software Tool for Soil Model Design” for more details about
this new module.

2.5 SES Object Library
SES Object Library is another new program included in this version of the software. This
program allows you to inspect the properties of a large number of components that can be part
of models for many SES Software engineering programs. It currently includes a comprehensive
database of conductors as well as several power cables; many other components will be added to
the library in the near future.
The user-interface of the program makes it easy to search the library for desired items. The
available items can be filtered in many different ways and sorted according to any of their
properties. A large number of properties are available to help identify the most suitable items for
the task at hand.
This module is currently available as a stand-alone program but will eventually be integrated as
a component in several other SES programs, where it will act as a replacement for the existing
SES Conductor Database and other libraries currently available in the software.
Consult the article “Introducing the New SES Object Library” for more details.

2.6 SESPlotViewer and the SES Plotting Engine
SESPlotViewer is a new SES Software program that can display data as plots of various types.
The program also displays that same data in tabular format, and optionally allows editing it.
The main use of this program is as a plotter for the computation results produced by some SES
Software engineering programs, notably SESEnviroPlus, but it can also be used to create plots
from scratch, using your own data.
The plots displayed by the program are interactive. You can zoom anywhere in the plot using the
mouse wheel; pointing the mouse at any point on a plot brings up a window showing the values
of the nearest data point.
Fonts and colors can be customized for many plot elements, including the axis labels and the
plot title. The plot can be copied to the Windows clipboard to be included in reports or saved to
disk.
This program is based on a new SES Plotting Engine component which will eventually be
integrated in all SES Software user-interface programs that need to plot data; it already plays
this role for the new SESFFT program described previously.

2.7 SESTralin
The SESTralin program that was introduced for the first time at the last CDEGS User’s Group
Meeting. This version adds a few enhancements:
 A View Report button was added in the Computation Results panel. This allows a
quick access to the output (F09) results file once the results have been computed.
 The program now uses the new Soil Model Editor component to display end edit the
properties of the soil model. This will make it easier to share soil model data between
SESTralin and other SES programs.
3 HIFREQ: Junctions between Metallic Plates and Wires

Support for the modeling of metallic plates was added to HIFREQ two years ago. Since then, the
capabilities of the HIFREQ computation module dedicated to plates have been steadily
increasing.
This version introduces the possibility of modeling direct contacts (junctions) between wires and
metallic plates. This new capability greatly increases the range of applications that can be

studied using combinations of wires and plates. Wire-plate junctions can be used, for instance,
to attach a ground wire to a transformer tank or other metallic container. They can also be used
to energize objects made of plates.
Consult the article “Modeling Network of Metallic Plates and Cylindrical Conductors in
HIFREQ” for more details about these new capabilities of HIFREQ and for practical examples of
the use of plates in HIFREQ.
4 ROWCAD/Right-of-Way: Total Interference under Steady-

State, Arcing Distance, KML files, and much more…
Many new features and capabilities have been added to the Right-of-Way and ROWCAD
programs since the last User’s Group Meeting.
Among the Right-of-Way improvements, the calculation of the total interference level is now
fully integrated and automated, making it possible to account for both inductive and conductive
effects under steady-state conditions. Plots of Touch Voltages and 1 cm2 Holiday Leakage
Current Density under steady-state were also added to display the results of this analysis.
Also, the calculation methods of soil maximum breakdown distance and of fault arcing and
flashover distance introduced last year were improved, and new options were added.
Another interesting enhancement is the option to define the enclosure of a Group path as a
“Dummy” or virtual conductor, making it possible to quickly define a group of cables running
parallel to each other, within a trench or duct.

See “New Features in Right-of-Way Pro” for all the details on these and other features.
Over the past few years, the ROWCAD module has simplified data specification for Right-of-
Way. This version of ROWCAD continues this trend.
ROWCAD is now outfitted with a convenient Keyhole Markup Language (kml) file import
feature. Latitude and longitude are converted via a Universal Transverse Mercator projection:
paths are converted to Polylines and placemarks to Entities. In addition, you can choose how the
system is centered to avoid large coordinates, thereby streamlining your work with GPS
coordinates and Right-of-Way, and resulting in a simpler, more accurate model creation
process.
ROWCAD now also allows you to define a minimum cut length for region generation. In some
cases, this will lead to far fewer generated regions, saving substantial computation time and
greatly reducing file sizes, all while maintaining computational accuracy.
The user experience in ROWCAD is enhanced with a new sleeker viewer toolbar, more
ergonomic polyline and path manipulation button locations in the Paths panel, along with
background gridlines to help visualize distances and positions of objects in the system.

See “New Features and Improvements in RowCAD” for more details.
5 CorrCAD: Coating Effective Resistance, Changes of Cross-

Section, and Improved Display
CorrCAD is a software package dedicated to the analysis and design of cathodic protection
system that made its debut at the 2015 Users Group conference. CorrCAD can solve a large
variety of cathodic protection design tasks and related issues for various onshore and offshore
projects.
Since it was first introduced, several enhancements were made to CorrCAD:
 The coating effective resistance of pipelines and other structures can now be modeled
using discrete or distributed elements. This can be useful, for instance, when modeling a
coated pipe for which polarization data in the presence of coating is not available.
 Different cross-sections can be assigned along the polyline defining the structure to be
protected. This makes it easier, for instance, to model cases where the characteristics of

the structure (e.g., radius of a pipeline) change along its length, to model the presence of
insulating joints, and so on.
 The SESCurveFit tool which is used in CorrCAD to fit polarization curves to analytical
formulas was enhanced with the addition of a new, more accurate form of polynomial fit.
This new Chebyshev polynomial option can be useful when trying to fit polarization data
that does not conform well to the traditional Butler-Volmer curves.
 Several options were added to control the display of the results. Also, background items
were added in the main drawing area to provide useful information about the drawing.
Read more about those enhancements and about CorrCAD in general in “New Advances and
Features in CorrCAD Software Package”.
6 SESShield-3D: Accuracy Level and Component

Identification
SESShield-3D is a software package that can be used to analyze and design shielding systems
against lightning. Many user-interface enhancements were introduced in this software package
over the past year to help quickly identify objects in a model. For example, it is now possible to
label components in a model with their name or their component number. The components
displayed in a model can also be shown or hidden based on their protection status. This can be
useful to quickly visualize all components of the model that are part of the protection system or
those that need to be protected. The way selected objects and components are displayed and
distinguished from other objects and components was also improved, with options to highlight
the entire object (group) to which a selected component belongs, only the component itself, or
both.

Also, geometrical data can now be imported from MALZ and HIFREQ files, in addition to MALT
and SESShield-3D files, as was already available, and the objects imported from the
SESShield-3D Object Database can now be scaled and positioned anywhere in the model.
From a computational point of view, the Accuracy Level with which the interception surfaces
are approximated with polygons can now be controlled, providing a trade-off between
computational accuracy and computation time.
See the article “Improvements and New Features in SESShield-3D” for the details.
7 SESThreshold: Undo/Redo and Copy Settings between

Touch and Step and between Zones
SESThreshold is a tool that can produce plots and reports that compare the computed values of
certain quantities against allowable threshold values, which can be different in different
geometrical zones covering the area of interest. This tool, which was introduced last year at the
User’s Group Meeting, can currently generate such plots and reports for touch and step voltages.
The threshold values assigned to each zone can be user-defined or can be computed by the
program based on relevant safety standards. This makes possible, for instance, to account for
the effect of different surface materials (e.g., asphalt, crushed rock, no surface material, …) being
installed in different areas of a substation on the applicable limits on touch and step voltages.
Several improvements were made to SESThreshold since the first version of the program was
presented, including:
 An Undo/Redo mechanism was implemented. This was done in both the main
interface of SESThreshold and in the Zone Editor component that is used to define the
geometry of the zones where different threshold values apply.
 The geometry and safety specifications of zones defined for the calculation of touch
voltages can be transferred to the calculation of step voltages, and vice-versa.
 The safety specifications for a zone can be copied to another zone defined for the same
quantity (touch voltages or step voltages).
Together, these improvements make the definition of zones and the related safety specifications
much more efficient and less error-prone.
You can find the details about these new features in the article “New Features and
Enhancements for the Calculation of Safety Limits in SESThreshold”.
8 SESCAD: Complete Editing Capabilities and Access to

Conductor Database
Several improvements were made to the SESCAD program since the last User’s Group Meeting.
First, the program can now act as a complete editor for MALT, MALZ, and HIFREQ input files.
The new Advanced Options dialog allow you to view and edit all those advanced settings that
were previously stored in the Command Template and that had to be edited either in
command mode or by using CDEGS – Specify.
Also, users of MALZ and HIFREQ can now specify the outer radius of conductors on their
conductor type, making it easier to verify that the specified outer radius matches the other

properties of the conductor. Conductor type definitions can also be imported from the SES
Conductor Database.
The detection of network gaps in the Ambiguity Viewer was also improved. Now, more gaps
are detected (such as gaps between conductors located in different planes) and the specification
of what is to be considered a gap has been simplified.
The article “Improvements in SESCAD” gives more details about these new features and
enhancements.
9 SESeBundle: Hollow Equivalent Conductor

SESeBundle is a tool that can be used to find the characteristics of a single conductor electrically
equivalent to any set of conductors connected in parallel (bundle). This tool was introduced for
the first time at last year’s Users’ Group Meeting. In that first version, the tool was restricted to
finding equivalent solid conductors.
This version of SESeBundle introduces the possibility to create equivalent hollow conductors.
This can be useful, for instance, when modeling power cables that include concentric wires
and/or a wire armor composed of several individual conductors: the characteristics of the

equivalent hollow conductor calculated by SESeBundle can be used to specify a single hollow
conductor as a sheath or armour in the cable model.
Read all about this new Hollow Equivalent Conductor capability of SESeBundle in “New
Features in SESeBundle: Finding an Equivalent Hollow Conductor Representing a Bundle of
Conductors”.
10 SESEnviroPlus: User-Defined Methods for Corona

Performance
The SESEnviroPlus software package evaluates the environmental impact of AC/DC
transmission lines. An important aspect of the effects of transmission lines on the environment
that can be evaluated using SESEnviroPlus is the levels of Audible Noise and Radio Interference
that they generate. The program can also evaluate the losses on the lines due to corona effects. A
large number of empirical methods have been proposed in the literature to compute those
quantities, and SESEnviroPlus supports several of those methods. However, the formulas and
empirical coefficients used in the formulation of those methods are not accessible and cannot be
modified, making it difficult to use SESEnviroPlus to calculate corona performance parameters
for methods that differ from the built-in methods in SESEnviroPlus.
This version of SESEnviroPlus introduces a new User-Defined method that takes care of this
problem. This new method makes it possible to:
 Change the coefficients of the existing (built-in) methods.
 Create new methods.
 Change, add or ignore corrections.
The article “User-Defined Calculation Methods for Corona Performance in SESEnviroPlus”

describes the User-Defined method of SESEnviroPlus in detail.
11 Concurrent Runs with a Multi-User Network License

Network-based licensing of SES Software is a convenient way to maximize the use of your SES
Software license by sharing it amongst several users. In this version, this licensing method has
become even more interesting: with a multi-user network license, it is now possible for one user
to run more than one case concurrently on a single machine.
This operation is referred to as reserving the license. At most one user can reserve the license in
this way. When a user reserves the license, the seats that are reserved are not available to other
users of SES Software using that same network license, until the license is restored.
The following steps show how to run the SESLicenseManager utility to reserve the license for
concurrent runs.
1. Run SESLicenseManager from the Start | All Programs | SES Software |
System menu.
2. Click the File menu, and check the Concurrent Runs Configuration… menu item.
3. Click the arrows to increase the Number of Concurrent Runs, enter your phone
number (or any information allowing other users of the network key to identify you) in
the Notes textbox, and then click on the Reserve button.
4. Click Yes when the ‘Please note that if other SES Software applications are currently
running with the key, they will be terminated.’ warning message shows.

It is important to make sure that no-one else is using the network key when you use
SESLicenseManager to reserve the license, since this operation will terminate the active sessions
of all other users that are using that key.
After reserving the license, you can use SESBat to run multiple concurrent runs. To do this,
increase the Number of concurrent runs in the Batch Options tab of SESBat, and then
start the runs.
Once the concurrent runs are completed, you should use SESLicenseManager again to
release the license so that other users can use the key. To release the license:
1. Run SESLicenseManager from the Start | All Programs | SES Software |
System menu.
2. Click the File menu, and check the Concurrent Runs Configuration… menu item.
3. Click on the Restore button to release the reserved license.
Note that the Restore command can be used by any SES Software user on the network when
the license is reserved, not only the user who reserved the license. Again, you should make sure
that no-one else is using the network key when you use SESLicenseManager to restore the
license, since this operation will terminate the active sessions of all other users that are using
that key. In particular, one should contact the user who reserved the license before clicking the
Restore button because all concurrent runs will be terminated once you release the reserved
license.

13 A NEW INTERFACE FOR CDEGS
Éric L. and Francis Gougeon

Email: info@email.com, Web Site: www.sestech.com
Abstract
The CDEGS application is currently being redesigned in accordance with the new SES User Interface standards.
This is the first phase of many in which the original CDEGS features will be migrated to a brand new UI.
1 Introduction
The CDEGS software package is a powerful set of integrated engineering software tools designed
to accurately analyze a wide variety of electrical and electromagnetic phenomena. This application
is now getting redesigned to a new user interface and will now comply will the rest of the newly
updated software packages.
The update will take place over multiple release phases. In this first phase, the focus was on going
forward with the actual user interface update. A particular effort was made to maintain the present
capabilities in order to minimize the impact on our customers.
The following sections provide a guided tour of the new interface.
2 Application Tour
2.1 Overview
As can be noticed by comparing Figure 1 and Figure 2, the overall aspect of the application has
been preserved.

Figure 1. Main display of the original CDEGS.
Figure 2. Main display of the new CDEGS.

2.2 Engineering Modules
All of the engineering modules that were present in the original CDEGS have been introduced in
the new version as well. The engineering module buttons can be found in the application’s ribbon
within the Modules tab. Their visual aspect has been preserved in order to facilitate adaption to
the newer graphical interface.
Figure 3. Engineering module buttons in the original CDEGS.
Figure 4. New CDEGS’s engineering module buttons.

2.2.1 Session Mode
The session mode combo box was removed from the interface and was integrated to the
engineering module buttons instead. This gives clear feedback about which sessions can be
launched for a specific module and JobID.
Figure 5. Session mode in the Original CDEGS.

Figure 6. The session mode in the new CDEGS. Only Specify and Compute are available for this
particular case.
Figure 7 New CDEGS’s session mode with Specify, Compute and Examine.
Based on the currently selected JobID, the session buttons will update themselves according to
the availability of the related files. In the case illustrated in the previous figures, the Specify and
Compute sessions are available at first. Once the Compute process is completed, the Examine
session will appear and become available.
2.2.2 New Session Mode
In the previous version of CDEGS, when a module had never run for a particular JobID, the button
would appear greyed out.

This behavior has been replaced by the New button. Once the New button is clicked, CDEGS will
launch the Specify session mode for the requested model.
Figure 8. Greyed module buttons in the original CDEGS.
Figure 9. The New button in the new CDEGS program.

2.2.3 Invalid Session Mode
Both the original and the new CDEGS have a mechanism in place to detect an error in the JobID
entry. This is achieved by completely disabling the engineering module buttons on the interface.
In addition in the new CDEGS, the textboxes that are used for the JobID entry will be highlighted
in red if the path or the JobID name is invalid. This gives a quick feedback about the validity of
the provided JobID information.
Figure 10 Original CDEGS’s disabled module buttons.

Figure 11. Disabled module buttons in the new CDEGS.
2.3 Tools
All of the tools that were present in the original CDEGS have been introduced in the new version.
The buttons to start each tool can be found in the Tools tab on the ribbon.
Figure 12 Original CDEGS’s tools.

Figure 13. Tools in the new CDEGS.
2.4 Legacy Tools
Another tab on the ribbon contains legacy tools. These tools consist of applications that were
replaced or updated in the new software packages. These tools are considered as deprecated.
Figure 14. Legacy tools in the new CDEGS.
2.5 JobID List
The JobID list of the new CDEGS uses the same source as the original CDEGS. This means that
you will be able to pick up your work in the new CDEGS from where you left off in the original
CDEGS. In addition, you can now drag and drop files or folders directly in the list to add the
corresponding JobIDs. Any JobIDs found into the dropped folders and their sub-folders are
added to the list.

Figure 15. JobID list in the original CDEGS.
Figure 16. JobID list in the new CDEGS.

3 Conclusion
In this first phase, the user interface was updated in accordance with the latest SES software
standards in order to minimize the learning curve for existing users. A particular effort was made
to maintain the present layout in order to minimize the impact on our customers. The application
improvement is ongoing and will expect to start embedding modules within CDEGS in upcoming
versions.

14 IMPROVEMENTS IN SESCAD
Simon Fortin
Safe Engineering Services & Technologies ltd

Abstract
This article describes the improvements made to the SESCAD program since the 2015 CDEGS User’s Group Meeting.
1 Introduction
This document describes the most important improvements that were made to the SESCAD
program since the 2015 CDEGS User’s Group Meeting. These are summarized as follows:
 SESCAD is now a complete editor for MALT, MALZ, and HIFREQ files: all settings for
those programs that could not be edited in previous versions of SESCAD can now be
defined on the new Advanced Options screen.
 The Outer Radius of conductors can be specified on their Conductor-Type for MALZ
and HIFREQ.
 In MALZ and HIFREQ, Conductor-Type information can be imported from the SES
Conductor Database.
 The Delete Unused Characteristics command was introduced to delete
characteristics that are currently not used in a document, re-adjusting the characteristics
type codes assigned to conductors and other objects so that they keep on pointing to the
same Characteristics Definitions.
 The detection of network gaps in the Ambiguity Viewer screen was improved with the
addition of a few options.
 Selection Handles were introduced to help select plates and soil volumes using the
mouse.
 Energizations can be specified on metallic plates in HIFREQ.
 The “Default” Conductor Type, Coating Type and Lead Type are now imported when
importing a file.
and more… The remainder of this document describes these features in greater detail.
2 SESCAD as a Complete Editor - Advanced Options

Since it was first introduced as a graphical editor for MALT, MALZ, and HIFREQ files, there were
always a few advanced settings of those programs that could not be edited in SESCAD. Editing
those settings required using CDEGS – Specify (Input Toolbox), which could be inconvenient.
This is now a thing of the past. This version introduces the new Advanced Options component
(Figure 1). This dialog is reachable using Define | Advanced Options. It allows you to define
many advanced settings controlling the computation process in MALT, MALZ, and HIFREQ.
Figure 1. The Advanced Options screen used to define advanced options affecting the computations for
HIFREQ in SESCAD. Similar screens are available for MALT and MALZ.
Consult the online help of SESCAD for a detailed list of available settings and a description of their
function.
With this addition, SESCAD is now a complete environment for editing, running, and accessing
results visualization tools for MALT, MALZ, and HIFREQ.
Note: The Advanced Options component is presently at an early Beta stage of development. If
necessary, it is still possible to use the editor that was available in previous versions of SESCAD.
This editor, which is basically a simple text editor for the command mode version of the options,
can be used by selecting the Built-In option for the Advanced Options setting on the
Interaction page of the General Settings screen in SESCAD (Figure 2).

Figure 2. Advanced Options setting in the General Settings screen in SESCAD.
3 Outer Radius on Conductor Type and Importing Conductor

Types from SES Conductor Database
3.1 Introduction
For historical reasons, the outer radius of conductors was traditionally specified at a different
location in SESCAD than other physical properties of the conductor, such as its inner radius and
the resistivity and permeability of the metal in the conductor. This could make it difficult to make
sure that the specification of conductor data was consistent, with the correct outer radius
matching the specified resistivity and permeability.
In this version of the program, the outer radius of a conductor can now optionally be specified on
the Conductor Type associated to the conductor, along with its inner radius, resistivity and
permeability. Moreover, these values can now be imported from the SES Conductor Database.
3.2 How This Works
Currently, SESCAD is the only program that allows the specification of the outer radius of
conductors on their conductor types. Moreover, this way of specifying the outer radius of
conductors is optional: the outer radius of conductors can still be specified in the
Characteristics screen, in the Radius box. However, if a conductor is assigned a Conductor
Type that defines an Outer Radius, the Radius box becomes unavailable (it is grayed out; see
Figure 3) and the value of the Outer Radius specified on the Conductor Type is automatically
transferred to the Radius. This way, all tools and programs that currently do not support the
specification of the outer radius of conductors on their conductor type can continue working
normally with files saved using the new version of SESCAD.
Figure 3: The Characteristics screen for a conductor whose Radius is specified on its Conductor
Type.
Usage Notes:
 In SESCAD, it is possible to edit the Characteristics of several conductors
simultaneously. Note that when the Radius of selected conductors is modified in this way,
only the radius of conductors whose Conductor Type does not define an Outer Radius
is modified.
 When loading a file, the radius information specified directly on the conductor (not on its
Conductor Type) takes precedence. If both values are defined and disagree, the Outer
Radius definition is automatically removed from the Conductor Type, to maintain
consistency. (You will be notified with a message similar to that shown in Figure 4.)

Figure 4: Message for mismatched radius definitions.
 Similarly, if the data specified on a Conductor Type is identified as having been

imported from the SES Conductor Database but no longer matches the data stored for
the corresponding conductor in the SES Conductor Database, then the link with the
SES Conductor Database is automatically removed for that Conductor Type, and a
message similar to the following is shown:
Figure 5: Message for mismatch with SES Conductor Database.
3.3 User Interface Changes
Figure 6 and Figure 7 show the new Conductor Types screen in SESCAD. The Conductor
Types grid has two new columns:
 Outer Radius: Specifies the outer radius of the conductor, in meters, feet, or inches
(according to the selected system of units). This value is optional. If a value is specified in
this screen, it overrides any value specified in the Characteristics screen for conductors
with this Conductor Type. Note that you can select this value from the Ground
Conductor Database (see Figure 8) by clicking on the … button in this column.
 Database ID: Indicates the conductor class and conductor name of the conductor from
the SES Conductor Database from which the Outer Radius, Internal Radius,
Relative Resistivity, and Relative Permeability data were imported. This is set
automatically by the program and cannot be modified.
Figure 6: The enhanced Conductor Types screen in SESCAD (Part I) showing the new Outer Radius
column.
Figure 7: The enhanced Conductor Types screen in SESCAD (Part II) showing the new Database ID
column.

Figure 8. The Ground Conductor Database screen. This screen can be used to select the Outer
Radius of a conductor from a pre-defined list. It can now be used by clicking on the … button in the
Outer Radius column in the Conductor Types screen.
The following new operations are also available:

 Import from Database: Imports the Relative Resistivity, Relative Permeability,
Outer Radius, and Internal Radius for the selected conductor types from the SES
Conductor Database. This also automatically sets the Impedance Specification
Method to Computed for the selected conductor types and fills the Database ID
column with a string identifying the database conductor from which the data was
imported. Note that the imported data is locked and cannot be edited until Break Link
to Database is used.
 Break Link to Database: Breaks the link between the selected conductor types and the
SES Conductor Database conductor from which the conductor type data was
imported. The data is not modified as a result of this operation, but can be modified from
this point on.
 Duplicate: Creates duplicates of the selected conductor types. This operation duplicates
the link to the SES Conductor Database, if any.
Notes:
 Importing data from the SES Conductor Database does not prevent you from
specifying other data (for example, Load Resistance, Load Inductance, Working
Potential, …) on the same conductor type.
 When changing the system of units, the Outer Radius and Internal Radius of
conductor types that are imported from the SES Conductor Database are always
converted, regardless of the option selected in the Data Conversion Options screen
(see Figure 9).

Figure 9. The Data Conversion Options screen. These options do not affect the properties of
conductors imported from the SES Conductor Database.
4 Delete Unused Characteristics

For a variety of reasons, it is not unusual for models built using SESCAD to include definitions of
characteristics (Lead Types, Conductor Types, Coating Types, Energization Types, and Cable
Types) that are not assigned to any conductor. This often occurs, in particular, when using the
Insert Right-of-Way Model command, since this command needs to define several such
characteristics that are ultimately abandoned in the final model.
At best this is a nuisance, but it can actually prevent programs from running. Moreover, it is not
easy to remove those unused characteristics definitions, since this may require renumbering the
characteristics codes assigned to the conductors in the model.
This is what the new Delete Unused Characteristics command does: it removes selected
characteristics definitions from the model, adjusting the characteristics codes assigned to all
conductors in the model so that they keep on pointing to the same characteristics.
To use this feature:
1. In SESCAD, click Define, and then click Delete Unused Characteristics.
2. In the Delete Unused Characteristics screen (see Figure 10), select the unused
characteristics definitions that you want to delete and click OK.
Note that only those characteristics definitions that are not currently used in the model are
available to be deleted in this screen. Note also that, as indicated on the screen, this operation is
currently not undoable, so ideally it should be used after all editing operations have been
completed on the model (and the Undo facility is no longer required).

Figure 10. The Delete Unused Characteristics screen showing Lead Types, Coating Types, Conductor
Types, Energization Types, and Cable Types that are current not used and can be deleted.
5 Network Gaps
The presence of poor connections or gaps in a conductor network can have a detrimental effect
on the performance of the network. This is true in the real world and is also true in MALZ and
HIFREQ computer models.
To help make sure that conductors in such models are properly connected, the Ambiguity
Viewer in SESCAD includes a Network Gaps detection option. However, in previous versions
of the program, the definition of what constitutes a network gap was rather restricted: a gap would
be considered to occur between two conductors only if one of them had both extremities floating
and if that conductor would come within a specified distance (the Maximum Gap Size) of another

conductor. Moreover, the definition of that “Maximum Gap Size” was rather awkward, being
specified as a factor to be multiplied by the sum of the radii of the conductors involved in the gap.
In this version, the following aspects of this feature were improved:
 The Maximum Gap Size can now be specified as an absolute distance, in meters or
feet. This is now the default way of specifying the gap size. The method of previous
versions is still available.
 Crossing Gaps (where two conductors form a “+”, without intersecting) can also be
optionally detected.
 The restriction that conductors must be floating in order to be candidates for gaps was
removed, although it can be re-instated by clearing the Allow Gaps on Connected
Conductors option.
Overall, the new algorithm tends to report more gaps, and they also tend to be more meaningful.
Figure 11. Options related to Network Gaps for the Ambiguity Viewer feature. More gaps are now
detected and the specification of what constitutes a gap is easier to control.

6 Cables in Object Database

Several objects defining power cables have been added to the SESCAD Object Database (see
Figure 12). These objects, located in the Power Cables section of the database, define a single
100 m (or 100 feet) long piece of cable along with the related Cable Type definition. These objects
can be inserted in the drawing as any other object, using Insert | Object from Database,
making it easy to define cables of those types.
Currently, all cables contained in the database are of coaxial type, with a sheath and nor armour.
They are regrouped by voltage class and ordered by core size. Cables of different types and of
different voltage classes will be added regularly to the Power Cables section.
Figure 12. The Power Cables section in the SESCAD Object Database.
7 Selection Handles
Models including finite soil volumes or metallic plates are more and more common. Those objects
can be edited graphically in SESCAD. They can be selected with the mouse by clicking on one of
their edges. There are circumstances, however, where this is inconvenient. This is the case, in
particular, when two or more such objects share a common edge: in this case, it is impossible to
predict which of the objects will be selected when clicking that edge.
This version of SESCAD introduces the notion of Selection Handles to deal with this problem.
Selection Handles are small “+” symbols displayed at the center of certain objects, such as metallic

plates and soil volumes (see Figure 13). When Selection Handles are displayed (they are by
default), those objects can be selected by clicking on their Selection Handle.
Figure 13. Selection Handles on metallic plates that can help select plates with the mouse. Selection
Handles are also available for Soil Volumes.
The display of Selection Handles can be controlled in the View Options screen (Common page
– see Figure 14). Click Options | General Options to reach this screen. You can also set this
option directly under the View menu (using View | Selection Handles).

Figure 14. Options related to Selection Handles.
8 Specification of Energization on Plates in HIFREQ

Support for the specification of metallic plates in HIFREQ was introduced last year in SESCAD.
This year, the possibility to directly energize a plate in HIFREQ was introduced, with the new Add
/ Edit Plate Energizations dialog (Figure 15).
Plates are energized by first identifying a specific patch of the plate in which current is to be
injected and then by specifying an Energization Type Code for the patch, in much the same
way as is done for wire segments. The identity of the patch that is energized can be specified in
one of two ways:
a) By giving the patch number along the two main directions of the plate at the intersection
of which the patch is located.
b) By giving the position of a point in space, in which case the energization will be applied to
the plate patch which is located closest to that point.

Figure 15. Illustration of Add / Edit Plate Energizations dialog used from the Edit Object screen.
(a)- Specification of the patch number along the two main directions of the energized plate.
(b)- Specification of the position of a point in space.
Note that the second method of patch specification is less sensitive to the detailed way in which
plates are subdivided into patches. Note also that several energizations can be specified on the
same plate, but it is not possible to specify several energizations on the same patch of a plate.
For each energization, the required data is as follows:
 Energization Type: Indicates the type of energization selected for the analysis. At this
moment, only Current (Lead) energizations are supported for metallic plates.
 Patch Selection: Indicates which method is selected to identify the energized patch.
When By Position is selected, the energized patch is identified by providing the X, Y, and
Z coordinates of a point located close to the patch. Therefore, the closest patch to this point
on the plate containing that patch is considered as being energized. When By Patch
Numbers is selected, the energized patch is identified by providing the explicit patch
numbers along the two principal directions of the plate.
 Patch Numbers: Specifies the subdivision numbers along the 'ab' and 'bc' directions that
identify the energized patch.
 Reference Point Coordinates: Specifies the X, Y, and Z coordinates (in meters or feet)
of the point used to identify the energized patch.
The same dialog can be used to apply the same energization to several plates simultaneously. To
do this, select the plates whose energization properties are to be modified and click Edit | Edit
Attributes of Plates. In the Edit Attributes of Plates dialog (Figure 16), check the

Energization box and click on the corresponding … button to modify the energization properties
of the selected plates.
Figure 16. Plate energization in the Edit Attributes of Plates dialog. This can be used to edit the
energization properties of several plates simultaneously.
9 Outward Normal and Edge Continuity for Solid Objects

Made of Plates
The generation of solid objects (disks, cones, cylinders, spheres, prisms, and portions of those
objects) has been available in SESCAD for a few years. Those objects can be generated in terms of
linear conductors or, in MALT and HIFREQ, in terms of metallic plates. In this version of
SESCAD, the generation of solid objects in terms of plates was improved to make it more useful
for the MALT and HIFREQ engineering programs.
When generating plates for HIFREQ, SESCAD now enforces perfect edge continuity between
connected plates in the generated objects. This condition is required by HIFREQ to properly
represent current flow between connected plates.
Also, the generated plates are now oriented in such a way that the normal of each plate (obtained
by applying the right hand rule to the first three vertices of the plate) points towards the outside
of the solid object (or towards the sky for the Disk object). This consistent definition of the normal
of the plate is useful when applying different coatings on the two sides of a solid object, since the
sides of a plate are identified by the direction of their normal vector in the Plate Characteristics
Editor dialog (Figure 17).

Note that it is now possible to edit the coating of several plates simultaneously using the Edit
Attributes of Plates dialog, and to leave any existing (and possibly different) coating definition
on selected plates unmodified in the Plate Characteristics Editor dialog. This option is useful
when it is desired to modify the coating definition on one side of a large number of plates while
leaving the coating definition on the other side unmodified.
(a)
(b)
Figure 17. (a): The Edit Attributes of Plates dialog used to specify the same coating on multiple plates.
(b): The Plate Characteristics Editor dialog used to specify coatings on metallic plates in MALT.
Notice the new Do not modify option which allows you to keep the existing coating definition for the
edited plates.
10 Importing Default Types when Importing Files

When importing a MALT, MALZ, or HIFREQ into another file, the Characteristics Definitions
(Coating Types, Conductor Types, Lead Types, Energizations, and Cable Types) can also be
optionally imported. In previous versions of SESCAD, the default type (i.e., type No. 0) of Coating
Types, Conductor Types, and Lead Types were not imported in this operation. This could lead to
some data being imported with incorrect properties in those rare cases where default types were
customized.
Now, the program imports the default types of Coating Types, Conductor Types, and Lead Types.
Note that it does so only when the default type has been customized either in the original file or
in the file being imported.
11 New Soil Model Editor Component

A preview of a new Soil Model Editor (Figure 18) was added to SESCAD as an optional
component. This component will eventually become the user interface for editing the properties
of the soil model in all SES programs.
Figure 18. Soil Model Editor screen in SESCAD.
Currently, this component offers similar functionality to that of the built-in Soil Structure
screen in SESCAD but it is slated to acquire more powerful features in future versions.

For the time being, usage of this component is turned off by default in SESCAD. You can turn it
on to get a feel of what’s coming by specifying Component as the Soil Model Editor in the
Interaction page of the General Settings screen in SESCAD (Figure 19). When this is done,
the Soil Model Editor component will be used from that point on when using Define | Soil
Model in SESCAD.
Figure 19. The Component value for the Soil Model setting in the General Settings screen in
SESCAD. Use this value to activate the use of the Soil Model Editor component for editing and
visualizing the soil model.

This article described the new features and enhancements introduced in the SESCAD program
since the 2015 CDEGS User’s Group Meeting.
As is the case for most of first-generation user interface programs in SES Software, SESCAD is
drawing towards the end of its lifetime. Nevertheless, several features are still being planned for
the near future, either in SESCAD itself or in its successor. Among others, the following features
are being considered:
 User-defined colors in Display Filters.

 Display of common electrical symbols in the main drawing.

 More powerful Node Editor, allowing graphical editing of several nodes simultaneously.
 …

15 SESRESAP: A NEW SOIL RESISTIVITY

MEASUREMENT EDITOR
Stéphane Baron, Stéphane Franiatte, Maryam Golshayan, Yanmei Jiang,

Zhiqiong Luo, Majid Siahrang and Luis Valcárcel

Abstract
The RESAP interface revamp is well under way and a first version, called SESResap, is now available and replicates
almost all the inputs traditionally specified through the CDEGS Input Toolbox. It was optimized for quickly and
efficiently inputting the data, inspecting it for quality and analyzing it, all without leaving the main window. New
features include a user-selectable labelling convention for the electrodes, the possibility to display optional
columns in the measurements data grid to obtain more information about the system, the addition of comments
for any data point, an improved plot of the data, a clearer way of specifying analysis parameters and a useful data
sheet that can be used for field measurements.
1 Introduction
Correctly interpreting soil resistivity measurements is an important step for any grounding study
and is accomplished using SES’s well-known RESAP computation module. This program
automatically searches for the optimized parameter values that best describe a soil’s electrical
structure as measured by any type of 4-pin surface measurements, namely the thickness and
resistivity of the various soil layers.
During the 2015 Users’ Group Conference [1], an announcement was made that the user interface
of RESAP was in the process of being revamped, and a preview was given for what it would look
like. A year later, it is time to unveil the first release of this new program, SESResap, which is
meant to become the successor of the traditional CDEGS Toolbox interface of that module.
For this first implementation cycle, the focus was put on replicating the existing and frequently
used inputs and accesses to the functionalities of the program, with the primary objective being
to make them more intuitive to find and reduce the back and forth between windows, while
enhancing the visibility of data, processing logs and results. The following sections provide a tour
of the new interface, highlight the novelties and give a glimpse of the objectives for the next
development phase.
2 Application Tour
2.1 Measurements Data
Soil resistivity measurement data is not especially complex. All that is needed is a description of
the electrode layout and an apparent resistance measurement obtained from the ratio of voltage
between a pair of electrodes to the current circulating through the soil via another pair of

electrodes. There are however different ways to specify these quantities and flexibility for doing
so has been given in the Measurements Panel, whose elements are described in this section.
2.1.1 Measurement Methods
Indeed, various types of electrode arrangements exist for performing the measurements. RESAP
has always supported many popular methods and the new interface continues to make them
available, as shown in Figure 1.
Figure 1: Electrode arrangements supported by the application. Probe depths can also be accounted for.
Schematics illustrating each of the available methods have been updated compared to those
shown in the previous version, as shown in Figure 2. As before, they show which spacings serve
as input for the program, but now also include spacing labels coinciding with those used in IEEE
standards [2]- [3], for the Wenner and Schlumberger methods, in order to avoid possible
ambiguities.
Figure 2: Schematic view of the Wenner measurement method. The inter-electrode spacing required for
data input is clearly indicated. C1 and C2 label the current electrodes, while P1 and P2 label the potential
probes. DC and DP represent the depth of the current and potential probes, respectively.
2.1.2 Alternate Electrode Labelling Convention
Instruments manufactured especially for soil resistivity measurements tend to have their
terminals connecting to each of the electrodes labelled either as C1, C2, P1 and P2 or as A, B, M

and N. To prevent interpretation mistakes, the two labelling conventions have been made
available as a preference for the user (Figure 3). The labels used in the schematics and elsewhere
in the interface update accordingly.
Figure 3: Electrode labelling convention choice (from the Preferences Panel).
2.1.3 Data Type
Similarly, for the electrical measurements, instruments often report either directly the Apparent
Resistance, expressed in units of ohms, or a derived quantity, the Apparent Resistivity, expressed
in units of ohm meters. For convenience, the interface continues to offer these two choices for the
type of data, as shown in Figure 4. The two quantities are simply related by a geometrical factor
that reflects the electrode arrangement. When using an instrument that reports apparent
resistivity, care must therefore be taken to make sure the instrument’s setting include a valid
specification of the inter-electrode spacings for each data point, otherwise the readings will be
distorted. For the same reason, when inputting the data in SESResap, distinguishing between the
two types is important and a frequent source of error that can invalidate a measurement
interpretation.
Figure 4: Data types accepted for the measurements.
2.1.4 Data Grid
For the simplest case of the Wenner electrode arrangement, only two columns are necessary to
record the data: the inter-electrode spacing, a, common to all adjacent pairs of electrodes, and the
mutual resistance between the current and potential electrodes. Figure 5 shows such a simple
grid.

Figure 5: Simplest view of the measurements data grid for the Wenner method.
However, for practical considerations, when performing measurements in the field, it is not
necessarily the a-spacing which is recorded. Indeed, when positioning the electrodes, it is much
easier to measure distances from the center of the traverse, since this is usually one of the fixed
point (the other one being C1) where the measurement equipment is installed, up to the potential
electrodes and up to the current electrodes. For this reason, the new interface was designed to
optionally display the half separation distance between the potential electrodes, P1P2/2, and the
half separation distance between the current electrodes, C1C2/2. This makes it easier to correlate
the field datasheet with the program’s input such as to minimize the chances for input error.
Alternately, this also provides an easy way of planning the electrode positions in advance for
desired inter-electrode spacings.
For measurement methods other than the Wenner arrangement, more spacings are obviously
needed: two for Schlumberger and three for the General, Unipolar and Dipole-Dipole methods.
In all cases, the data grid can be customized to show only the columns in the data grid where input
is expected or to show any other of the inter-electrode spacings that could have been used to
measure the electrode locations while on the field. These are shown in Figure 6, where the
electrode pairs are labelled as EiEj, where E can be either C or P for the current or potential
electrode type, respectively, and i, and j can be 1 or 2 for the electrode number.
Figure 6: Selection, in the Preferences Panel, of the electrode spacings to display in the measurements
data grid, for each measurement method.
The checked and disabled items represent the required inputs for the different methods and are
therefore mandatory. The other columns are optionally displayed for convenience, e.g. to allow
verifying the data input against that recorded in a datasheet using a different spacing basis, like
P1P2/2 and C1C2/2 for the Wenner method.

Similarly, there are other columns that are optional. Figure 7 shows the availability of columns
for the measured apparent resistance, the corresponding apparent resistivity and the geometrical
factor that relates them. Of these, the one corresponding to the chosen data type for input is
mandatory, while the other two can be shown on demand, for verification purposes. This helps
minimize the chances for mistakes and highlights the relationship between the different
quantities. Indeed, the extra columns can be used to confirm what the instrument was actually
reporting and if its spacing setting was correctly used, and also allows examining simultaneously
the trends for the apparent resistance and apparent resistivity, which are typically different.
Figure 7: Additional options for controlling the display of optional columns.
A new feature that Figure 7 also shows, is that it is now possible to include a comments column
that allows recording any special circumstance that occurred while measuring a given data point.
This can be useful when interpreting the measurements, as they can provide a basis for deciding
to eliminate doubtful points or they can suggest that a corrective measure be applied.
The grid of Figure 5 is shown again in Figure 8, but showing optional columns that allow for easier
cross-examination of the data.
Figure 8: Measurements data grid for the Wenner method with optional columns.
The primary parameter for the method, the a-spacing is available for input in a column that is
enabled for edition. Half-separation distances between sets of electrodes are shown in read-only
columns to help correlate with field measurements of electrode locations. Depths of electrodes
complete the electrode arrangement specifications. Other columns show the measured apparent
resistance, the corresponding apparent resistivity, the geometrical factor that relates them, and
comments recording the peculiarities of some data points.

2.1.5 Conversions between Measurement Methods
Even though SES recommends the Wenner arrangement for obtaining reliable soil resistivity
measurements that are less susceptible to local soil inhomogeneity [4], field conditions can make
it such that it is not possible to adhere to the prescribed a-spacing for each of the data points.
Physical obstacles can preclude some electrode positions from being safely reachable (road
crossings, construction or no-trespassing zones) or can downright be impossible to reach (flooded
zone, fenced area, dense vegetation, etc.). Other times, locations can be reached, but are deemed
undesirable because some of the electrodes would end up being in proximity to grounded metallic
objects that can provide an alternate path for the current compared to the through-earth path that
is sought to be measured. Such electrode placements can distort the measurements and should be
avoided when possible to reduce the need to take corrective actions [5]. To that end, a Wenner-
type initiated measurement traverse can be modified during the course of the measurement to
accommodate exceptional irregular electrode spacings, e.g. akin to that of the Schlumberger
method (which can be useful when measuring a traverse in close proximity and parallel to a power
line in order to avoid electrodes near tower footings for instance), or even completely arbitrary
spacings (General method).
Clearly, if a laptop running SESResap is used when recording the data in the field, conversion
options are desirable such as to avoid having to re-input the data in the case that the electrode
arrangement method changes. The new interface achieves this very simply. The fact that any of
the EiEj columns for the different spacings can be made visible for any of the methods makes the
conversion process completely transparent: the columns do not change meaning as was the case
for the classic interface, it is just which columns that are available for input that changes state.
Of course, this automatic conversion between measurement methods is only possible when the
data is compatible. When the method to be converted from contains more parameters than the
method to be converted to, the program warns the user about the data loss implications before
committing to the action. Figure 9 shows an example of this.
Figure 9: Message warning against data loss during measurement method conversion.
2.1.6 Plot of Input Data
Plotting the data and examining its trend is obviously important for the analysis as it allows
determining starting values for the RESAP algorithm, such as the number of soil layers to expect,
the top soil layer’s resistivity and its approximate thickness. It is also especially important to plot
the data while still on the field, because it allows detecting anomalies in the measurements and

permits corrective measures to be attempted before it is too late. For instance, for a horizontally
layered soil, the apparent resistance is expected to monotonically decrease with electrode spacing
(except for some unusual soil type scenarios), whereas the apparent resistivity curve (when
plotted on a log-log scale with identical decade lengths for the X and Y axes) can have a slope that
varies between positive and negative values, as long as it sustains an angle within the +45° range
for low to high resistivity contrasts soil types.
In order to examine those trends, a graph of the measurement data was always available in
RESAP. In the new interface, a dedicated Plots Panel was introduced for this, which can be made
as large as necessary for proper inspection and can be made to be visible at all times if desired,
not just when the measurement data grid is shown. As before, the type of axis (linear or
logarithmic) can be selected, as well as which quantity to plot (apparent resistance or apparent
resistivity). The plot now offers zooming capabilities and tooltips appearing when clicking on data
points that make it easy to read accurately the graph. Figure 10 shows an example.
Figure 10: Plot of the measurement data. Examination of the trends of apparent resistance and apparent
resistivity prior to performing the analysis allows evaluating the quality of the data and helps select
appropriate parameters for an accurate interpretation.
2.2 Computation Panel
The measured data is only part of the inputs required for the soil interpretation. Indeed, SESResap
offers several computation options from which to select to perform the analysis. These are
grouped under the Computations Panel.
2.2.1 Soil Type
Figure 11 shows the choice for the desired soil structure.

Figure 11: Selection of the desired soil structure and number of layers (in the case of the horizontal soil
type) for the interpretation of the measurements.
It is the goal of the RESAP analysis to determine the value of the parameters of a given soil
structure (e.g. resistivity and thickness of layers for a horizontally layered soil). It does this
through a complex optimization process that sometimes benefits from guidance from the user.
For each type of soil, a data grid therefore allows specifying initial values for its parameters. Figure
12 shows this data grid for the horizontally layered soil type. As for the previous version of RESAP,
it is possible to lock the values of chosen parameters such as to instruct the program not to
optimize them. This is now simply done by checking the desired cells under the columns identified
with a padlock icon.
Figure 12: Datagrid allowing the specification of starting values for the parameters’ optimization process.
Empty cells let the program determine the starting values. Locked parameters will not be optimized
during the analysis. Clicking the padlock icons will lock/unlock the entire columns at once.
2.2.2 Computation Options
Other than the specification of the modeling function (soil type) and corresponding parameters
that are to be optimized for, the optimization algorithm itself and its controlling variables can also
be chosen by the user. The most often used options for these choices have been made more
accessible and can now be found directly at the level of the main interface as can be seen in Figure
13. These include the algorithm’s methodology, the targeted accuracy, the maximum number of
iterations allowed, the desired step size and the reference small number to evaluate convergence
of the optimization function.

Figure 13: Frequently used computation options have been made more accessible.
2.2.3 Advanced Options
As for the less frequently used computation options, Figure 14 shows that they can still be
specified, but they are grouped under a different, higher-level window in order not to clutter the
main workspace. More information about the advanced options can be obtained from the
application’s help files.
Figure 14: Rarely used computation options have been regrouped in an Advanced Options window.
2.3 Processing
Computations are enabled provided that no errors and no tasks pending are displayed in the
Issues List (see section 3.2). Launching the analysis can be done by pressing either of the
Compute buttons shown in Figure 15.
Figure 15: Compute buttons are made available in the Home ribbon tab and in the application’s titlebar
(the latter being always accessible).
Contrary to its predecessor, launching the computation in this updated interface does not require
exiting the input screens. Indeed, the computation progress can be monitored in the
Computation Trace Panel shown in Figure 16, which can be docked alongside the other panels
mentioned previously.

Figure 16: Progress of the computation can be monitored from the Computation Trace panel.
2.4 Basic display of results (F09)
More than just an input session editor, the new SESResap interface will be a one-stop shop for a
complete analysis from input to output. After the processing completes, the F09 output file
becomes available for examining the results (Figure 17). A complete “Examine” session that
includes plotting capabilities is planned for future development. For now, the traditional CDEGS
Output Toolbox can however be invoked. The key novelty is that all sessions (input, computation
and output) are always visible simultaneously, making it easier to iterate the analysis until a
satisfactory solution is found.
Figure 17: The F09 output file and CDEGS Output Toolbox are easily accessible to examine and plot the
results.
3 Other Features
3.1 Project Information
Basic information about the project can be specified via the Project Info page shown in Figure
18. This includes a case description, the system of units to use, as well as the Run ID.
Figure 18: Project Info page of the application.
3.2 Improved, Stronger Validation
The application integrates the Issues List, which has now become omnipresent in the newer SES
interfaces (Figure 19). It guides the user by defining tasks that need to be accomplished before the
computations can be launched. For example, it will request that at least 5 data points be entered
for the analysis to be possible. It will also flag data input errors such as negative values for
spacings, depths and apparent resistances or apparent resistivities.
Figure 19: The Issues List informs about errors, tasks, warnings and messages.

3.3 Interface Options
As explained above, interface elements are arranged inside panels. The panels can be hidden,
docked at different places inside the main window or even undocked (floating). Figure 20 shows
the Options tab of the ribbon. This option allows restoring panel visibility or different layout
options, and control other interface elements.
Figure 20: The Options ribbon tab.
3.4 More Accessible Help and Guidance
As for most of the other new SES applications, the ribbon includes a Help tab to give easy access
to standard help and How To documents that demonstrate the use of SESResap (Figure 21). It
also gives access to the list of SES publications and to the Frequently Asked Questions (F.A.Q.)
portion of our website related to the topic of soil resistivity measurements and their
interpretation. This tab further contains useful shortcuts for contacting our support team, for
sending feedback about the application and for checking the version number, and whether
updates are available.
Figure 21: The Help ribbon tab.
3.4.1 Datasheets to Bring to the Field
Although bringing SESResap to the field certainly has its advantages, it is not always possible to
do so. After all, a security key need to be available for travel in order to save the data and run the
program, which not all companies will allow. Adverse weather conditions or limited battery
lifetime can also prevent the normal use of a laptop. For these reasons, it is sometimes preferable
to record data using the time-proven method of pen (or pencil, since it resists wet conditions
better) and paper. A well thought-out data sheet helps tremendously speed up the measurements
and record the essentials about the site conditions. Such a sheet is made available through the
Help ribbon tab of the interface and is shown in Figure 22.
The sheet is available in PDF format, for printing, or in Microsoft Excel format, since this may be
an alternative for people wanting to bring a laptop to the field, but not having a security key

dongle. Future developments should allow importing the sheet’s data directly to the SESResap
application.
Figure 22: Datasheet for field use.
4 Example
The example file used for creating the figures of this article can be found under the UGC
2016\Example Files\SESResap subfolder of the SES Software conference distribution. It specifies
a Wenner-type measurement and includes comments for some of the data points, as well as the
specification of initial values for the soil layer parameters.
5 Conclusion
The new SESResap interface released this year succeeds in reducing the amount of switching
between the different input screens and between the traditional Specify, Compute and Examine
sessions. It makes the data more visible and provides improved ways of relating it to the
recordings made on the field, such as to reduce the risk of mistakes or misinterpretations.
Since practically all the program’s input have been duplicated, the new interface is almost ready
to effectively replace the previous Input Toolbox interface, except for a few rarely used options. It
is released in beta form at this time because we want to improve its robustness and test it further
before it becomes an official production tool. After all, a proper soil resistivity analysis is a key
element to a sound grounding study.
As the program already integrates together the input and computation aspects, the main objective
for the next year will then be to also bring the output session within the same interface rather than
as an external component. Then we want to concretize the other objectives listed last year in [1],
namely to integrate the module within all of the SES applications/modules that will need an
interpretation of a soil structure directly from resistivity measurements, make the data visualizer
(graphs) interactive, create new tools to help combine low-contrast layers and achieve soil
structures with minimal layering that still represent adequately the data, and handle multiple
analysis runs side-by-side on the same data set in order to compare them better.
6 Acknowledgements
The authors would like to thank their colleague Mr. Robert Southey for contributing his
experience with soil resistivity measurements and for his useful suggestions.
7 References
[1] SES R&D and Software Development Team, "Progress Status on Future SES Module Interfaces," in
CDEGS Users' Conference Proceedings, San Diego, California, USA, 2015.
[2] IEEE Power and Energy Society, "IEEE Std 80-2013, IEEE Guide for Safety in AC Substation
Grounding," New York, 2013.
[3] IEEE Power and Energy Society, "IEEE Std 81-2012, IEEE Guide for Measuring Earth Resistivity,
Ground Impedance, and Earth Surface Potentials of a Grounding System," New York, 2012.
[4] W. Ruan, R. D. Southey, S. Fortin and F. P. Dawalibi, "Effective Sounding Depths for HVDC Grounding
Electrode Design: Wenner versus Schlumberger Methods," in IEEE/PES T&D 2005 Asia Pacific,
Dalian, China, 2005.
[5] R. Southey, G. Berlin, S. Fortin and W. Ruan, "Correcting Soil Resistivity Measurements for Distortions
Introduced by Nearby Bare Buried Metallic Structures," in CDEGS Users' Conference Proceedings,
Montréal, Québec, Canada, 2010.

16 SOILMODELEDITOR: NEW SES SOFTWARE TOOL FOR

SOIL MODEL DESIGN
Michel Chami, Eric Dawalibi, Greg Noel, Mark Boisjoli, Peter Zhao, Nina Mitskevitch, and
Alexandre Mailhot

Abstract
This article introduces the new SoilModelEditor program. This program is a new unified tool used to specify, edit,
import and export soil structures for various SES software computation modules and packages. This tool can also
be used as a standalone module, if required.
1 Introduction
SoilModelEditor is a new software tool that can be used to define a soil structure in SES Software.
This new tool preserves all existing functionalities pertinent to soil structure editing in various
software modules and packages of the current version of SES Software and also introduces a
number of useful features, more user-friendly screen design, interactivity, simplified data input,
etc.
SoilModelEditor can function as a standalone application or can be integrated as a component
within an engineering module or a software package. This article describes the interfaces and
features of SoilModelEditor when it functions as a standalone tool. Also, some examples are
presented when SoilModelEditor is integrated into SES software modules as a component.
2 SoilModelEditor as a Standalone Application

The main screen of SoilModelEditor is shown in Figure 1. In the middle of the screen, there are
two main panels: the Soil Structure Properties panel (on the left) that allows specifying or
editing a desired soil type for the selected engineering module or package and the 3D Viewer
panel (on the right) that displays the defined soil type schematically.
The Soil Structure Properties panel consists of two sections (Soil Type and Soil
Characteristics) dedicated to data specification for the soil structure. In the Soil Type section,
you can select a soil type and give it a name. The list of available soil types is consistent with the
engineering module selected in the Home tab on the ribbon. In the Soil Characteristics
section, the data specifying the properties of each soil layer (soil resistivity, relative permeability,
relative permittivity and other geometrical data if needed) can be defined.

Figure 1: Main screen of SoilModelEditor.

Soil volumes can be embedded into the Uniform and Horizontal soils in the MALT and MALZ
engineering modules. To add soil volumes to a soil model, check the Add Embedded Soil
Volumes box and specify the soil volumes characteristics using the Volumes tab in the Soil
Characteristics panel.
The application ribbon, available only in standalone mode, consists of the following tabs:
 Project
 Home
 Options
 Help
Details of the Home, Options and Help tabs are shown in Figure 2. The Home tab, allows you
to select the engineering module or software package for which the soil structure data is to be
specified. You can also select the system of units and access Import and Export features on this
tab. The Options tab allows you to customize the display language and font, as well as the location
and size of the various panels of the program. The Help tab allows you to access available
technical documentations, FAQ and other useful features.

Figure 2: Home, Options and Help ribbon tabs details.
Finally, the Project tab, shown in Figure 3, allows you to specify the module or project
description and the Run Identification for the case and to set the system of units. Also in this
screen, it is possible to start a new project, open and save the project, and exit the application.
Figure 3: Project tab on the ribbon.
2.1 New Unified Command Structure
SoilModelEditor uses a new unified command structure which is designed to store soil model
parameters for all engineering modules. An example of such a command structure for a horizontal
multilayer soil is shown in Figure 4.

Figure 4: Example of new command structure for a horizontally layered soil model.
Having such a unified command structure will make it easier to import (or export) a soil model
from one engineering module to another. Currently, input files with this new command structure
(SS_JobID.F05) can be used only by the standalone version of SoilModelEditor. Gradually, this
new unified command structure will be incorporated into each engineering module as well.
3 SoilModelEditor as a Component
3.1 SESTralin
In this section, SoilModelEditor is presented as a component integrated into the SESTralin

engineering module. Figure 5 shows examples of uniform and horizontally layered soils
specifications in SESTralin.
Figure 5: Uniform and Horizontal soil types in SESTralin.
3.2 SESCAD
SoilModelEditor is also integrated in SESCAD, where it will replace the legacy Soil Structure
screen to define a soil structure for the MALT, MALZ, or HIFREQ engineering module. Figure 6
shows an example of the SoilModelEditor component to specify the soil model in SESCAD.

Figure 6: Example of SoilModelEditor screen in SESCAD for the MALT engineering module.
4 Modifications to Soil Type Specification

The overall behavior of the new SoilModelEditor tool is quite similar to that of existing screens
dealing with soil structures. However, there are a few differences, notably with respect to the
Finite Volume soil model.
4.1 Soil Volumes
The current interface to define soil structures for MALT and MALZ offers the Arbitrary
Heterogeneities soil type, as shown in Figure 7. This soil type allows specifying finite soil
volumes of different resistivities embedded into a uniform native soil. In SoilModelEditor, this
option does not exist as a separate soil type. Instead, soil volumes can be specified when defining
a Uniform or a Horizontal soil type. To add soil volumes, check the Add Embedded Soil
Volumes box: the Volumes tab becomes available and can be used to specify the properties of
the volume. Figure 8 shows an example with soil volumes embedded into a uniform soil.

Figure 7: Arbitrary Heterogeneities soil type option in the legacy Soil Structure screen in MALT or MALZ.
Figure 8: Specification of soil volumes embedded into a uniform soil in SoilModelEditor.

5 Conclusion and Future Development

The new software tool SoilModelEditor facilitates and unifies the soil model specification for all
pertinent engineering modules and software packages. It is currently used to specify soil models
in SESTralin and SESCAD, and will soon be used for that purpose in all SES programs requiring
the specification of soil models. It can also be used as a standalone program, dedicated to the
editing of soil data.
Future versions of SoilModelEditor will be used not only for the specification of a single soil type
but also for definition of soil regions where each region may have a different soil type. Such a
feature will be useful for cases where a conductive network is distributed over large distances.
This typically takes place in AC interference studies where an exposed metallic line, for example,
a pipeline or a railway, shares the same corridor with energized transmission line for many miles.
The significant soil structure variations along such large distances strongly affect the results in
such studies and it is important to account for those variations. Therefore it is necessary to
partition the whole area of interest into geometrical regions each having different soil
characteristics. One such approach is currently employed in the Right-Of-Way software package.
In the future, SoilModelEditor will be able to generate soil regions based on the conductive model
geometry, soil resistivity measurement locations and a selected region-generation strategy.

17 NEW FEATURES AND ENHANCEMENTS FOR THE

CALCULATION OF SAFETY LIMITS IN SESTHRESHOLD
Said Touimer, Aditya Choubey, Martin Deslongchamps, Simon Fortin, Amir Hajiaboli, Sylvie
Lefebvre, Nina Mitskevitch, Robert D. Southey and Farid P. Dawalibi

Abstract
This article presents the new features and enhancements introduced during this year in the new SESThreshold
interface. Along with the integrated Zone Editor tool, SESThreshold allows to calculate the thresholds of multiple
zones. It also plots the results against these thresholds, in a single convenient plot. SESThreshold is presently
developed with safety limits in mind, but is intended to be generalized to other limits, such as the potential or the
electric and magnetic fields.
1 Introduction
Safety thresholds for touch and step voltages strongly depend on the resistivity of native soil and
presence or absence of the insulating soil material. Using at least two safety thresholds is very
typical for the analysis of the performance of a grounding system. The reasons for a larger
number of safety thresholds can be related to different scenarios, such as crushed rocks not
covering the entire area of interest, presence of different soil materials inside the site, presence
of buildings with concrete floor, presence of transformer pads, etc.
SESThreshold allows to create geometrical zones that represent areas with different safety
thresholds and computes the safety threshold value for each zone. It generates a 2-D spot plot
that displays the network computation results of the investigated quantity (touch or step
voltage), with respect to the thresholds of all selected zones in one single graph.
The drawing of the zones, done in the integrated Zone Editor application, is left entirely up to
the user having the grid and the profiles in the background as reference. The zones may
correspond to surface material variations, fence enclosed area, buildings (etc.). It is also possible
to exclude zones from the analysis, such as locations far outside a substation.
Once established, SESThreshold will replace the Safety module which is currently accessible
through CDEGS-Examine. Unlike the Safety module which reports the safety limits for both
touch and step voltages in one shot, SESThreshold has separated these two quantities so that the
regions of interests (zones) and their threshold specifications can be defined separately, based
on their own needs.
The application can presently perform threshold calculations for the touch and step voltages for
the IEEE80-2013 standard. Other standards will soon follow. Meanwhile, a few enhancements
and features have been developed, as summarized below:

For SESThreshold:
 Drag-and-drop of input (F05) and database (F21) files is now possible.
 The panels in the SESThreshold interface are now improved.
 Zones and specifications can now be transferred from Touch to Step and vice-versa.
 Within a given quantity (Touch or Step), specifications can now be copied from one
zone to another.
 The IEEE80-2000 standard options are now updated to IEEE80-2013.
 A Decrement Factor Tool is offered to help explain and visualize the decrement factor
concept.
 A Typical Range of Resistivities window is now available to give information about
the typical resistivity of different soil materials.
For Zone Editor:

 Zone Editor now allows the display of the vertex numbers and bounding box of a
shape.
 Primitive shapes can now be created and exploded in Zone Editor.
 Characteristics of a shape can be controlled during its construction.
For both SESThreshold and Zone Editor:

 Undo/Redo actions are now possible.
 A Quick-start Guide is now available when SESThreshold or Zone Editor help is
invoked.

2 New Features and Enhancements for SESThreshold
2.1 Drag-and-drop of Files
Drag-and-drop operations can now be used for opening an input Th_*.F05 file or for linking a
MALT, MALZ or HIFREQ database F21 file. A tooltip is displayed when the dragged file is not
supported by the application.
There are two ways to drag and drop an input file in SESThreshold. One is to go to the My Job
IDs folder in the Open tab of the backstage window, where the file can be dropped in the JobID
list (Figure 1). The other is to drop the F05 file directly into the zone viewer of SESThreshold
main window.
Figure 1: Drag-and-drop of an F05 input file.
When linking a new case to an F21 file, you can simply drag and drop the file as shown in Figure
2 below. However, the Link option must be selected prior to the drag-and-drop operation. Note
that it is now allowed to have a linked F21 file with a JobID name and folder location that differ
from the Th_*.F05 input file.
Figure 2: Drag-and-drop of an F21 linked database file.
2.2 Enhancements to the Panels of SESThreshold
The interface of the SESThreshold module has been re-organized and updated with some new
features (Figure 3).
 Network Specification panel: Since it is common to the whole project, this panel is
now detached in a separate expandable panel. Once the data are entered, the panel can
be collapsed to give more room to the Zone Specification tab. In the collapsed form, a
summary of the network data is displayed for quick information in the expander’s
header.
 Site data and Human data tabs: Since they are related to the selected zone, these
specification tabs have been grouped under the Zone Specification collapsible tab.
 Issues List panel: In order to offer more visibility to the important specification tabs of
the interface, this panel is now hidden by default. It can be opened anytime by clicking
on the issues summary located at the bottom right corner of SESThreshold. When an
issue is published, it will be notified in red in this summary. Moreover, if the issue is an
error or a warning, it will blink to draw the user's attention.
 Results panel: This panel has been improved to include a color level for the display of
the thresholds. When the option By Threshold Values is clicked, the threshold values
are sorted in ascending order and each zone is displayed with its corresponding
threshold color level in the zone viewer.

Figure 3: Enhancements to the interface of SESThreshold.
2.3 Copy of Zones and Specifications
After having defined the zones and specifications for either quantity, Touch or Step, it is now
possible to quickly define the other undefined quantity. When the selection has changed to the
undefined quantity, the program will ask whether the zones and specifications should be copied
to the undefined quantity:
Figure 4: Copy zones and specifications from one quantity (Touch or Step) to the other.
Take, for example, a scenario where the zones and specifications have been fully defined for the
Touch quantity while those for the Step quantity have not been defined yet. Upon switching to
the Step quantity, the dialog shown in Figure 4 appears. When Copy is clicked, the zones will
be copied along with their associated specifications from Touch to Step. The reverse is also
true if one defines the data for Step before the data for Touch.
Within a given quantity (Touch or Step), it is possible to copy the specifications of one zone to
another zone. This new copy feature is available in the Home tab of the ribbon in the
Specification group (Figure 5):
Figure 5: Copy specifications from one zone to another, for a given quantity.
After the selection of the target zone, the Copy From list can be used to select the zone from
which the specification should be copied. If the copy is possible, the Apply to Selected Zone
button will become available and clicking on it will perform the copy.
2.4 Options Updated to the IEEE80-2013 Standard
The IEEE80-2000 standard options are now updated to IEEE80-2013. Regarding safety, the
two standards offers the same formulation for the calculation of the foot resistance and safety
limits. Therefore, all references to IEEE80-2000 were simply replaced by references to the more
recent IEEE80-2013 standard.
2.5 Decrement Factor Interactive Tool
A decrement factor tool is currently accessible through the Interactive Tool… button of the
Network Specification section (see Figure 6).
Figure 6: Access to the Decrement Factor Interactive tool.

The main purpose of this tool is to explain the decrement factor concept and how it can be
included in a safety study. It calculates the decrement factor according to the formulation
described in IEEE80-2013 and plots a transient response for illustration purposes. It also
provides the peak voltage factor (per unit of symmetrical rms value), which can be useful to
study the zone of influence in the case of interference studies of telecommunication lines and
equipment [1-2].
As shown in Figure 7, some information about the decrement factor concept can be found under
the Help expander located at the top of the window. The X/R ratio can be entered directly in the
input field or can be specified with the slider located below. Alternatively, the decrement factor
can be the input, in which case, X/R ratio will be computed. The frequency and time duration
used in the calculations are those set in the SESThreshold main window (Figure 6). Once all
inputs have been specified, the graph and the output fields will then show the corresponding
curve and computed results instantaneously.
Figure 7: Decrement Factor Interactive Tool.
The results obtained by the Decrement Factor Tool can be exported to the main window of
SESThreshold in order to be used in the safety limits calculations. This can be done by simply
clicking on the OK button. The Cancel button can be used when the results are to be ignored;
the values specified in the main window will then be preserved.

Finally, it is important to mention that use of this tool is not required to calculate or to define
the decrement factor; they can be specified directly in the main window of SESThreshold
(Figure 6).
2.6 Typical Range of Resistivities Window
A Typical Range of Resistivities window is now providing information for the typical
resistivity of different soil materials. The information given in this window can help to choose an
adequate surface material in order to achieve safety. Figure 8 shows a screenshot of this window
(see Figure 8).
Figure 8: Typical Range of Resistivities for different soil materials.
The window can be accessed by clicking on the Typical Ranges button in the Surface Layer
panel under the Site Data tab as shown in Figure 9. The typical ranges shown can become
useful whenever the calculation method for Foot Resistance is set to something other than
None or User-Defined (e.g., IEEE80-2013 (Two Layers)).

Figure 9: Access to the Typical Range of Resistivities window.
3 New Features and Enhancements for Zone Editor
3.1 Display of Vertex Numbers and Bounding Box
The vertex numbers and the bounding box of any shape can be displayed by using the Show
Vertices or Show Bounding Box options located in the View group of the ribbon’s Home
tab:
Figure 10: Show Vertices and Show Bounding Box options.
This feature is very helpful to locate a particular vertex when the concerned shape has many
vertices (see Figure 11 ).

Figure 11: Zone Editor interface.
3.2 Primitive Shapes
In Zone Editor, primitive shapes like rectangles, ellipses and circles1 can now be drawn in their
original form (Figure 12). The properties of the shape (radius, segments, etc.) are shown in a
grid located at the right of the interface and can be modified easily. However, after closing the
Zone Editor window or after clicking on Explode, the shapes behave like ordinary polygons
with vertices. Once converted to polygons, the vertices can be edited in the Vertices Editor
located on the right side of Drawing Area.
1 To draw a circle, click on shift key while drawing an ellipse.

Figure 12: Properties of the primitive shapes.
When drawing a primitive shape with the mouse, the main characteristics of the shape are
reported instantly in an adjacent tooltip (see Figure 13). In the case of regular polygons, the
same tooltip is available and allows to adjust the length and orientation (angle) of the segment
under construction.

Figure 13: Drawing tooltips.
4 New Features and Enhancements for both SESThreshold

and Zone Editor
4.1 Undo/Redo
All operations in SESThreshold and in Zone Editor can now be undone and redone using
the Undo and Redo buttons in the Home tab on the ribbon.
An Undo/Redo list of actions is also available to see the list of actions that have been
performed in the main window of SESThreshold or, in Zone Editor, starting from the time
that Zone Editor has been loaded. Using this list, it is possible to undo or redo a subset of
actions as shown in Figure 14 below.
SESThreshold Zone Editor
Figure 14: Undo / Redo list of actions.

4.2 Quick-start Guide
A Quick-start Guide is now available when SESThreshold or Zone Editor help is invoked
(see Figure 15 below). This guide describes briefly how to use SESThreshold and Zone
Editor in order to calculate the safety limits and to plot the touch and/or step voltages with
respect to these limits. It provides a short description of the main features that are accessed
during this step-by-step process.
Figure 15: Access to the Quick-start Guide from the Help menu.
5 Conclusion
This year’s improvements address some of the most demanded features for SESThreshold. In
order to get started quickly, a Quick-start Guide can now be accessed from the Help menu
and explains step by step how to use SESThreshold and ZoneEditor. The software is
currently offering the calculation of the safety limits within the guidelines of the more recent
IEEE80-2013 standard. Some other enhancements were introduced to make the interface more
user-friendly, namely the drag-and-drop of files, the reorganization of the panels and input data,
the capability to copy specifications from one zone to the other and, finally, the possibility to
define quickly the inputs for the Step quantity once the Touch inputs are defined. Creating
zones also becomes easier with the help of the new drawing tooltip, the primitive shapes and the
Show Vertices option. Moreover, any operation can now be undone or redone. Other features
were also implemented in order to provide a useful complementary information about the
decrement factor and the typical range of soil resistivities.
6 References
[1] J. Liu and F. P. Dawalibi, "Zone of Influence around Electrical Installations Subjected to Ground
Faults", International Conference on Electrical Engineering 2007 (ICEE), Hong Kong, 2007
[2] N. Mitskevitch, F. P. Dawalibi, J. Ma and J. Liu, "GPR Zone of Influence of a Typical Electric Power
Network", International Conference on Power System Technology (POWERCON2006), Chongqing,
China, 2006.

18 INTRODUCING THE NEW SES OBJECT

LIBRARY
Sheng Wei, Eric Dawalibi, Greg Noel, Natela Botchorichvili,

Jie Liu, Majid Siahrang, Yixin Yang, Tapabrata Mukherjee, and Sofiane Chabane

Abstract
This article introduces the new SES Object Library. This library includes the existing SES Conductor Database and
also introduces a cable database. This new tool offers most of the useful features of the existing conductor database
as well as many new features, such as sorting by the characteristics of the conductors or cables, and categorizing
the items by industry, country, voltage level, application, conductor type, etc., providing a straightforward and
efficient way to access the information in the database.
1 Introduction
The new SES Object Library which is unveiled at this year’s UGM, is an enhancement to the
existing SES Conductor Database. In addition to the conductor database, the new library offers a
cable database and will progressively include several other component databases, such as a GIS
database, a database of transmission line structures, a materials database, a transformer
database, etc.
The new SES Object Library is designed to be used as a standalone independent program or to be
called upon from the computation modules and related tools by importing data from a database
of interest. Typical modules and tools are, HIFREQ, MALZ, MALT, FCDIST, TRALIN, SPLITS,
SESCAD, ROWCAD, Right-of-Way, SESImpedance, etc.
This article introduces the preview version of SES Object Library in greater detail in the following
sections.
2 Conductor Database
The conductor database stores information about various types of conductors. Figure 1 shows the
four main types of conductors which are stored in the conductor database of SES Object Library.
The conductor database stores a large amount of data regarding the geometrical and electrical
characteristics of the conductors, including their cross section (area), inner and outer radii,
relative resistivity (with respect to copper), relative permeability (with respect to free space), DC
and AC resistances, geometric mean radius, reactance at one foot spacing, number of strands,
strand radius, and more.

Figure 1: Typical types of conductors in SES Object Library.
The new SES Object Library introduces a mechanism of categorization of all items stored in the
library. The items are arranged in groups so that they can be located quickly. As shown in Table
1, conductors are classified in five main groups: industry, country, voltage level, application, and
conductor type.
Table 1: Categorization Options for Conductor Database
Category
Category Options
Groups
Power Electric, Pipeline, Railway, Telecommunications, Civil (or

Industry
Construction), …
Australia, Canada, China, France, Germany, Japan, South Korea, United

Country
Kingdom, United States, …
Voltage Level EHV (230+ kV), HV (69+ -230 kV), MV (1-69 kV), LV (below 1 kV), …
Overhead Phase Conductor, Overhead Neutral Conductor, Overhead Shield

Wire, Messenger Wire, Grounding Conductor, Bonding Conductor, Bus Bars,
Application
Reinforcing Steel, Mitigation Conductor, Counterpoises, Lightning
Protection Conductor (including mast), Fence Post, …
Aluminum: 3M ACCR, AAC, AAAC, AACSR, ACSR, AASC, ACAR,

ALUMOWELD, ASC (A1), …
Copper: COPPERWELD – 30, 40, COPPERWELD CCS – 40, SOLID
COPPER, STRANDED COPPER …
Conductor
Type Optical: OPT – GW
Steel: SC/AC, STEEL, STEEL (ESKOM), ZINC PLATED STRANDED STEEL
(GB1200-75), …
Others: PIPE-STEEL, RAIL, RIBBON ANODE …

These categories act as filters: selecting a specific category in a group restricts the list of displayed
conductors to those that are included in that category.
Note that the following general options are available for each group:
 All: All the items will be displayed in the database table.
 Non-categorized: The items to which no category option is assigned will be displayed in
the database table.
 Generic: The items that belong to all category options under the selected group will be
displayed in the database table.
3 Cable Database
Defining cables in the applicable modules of SES Software is quite intuitive and straightforward.
However, extracting the required data regarding the geometrical and the relevant physical
characteristics of the cable components from cable data sheets (if they are available at all!) may
sometimes become a demanding task. The long-awaited cable database of SES Object Library
addresses this issue.
The cable database of SES Object Library is planned to include a wide range of cables of different
types, structures and specifications. This database stores data regarding the geometrical and
physical characteristics of the cables. Each cable available in the cable database is linked to a
SESCrossSection file in which the cable is modeled as a coaxial single core or a multi-core
structure. Figure 2 shows the two mentioned general structures for cables in the cable database.
The data stored in the SESCrossSection file will be transferred from the database to the calling
module to define a selected cable.
Figure 2: Single-core and multi-core structures for cables.
Note that the cable database is still under development and the unveiled version is introduced
mainly for testing purposes and for gathering feedback.

Although not implemented yet in the current release of the software, the cable database will be
equipped with search tools and categorizing functions specific to cables to help users find what
they look for in a systematic and straightforward manner.
The planned categorization options for the cable database are shown in Table 2.
Table 2: Categorization Options for Cable Database
Category
Category Options
Groups
Industry Power Electric, Telecommunications.
Australia, Canada, China, France, Germany, Japan, South Korea,

Country
United Kingdom, United States, etc.
Voltage Level Nominal Voltage of Cables.
Insulation
100%, 133% or 173%
Level
Insulation
PE, PVC, XLPE, EPR, Impregnated paper, etc.
Type
Conductor
Aluminum, Copper, Special conductors, etc.
Type
Power transmission, Power distribution, Grounding Cable, Bonding

Application
Cable, Mining, Submarine, etc.
Shield Type Tape, Corrugated, Concentric neutral, Composite laminated, etc.
4 The Graphical User Interface

This version of the new SES Object Library is a standalone application. The main window of the
program is shown in Figure 3. It contains a ribbon and five panels (i.e. Databases, Properties,
3D Viewer, Issues List and Search/Search Results panels).

Figure 3: Main interface of SES Object Library.
4.1 Ribbon
SES Object Library shares the general template of all the new user-interface programs developed
by SES over the past few years. The ribbon is designed to help you quickly find the commands that
you need to complete tasks related to components. Commands are organized in six groups under
the Home, View, Tools, Library, Settings and Help tabs on the ribbon, as shown in
Figure 4.
The Home tab contains the most important commands for the program. The Display group
contains the Categories and Display Units commands. The Categories command leads to the
Category Filter dialog window, which is used to customize the display of components. The
Display Units list offers a list of systems of units for displaying the data in the database table
and the properties panel.
The View tab on the ribbon contains options that control the way you want to see the display.
The Panel group controls the presence of the panels. Panel Visibility can be used to select the
panels you want to see, Show Default Layout restores the standard layout of the panels and
Show All will show all panels.
The Library tab on the ribbon contains the Categories command, which launches the
Category Filter dialog window. The library and category selection boxes in the Category Filter
group are synchronized with those in the Category Filter dialog window. They can also be used
to change the filter selection.
The Help tab on the ribbon contains help resources such as the Examples, Manuals, Video
Tutorials, FAQ and Publications. The SES Support Services group links to software
support from SES, such as SES Web Site, Contact SES Support, About, Release Notes,
Check for Updates, and Browse to SES Folders.

Figure 4: Ribbon of SES Object Library.
4.2 Dialog Windows and Panels
The Category Filter dialog window allows you to categorize the database according to pre-
defined classifiers. Different categories are available for different databases. When selecting a
database from the Component Library Selection list, the categories applicable to that
database are shown in the lower section of the screen, allowing you to restrict the items that are
to be displayed in the main panel. For example, there are five categories for the conductor
database, i.e. Industry, Country, Voltage Level, Application, and Conductor Type.

Figure 5: The Category Filter dialog window for the conductor database.
The panels are the main viewing and editing areas of the program. Each panel is detachable and
can be hidden. The default look of SES Object Library has four panels as follows:
(1) The Databases panel displays the list of database items that match the settings specified
in the Category Filter dialog window. The names and the main parameters of the components
are listed in the database table.
Note: The icon in the Status column indicates that the item is currently being edited.
Different databases are displayed in this panel. You can switch between the databases by clicking
on the corresponding tabs as shown in Figure 6. The Category Filter dialog window will
automatically allow you to filter the items to be displayed.

Figure 6: The Databases panel.
(2) As shown in Figure 7, the Properties panel is used to display or edit the configuration of
a selected component in the Databases panel. The SESImpedance application can also be run
from this panel. However, presently there is no direct data transferring between SES Object
Library and SESImpedance. In the future, the parameters specified in this panel will be
automatically saved to a SESImpedance file.

Figure 7: The Properties panel.
(3) The 3D Viewer panel displays the selected component graphically (see Figure 8).

Figure 8: The 3D View panel.
(4) The Issues List panel lists all the errors, tasks, warnings and messages regarding the
input data and operations.
Figure 9: The Issues List panel.
5 Using SES Object Library
5.1 Finding Information about Conductors or Cables
SES Object Library contains a large number of items. To help finding items in a straightforward
and efficient manner, filtering options are available to narrow down the list of items displayed in
the database table. The filtering features can be controlled from the Category Filter dialog
window, as shown in Figure 10. All ACSR conductors having Canadian standard sizes used as
overhead shield wires in the electric power industry are displayed in the table as shown in Figure
11. The category filters are also available in the Library tab of the ribbon.

Figure 10: The Category Filter dialog window for filtering database items.
Figure 11: The Category Filter in the Library tab on the ribbon.
Note that if the program cannot find any item matching the selected filtering options, the Apply
button will be unavailable.
5.2 Sorting the Data in the Database Table
The data shown in the database table can be sorted according to any column of the table. This can
be useful to help selecting a target item when there is a large number of items listed in the database
table.

For example, if you need to find the ACSR conductor with a cross section of 2156 kcmil, you can
use the Categories filter to select ACSR from the Conductor Type category. Next, click on the
Cross Section (kcmil) column header to sort the listed conductors from largest to smallest cross
section area or vice versa. You can then easily find the 2156 kcmil conductor among all ACSR
conductors, as shown in Figure 12.
Figure 12: Sorting the data in the database table.
5.3 Displaying the Database in Different Systems of Units
SES Object Library stores and saves all data internally using the metric (SI) system of units. When
data is displayed in the table or the property panel, the data is converted according to the user-
specified display units.
The system of units can be chosen in the Home tab on the ribbon as shown in Figure 13. There
are four options available:
 Imperial,
 Metric,
 Centimeter Radius,
 Inch Radius.
You can simply select any one of them and the database table will display the data accordingly.

Figure 13: Displaying data in different systems of units.
5.4 Calculating the Electrical Properties of Conductors through

SESImpedance
Quite often, the electrical characteristics of conductors and cables are known: they may be
obtained from a manufacturer data sheet or correspond to a well-known conductor. However,
there are situations where you have the data regarding the geometrical and physical
characteristics of a conductor, but where the electrical properties of the conductor, such as its
equivalent relative resistivity and permeability, are not available. In this case, SES Object Library
provides a quick link to run the SESImpedance tool, as shown in Figure 14. In the SESImpedance
tool, you can define the geometrical and electrical characteristics of the conductor and calculate
its electrical properties. Furthermore, the computation results can be manually brought back to
the database to add your own item in SES Object Library.
Figure 14: The interactive tool SESImpedance called from SES Object Library.

As an example, let us consider conductor 7 No. 10 in the COPPERWELD CCS-40 (i.e. CW CCS-
40) group. This conductor consists of seven identical copper clad steel conductors. Figure 15
shows one of strands of this 40% copper clad steel conductor.
Figure 15: The configuration of a single strand conductor in 7 No. 10 of COPPERWELD CCS-40.
The corresponding geometrical configuration and physical properties of the conductor are listed
in Table 3.
Table 3: Characteristics of 7 No. 10 of COPPERWELD CCS-40
Specification Characteristics
Outer Radius 0.39 cm
Number of Outer Strands 7
Outer Strands Radius 0.13 cm
The core is steel and the copper cladding

Materials
represents 10% of the overall wire diameter.
Figure 16 shows the SESImpedance environment in which this conductor is defined. The
computed equivalent relative resistivity and permeability are listed in the Computation
Results panel.

Figure 16: Computation of equivalent electrical characteristics of conductors in the SESImpedance tool.
5.5 Specifying User-Defined Categories
In SES Object Library, you can specify different available category options to classify items. First,
select the items you want to categorize. Click on the Edit button in the Properties panel and
select the categories in the Specified Categories list. Note that you can hold the Ctrl key to
select several categories at once. The screen is shown in Figure 17. Please note that, selecting the
Generic option is equivalent to selecting all the available items under a category group.

Figure 17: The category specification panel.
6 Conclusions and Future Developments

This article has introduced the new SES Object Library. The introduced version is still a preview
version of the product and the official version of the product is yet to be released with more
features and functionalities. The following major features have been planned for the official
release of SES Object Library:
 Search tools (Use keywords or values to locate the items in the database).
 Integration between SES Object Library and applicable modules of SES software (such as
the HIFREQ module and the Right-of-Way and AutoGrid Pro packages).
 Online context sensitive help.
 Automatic data interaction between SESImpedance files and the conductor database.
 Inclusion of GIS and GIL databases.
We encourage SES Software users to provide us with their feedback and any suggestions they may
have about SES Object Library by sending us emails at feedback@sestech.com.

19 NEW FEATURES IN SESEBUNDLE: FINDING AN

EQUIVALENT HOLLOW CONDUCTOR REPRESENTING A
BUNDLE OF CONDUCTORS
Octavio Ramos-Leaños, Jie Liu, Carleen Cheng, Eric Dawalibi, Stéphane Franiatte,
and Yixin Yang
Safe Engineering Services and technologies ltd.

Abstract
This article describes a new feature of SESeBundle which allows representing a circular formation of solid
conductors on the same phase as a single hollow conductor, allowing its use in existing cable models, such as those
available in the HIFREQ and TRALIN modules. The validity of this “hollow equivalent conductor” approach is
demonstrated using an induction study in the HIFREQ module.
1 Introduction
SESeBundle is a tool that can be used to find the characteristics of a single conductor electrically
equivalent to any set of conductors connected in parallel (bundle). This tool was introduced for
the first time at last year’s Users’ Group Meeting. In that first version, the tool was restricted to
finding equivalent solid conductors.
In this version of SESeBundle, a new functionality has been introduced to create equivalent hollow
conductors. Such equivalent hollow conductors can be useful when modeling power cables for
grounding and AC interference studies. Some commonly encountered types of power cables
include concentric wires and/or a wire armor composed of several individual conductors. Figure
1 shows an example of such a cable. Direct modeling of these types of cables can be complicated
given that the cable models currently available in SES Software only consider continuous sheaths
and armors.
Figure 1: Cross section of 400 kV underground cable. Photo credit to [2].

This new functionality not only allows creating cable models that would otherwise be impossible
to build in TRALIN, RightOfWay, and HIFREQ but it can also help to reduce memory usage and
increase computational performance.
Note: For consistency with the general terminology used in SESeBundle, concentric wires and
wired armors will often be referred to as “bundles” in the remainder of this article.
2 Methodology
The objective of the hollow equivalent feature of SESeBundle is to find the external radius 𝒓𝒆𝒙𝒕 ,
internal radius 𝒓𝒊𝒏𝒕 and relative resistivity 𝝆𝒓 of a single equivalent conductor that accurately
replaces a bundle of conductors in the sense that the equivalent conductor and the bundle yield
the same series impedance. The characteristics of the equivalent hollow conductor calculated by
SESeBundle are selected to reproduce not only the self-impedance of the bundle but also that of
the core conductor and the mutual impedance between the core and the bundle. This currently
needs to be done without any knowledge of the core conductor’s electrical and geometrical
characteristics. Tests have shown that the safest way to do this is to require the relative
permeability 𝝁𝒓 of the equivalent conductor to always be set to 1. The cross-sectional area and
resistivity of the equivalent conductor are then adjusted to reproduce the self-impedance of the
bundle. To obtain the equivalent, the TRALIN module is used to compute the total impedance of
the bundle and the internal impedance of the individual conductors in the bundle. Then, the
external impedance of the bundle can be calculated and the external radius 𝒓𝒆𝒙𝒕 can be obtained.
A single conductor of a given internal radius 𝒓𝒊𝒏𝒕 can then be found having the same external
impedance, internal resistance and internal reactance as the bundle, although its equivalent
resistivity will generally be different than those of the conductors in the bundle.
SESeBundle carries out the following computation steps:
(1) Use the TRALIN module to compute the total impedance of the bundle and the internal
impedances of all sub-conductors in the bundle.
(2) Calculate the internal resistance, and the internal and external reactance of the bundle.
Use the external reactance to derive the external radius of the equivalent hollow
conductor; then, use the internal impedance with provided external radius to calculate
the internal radius and equivalent resistivity.
3 The User Interface

The new user interface of SESeBundle, shown in Figure 2, is very similar to that presented last
year [1], with a few changes to accommodate the hollow equivalent feature. A new Equivalent
Type dropdown list has been added on the Home tab. This list defines the type of equivalent
conductor that will be computed and offers two options: Solid Conductor or Hollow
Conductor.

Figure 2: New main window of SESeBundle.
The Hollow Conductor option allows representing a bundle of conductors as a hollow

conductor, as shown in Figure 3. This option is only available when using a Circular Pattern
Type, in Simple Specification Mode. This limitation is due to the fact that the implemented
method requires a well-defined external radius. Note that a hollow equivalent can also be obtained
for a solid conductor by setting 1 as the number of sub-conductors in the bundle.
The next section shows some examples of the use of this tool.
4 Examples and Validations

The examples described in this section demonstrate that the equivalent hollow conductor
computed with SESeBundle can properly replace bundles while maintaining the self and mutual
impedances of the coaxial conductor system of which they are part.

Figure 3: Concentric wires and their hollow equivalent.
4.1 Single Core Cable
This example consists of a single core 3.3 kV underground cable. The cross-sectional configuration
of the cable is shown in Figure 4 and its characteristics are given in Table 1. Direct specification
of this type of armors is difficult and sometimes impossible in many SES software programs. This
example can be find under UGC 2016\Example Files\BE_CableEquivalent.f05.
In order to introduce the characteristics of these conductors in SESeBundle, select the Simple
option under Specification Mode on the ribbon (see Figure 2). Introduce the electrical
characteristics of aluminum in the Conductor Characteristics section of the Bundle
Definition panel. In the Replication Details section, select Circular for the Pattern Type
and enter the values shown in Figure 5. Once the bundle is specified, click Compute on the
ribbon. The results are given in the Computation panel, as shown in Figure 6.
The properties of the equivalent hollow conductor of Figure 6 computed by using SESeBundle can
be easily introduced in all SES software cable specification tools.
Figure 4: Cross-sectional view of a 3.3 kV single core cable.

Table 1: Underground Cable Characteristics
Component Dimension/Quantity
Copper core radius 4.6 mm
Insulation thickness 2 mm
Aluminum armor wire radius 0.625 mm
Number of aluminum wires 36
External insulation thickness 1.8 mm
Figure 5: Bundle definition.

Figure 6: Computed characteristics of equivalent hollow conductor.
4.2 Validation
4.2.1 Self and Mutual Series Impedances
The self and mutual series impedances for the cable were computed with TRALIN, at power
frequency, using both an exact model including all the wires of the armor and a model where those
wires are replaced by the equivalent hollow conductor whose characteristics were computed with
SESeBundle. The results, given in Table 2, show that the series impedance of the core is the same
in the two models, while the series impedance of the armor is larger by 6% (and the mutual
impedance between the core and armor by 10.1%) in the model using the equivalent hollow
conductor. These results can be computed using UGC 2016\Example Files\
TR_UGMRealCableImpedance.f05 and TR_UGMEquvalentCableImpedance.f05. Table 2 also
presents a comparison when the equivalent is obtained by using a simple geometrical equivalency.
The equations for this simple method are:
𝑟𝑒𝑥𝑡 = 𝑟 + 0.5𝑡 (1)
𝑟𝑖𝑛𝑡 = 𝑟 − 0.5𝑡 (2)

𝑁𝑟𝑠2
𝑡= (3)
2𝑟
where
𝑟𝑒𝑥𝑡 , equivalent external radius.
𝑟𝑖𝑛𝑡 , equivalent internal radius.

𝑡, equivalent’s thickness.
𝑟, bundle radius.
𝑁, number of strands in the bundle.
𝑟𝑠 , radius of a bundle strand.
Note that the relative permeability of the equivalent obtained with the simple method needs to be
set to 1. The obtained results from the simple method equivalent present a relative error
comparable to that obtained with SESeBundle for this case. Results for this case can be computed
using the example file UGC 2016\Example Files\TR_UGMEquivalentSimpleMethod.f05.
Table 2: Underground Cable Characteristics
Component Series Impedance (Ω/km)
Case Self-Core Self-Armor Mutual
Original 0.9912 ∠ 72° 1.0621 ∠ 49° 0.8066 ∠ 86°
SESeBundle Equivalent 0.9912 ∠ 72° 1.1254 ∠ 51° 0.8877 ∠ 86°
SESeBundle Error 0% 6.0% 10.1%
Simple Method Equivalent 0.9912 ∠ 72° 1.1247 ∠ 52° 0.8893 ∠ 87°
Simple Method Error 0% 5.9% 10.3%
4.2.2 HIFREQ Current Induction
A test circuit was created in HIFREQ to validate the results obtained when the Equivalent Hollow
Conductor is used. The circuit is shown in Figure 7. The cable is 5 m above ground so the bundle
can be created as shown in Figure 8. The cable length is 1000 m and it is subdivided into 33.34 m
long sections. The substation grids are 50 by 50 m squares 0.5 m deep in the ground. The
characteristics of the core and armor conductors are taken from Table 1. The same test was
performed using the cable model of HIFREQ with a sheath having the characteristics of the
equivalent hollow conductor computed by SESeBundle given in Figure 6. Files to this example are
UGC 2016\Example Files\HI_UGMEquivalentCable.f05 and HI_UGMRealCable.f05.
The results for this test are given in Table 3. These results show that the obtained induction and
conduction properties of the equivalent cable model are very close to those of the original model,
with only 0.1 % difference in the results. As a result of the reduction in the number of conductor
segments used in the model, the computation time is about 70 times smaller in the equivalent
cable model case.

Figure 7: HIFREQ test set up.
Figure 8: HIFREQ wired armor cable model.
Table 3: HIFREQ Test Results
Case Receiving end Armor Computational

Current (A) Time (s)
Original wired
770∠6° 156.7
armor cable
Equivalent 769∠6° 2.3
5 Limitations
The characteristics of the equivalent hollow conductor calculated by SESeBundle are selected to
reproduce not only the self-impedance of the bundle but also that of the core conductor and the
mutual impedance between the core and the bundle. This currently needs to be done without any
knowledge of the core conductor’s electrical and geometrical characteristics. This leaves only the
internal radius as a variable that can be adjusted to try reproducing all required impedances.
There are cases where this is impossible, or where the resulting equivalent hollow conductor has
an internal radius that is too small (smaller than the radius of the core conductor). In the former
case, SESeBundle returns -1 for all parameters of the equivalent conductor to indicate the
limitation, as shown in Figure 9. The condition on the inner radius must be validated by the user,
to verify if the equivalent hollow conductor is suitable for a given cable model.

Figure 9: Results for a limiting case.
6 Future Work
In order to avoid the issues mentioned in Section 5 information about the core conductor will be
introduced to help obtain a precise equivalent with valid geometrical characteristics.
7 Conclusion
This article has introduced the new hollow conductor equivalent feature of SESeBundle. An
example was given to show its capabilities and limitations and to demonstrate the usefulness of
such a tool when it comes to model bundles in concentric disposition. It was shown that models
using an equivalent hollow conductor can produce very accurate results. More importantly, it was
shown that this feature can be used to approximate cable configurations that would otherwise be
impossible to represent in cable models such as that in HIFREQ.
8 References
[1] Carleen Cheng, Eric Dawalibi, Stéphane Franiatte, Jie Liu, Octavio Ramos-Leaños,
Yixin Yang, "SESeBundle: Finding an Equivalent Single Conductor Representing a
Bundle of Conductors," in CDEGS Users' Conference Proceedings,San Diego, California,
USA, 2015.
[2] By Wdwd - Own work, CC BY-SA 3.0,

https://commons.wikimedia.org/w/index.php?curid=10282825

20 NEW FEATURES AND IMPROVEMENTS IN ROWCAD
Maxime Daigle, Yexu Li, Luis Valcárcel, Christian J. Voyer

Abstract
This paper presents the new features and improvements introduced in ROWCAD since the last User’s Group
Meeting. The most salient of these new features are the capability to directly import kml (Google Earth™) files, the
user control over the minimum region cut length used in the region generation process, and the modifications made
to the cable group definition, allowing multiple cables to be defined in a single cross-section, without the need to
include a pipe enclosure.
1 Introduction
RowCAD is continuing its evolution as the preferred input mode for Right-of-Way Pro projects.
This year, the new functionalities mainly aim at increasing productivity and flexibility. The direct
import of kml files will undoubtedly prove to be a great time saver for many projects where initial
data is already prepared in that format. A restriction has been lifted for cable paths definition; a
pipe enclosure is no longer necessary in order to define more than one cable on a given path. With
the new implementation of the minimum region cut length, more control on region generation is
given to reduce project size without sacrificing accuracy.
2 New Features
2.1 Direct Import of Keyhole Markup Language Files
Google Earth™ is a tool widely used in the planning of Right-of-Way studies and sometimes, when
no precise GPS surveys have been made, the only available source of information regarding the
geographical location of power line structures or pipeline staking, or regarding the dimensions of
substations.
The latitude and longitude coordinate system used by Google Earth files is well adapted for
locating positions on a round Earth. The coordinate system used in SES software, however, is not
spherical, but planar (for simplicity), since systems to be modeled are local enough for the
curvature of the earth to be ignored. It is therefore necessary to transform the latitude and
longitude onto a flat 2-dimensional surface for use in SES Software.
RowCAD can now import a keyhole markup language file (Google Earth kml file) in a project and
uses the Universal Transverse Mercator (UTM) projection with the World Geodetic System 1984
(WGS84) datum surface (an oblate spheroid) to convert the curved system into SES Software
compatible 2-D coordinates. The elevation (altitude) of points is not considered in the conversion,
following the equations derived by Karney [1]. Note that if you are in possession of a kmz file, it
can easily be converted to kml by opening the file in Google Earth and saving as kml.

2.1.1 Google Earth Paths into RowCAD Polylines
Paths in the kml file are naturally transformed into polylines for use in RowCAD. The prefix ‘ckt:’
(not case sensitive, make sure to include the space) can be added to the name of paths in order to
have RowCAD automatically assign the Tower point type to the associated polyline created.
Otherwise, the point type will be set to the standard default Undefined.
2.1.2 Google Earth Placemarks into RowCAD Entities
Placemarks are used to mark important locations of various types. Soil measurement, substation,
valve station, and line characteristics change (tower configuration, pipe diameter, etc.) locations,
among many others, are conveniently imported into RowCAD alongside polylines as entities. The
default entity type is a Note, but if the prefix ‘custom:˽’ (not case sensitive, make sure to include
the space) is added to the placemark name, the entity type is set to Custom Entity allowing you
to associate a MALZ file of a grid or valve station for the ultimate purpose of creating a MALZ
template file.
As the RowCAD model is being created, locating cross section or phase leakage change locations,
as well as making soil assignment decisions, becomes substantially simpler and more precise.
Once Generate Regions has been run, it similarly is much easier to find which circuit sections
correspond to these locations.
2.1.3 System Centering and Multiple Imports
By default, when a set of Google Earth paths and placemarks are imported, RowCAD uses the
average longitude of all objects to determine a UTM zone for conversion for all points in the
system, rather than convert each object’s coordinate based on its own longitude. This way, if an
import includes elements straddling across a UTM zone boundary, the system remains
continuous and maintains the real world relative coordinates.
Also inherent to a standard UTM conversion, the system origin is assumed to be at the intersection
of the zone center meridian and the equator. Consequently, for the vast majority of projects, the
x and y coordinates of a system imported into RowCAD will be very large (on the order of
thousands of km). This is generally not advisable for MALZ models as it can lead to some
numerical round-off errors. Moreover, from a user-interaction standpoint, this can be confusing.
Consequently, it is convenient to be able to set the origin in such a way as to simplify the
interpretation of results.
The kml import feature of RowCAD includes a special recognition for a placemark with the name
origin. The polylines and entities will be automatically shifted such that the origin entity location
be (0, 0), thereby avoiding large coordinate issues completely. The origin location, if included in
the Google Earth file, is also used to set the UTM zone for conversion instead of the average
longitude of all objects.
If the importing of various elements needs to be done in multiple import actions, simply copy the
origin placemark to ensure that each subsequent import is properly centered. It is essential that
the same origin location be used, otherwise each portion from different imports will be shifted
from one another.

Figure 1: Original Google Earth file and resulting imported project using the example file Laval System
with Origin.kml (see example folder).
2.1.4 Special Case: Structure Locations Folder and MultiGeometries
Some third-party software packages produce kml files that model transmission lines with
sophisticated drawings including catenaries that describe shield and phase wires, along with
detailed drawings of towers or poles. Such a circuit is successfully imported as a polyline, although
in these circumstances, there are often nodes for every small segment used to describe the
catenary, making a direct interpretation impossible. In some cases, a special polyline is exported
with nodes only at tower locations. In such a case, this polyline can more easily be imported to
represent the circuit in question.
In the absence of such a polyline, one redeeming feature is the possible existence of a folder
containing a series of placemarks that indicate the location of each tower. As per section 2.1.2, by
default a separate entity would be created for each placemark. If the series of placemarks is
however placed in a folder with the name structure locations (not case sensitive), upon import
into RowCAD a polyline will be created with nodes at each placemark location, with a point name
corresponding to the placemark name, and with a Tower point type. Note that every placemark
occurring after the structure locations folder will be interpreted as part of the transmission line,
whether inside the folder or not. If an origin placemark is to be defined, it should therefore be
placed just before the structure locations folder. Moreover, any kml file that includes a structure
locations folder will be imported to extract these tower locations.
As a best practice, if a structure locations folder needs to be imported, simply create a folder to
include first the origin, then the structure locations folder with nothing else. This will not only
ensure a proper interpretation of the placemarks but also save time; third-party generated kml
files can be very large and lead to long import time.

Figure 2: Recommended Structure Locations case organization.

Finally, these types of files often contain composite objects called MultiGeometries. These can
only be created externally from Google Earth. The RowCAD kml import feature currently ignores
objects included in MultiGeometries.
2.1.5 Import Summary Message
Upon successful import, a message summarizing the number of polylines and entities imported
is shown. The message will indicate if the conversion was limited to the interpretation of a
structure locations folder. The message also indicates the latitude and longitude of a user-defined
origin, if it was found, as well as the UTM zone and hemisphere used for the conversion.
Figure 3: Message summarizing the kml file import.
2.1.6 Usage Summary and Tips
Using this feature should in most cases be quite simple: Google Earth paths are transformed into
RowCAD polylines, and placemarks are transformed into Entities. The system is recentered
around the placemark with the name origin. Google Earth paths whose name have the prefix ckt:˽
will automatically be assigned the Tower point type in RowCAD, others will have the standard
default Undefined point type. Placemarks whose name have the prefix custom:˽ will be
converted to Custom Entities, others to Notes.
For subsequent imports for additions or modifications, simply make sure the same origin
placemark is present.
As is the case in any import for any software, keep the kml to be imported as simple as possible.
The import should be fast even for complex models, as long as only the polylines and placemarks
useful to RowCAD are included.
With the exception of folders named Structure Locations, the import feature ignores the folder
structure entirely. The order in which elements are imported is the order in which they appear in
the Google Earth tree view.

If you need to make use of the special Structure Locations interpretation, it is best to organize a
separate kml file, as shown in Figure 2.
2.2 User Editable Minimum Region Cut Length
Adequate region matching is an essential aspect of the ROW computational strategy. One has to
make sure that sufficient regions are defined to faithfully represent the physical reality of the
modeled system while maintaining the total number of regions as limited as possible to keep
computation times low. In previous versions, a minimum section length was automatically
computed by the software during the region generation process to ensure that each section was
longer than the largest conductor radius on all paths, thus preventing short conductor errors in
the MALZ Total Interference models. Alternatively, you can now directly specify the minimum cut
length in order to better control the region matching behavior. This value can be entered through
Preferences | ROWCAD File Options where you are proposed 4 options to control the
minimum cut length:
 Automatic: In this mode, the software searches for the largest conductor radius among
all the defined Cross-Sections and uses a safety factor (currently adding 10% to the largest
radius) to determine the minimum cut length used for the region generation.
 Legacy: It has been found that the automatic minimum cut length calculated by previous
versions of the software was unit type dependent, i.e. the region matching could be
different depending on the system of units in which the system was defined. This has been
corrected with the new Automatic mode. Nevertheless, the former calculation has been
maintained to ensure backward compatibility.
 User-Defined: This mode will let you directly define the minimum region cut length in
the project’s system of units.
 Off: This mode prevents additional points from being created on a non-Main path.

Figure 4: The Minimum Cut Length options are available in RowCAD’s Preferences panel.
The Automatic mode is used by default and will result in the most precise region matching, but
also the largest number of regions. Astute users can use the User-Defined option to reduce such
high number of regions, without sacrificing precision.
As an example, consider the following system where two transmission lines run parallel to each
other. Some of the tower locations show only a slight offset between the two lines while others are
well separated.
Figure 5: System under study, some towers on a principal path are slightly offset from the towers on the
Main path.
Using the Automatic mode with the Force Region Cuts at Tower Locations option selected,
all the towers are correctly detected and matched to a region’s end. However, short regions are
created due to the small X-axis offset between towers on the main and towers on the principal
path.

Figure 6: Region cuts obtained using the Automatic mode: unnecessary short extra regions are
produced.
In these cases, you could either keep the high number of regions or move the principal path tower
locations to precisely match the tower locations on the Main path, leaving you with the difficult
choice of favoring precision over expediency, or vice-versa.
When the Minimum Cut Length mode is set to Off, no new point can be created on a principal
path, which means that the region will snap to the closest node. This can easily lead to erroneous
matching; this option should be used with caution.
Figure 7: Region cuts obtained using the Off mode: erroneous matching results.
Finally, with the Minimum Cut Length option set to User-Defined, and using a minimum
cut length value slightly larger than the towers offset, unnecessary regions can be removed while
keeping all the tower locations properly matched on the principal path.
Figure 8: Region cuts obtained using the User-Defined mode with appropriate minimum length: good
match with minimum number of regions.
3 Improvements
3.1 Grid Lines in the Viewer
A light grey grid with graduated axes has been added to the drawing canvas. The mesh is
automatically resized upon zooming actions. Light green lines indicate the origin axes, as in
SESCAD.

Figure 9: The new Viewer grid.
3.2 More Flexible Cable Group Definition
Making use of Right-of-Way’s new ability to consider cable groups without a pipe enclosure [2],
RowCAD now allows for multiple cables to be defined within the same cross-section. A cable duct
for example can now be modeled more easily.
Figure 10: 4-circuit cable duct, now a valid cross-section for use in Right-of-Way and RowCAD.

Individual cable cross sections therefore no longer require the cable to be centered. Off-centered
cables will simply be treated as a cable group containing a single cable and no enclosure. Pipe-
type cables (with an enclosure) continue to require being centered and stronger validation was
implemented in this respect.
3.3 Project files
While in previous versions the copying of the RowCAD project to the ROW scenario folder was
convenient, advances in ROW and RowCAD include changes that make this copy no longer
particularly useful, as it would likely be incomplete. The project is therefore no longer copied
under the ROW scenario folder where it is imported. This means you must take care not to delete
the original copy.
3.4 Example Files
The “UGC 2016\Example Files\NewFeaturesRowCAD” subfolder of the SES Software conference

distribution contains example files that illustrate the new features presented in this article.
4 Conclusion
This year’s development cycle included important work that will undoubtedly please many Right-
of-Way Pro users. Indeed, the kml file import feature will make it much more straightforward and
easy to define accurate geometries within RowCAD, without resorting to a complex and tedious
external conversion process that was used before. The import capabilities already address the
most common use cases, but we have plans to increase the flexibility even more during the next
year. Similarly, the minimum cut length option of the region generation algorithm also has the
potential of drastically speeding up studies of larger projects. With this option, accuracy in the
geometries is preserved while the number of regions necessary to describe the system can be
minimized. Furthermore systems comprising multiple parallel cables should now be easier to
specify, since these will now require a single polyline rather than multiple ones.
5 Acknowledgments
The authors would like to thank John Xie for the work on the grid lines and Eric Dawalibi for his
help with the kml import and his work on the cable groups.
6 References
[1] C. F. F. Karney, "Transverse Mercator with an accuracy of a few nanometers," Journal of Geodesy,
vol. 85, no. 8, pp. 475-485, 2011.
[2] Y. Li, "New Features in Right-of-Way Pro," in CDEGS Users' Conference Proceedings, Boulder,
Colorado, USA, 2016.

21 NEW FEATURES IN RIGHT-OF-WAY PRO
Yexu Li and Farid P. Dawalibi

1 Introduction
Since our last Users Group conference in 2015, SES’ technical and software development teams
have been relentless in working on improving and adding new features and capabilities to the
Right-of-Way software package, as summarized hereafter.
 The SES TransposIT utility has been integrated into ROW to allow users to implement
automated and flexible transmission line phase transpositions and cable cross-bondings.
 The pipe-enclosure of a Group path can be defined as a “Dummy” or virtual conductor
allowing users to quickly define a group of cables located within a trench or duct.
 Users can now import conductors, profiles and any other MALZ object into a total
interference MALZ file directly from the ROWCAD Entity definition as a “Total
Interference” template file.
 The total interference level for steady state conditions is now fully integrated and can be
automatically computed without user interactions allowing users to account for both the
inductive and conductive coupling simultaneously in ROW. Furthermore, to better
support this feature, additional plot options have been added in the Advanced
Process… window of the “Steady State” option.
 In the Total Interference module, users can apply a global shift on the entire network
system. Once requested, this shift consisting of a global Xs and Ys shift value will be added
to all conductors’ X and Y coordinates in the total interference MALZ model file.
 The existing maximum soil breakdown distance and various fault arcing and flashover
distance options have been improved and additional new options have been added.
Therefore, it is possible to estimate critical soil breakdown and flashover distances due to
fault conditions, whether initiated by lightning or not, based on different approaches.
Users can select the option that best matches the situation being examined.
This article discusses these new improvements, features and capabilities and briefly explains how
to use them.
2 Integrated TransposIT Software Tool

It is well known that substantial current unbalance can develop within and between the circuits
of a transmission line when the phases are not regularly transposed (Figure 1), even when the
loads are balanced. Therefore, phase transpositions are designed and installed along a
transmission line or along cable routes in some cases. On the other hand, to minimize the heat
losses associated with induced currents in cable sheaths or pipe-type cable steel walls, cross-
bonding is applied to the sheath and armors of power cables (Figure 2). Therefore, realistic phase

transposition and cross-bonding modelling is critical to compute correctly load and fault currents
in order to determine accurate AC interference levels on exposed lines, circuits and metallic
utilities, especially under steady state conditions.
C
B
A
Figure 1: Phase transposition in transmission line.
Sheath Voltage Limiting Device Joint Splice with Discontinuous Sheath
Figure 2: Cross-bonding in underground power cable.

Previously, to model a transmission line phase transposition or cross-bonding in ROW, you could
proceed as follows:
 Method #1: Define different Cross-sections in ROWCAD, or different Attribute Sets
in ROW. This method is easy and intuitive for modeling transmission line phase
transpositions, but is not applicable to cable cross-bonding.
 Method #2: Introduce artificial phases in the SPLITS circuit and apply various
operations in Modify Circuit so that transposed phases can be defined and
implemented. This method is good for both transmission line phase transposition and
cable cross-bonding. However advanced planning is necessary because additional phases
are required and need to be added and defined adequately.
 Method #3: Use the ROW generated SPLITS SICL input file (SP_jobid.F05) from ROW
as input to the TransposIT utility, apply phase transpositions, and produce a new modified
SPLITS input file with transpositions applied at the specified locations on the circuits.
Then, this new SPLITS input file need to be reloaded back to the ROW project scenario to
continue the computation process. This method is applicable for both transmission line
phase transposition and cable cross-bonding. However users need to be familiar with the
standalone software tool TransposIT and some manual file operations are required.
With the new version of ROW, the TransposIT tool is fully integrated into the Modify Circuit
module of ROW. Users are now able to model phase transpositions or cable cross-bonding
scenarios easily. See the example project: “.\ROW\Transposition\” in the example folder.
Here is how this is done in a few simple steps.
Step #1:
Start the Modify Circuit module, and specify Reference Circuit (SPLITS) ID & New
Circuit (SPLITS) ID.

Step #2:
When the Transposition… button is clicked,
 The TransposIT tool is called;
 An empty TransposIT project is created with the SPLITS circuit
“Sp_xxx_ReferenceCircuitID.f05” as input, if the corresponding TransposIT project
(TransposIT_ReferenceCircuitID) doesn’t exist. Otherwise, the existing corresponding
TransposIT project is loaded.
Step #3:
Click System to access the System window and define the transposition or cross-bonding:
 Define the circuits that exist in your network.

 Specify the transpositions that will be applied along the right-of-way for all circuits
defined in the previous screen (Circuit Configuration).
Step #4:
Click Save Data & Close buttons, respectively, and go back to the main screen of the
TransposIT tool.

Step #5:
When the Process button is clicked, the following output file is generated automatically: A new
circuit Splits file in the scenario folder “…\ScenarioName\Sp_xxx_NewCircuitID.f05” that
includes the transposition and cross-bonding operations.
3 Dummy Pipe-wall Conductor

A high voltage power underground line often contains one or more 3 single phase cables circuits
with a few redundant cables. For expediency, it is desirable to define these 3 single cable circuits,
in as belonging to one path instead of defining them as 3 or more independent paths. Users may
want to define them as a pipe-type cable group path first and then, dummy the pipe-wall.
Presently, a true dummy (non-existent) pipe-wall cannot be specified. Previously users had to use
an approximate method if they wanted to define multiple cables using the Pipe-type option. This
approximate method consists of assigning the pipe-wall phase in the Terminal Energization
screen to have a “Dummy” status. As already mentioned, this method eliminates the pipe
enclosure from the circuit model but only after the computation of the line parameters of all
conductors including the effects of the pipe enclosure wall. This may affect the accuracy of the line
impedances (self and mutual impedances) of the other conductors. This inaccuracy is typically
dependent on the network configuration as well as soil characteristics. In practice, this inaccuracy
is not significant in most cases.
An alternative and accurate method would be to define the pipe-wall as a DUMMY phase in the
Attribute Set – Phase Leakage, Status… screen. This approach would prevent including the
pipe wall in the self and mutual computations alleviating any concerns regarding possible
computation inaccuracies. Unfortunately this was not possible in previous ROWCAD and ROW
versions.
An easier but still accurate method would be to define the pipe wall as a DUMMY conductor. With
the new version of ROW, pipe walls in a Group path can be defined as a real DUMMY (non-
existing) conductor. A dummy conductor will be omitted in the calculations of impedances
required by the SPLITS circuit model. The definition of a dummy (non-existing) pipe wall
conductor is now part of the available options that can be specified in ROWCAD. See the example
project folder “..\ROW\Dummy Wall\” in the example folder.
4 Import ROWCAD MALZ File Entity into Total Interference

In the Total Interference module of ROW, you can specify a MALZ file that will be appended to
the total interference model that is generated by ROW. This file can contain any valid MALZ
commands. For example, it is possible to specify conductors that represent an extra grounding
system, conductor-types that represent special conductor characteristics and observation profiles

along specified paths. This is quite a useful feature in order to add elements to the conductive
interference module that cannot be specified in the induction model of ROW.
The difficulty in creating a MALZ template is not-knowing precisely where to place the missing
elements such that they end up being connected correctly to the rest of the network elements.
Furthermore, it is important to define the CONDUCTOR, PROFILES, ENERGIZATIONS,
CONDUCTOR-TYPES, COATING-TYPES and LEAD-TYPES accurately in the template file.
Some existing documents (check with SES technical support group for those documents)
describing the manual procedures or steps to follow are provided to help users create a template
file for the total interference model. In this new version of ROW, users are allowed to import these
elements from ROWCAD as defined Entities. Three options are available:
- No Template: no additional components are appended to the total interference MALZ
file.
- External Template: Users can browse to an existing specially-prepared MALZ template
(the method used in previous versions of ROW).
- Use Entities (new): Automatically brings in the MALZ file the entities specified
ROWCAD when a ROWCAD project is saved and imported into the ROW project. The
MALZ file entity created by ROWCAD can be found at:
“..\ScenarioName\ROWCAD Project\MZ_Template_ScenarioName.F05"
See the example project: “..\ROW\ROWCADEntity TotInterfTemplateFile” on the example
folder.

5 Integrated Steady State Total Interference Accounting for

Inductive and Conductive Components
It is well known that under power line steady state conditions, an exposed pipeline or railway is
subject to inductive interference. However, it is less known (except by an increasing number of
SES Sofwtare users) that conductive interference under steady state conditions could be
significant and should not be ignored, as is often the case by the overwhelming majority of the
engineering community using other software packages.
The shield wire or neutral conductors of long transmission or distribution power lines can be the
subject of significant induced voltages and currents under normal operating conditions. A non-
negligible portion of such induced currents will discharge into the soil via the power line structure
footings and grounds as well as at the substation grounding systems. As a result conductive
interference will be transferred through soil to nearby metallic paths such as pipelines and railway
tracks. Therefore, under steady state conditions, pipelines or railway tracks will be simultaneously
subject to both inductive and conductive interference. The inductive and conductive effects are
generally additive, producing more severe interference levels than in the case of inductive
interference only. In this case, ignoring conductive interference effects can result in a dangerously
underestimated interference levels. In some cases of steady state interference scenarios, it is also
possible for the inductive and conductive effects to be subtractive. In this case, ignoring
conductive interference can result in an overestimated design.
In previous versions of ROW, it was possible for users to compute the total interference level
under steady state conditions that included both inductive and conductive components, by
following manual steps using the Monitor Fault and Total Interference modules. The major
steps required were as follows (check with SES technical support group for the detailed document
describing these steps):
1. Complete the SPLITS input file (.F05) that represents steady-state conditions. This is done
in the ROW Create Circuit and Modify Circuit (if necessary) modules.
2. Run the Monitor Fault Module by specifying a large Faulted Connection
Impedance (ohms), e.g., 999999+j999999 in order to prevent creating a real fault
condition in the network. Furthermore, it is recommended that the “fault” locations be
selected to ensure that at least one “fault” section is included for each soil region. Since the
conductive interference is sensitive to the soil structure, it is important to compute the
conductive coupling based on the local soil structures of each section to account for the
soil variations along the right-of-way.
3. Run the Total Interference Module by specifying a small value (e.g., 0.0001) for the
energization threshold for the tower GPR or current and defining appropriate profiles. As
a result, all towers are exported and modeled and the specified computation profiles are
able to account correctly for the conductive component based on soil variations in the total
interference model (MALZ input files).
4. Run all MALZ files, and plot the Reach Touch Voltages envelopes in the Report/Plot
Advanced window.
In the new version of ROW, the above procedure required to compute the total interference levels
for steady state conditions is now fully integrated. The total interference level under steady state
conditions can be automatically computed without user interventions to account for the inductive
and conductive coupling simultaneously in this new version of ROW. After the SPLITS input file

representing steady state conditions is completed users who wish to compute the total
interference under steady state condition can directly go to the Total Interference module, and
proceed with the Steady State option. Note that this Steady State option is only available for a
ROWCAD project. As a result, the program will automatically complete the above manual Steps
2, 3 and 4, i.e.:
 Automatically create the SPLITS input files for each soil region based on the soil region
assignment defined in ROWCAD;
 Run the SPLITS models required to produce the inductive computation files;
 Generate the Total Interference MALZ files for each soil region;
 Optionally run all MALZ input files to produce the total interference computation files.
All files will be stored in the folder “..\SS_TotalInterf_Store-xx\” (where xx represents the
SPLITS circuit number: Sp_ScenarioName_xx.F05”), including SPLITS & MALZ computation
files.
Furthermore, additional plot options are added in the Advanced Process… window for the
“Steady State” option. They are:
 Touch Voltage (Total Interference)
 1 cm2 Holiday Current Leakage Density (Total Interference)
All report output files obtained from “Steady State” in Advanced Process… window are
stored in the “..\SteadyState\” subfolder of the scenario. See the example project
“.\ROW\integrated SS\” in the example folder.

6 Global Coordinates Shift in Total Interference Module

In some cases, the coordinates of the right-of-way network can be very large. As long as the
relative locations among the network components are correct, the AC interference level will be
computed correctly. However, you might already have experienced that in some cases where the
coordinates are imported from Geographical Information Systems (GIS) based utilities, large
UTM coordinate values will cause round-off errors and numerical computation problems in the
MALZ models and sometimes even MALZ computation failures. When this occurs, users have to
go back to ROWCAD and shift the entire network to make it “centered” at the (0, 0) origin and
then re-import the ROWCAD project. In the new version of ROW, the Total Interference
module offers the ability to apply a Global Coordinate Shift on the entire network. As a result,
the defined global shift Xs and Ys coordinates will be added to all conductors and observation
points in the total interference MALZ model file whenever a MALZ file is created. This feature is
accessed in the Advanced tab of the Total Interference screen.
See the example project “..\ROW\GlobalShift_TotInterf\” in the example folder.
7 Evaluating the Vulnerability of Arcing to Pipelines under

Fault Conditions in Right-Of-Way
Electric breakdown and electric ionization occur in the soil when the current density discharged
by grounded structures subjected to lightning or fault currents exceeds a critical value resulting
in soil ionization due to electric fields exceeding the electric breakdown of the soil (Eb). As a result,
the low-current, low frequency ground resistance of the structure decreases.

The resistance of grounded structures subjected to high fault or lightning currents can be obtained
by a current dependent grounding model considering soil ionization. The resistivity of the
ionization zone depends on the total energy stored in the zone. Several authors have developed
analytical expressions to estimate the dynamic behavior of the ionized zone resistivity. Two
methods have been proposed to calculate the nonlinear grounding resistance considering the soil
ionization. One method considers the resistivity and dimensions of the ionization zone. The other
one uses an empirical relation between the voltage and current curve. Bellaschi [4], Liew and
Darveniza [6] and more recently Sekioka [15] among others have proposed such models.
The soil ionization and electric breakdown Eb vary with soil moisture, grain size, compactness and
presence or absence of organic material. It is also influenced by the applied wave shape. The
initiation of the electric breakdown discharge begins when the electric field becomes large enough
to ionize the air present in the voids between the soil grains. The electric field across the whole
soil gap at the time of the breakdown inside the air voids can be much smaller than the breakdown
field in an equivalent air gap. This is because there is an electric field enhancement due to the
voids’ irregular shapes and the relatively large dielectric constant of the soil.
As already mentioned, the main consequence of this breakdown is the reduction of ground
impedances during high fault current magnitudes and transient conditions caused by high surge
currents such as lightning strikes. A literature survey of research work published since the early
twenties reveals that a large number of measurements using large magnitude currents have
demonstrated this ground impedance reduction.
Some of the first experiments were carried out in 1929 by Towne [1] who has shown that loose
gravel soils exhibit a breakdown when the electric field is between 160 kV/m and 520 kV/m. Other
researchers have reported similar experimental results. Table 1 provides a summary of their most
salient results.
Table 1: Experimental soil breakdown electric field results
Authors ρ (Ω-m) Eb (kV/m) Reference

470 1200
Korsuntcev 180 1000 2
100 800
100 300
Bellashi 75 220 3&4
300 425
Petropoulous 290 830 5
50 110-300
63 50
Liew and Darveniza 87 127 6
157 200
310 70
Loboda et al. Various 560-900 7
Espel et al. Various 800-1700 8
Oettle Various 600-1850 9
Mousa From many tests, an average value 300 10
Dick et al 12-25
is suggested 13-221 11
IEEE Working Group Uses suggested value by Oettle 1000 12
CIGRE Lightning Performance No references or justifications 400 13

It is important to note that the experiments can be classified in two types, i.e., those that were
carried out using an impulse source (transient wave) and those involving power system
frequencies, i.e., 50 Hz or 60 Hz.
Sunde [14] has proposed two equations to calculate the critical distance D between a structure hit
by a lightning current If and a nearby metallic structure held at a low potential and buried in a
soil with a resistivity :
D = 0.08√𝐼𝑓  for   100 Ω-m (1)
or,
D = 0.047√𝐼𝑓  for   1000 Ω-m (2)
SES proposes instead to use one additional equation to cover the intermediate range:
D = [0.08367 − 3.67 ∗ 10−5 ]√If  for 100 Ω-m    1000 Ω-m (3)
Oettle [9] suggested a power based regression between soil resistivity 𝜌 and dielectric strength Eb,
giving:
𝐸𝑏 = 241[𝜌0.215 ] (4)
where Eb is in kV/m and  is in Ω-m. Figure 3 compares the computed results applicable to Oettle
against the experimental results given above and results provided by Espel [8] and his co-authors.
Figure 4 shows additional measurements as a function of moisture content confirming that the
increase of moisture lowers the dielectric strength of soils.
It is clear that the average values based on these experimental results are not consistent among
each other. In fact, significant disagreement subsists between researchers. This is not surprising
considering the highly non-homogeneous nature of natural soils. Note that the lowest measured
value is as low as 13 kV/m and the highest value approaches 2,000 kV/m which is close to the air
dielectric breakdown of 3,000 kV/m. Mousa [10] and others recommend a conservative value of
300 kV/m although he refers in his work to a situation where the measured electric breakdown
value was as low as 50 kV/m.
Figure 3: Average dielectric strength of soils as a function of soil resistivity.

In case of doubt, SES recommends to use an extremely conservative value of 30 kV/m

(0.3 kV/cm). This value is lower than the overwhelming majority of electrical soil breakdowns on
this planet.
Figure 4: Average dielectric strength of soils as a function of soil water content.
The Canadian Electricity Association (CEA) report 239T817, entitled Power Ground Fault Effects
on Pipelines is still one of the most widely used references when arcing distances are evaluated.
It distinguishes between two situations, one involving fault conditions at tower structures without
lightning discharges and lightning discharges that initiate faults to ground at tower structures.
Measurements indicated that the voltage required sustaining a lightning initiated arc to a pipeline
located at a distance D from the electric arc originating from a tower raised at potential Vs is
approximated by the following linear regression formula:
Vs = 5.801 + 0.0703D (5)
Typical results from this equation are shown in the following table.
Similar tests to determine the flashover distance assuming that the tower structure is raised at a
potential Vflash assuming that there is no lightning initiated arc were performed and the following
linear regression equation was developed.
𝑉𝑓𝑙𝑎𝑠ℎ = 18.01 + 0.1082𝐷 (6)
A more accurate geometric equation is also proposed as follows. Note that the geometric
regression provides more accurate results than the linear regression formula when the tower
potential rise is low (i.e. less than 20 kV), but provides unrealistic results at higher voltages.
𝑉𝑓𝑙𝑎𝑠ℎ = 8.086𝐷0.3056 (7)
Typical results from the above equations are shown in the following tables 2 & 3.

Table 2: Typical results of arcing distances with CEA suggested equations: Lightning initiated
CEA - Lightning Initiated Arc (Equation 5)

Computed Rounded-up
Transmission Line Breakdown Distance
Breakdown
Voltage (kV)
Distance (m) (m) (Feet)
25 1.2 2 4
40 2.5 3 8
69 4.8 5 16
115 8.6 9 28
138 10.5 11 35
230 18.1 19 60
345 27.5 28 90
500 40.2 41 132
735 59.5 60 195
Table 3: Typical results of arcing distances with CEA suggested equations: without Lightning
CEA - Flashover Arc without Lightning (Equation 6 & 7)

Computed Rounded-up
Transmission Line Breakdown Distance
Breakdown
Voltage (kV)
Distance (m) (m) (Feet)
69 2.0 3 6.5
115 4.5 5 15
138 5.7 6 19
230 10.6 11 35
345 16.7 17 55
500 25.0 26 82
735 37.6 38 123
Although the CEA suggested equations have been criticized occasionally, they provide another
way to evaluate the vulnerability of nearby pipelines to electric line structures subjected to phase
to ground faults whether initiated following a lightning strike or not.
Right-Of-Way now offers various options based on the above approaches. They consist of the
following choices:
1. Separation Distances Based on a Specified Electric Breakdown Value
2. Separation Distances Based on Sunde formulas - Equation 1 to 3
3. Separation Distances Based on Oettle Geometric Regression - Equation 4
4. Separation Distances Based on CEA Assuming a Lightning Initiated Arc - Equation 5
5. Separation Distances Based on CEA Assuming no Lightning Initiated Arc - Equation 6
6. Separation Distances Based on CEA Assuming no Lightning Initiated Arc Equations 6 & 7
7. Separation Distances Based on Espel Experimental Results - Figure 3
Users should select the option that best matches the situation being examined. Additional useful
references [16-21] are given hereafter. See the example project: “..\ROW\BreakdownDistance\”
on the example folder.
8 References
[1] Towne, H.M.: “Impulse characteristics of driven grounds”, General Electrical Review, November
1929, pp. 605–609.
[2] Korsuncev, A.V. "Application of the theory of similitude to the calculation of concentrated earth
electrodes'", Elektrichestro, no. 5, pp. 31-35, May 1958.
[3] Bellaschi, P. L., Armington, R. E., and Snowden, A.E.: “Impulse and 60-cycle characteristics of
driven grounds Part I and II”, Transactions American Institute of Electrical Engineers, AIEE Trans.,
1941, vol. 60, pp. 123–128,.and AIEE Trans., 1942, pt. II, vol. 61, pp. 349–363.
[4] Bellaschi, P. L., Armington, R. E., and Snowden, A.E.: “Impulse and 60-cycle characteristics of
driven grounds Part I and II”, Transactions American Institute of Electrical Engineers, AIEE Trans.,
1941, vol. 60, pp. 123–128,.and AIEE Trans., 1942, pt. II, vol. 61, pp. 349–363.
[5] Petropoulos, G.M. "The high voltage characteristics of earth resistances", Proceeding of the IEE,
vol. 95, Part II, pp. 59-70, 1948.
[6] Liew, A.C, and Darveniza, k. L., "Dynamic model of impulse Characteristics of Concentrated
Earths", Proceedings of the lEE, vol. 121, No. 2, pp. 123-135, I 974.
[7] Loboda, M., and Scuka, V., “On the transient characteristics of electrical discharges and ionization
processes in soil”. Proceedingsof the 23rd International Conference on Lightning Protection,
Florence, 23–27 September, 1996, pp. 539–544.

[8] Espel, P., Bonamy, A., Diaz, R., and Silva, J., “Electrical parameters of discharges in resistive soils’.
Proc. 26th Int. Conf. on Lightning Protection, Cracow, 2–6 September 2002, pp. 346–351
[9] Oettle E. E. "A new general estimation curve for predicting the impulse impedance of concentrated
earth electrodes". IEEE Transactions on Power Delivery, vol. 3, No. 4, October 1988.
[10] Mousa, A. M., “The soil ionization gradient associated with discharge of high currents into
concentrated electrodes”, IEEE Transactions on Power Delivery, vol. 9, No. 3, July 1994.
[11] Dick, W. K., Holiday H. R., “Impulse and AlternatingCurrent Testson Grounding Electrodes in Soil
Environment”, IEEE, Trans., Vol. PAS-97,1978, No. 1, pp 102-108
[12] IEEE Working Group Report, “Estimating lightning performance of transmission lines II- Updates
to analytical models”, IEEE PWRD, 8, 3 July 1993, pp. 1254–1267.
[13] CIGRE Working Group on Lightning. "Guides to Procedures for Estimating the Lightning
Performance of Transmission Lines", CIGRE, Paris, France, 1991.
[14] Sunde E. M., “Earth Conduction Effects”, New York, NY, Dover Publications, 1968, pp 296-298.
[15] Sekioka S., Lorentzou M. I., Philippakou M. P., Prousalidis J. M., “Current-Dependent Grounding
Resistance Model Based on Energy Balance of Soil Ionization”. IEEE Transactions on Power
Delivery, January 2006, Vol. 21, No. 1 pp 194 - 201.
[16] T. Matsui, M. Adachi, H. Fukuzono, S. Sekioka, O. Yamamoto, and T. Hara, “Measurements of

grounding resistances of a transmission-line tower base connected with auxiliary grounding
electrodes for high impulse currents,” in Proc. 10th Int. Symp. High Voltage Engineering, vol.5,
Montreal, QC, Canada, 1997, pp. 257–260.
[17] S. Sekioka, H. Hayashida, T. Hara, and A. Ametani, “Measurements of grounding resistances for
high impulse currents,” Proc. Inst. Elect. Eng., Gen., Transm. Distrib., vol. 145, no. 6, pp. 693–699,
1998.
[18] S. Sekioka, T. Sonoda, and A. Ametani, “Experimental study of current-dependent grounding

resistances of rod electrode,” IEEE Trans. Power Del., vol. 20, no. 2, pp. 1569–1576, Apr. 2005.
[19] J. M. Prousalidis, M. P. Philippakou, N. D. Hatziargyriou, and B. C. Papadias, “The effect of

ionization in wind turbine grounding modeling,” in Proc. Mediterranean Electrotechnical Conf.,
Cyprus, 2000, pp. 28–31.
[20] M. I. Lorentzou, N. D. Hatziargyriou, and B. C. Papadias, “Time domain analysis of grounding

electrodes impulse response,” IEEE Trans. Power Del., vol. 18, no. 2, pp. 517–5124, Apr. 2003.
[21] “Perspectives on soil ionization,” in Proc. International Conference on Grounding and Earthing,
Belo Horizonte, Brazil, 2000, CIGRE Task Force on Soil Ionization—WG 33.01, pp. 105–106.

22 NEW ADVANCES AND FEATURES IN CORRCAD

SOFTWARE PACKAGE
John Xie(1), Cathy Cheng(1), Adrian Ngoly(1), Liqian Feng(2), Mandy Zhang(2), Simon Fortin(1),
Farid P. Dawalibi(1) and Yexu Li(1)
(1) Safe Engineering Services & Technologies ltd. (2) Beijing Safe Engineering Services & Technologies
Email: info@sestech.com, Web Site: www.sestech.com Email: info@seschina.cn , Web Site: www.seschina.cn
1 Introduction
CorrCAD is the new software package that was officially introduced last year during the 2015
Users Group conference. Presently, CorrCAD can solve a large variety of cathodic protection
design tasks and related issues for various onshore and offshore projects. For example, it can
predict the degree of corrosion control provided by a system, such as impressed cathodic current
protection systems (ICCP) and sacrificial anodes employed in anodic protection systems.
Moreover, it can estimate the effects of stray currents such as those produced by HVDC electrodes
or DC rail traction systems on the corrosion rate of buried metallic structures such as pipelines.
It can also evaluate the corrosion status of a metallic structure and help optimize the location and
characteristics of the corrosion protective system (such as ICCP) to minimize stray current
interference effects on protected structures such as pipelines.
Since then, the CorrCAD technical and software development team has focused its efforts on
stabilizing and enhancing the robustness of the product. Furthermore, a number of new features
and capabilities have recently been implemented in addition to bug fixes and other miscellaneous
improvements. This short article describes the major advances made in CorrCAD since our last
User Group Meeting.
 The structure (e.g., pipeline) coating effective resistance in the presence of holidays
(coating defects) can now be modeled using discrete as well as distributed elements. The
polarization potential and the associated leakage current are then computed accordingly,
based on these two coating effective resistance computation methods.
 In addition to the existing user-defined Regular polynomial function, a new automated
Chebyshev polynomial regression method is now available to accurately approximate
experimental polarization curves associated with cathodic and anodic electrochemical
reactions. The experimental polarization curves can be digitized or defined by importing
the appropriate Excel or CSV file or simply by pasting the polarization data in the CorrCAD
table. Two alternate forms of the polynomial can be selected, namely the original
Chebyshev polynomial or an alternate Normalized polynomial.
 The new version of CorrCAD now supports multiple cross-sections along a polyline
defining the centerline of the protected structure.
 A few customizations and options have been added to the polarization potential plot
engine:
 It is possible to directly export the results to an Excel CSV compatible format file;
 Users can select a target polyline to be plotted, instead of always plotting all computed
polylines on the same plot;

 Users can display the results obtained from any selected iteration, instead of the last
iteration or the final results only;
 The curve can be plotted using different styles, e.g., a continuous line or scattered dots;
 Users can request to optionally add the target Design Objective Polarization Line to
the plot.
 Background items have been added to the graphical work area. These items occupy the
background of the drawing area and provide useful information about the drawing, such
as the Coordinate Axes, the Drawing Grid, etc…
This article discusses these new improvements, features and capabilities in more detail and briefly
explains how to use them.
2 Coating Resistance Effects – Discrete and Distributed

Holidays
The leakage current and potential distribution on a pipe are critical parameters for CP design.
Correct CP design will depend strongly on the types of holidays assumed to exist along a coated
structure such as a gas pipeline. In general, there are two methods to model the influence of
holidays in coated structures, namely, the Discrete Single Coating Holiday (or other coupon
sizes) and the Uniform Coating Efficiency methods.
The Uniform Coating Efficiency is defined as:
Icoated
ℎ = (1 − ) × 100% (1)
Ibare
where Ibare is the current required to protect a bare surface and Icoated is the current required to
protect the coated surface.
In general, the design constraints for a discrete single coating holiday are much more severe than
those for a pipe with uniform coating efficiency. Therefore, the CP design must account for
discrete holidays if localized damage to the coating is anticipated. However, the design based on
a uniform coating efficiency is still the most often used method by corrosion engineers.
Polarization occurs when current flows in or out of the pipe wall metal. The polarization potential
is the potential difference between both sides of the interface in contact with the pipe wall and the
soil (electrolyte). The polarization potential is a measure of the CP effectiveness and provides a
direct characterization of the stray current interference level.
The polarization potential and leakage current have a nonlinear relationship, which is referred to
as the Polarization Curve. A polarization curve is usually obtained from an experimental setup.
Under the same environmental conditions (such as surrounding soil), the differences of material
and surface conditions of a coated metal structure or a bare metal surface, such a pipe, will lead
to discernible differences between polarization curves. See Figure 1 for typical polarization curves.

Figure 1: Example of Polarization curves for a coated and a bare pipe immersed in the same soil
environment.
CorrCAD computes the polarization potential accounting for the applicable polarization curve of
the target system. During a design, if polarization curves are available for the coated and bare
metallic systems as in the above figure, then CorrCAD can simply:
 Use the Coated Polarization Curve for the coated metallic systems such as pipes;
 Use the Bare Polarization Curve for the bare metallic systems such as anode
electrodes of an impressed dc cathodic protection system or a system that uses
sacrificial anodes instead.
However, in many cases, the Coated Polarization Curve or equivalent data may not be readily
available. As an alternative solution, it is necessary to use the Bare Polarization curve instead
and adapt it to handle a coated metallic structure in the computations. Two approaches have been
implemented in this new version of CorrCAD, depending on the options selected to determine the
coating resistance effect models. They are the Discrete Holiday and Coating Efficiency
methods.
Discrete Single Coating Holiday (or Coupon) Model

This method is based on Equation (2) that links the current density to the steady-state coating
stress voltage and the coupon resistance
Coating Stress
Holiday Current Density = (2)
(Coupon Resistance) × (Coupon Area)
Here, the coupon resistance is equal to the resistance of the holiday used to emulate the
distributed holidays along the pipeline. Since the coating stress voltage is normally not
significantly affected by the presence of individual holidays, the holiday current density can be
estimated fairly accurately along the entire pipeline from the coating stress voltage alone, without
necessarily modeling the actual holidays at each location of interest. This approximation is valid
if, as assumed here, there are “uncoated” locations on the pipeline (e.g., the anode electrode beds)
that establish the external potential (which is the most common case in CP coated pipeline
models). In this case, a small bare coupon, representing a holiday, introduces only a small
perturbation of the external potential. The holiday current density is then estimated as being the
ratio of the coating stress voltage to the product of the coupon resistance and its area, as described
in Equation (2).
An actual discrete equivalent holiday (or coupon) resistance can be Computed or Specified. If
the Holiday Specification model is defined as Computed, the holiday size must be specified
in CorrCAD. In this case, the discrete holiday resistance in Equation (2) will be computed based
on the local soil by CorrCAD. See Figure 2(a).
The polarization curve The pipeline is Discrete single holiday The holiday resistance The holiday size is
is for bare structures. coated. method is used. is “Computed”. specified as 1 cm2.
(a): Interface for modeling a discrete holiday as “Computed”
If the actual Holiday Specification is defined as Specified, the known (or user’s computed
value) holiday (or coupon) resistance as defined in Equation 2 or any other equation is directly
entered in the program. See Figure 2(b).
The polarization curve The pipeline Discrete single holiday The holiday resistance The holiday resistance
is for bare structures. is coated. method is used. is “User Defined”. is entered directly.
(b): Interface for modeling a discrete holiday as “User Defined”

Figure 2: Modeling discrete holidays.

Coating Efficiency Method

Equation (1) is used to compute Ibare , provided the user has defined the coating efficiency h and
that Icoated was obtained from a MALZ or HIFREQ run. See Figure 3.
The polarization curve The pipeline is Distributed single The pipeline coating
is a bare structure. coated. holiday method is used. efficiency is specified.
Figure 3: Interface for coating efficiency method.
In summary, the new version of CorrCAD provides the following capabilities:

 When the Polarization Curve is for a Coated Conductor,
o If the pipe is coated, carry out the MALZ iterative computation directly using
the current density obtained from a previous MALZ run;
o If the pipe is bare, no computations can be made.
 When the Polarization Curve is for a Bare Conductor,
o If the pipe is bare, carry out the MALZ iterative computation directly using
the current density obtained from the previous MALZ run;
o If the pipe is coated, do the MALZ iterative computation using one of the
effective holiday type method: Discrete or Distributed.
3 Polynomial Functions for Polarization Curves

Polarization curves play a central role in electrochemical corrosion and a key role in
computational corrosion analysis. A polarization curve encapsulates various physical processes
that lead to corrosion and it is precisely a two-terminal electrical device with a nonlinear current-
voltage characteristic. It is this nonlinear relationship that provides the ability to accurately model
polarization curves in a corrosion program.
On the other hand, polarization curves can have very “odd” shapes depending on the chemical
reactions on-site. Theoretical analyses and accurate modeling of these curves are not always easy.
The actual corrosion electrochemical reaction theories are beyond most SES software users
engineering interests. The goal of the CorrCAD program is to hide the theoretical aspect of the
corrosion process and provide a variety of practical techniques to account for the polarization
curves in the analysis, in order to carry out the necessary MALZ iterative computations. There are
presently four different techniques that have been implemented in CorrCAD in order to account
for such polarization curves accurately. They are based on using:

1) A Butler-Volmer function;
2) A Butler-Volmer function accounting for the concentration polarization phenomenon;
3) Polynomial function to approximate the curves;
4) Lookup tables providing the required data.
Polynomial functions that fit experimental curves are popular methods since they are easy to
manipulate, and in general, can approximate any curve, regardless of the curve’s physical
meaning.
Last year, Regular polynomials that fit the experimental curve were introduced and could be
specified in CorrCAD. However, this kind of polynomial fit may have some unpredictable effects
on the results. For example, it may exhibit very wild numerical variations that could affect the
stability of the iterative process. Furthermore, it was the responsibility of the user to provide the
appropriate order and coefficients of the polynomial which is not always an easy process. This
type of polynomial is defined as shown in Equation (3):
a) Regular polynomials:
The regular polynomials are given by
N
P(𝑥) = ∑ 𝑎𝑖 𝑥 𝑖 , (3)
𝑖=0
where 𝑎𝑖 are the coefficients of the regular polynomial.

In the new version of CorrCAD, an automated Least-Squares Polarization Curve Fitting using
Chebychev Polynomials was implemented. The computed polynomial is provided in two
flavors, namely, the original Chebyshev polynomial form and the so-called Normalized
polynomial form.
b) Chebyshev polynomial (Default):

The Chebyshev polynomial 𝑃 is defined as:
N
P(𝑥) = ∑ 𝑎𝑖 T𝑖 (𝑢) (4)
𝑖=0
where 𝑎𝑖 are the polynomial coefficients and T𝑖 (𝑥 ′ ) is the Chebyshev polynomial of order 𝑖
evaluated at point 𝑥. The value of 𝑥 ′ is given by Equation (5).
2𝑥 − (𝑥max + 𝑥min )
𝑢= (5)
𝑥max − 𝑥min
The recursive formula for the Chebyshev polynomials is given by:

T0 (𝑢) = 1
T1 (𝑢) = 𝑢
T2 (𝑢) = 2𝑢2 − 1
T3 (𝑢) = 4𝑢3 − 3𝑢
⋮
T𝑛+2 (𝑢) = 2𝑢T𝑛+1 (𝑢) − T𝑛 (𝑢).

c) Normalized polynomial:
The Normalized polynomial 𝑃 is defined as:
N
P(𝑥) = ∑ 𝑛𝑖 𝑢 𝑖 (6)
𝑖=0
where 𝑛𝑖 are the Normalized polynomial coefficients and the value of 𝑢 is given by Equation
(5). For 𝑥 = 𝑥min , this gives 𝑢 = −1. For 𝑥 = 𝑥max , this gives 𝑢 = +1, and for 𝑥 = 𝑥max +
𝑥min )/2, this gives 𝑢 = 0.
So, 𝑢 is restricted to the range −1 ≤ 𝑢 ≤ +1. Here, 𝑥min and 𝑥max represent the minimum and
maximum values of 𝑥 for which data can be fitted. However, the polynomial can be evaluated
outside the [𝑥min , 𝑥max ] domain as well.
Figure 4 shows the Window interface that accesses the above three types of polynominal
representations.
Figure 4: Accessing the three types of polynomial function definitions.
The steps required to obtain the polynomial function to model polarization curves are described
hereafter.
Step #1:
Click on the Digitize button shown in Figure 4 to access the SESCurveFit tool.
 If the polarization data is in an Excel or CSV compatible file, select the option Get
Numerical Data, and Import the data (Figure 5). Then go to Step #2.

Figure 5: Importing polarization data from a data file.
 If the polarization data is provided as a polarization curve in a bitmap or JPG format proceed
as follows (Figure 6).
a) Select the Digitize option, then;
b) Click on the Load Image ribbon button to load an image containing a polarization curve.
You should know the curve format type (i.e., Potential VS Current or Current VS Potential)
and the appropriate units corresponding to the X and Y axes if not shown on the loaded
image.
c) Set the axes reference minimum (green) and maximum (red) lines to the appropriate
locations on the curve in order to specify the numerical values of the limits of the desired
digitized area;
d) Define the Polarization Curve settings corresponding to the appropriate curve format,
i.e., select the appropriate definitions for the X and Y axes as well as the applicable units
and specify the values of the reference line axes limits.
e) Click on the Start Samples ribbon button to mark the sample points first, by selecting
the anodic points (red dots) and then the cathodic points (green dots);
f) Click on the Extract Samples ribbon button to extract the sample points.
g) The digitized data will be stored in the Samples table.

a
b d f
e
c
g
b c
Figure 6: Digitize the polarization curve.
Step #2:
Refer to Figure 7.
a) Go to the Numerical Fit panel by clicking on the Numerical Fit tab;
b) Make sure that the Polynomial Polarization is selected on the ribbon button;
c) Define the Polarization Curve settings;
d) Click on the AutoFit on the ribbon button;
e) The Normalized and Chebyshev polynomial coefficients are computed and listed in the
‘Parameters’ table. Furthermore, the RMS error (in %) of the computed and sampled
data values is displayed.
Finally, when SESCurveFit is closed, the polynomial coefficients are returned to CorrCAD as
shown in Figure 4, and are used to evaluate the polarization potential for a given current leakage
density.

d
b
Figure 7: Autofit the polarization curve with a Chebyshev polynomial.
4 Multiple Cross-sections
A cross-section defines the characteristics of the conductors defined along a polyline path. The
Cross-Sections panel in CorrCAD allow users to define Polylines cross-sections as well as
conductor and coating material characteristics. This panel is split into two sections, Cross-
Section Warehouse and Cross-Sections which allows the user to temporarily recall and store
commonly used cross-sections files so that they are ready for the next assignment.
Previously, a single polyline (e.g., pipeline) could only be associated with one cross-section. This
limitation made the modeling unnecessarily cumbersome. For example, if the pipeline changes
its characteristics (e.g., the radius), the user was required to model the new section as a different
polyline. When an insulating joint (IJ) is installed along a pipeline, the user was required to model
the IJ as an open gap while the pipeline was defined as two independent polylines.
In the new version of CorrCAD, a polyline can be assigned with different cross-sections, or several
segments can be associated to a given cross-section even if they are in different polylines (see
Figure 8).

Figure 8: Multiple cross-sections defined along the same polyline.
5 Customizations and Options in the Polarization Plot

The relevant computed values are displayed once the requested computations are completed.
CorrCAD provides useful information about the polarization computation process such as the
number of iterations required to complete it, the RMS error at each iteration step, the conductor
over-potentials, as well as the final detailed numerical results. In the new version of CorrCAD, a
background grid is added on the plot for a better look and to provide additional plot information.
Furthermore, a few customization and options are available (refer to Figure 9):
a) Users can now directly export the over-potential results to an Excel compatible CVS format
file.
b) Multiple polylines can be computed in a single run. Users can now select a specific polyline
to be plotted, instead of plotting the results for all computed lines on the same graph.
c) The computation modules of CDEGS such as MALZ and HIFREQ do not have the ability
to compute problems with nonlinear polarization curves. The approach used in CorrCAD
is to use numerical approaches such as the simple fixed point iteration. It starts with an
initial value of either potential or current and solves the linear boundary conditions at each
iteration. The polarization curve (which corresponds to a nonlinear boundary condition)
is used during the iterative process to compute the new values of the potential or current
for the next iteration. The iteration is stopped when the difference between two successive
steps is smaller than a pre-specified tolerance threshold or if the maximum number of
iterations has been reached.
In the plot, you can now select and plot the results obtained during any iteration, instead
of having access to the final results only. In some cases, it is possible that an earlier
iteration gives more accurate and reasonable results than the latter iterations.
d) Two plot curve line styles are available: a continuous line or a scattered line.

e) You can now optionally request to add the Design Objective line (e.g., -0.85 V) on the plot,
which clearly indicates the adequacy of the CP system or the severity of the corrosion
process on the system.
(a) (b) (c) (d) (e)
Figure 9: Polarization plot customizations and options.
6 Background Items in the Working Space Area

Background items are added on the graphical working area. These items occupy the background
of the working drawing area and provide useful information about the drawing, such as the Axes
Coordinate system, the Drawing Grid, etc.
Figure 10: Background information in Working Space.

23 A NEW INTERFACE FOR FFTSES
FFTSES team

1 Introduction
As part of the overall restructuring of the user-interface of SES software programs, we present
this year a new version of the FFTSES module that has been completely revamped to provide a
more streamlined and integrated interface. Improvements and new functionalities have also been
introduced. This article first summarizes the new features with respect to the previous version 1.
Then, the main screens are presented and described. Finally, a simple example is provided to show
how to use the program. Note that this article assumes some familiarity with the previous version
of FFTSES.
2 Summary of the New Features

In addition to the revamped interface, new functionalities have been added. This section provides
a summary of these new functionalities.
1. Time domain windows have been implemented in FFTSES to smooth the beginning and
end of input signals, in order to remove artificial high frequency components that could
be generated during the FFT (Fast Fourier Transform) process. These functions can be
illustrated using the original curve (blue dots) of Figure 1. Since the value of this signal is
not zero at its beginning, we may want to multiply it by a function (called a window) such
that the curve starts and ends at Y = 0. If you choose to apply windows, you have the
following options:
 Rectangular;
 Hanning;
 Flat Top.
The Rectangular window is used only to truncate the signal whereas the Hanning and
Flat Top windows multiply the input signal such that both extremities reach zero in a
smooth fashion, following different patterns. As an example, a Hanning window is
applied to the original signal and the resulting curve (green) is shown in Figure 1.
2. Similarly, instead of using windowing functions, you can add regularization functions at
one or both extremities of the signal. The functions that are available are:
1The new features presented are not supported in the “Classic” version. As a result, you may lose your settings if you take an .F05 file
generated from this new program and save it in the old FFTSES module. Also note that the executable name of this new program is
SESFFT.

PART I: NEW FEATURES OR PART II: USER CONTRIBUTIONS OR PART III: SES CONTRIBUTIONS
o Cosine;
o Exponential;
o Power function.
The example shown in Figure 1 is repeated in Figure 2, but with a cosine regularization function
added at the beginning of the original signal.
Figure 1: Curves of the original signal (blue dots) and its windowed version (green curve).
Figure 2: Curves of the original signal (blue dots) and its version with a cosine regularization function at
the beginning (green curve).

3. In addition to the time domain waveform of the input signal, its frequency spectrum is
now shown in the Forward mode screen. Both plots are interactive, allowing you to zoom
onto the curves and to visualize the X and Y coordinates of a point on a curve by clicking
this point.
4. In the previous version, the recommended frequencies were always generated based on
the whole spectrum, which can result in superfluous recommended frequencies to run in
HIFREQ, especially when the frequency sampling of the input time domain signal is high.
In this version, it is now possible to apply the frequency recommendation algorithm only
within a user-defined range.
5. In the new interface, it is now possible to request the computation of multiple quantities
for the conductor segments and profiles at the same time. In this case, the frequencies
generated by the program are based on all those requested quantities.
3 Program Functionalities and Interfaces

As with all new SES products, this new FFTSES interface offers a more streamlined interface
where most settings are accessible from a single screen. Important commands, such as Save and
Process, are directly accessible at all times. All input errors are immediately reported in the
Issues List panel, providing immediate feedback.
The following sections present this new interface in more detail. The overall structure is first
presented and then, the Forward and Inverse screens are described with their related panels.
3.1 Interface Overall Structure

The FFTSES program uses the same shell as most of the new CDEGS interfaces. As a matter of
fact, the backstage (the first screen seen when the application is running) and the standard
functionalities included in the ribbon are common to many new SES modules and therefore, they
will not be presented herein. You may refer to [1] for more information on these functionalities.
When a new project is created, the main FFTSES interface is as shown in Figure 3, for which the
description is provided below:
1. Ribbon: The ribbon has four tabs: Home, Options, Help, and Results (the Results
tab is available only when the Results panel is selected – see point 7 and Figure 11). The
functionalities included in the Options and Help tabs are common to other new SES
application interfaces whereas those in the Home and Results tab are dedicated for
FFTSES.
2. FFT list: This list includes the Forward and the Inverse modes.
3. Data Input panel: On the left side of this panel, you can specify the input data.
Additional information is available on the right side of this panel, depending on the
computation mode (Forward or Inverse). More details will be given about this panel in
the following sections.
4. Issues List panel: Displays the error and warning messages.

5. Computation Trace panel: Displays the progress messages printed during the
computations. Also, the two buttons below allows you to open directly the .F09 and .F27
files.
6. Status Bar: Shows quick information about the project directory and the number of
Errors, Tasks, Warnings, and Messages generated by the program.
7. Results tab: Click this tab to make the Results tab appear in the ribbon to display the
results. This tab is available only when an .F21 file exists.
1
2
4
4
Figure 3: Main screen of the Forward mode. This screen shows an example of how the Lightning (Heidler)
signal can be defined.
3.2 Forward Mode Interface

Once you have launched the program and created a new project, the first screen that appears is
that of the Forward mode previously seen in Figure 3 and detailed in Figure 4. This screen allows
you to define the input signal of the system and is described as follows:
1. Definition of the Transient Input Signal;
2. Definition of the Surge Parameters of the chosen signal;

3. Display of the time domain signal and its spectrum.

Once the data are filled in the Forward mode screen, Compute Forward FFT can be clicked
to run FFTSES in order to obtain the recommended frequencies to which the HIFREQ model will
be run.
1 3
Figure 4: Data input panel in the Forward mode.
Additional Notes
In cases you need to manually define the input signal (using the Specified-Surge generator),
you may want to apply pre-processing on it, which can be done with the dialog shown in Figure 5,
obtained by clicking Regularizations/Windowing (below the plots in Figure 4). Using the left
portion of this screen, you can multiply the signal by different types of windows (Rectangular,
Hanning, or Flat Top), as explained in Section 2. This pre-processing step is a new feature in
this FFTSES version and allows, for instance, to reduce the artificial high frequency components
that can be generated when the time domain signal has a non-negligible value at its beginning
and/or its end. Similarly, you also have the alternative to define regularization functions on the
right of the dialog box. These different functions are used simply to ensure one or both ends of
the signal to be zero.

Figure 5: This dialog box shows how the signal can be modified by multiplying it by a window (defined on
the left) or by adding regularization functions (defined on the right).
Also, by default, the program does not recommend more than 30 frequencies per iteration and
the algorithm used to find those frequencies is Optimized. However, by clicking Advanced
Options, the appearing dialog shown in Figure 6 allows you to change these data (on the left).
On the right of this dialog box, it is possible to specify a reduced frequency range for which the
program will apply the frequency recommendation algorithm. This can be done by choosing the
User-Defined option.
Figure 6: This dialog box shows the options available to control the algorithm used for the generation of
the recommended frequencies (on the left) and the fields to modify the frequency range in which those
recommended frequencies will be restricted.
3.3 Inverse Mode Interface

Once the HIFREQ model has been run at the frequencies generated in the Forward mode, you
can get back to FFTSES and run the software in the Inverse mode by selecting this option from
the FFT list. This Inverse screen is shown in Figure 7 and Figure 8 and is described as follows:
1. System Response Options: The Response list can be set to Computed (see Figure
7) or User Defined (see Figure 8). If you select Computed, you will need to specify the
MALZ or HIFREQ files from which the computation results are extracted (the
corresponding screen is shown in Figure 9). On the other hand, if you select User-
Defined, you need to input the system response manually.

2. Conductors and Profiles Selection: If the Computed option is selected, the data grid
is used to indicate the conductor segments and profiles for which you want to compute a
specified quantity (GPR, touch voltage, etc.). On the other hand, if the User-Defined
option is selected, you can specify the system response manually in the data grid seen in
Figure 8.
3. Graphical Zone: In future versions, if the Computed option is selected, this screen will
show the MALZ/HIFREQ model. On the other hand, if the User-Defined option is
selected, the spectrum of the user-defined response will be displayed there.
4. Filter Zone: In future versions, a filter allowing you to easily select the conductor
segments and profiles will be available there.
1
3
Figure 7: Data Input panel in the Inverse mode with the Computed option selected.

Figure 8: Data Input panel in the Inverse mode with the User-Defined option selected.
Figure 9: Dialog allowing the specification of the HIFREQ or MALZ computation database (.F21) files to
be included when creating the FFTSES input databases.

Additional Notes
In the Home tab of the ribbon visible in Figure 7, you can check the option Do not Modulate
Signal, which instructs the program to use only the HIFREQ response to obtain the
recommended frequencies. The result is also that the inverse FFT will be performed only on the
unmodulated frequency response, i.e., the input surge is not taken into account.
Unless Do not Modulate Signal is checked and the sampling exponent is specified, FFTSES
cannot be run in Inverse mode prior to being run in Forward mode. As a result, Compute
Inverse FFT is available only when the system has been run in the Forward mode.
Also, you can control the way recommended frequencies are generated by FFTSES in Inverse
mode by clicking Advanced Options at the bottom right of the Inverse screen to display the
dialog box shown in Figure 10. On the left side of this screen, you can modify the parameters for
the frequency recommendation algorithm whereas a reduced frequency range can be applied by
selecting the User-Defined option on the right side of the dialog box.
Figure 10: This dialog box shows the fields to modify the frequency range in which the recommended
frequencies will be restricted.
3.4 Display of the Results

Once the program has been run (either in the Forward or in the Inverse mode), you can select
the Results panel shown in Figure 11 (point 1) and then, the Results group of tab (the Plot
Options tab) appears (point 2). This screen is described in the following:

2 3
Figure 11: Screen showing how to define the data in the Results tab of the ribbon.
1. Results panel: Click this tab to display the Plot Options in the ribbon.
2. Plot Options: When this tab is selected, you can choose the Plot Type (2D, Contour-
2D, Spot 2D, and 3D-Perpective) from the corresponding list, as well as the Type of
Axis (XY-Linear, XY Logarithmic, Y-Logarithmic, and Y-Percent). Also, using
the check boxes, you can ask for the Unmodulated and Modulated spectrums, for
which you can choose to plot the Magnitude or display the Real and Imaginary parts
of the response using the corresponding boxes. Similarly, the output response can also be
plotted in the Time Domain by selecting the corresponding check box.
3. Plot button: The plots are generated by clicking this button, which is available once the
databases are generated (if the Computed option was chosen in the Inverse mode) or if
at least two points are defined in the User-Defined table (if the User-Defined option
was chosen in the Inverse mode).
4. Display Options: In this panel, you can specify a Plot Title for the graph. It is also
possible to zoom on different portions of the graph using the Type of Axes Scaling table.
5. Embedded Plots: In future versions, the plots will be displayed at this location.

4 A Simple Example
To help you getting started with this new FFTSES program, a simple example of a metallic pole
being struck by lightning is demonstrated. In this example, we are interested in the Transient
Ground Potential Rise (TGPR) on the pole and on the soil surface close to it. The steel pole is 14 m
high with an embedded depth of 4 m in a 100 Ω-m soil. It is modeled as a steel conductor having
a radius of 1 cm, subdivided into 19 equal-length segments. Also, in order to compute the soil
potentials in the vicinity of the pole, 10 equally spaced individual observation points are defined
between 1 m and 10 m from the pole.
This example only focuses on the preparation of the FFTSES file using the new interface. It is
assumed that the user is familiar with HIFREQ and is able to prepare the corresponding model.
The FFTSES and HIFREQ files used to generate the results of this example can be loaded from
the following location: UGC 2016\Example Files\A New Interface for FFTSES.
4.1 Forward Mode

In transient studies performed with FFTSES, the input signal of the system is defined in the time
domain using the Forward screen of FFTSES. In this example, the input signal is a lightning
surge of the Heidler type. The steps to define this surge in the Forward screen of FFTSES are
described below whereas the corresponding screen is provided in Figure 12.
1. In the Forward screen, choose the Lightning (Heidler) for the Type of Signal.
2. Set the Time Duration (in seconds or microseconds) to 150 µs and the Sampling
Exponent to 11 i.e., 2,048 samples (this number must be a power of 2 for the FFT
algorithm). Following this, the Sampling Frequency and the Sampling Time are
updated. Note that you can also manually change them, leading to a recomputed value for
the Time Duration.
3. Set the Surge Parameters of the Heidler equation as they are defined in Figure 12.
Observe that the effect of modifying the coefficients can be seen in real time on the time
and frequency domain plots on the right side of the window.
4. Once the parameters of the input signal are defined, you can click Compute Forward
FFT in order to perform the FFT operation and generate the recommended frequencies.
Each step of the computations is displayed in the Computation Trace panel and the
error or warning messages, if any, appear in the Issues List. Also note that at the bottom
of this panel, you can click the links to open the .F09 and .F27 files directly, including the
model and computation information, as well as the recommended frequencies,
respectively.
Once the FFTSES computations are completed in the Forward mode, the recommended
frequencies must be imported in the HIFREQ file and the HIFREQ model should be run at those
frequencies. The corresponding HIFREQ file is hi_Pole_Run1.f05

Figure 12: Definition of the input surge in the data input panel of the Forward mode screen.
4.2 Inverse Mode

When the HIFREQ model has been run at all recommended frequencies generated by FFTSES,
you can extract the frequency domain response of the system to a unit current energization. This
frequency response will be loaded in the Inverse screen of FFTSES to obtain the time domain
response of the desired quantities at specific locations. In this example, we are interested in the
GPR on the pole and on the soil surface at 1 m, 2 m, and 3 m from the pole. The following steps
explain how to input these data into the FFTSES program and generate the necessary FFTSES
Input Databases
1. In the Inverse screen, select the Computed item from the Response list (see Figure
13).

Figure 13: Selection in the Inverse mode screen of the system response computed with HIFREQ.
2. To create the FFTSES Input Databases from the HIFREQ file, click HIFREQ/MALZ
Response Files to load the FFTSES Database Management dialog box displayed in
Figure 14. Then, click Add to load the hi_Pole_Run1.f21 file and click Create to generate
the databases. Note the progress bar showing the status of the building process. When
multiple .F21 files are used to generate the databases, this process can be aborted at any
time by clicking Cancel.
Figure 14: Dialog allowing the specification of the HIFREQ or MALZ computation database (.F21) files to
be included when creating the FFTSES input databases.
3. Once the FFTSES Input Databases are created, you can select the quantities to be
computed and the profiles and segments for which they will be computed. In our example,
we want to compute the GPR on segment #1 of the pole and the scalar potentials at the
individual observation points #1, #2, and #3. These data must be entered in the Inverse
mode screen table, as shown in Figure 15-a) and b) using the Conductors and Profiles
tab.
4. Once all data is filled in, you can click Compute Inverse FFT to obtain the time domain
response of the system for the selected quantities and conductors/profiles.
5. Once the run is completed, you may open the .F27 file (or the .F09 file) to see if there are
other frequencies recommended by the program. Then, you have the following options:
 Whether the program suggests further frequencies or not, you can always display the
results, as will be explained in the next section.
 If the program recommends more frequencies, it is strongly suggested to run HIFREQ
at those frequencies. In principle, you should do the following:
a) Save the HIFREQ file under a different name (hi_Pole_Run2.f05 for instance) and
run it with only the new recommended frequencies.
b) In the Inverse mode screen, load the FFTSES Database Management dialog
box and click Add to load the hi_Pole_Run2.f21 file in addition to the
hi_Pole_Run1.f21 file already there and click Create New Databases. Once this
process is done, close the dialog box and click Compute Inverse FFT.
c) When the FFTSES run is completed, verify the .F27 file (or the .F09) to see if there
are new recommended frequencies. If so, repeat this process from step a).
Otherwise, you can plot the results, as will be explained in the next section.
Following through this entire process requires 4 HIFREQ runs, resulting in a total of 84
frequencies. Once this whole process is done, you can display the results as shown in the next
section.
a)
b)
Figure 15: Conductors and Profiles Selection tables used to specify the conductor segments and
profiles in the Inverse mode screen. Note that the Reference Segment in the Conductors tab is used
only when the GPD of Conductor Segment is selected or when All Quantities is selected.
4.3 Results
Once FFTSES has been run in the Inverse mode, the system response in the time domain can be
plotted. Therefore, by clicking the Results tab, you can display the results using the Plots
Options group tab. As seen in Figure 16, you can select only Time Domain and choose the Real
Part. Then, by clicking Plot, the graphs shown in Figure 18 and Figure 19 are generated.

Figure 16: Results tab of the ribbon used to generate the plots.
Since most of the variation in the signal is seen at the early time of the transient curves, you can
zoom on the first ten microseconds by specifying this value in the table. For instance, the
coordinates shown in the table of Figure 17 have been used to generate Figure 20. Figure 21 has
been obtained the same way, but with Ymax = 100,000.
Figure 17: Table of the Results panel used to zoom on the plots. In this table, it is requested to show the
time domain plot only between 0 and 10 µs.

Figure 18: Conductor GPR of segment 1 of the pole hit by the lightning strike.
Figure 19: Soil potential (Scalar potential) of the profile points 1 to 3 at the soil surface.

Figure 20: Close-up view on the early time response of the curves shown in Figure 18.
Figure 21: Close-up view on the early time response of the curve shown in Figure 19.

5 Conclusion
In this article, a new interface for the FFTSES program was presented, showing its new
appearance and improvements. Also, the following new features of the program were presented:
 In addition to the time domain input signal, its frequency spectrum is now shown in the
Forward screen, avoiding back and forth operations when you want to see quickly the
effects of generator modifications on the frequency spectrum.
 Time domain windows can be defined in the user-interface in the order to smooth the
extremities of the input signal, avoiding artificial high frequency components in the
transient analysis.
 Similarly, regularization functions can be added at one or both ends of the signal to
attenuate the artificial high frequency components.
 It is possible to define a frequency range in which the recommended frequencies will be
included, allowing to restrict the selected frequencies within the range of interest or reduce
the number of frequencies to be run, if needed.
 In the new interface, it is now possible to request the computation of multiple quantities
for the conductor segments and profiles at the same time. In this case, the algorithm to
compute the requested frequencies is based on all specified quantities.
Since this is the first version of this revamped FFTSES program, further improvements are
expected in future years. For example, one of the enhancements being considered is the possibility
to view the network in the FFTSES interface. From this network, you will be able to define zones
in which conductor segments and profiles will be loaded and input in the Inverse mode screen,
instead of having to manually specify them in the table. Tools such as filters will allow you to keep
only a portion of the selected segments/profiles more easily. Similarly, the plotting of the results
obtained in Inverse mode will be embedded in the program, as is the case for the Forward
mode. Furthermore, more flexibility in the plotting will be provided, such that you will be able to
choose which curves to plot in a graph. Finally, improvements in the frequency recommendation
algorithms and the user-interface are also considered in the future years.
6 References
[1] Chami, M., Daigle, M., Dawalibi, E., Franiatte, S., Noel, G., Siahrang, M., Valcarcel, L., and Voyer, C.,
"Improvements in SESTralin," in CDEGS Users' Conference Proceedings, San Diego, California, ,
USA, 2015.

24 IMPROVEMENTS AND NEW FEATURES IN SESSHIELD-

3D
Stéphane Baron
Safe Engineering Services & Technologies ltd

Abstract
This article describes in detail the new features implemented and the major improvements made to the SESShield-
3D software package since the 2014 User’s Group Conference [1].
1 Introduction
Several new features have been implemented in the SESShield-3D software package since the
Users Group Meeting in 2014 [1]. On the technical side, the most important addition is the
Accuracy Level, which can help reduce the computation time. The user interface was also
improved, with many new CAD capabilities designed to allow you to quickly build the system to
be analyzed.
The major enhancements are:
 The Accuracy Level concept has been introduced in order to maximize the computation
speed by minimizing the number of sides in each polygons used to approximate the
interception surfaces.
 All rendering modes are now available for all the analysis methods. Buttons for selecting
the rendering mode for the generated volumes are now available on the toolbar.
 Components can now be labeled with their name or component number in the graphical
area.
 The location of objects imported from the SESShield-3D Object Database can now be
specified. A scaling factor can also be applied to the imported objects.
 SESShield-3D now provides options for highlighting components and objects in the
graphical area according to whether they belong to a group or not.
 The Construction Tree now has its own toolbar and its overall behavior was improved.
In particular, you can now control which objects are displayed in the graphical area (all
objects, only the objects to be protected or only the objects part of the protection system)
from the Construction Tree toolbar.
 Some options in the SESShield-3D Settings screen have been moved and new options
have been introduced. For example, the maximum number of recent files can be specified
from the On Startup section.

 In addition to the possibility to import geometries created with SESShield-3D and with
SESCAD (saved as MALT files), existing geometries saved as MALZ and HIFREQ files can
now also be imported in SESShield-3D.
 Finally, a new How To manual for designing the shielding system of a simple substation
has been added to the numerous existing SES How To manuals.
This article describes these and other improvements in greater detail.
2 User Interface Changes

The main screen has been slightly changed in order to provide additional accesses to existing
commands such as the rendering modes and to offer a toolbar entirely devoted to the
Construction Tree.
Hence, changing the rendering mode of the generated collection surfaces whatever the selected
computation method is now quickly accessible from the toolbar. Also, applying operations directly
to Components and Objects through the Construction Tree
Finally, the status bar at the bottom of the main screen clearly indicates the running status of the
application. Figure 1 shows these improvements applied to the main screen.
Figure 1: The main screen provides quick access to important commands and indicates whether the
application runs in Demo mode or not.
2.1 Rendering Modes Toolbar Buttons
The Rendering Modes, already accessible from View | Collection Surface Rendering State,
are also now quickly accessible on the toolbar. In addition, all modes can be applied whatever the
selected computation method. Table 1 lists the four collection surface rendering modes which can
be used when a lightning shielding analysis is performed.
Table 1: Available rendering modes
Buttons Description
Sets the collection surface rendering style in solid mode.
Sets the collection surface rendering style in transparent mode

without the hidden volume.
Sets the collection surface rendering style in transparent mode with

the hidden volumes.
Sets the collection surface rendering style in wire-frame mode.
2.2 Construction Tree Toolbar
The Construction Tree now has its own toolbar and provides functionalities which directly
control the items listed in the Construction Tree. Table 2 lists the new tools provided by the
Construction Tree toolbar.
Table 2: Icons used in the Construction Tree toolbar
Buttons Description
Collapses all the object folders so that you can no longer see all the
sub-objects and components under each one.
Expands all the object folders so that you can see all the sub-objects
and components under each one.
Expands the selected object folder so that you can see all the sub-
objects and components under it.
Shows only the object folders of level 2; i.e.; the level which shows
the editable characteristics of composite objects.
Renames the selected item. Note that the F2 shortcut key can also
be used to rename an item.
Limits the objects and components that are displayed in the

graphical area by their Protection Role. See Component and
Object Filter for more details.
Note that these functionalities are also available from the context-sensitive menu when you right-
click a node in the Construction Tree.

3 Creating, Editing and Selecting Objects

Minor additions in this version of SESShield-3D regarding the manipulation of objects in the
graphical area. However, useful functionalities such as the possibility to display the name or the
identification number of the components have been introduced.
3.1 Number and Name Labeling of Components
Labels showing either the identification number or the name of components can be displayed next
to components that are visible in the graphical area. This capability allows you to quickly identify
the components describing the system to be analyzed. It can be turned on from the Labeling
option of the View menu. Figure 2 shows a wind turbine for which all components are identified
by their identification numbers (a) and their names (b).
(a) (b)
Figure 2: Components in the graphical area can now be identified by (a) an identification number and (b)
their name.
Note that options allowing you to control the appearance of the labels (size, color, font, etc…) will
be available in a future version of SESShield-3D.
3.2 SESShield-3D Object Database
The SESShield-3D Object Database screen now allows you to specify the location of the object
to be imported. When the database is loaded in Use Mode (File | Database Manager), the
Actual Location zone is enabled and provides the current position of the geometrical center of
the selected object. Simply specify the desired X, Y, and Z coordinates of the selected object and
click Paste in Document to insert it in the current system.
In addition, the Desired Scaling Factor value allows you to apply a scaling rate by which the
selected object should be scaled during the insertion operation.

Figure 3: The coordinates of the center of the object and a scaling factor can now be specified when
inserting objects in the current system.
Note that in order to insert an object, you can also use the drag-and-drop feature: simply select it
among those offered in the list, drag it on the graphical area and drop it. The object will be inserted
in the current system at its default location.
3.3 Component and Object Filter
The Component and Object Filter (Figure 4) located above the Construction Tree is one of the
new functionalities provided by the Construction Tree toolbar (see Construction Tree Toolbar
for information about the other functionalities).
Four filter types are presently supported by SESShield-3D. They allow you to separate the
components and the objects used in the current system to analyze based on their Protection
Role.
Table 3: List of the four filter types supported by SESShield-3D
Filter Type Description
Show All Displays all the components and objects describing the
current system to be analyzed.
To Be Protected Displays only the components and objects defined as

requiring protection.
Protection System Displays only the components and objects defined as

offering protection against lightning strikes.
Ignored Displays only the components and objects that are

considered to be ignored from the point view of the
lightning protection analysis.

Figure 4: Four filter types allow displaying the components and the objects based on their Protection
Role.
In order to apply a filter, just select it among the suggested ones in the list. Automatically, only
the components and objects whose Protection Role matches the selected filter are displayed in
the graphical area.
For example, Figure 5-a shows the 3D model of a simple substation for which equipment such as
transformers are protected using lightning masts. In this 3D model, pre-built objects (trees, car,
etc…) have been added as decoration. The Component and Object Filter is set for displaying only
the equipment to be protected (in Figure 5–b), only the system of protection (in Figure 5–c) and
only the decorative objects (in Figure 5–d).
Whatever the selected filter type, a zoom extending to the components and objects being displayed
is performed. In addition, when one of the To Be Protected, the Protection System or the
Ignored filter is selected, SESShield-3D automatically selects the related components and
objects. Note however, that the selected objects cannot be grouped or ungrouped.
(a) (b)

(c) (d)
Figure 5: Filter (a) Show All (b) To Be Protected (c) Protection System and (d) Ignored applied to a
substation.
3.4 Highlighting Options for the Selected Components and Objects
In previous versions of SESShield-3D, it could sometimes be fastidious to distinguish the

highlighted components and objects for very large systems containing an important number of
items. In addition, when all the components and objects of such a large system are selected, the
fluidity of operations such as rotating and shifting can be slowed-down.
SESShield-3D now provides options for highlighting the components and objects in the graphical
area according to whether they belong to a group or not. Three options (Component Level,
Object Level and Component and Object) can be selected from the SESShield-3D Settings
screen (Figure 6).
Figure 6: SESShield-3D provides options for highlighting components and objects in the graphical area.
Three options are available: a) Component Level, b) Object Level and c) Component and Object.
 Selected Objects <Default = ‘Component Level’>: Defines the type of selection mode.

o Component Level: Creates a bounding box around only the selected components
using the Component Color option.
o Object Level: Creates a bounding box around only the selected composite objects
using the Component Color option.
o Object and Component: Creates a bounding box around both the selected
components using the Component Color option and the selected composite
objects using the Object Color option.
o Component Color <Default = ‘Green’>: Controls the color of the lines used to
highlight the objects.
o Object Color <Default = ‘Purple’>: Controls the color of the lines used to highlight
the components.
o Line Thickness <Default = 1>: Controls the thickness of the lines used to highlight
the components and the objects.
For example, Figure 7 shows how components and objects are highlighted when they are selected.
Figure 7–a shows that a bounding box is created around all the components while Figure 7–b
shows that a bounding box encapsulates only the selected objects. Figure 7–c shows that a
bounding box is created around all the components and around each of the selected objects.
(a) (b)
(c)
Figure 7: Three type of selection modes used to highlight the selected items are now available: (a)
Component Level, (b) Object Level and (c) Object and Component.
4 Calculation and Results

SESShield-3D now allows you to control how finely round surfaces are subdivided in order to
maximize computation speed or to maximize the computation accuracy when performing a
lightning shielding analysis. This section describes in details the new Accuracy Level setting
available from the Display category of the SESShield-3D Settings screen (Figure 9).

In SESShield-3D, round objects, shielding elements, and their respective lightning interception
surfaces are approximated by polygonal surfaces. The accuracy of a lightning shielding analysis
can be maximized by increasing the number of sides in each polygon, at the expense of
computation speed and memory usage.
This version of SESShield-3D offers an option to specify an Accuracy Level which allows you to
control the number of triangles used to generate the polygonal surfaces when performing a
lightning shielding analysis.
This value is expressed as a distance and is converted as a number of Slices (number of
subdivisions around the main axis; i.e.; around the Equatorial plane) and Stacks (number of
subdivisions around the main axis; i.e.; around the Meridian) when rounding the volumes
associated to components that are generated when a lightning shielding analysis is performed.
Figure 8 illustrates the effect of this accuracy setting on the interception surfaces generated when
using the Lightning Interception Surface Rendering and Intersection Method 1 for a single
cylinder object, using an accuracy of (a) 10.0 m, (b) 1.0 m and (c) 0.1 m.
(a) (b)
(c)
Figure 8: Accuracy Level of (a) 10.0 m, (b) 1.0 m and (c) 0.1 m applied to a single cylinder object.
The Accuracy Level value defines the distance between the different Slices and the distance
between the different Stacks used to round the generated volumes. This value can be specified
either in meters or in feet from the SESShield-3D Settings screen as illustrated Figure 9.
1Alternative to the Rolling Sphere Method used by SESShield-3D in order to create lightning interception surfaces for each object,
where the surfaces correspond to the loci of the centers of the rolling spheres. The concepts behind this method are discussed in [2].

Figure 9: The Accuracy Level allows you to control the number of triangles used to generate the
polygonal surfaces when performing a lightning shielding analysis.
5 SESShield-3D Related Documentation

A new manual showing how to design a shielding system for a simple substation has been added
to the numerous existing SES How To manuals. The objective of this manual is to show how to
carry out a typical lightning protection analysis for a simple substation using the SESShield-3D
software package.
This manual is also intended to help you understand the basic concepts behind SESShield-3D and
the main operations that can be performed in the program through simple step-by-step exercises
that teach you how to input data, run the analysis and explore the computation results. These
exercises take a short amount of time to complete and can be carried out in any order.
The lightning shielding analysis discussed in this manual deals with the substation described in
the IEEE-998-2012 standard and the objective of the analysis consists in properly positioning
lightning shielding devices (lightning masts and shield wires) to achieve a full protection of the
substation with the help of the SESShield-3D software package.
The manual provides a full description of the system to be analyzed, highlights the main
capabilities and features available in the program and offers practical exercises.
The manual can be open from the Help menu by selecting the Lightning Shielding Manual
option of the How To sub-menu.

Figure 10: A new How To manual devoted to SESShield-3D is now available and provides details for
completing a lightning shielding analysis for a simple substation.

This article has described in detail the new features and enhancements implemented in
SESShield-3D since the 2014 CDEGS Users Group Meeting. Most of the improvements have been
made to the CAD capabilities of the program in order to provide efficient tools to quickly create
the system that requires analysis against lightning.
This year, the Accuracy Level has been introduced to help reduce the computation time by
minimizing the number of sides in each polygons used to approximate the interception surfaces.
Several new features are presently under development, such as the possibility of subdividing the
system to be analyzed in zones where different criteria could be used to implement the shielding
system and the possibility to provide an alternative way to show the unprotected zones.
Finally, a new How To manual showing how to design a shielding system for a simple substation
has been added to the numerous existing SES How To. This document covers all the
functionalities of SESShield-3D in order to quickly create the three dimensional model of the
system that requires protection, design the protection system based on the available calculation
methods and optimize the shielding system.

7 References
[1] S. Baron et al, "Improvements and New Features in SESShield-3D," in CDEGS Users Group
Conference, Newport, Rhode Island, USA, June 16 to 19, 2014.
[2] Q. Xie et al, "3D Computer Graphics Enhanced Shielding Failure Evaluation by Collection Surface
Method," in Electric Power Systems Research, 2015, EPSR-D-15-00590R1.
[3] S. Baron et al, "Lightning Shielding Design Study for a Substation with Low Insulation Level," in
CDEGS Users Group Conference, San Diego, California, USA, June 8 to 11, 2015.
[4] SESShield-3D, "SESShield-3D," Safe Engineering Services and Technologies ltd, [Online]. Available:
www.sestech.com/Products/SoftPackages/SESShield-3D.htm.

25 USER-DEFINED CALCULATION METHODS FOR

CORONA PERFORMANCE IN SESENVIROPLUS
Rachid El Hani and Simon Fortin

Abstract
This article presents the new “User-Defined” method for the calculation of the corona performance of AC/DC
transmission lines in SESEnviroPlus. A detailed description of the parameters that can be customized when
specifying a User-Defined method is given. Practical examples of specification of User-Defined methods are
presented, along with numerical results for the corona performance of a typical transmission line calculated using
such methods.
1 Introduction
The SESEnviroPlus software package evaluates the environmental impact of AC/DC transmission
lines. Among other quantities, it can evaluate the corona performance of the lines by computing
the levels of Radio Interference (RI) and Audible Noise (AN) generated by transmission lines and
the Corona Losses (CL) of the lines. Knowing and understanding of corona effects on transmission
lines is an essential step for design consideration, in order to choose the appropriate diameter of
conductors and the number and size of sub-conductors (when bundled conductors are required).
These corona effects are sensitive to a large number of variable parameters which are difficult to
measure. These parameters are also responsible for fluctuations which calculations are unable to
take into account and that can only be understood statistically.
Based on statistical studies of experimental measurements, empirical methods are used for
predicting the corona performance of transmission lines. A large number of such empirical
methods have been proposed in the literature, based on different experimental conditions.
SESEnviroPlus supports several of those methods. However, as shown in Figure 1, only the
method name can be specified for those methods: the formulas and empirical coefficients used in
the formulation are not accessible and cannot be modified.
Many requests have been made over the years for methods that are either not directly supported
by SESEnviroPlus or that represent small variations on supported methods. Up until now, it could
be rather difficult (if not squarely impossible) to calculate corona performance parameters for
such methods using SESEnviroPlus.
This version of SESEnviroPlus introduces a new User-Defined method that takes care of this
problem. As shown in Figure 2, the input data to this method includes all the necessary empirical
coefficients and corrections. This new method can be used to:
 Change the coefficients of the existing (built-in) methods.
 Create new methods.

 Change, add or ignore corrections.
Figure 1: Built-in methods in SESEnviroPlus.
Figure 2: User-Defined method in SESEnviroPlus.
This article describes this new User-Defined method in detail. Section 2 presents the general
formulation of the method. Next, the usefulness of this formulation is illustrated by an example,
in Section 3.
Note: The User-Defined method is currently available in the command-mode version of
SESEnviroPlus only. The user interface for this feature is currently under development.
2 General Formulation
The corona performance of transmission lines is influenced by many factors, including:
 The line voltage
 The diameter of the conductors
 The relative geometrical position of the conductors
 The height of the conductors
 The weather conditions
 The ground resistivity

 The altitude
 The air density
The RI and AN depend also on the distance between the conductors and the observation point.
In the User-Defined method of SESEnviroPlus, the contributions of these parameters are
regrouped into two sets: a General Formulation that accounts for the effects of the voltage and
geometry of the line and Corrections that account for the generally smaller effects due to
environmental variations.
2.1 General Formulation
The general formulation of the corona methods has the following form:
𝐶 = 𝐶0 + 𝑓𝑔𝑎 (𝑔𝑎 ) + 𝑓𝑔𝑚 (𝑔𝑚 ) + 𝑓𝑑 (𝑑) + 𝑓𝑛 (𝑛) + 𝑓𝐷 (𝐷) + 𝑓𝑋 (𝑋)

(1)
+ 𝑓𝐻 (𝐻) + 𝑓𝑆 (𝑆) + 𝐶𝐶𝑜𝑛𝑠𝑡
where,
𝐶0 : The reference coefficient.
𝐶𝐶𝑜𝑛𝑠𝑡 : A constant.
𝑓𝑝 (𝑝): A function of parameter 𝑝.
The various parameters (and corresponding functions) entering this expression are as follows:
𝑔𝑎 : The average of the maximum gradient at the surface of the conductors in a bundle.
𝑔𝑚 : The maximum gradient at the surface of the conductors in a bundle.
𝑑: The diameter of the conductor.
𝑛: The number of sub-conductors in a bundle.
𝐷: The radial distance between the bundle and the observation point.
𝑋: The lateral distance between the bundle and the observation point.
𝐻: The height of the bundle.
𝑆: The distance between two poles, in case of bipolar DC circuit.
Generally, 𝐶0 is the value corresponding to a reference transmission line configuration. The term
𝐶𝐶𝑜𝑛𝑠𝑡 is a constant, i.e. a value that is independent of the above parameters.
The functions 𝑓𝑝 for each parameter 𝑝 are expressed by the following equation:
𝑘𝑑𝑖𝑣
𝑓𝑝 (𝑝) = 𝑘𝐶𝑜𝑛𝑠𝑡 + 𝑘𝑙𝑜𝑔 𝑙𝑜𝑔10 (𝑝) + 𝑘𝑡𝑖𝑚𝑒𝑠 𝑝 + (2)
𝑝
The coefficients 𝑘𝐶𝑜𝑛𝑠𝑡 , 𝑘𝑙𝑜𝑔 , 𝑘𝑡𝑖𝑚𝑒𝑠 , and 𝑘𝑑𝑖𝑣 define the function 𝑓𝑝 related to the parameter 𝑝 and
can be specified for all parameter function 𝑓𝑝 in equation (1).
In some cases, these coefficients are not constant, and can be functions of other parameters.
Several such functions are available; consult the SESEnviroPlus command help for a detailed list.
2.2 Corrections
Depending on the corona performance and the predictive methods, some corrections are added
to take into account of the parameters that are not included in the formula. Here are the most
often used corrections:
 Weather Corrections: Corona performance is strongly affected by weather conditions.
Some methods are applicable only under some specific weather conditions, and others
suggest to apply a correction to take weather conditions into account. The most often used
weather conditions are: Heavy Rain, Light Rain (or Wet Conductors) and Fair Weather
(dry Conductors):
- Heavy Rain: The rain intensity is greater than 7.7 mm/hr. This condition is also
referred to as 𝐿5 level in rainy weather.
- Light Rain: The rain intensity does not exceed 2.5 mm/hr. This condition is also
referred to level 𝐿50 under rain and wet conductor conditions.
- Fair Weather: The condition where there are no precipitations which results in dry
conductors.
 Altitude Correction: This correction term accounts for the effect of altitude above sea level.
 Operating Frequency Correction: This correction is applied, for example, when the
selected method is derived from 50 Hz transmission line measurements and the corona
performance is to be evaluated for a 60 Hz transmission line.
3 Application
3.1 EPRI method
This section illustrates the use of the User-Defined method, using an example computing RI by
the EPRI method. The Built-in EPRI method, for heavy rain weather conditions, is given in [2],
by:
580
𝑅𝐼 = 55.968223 − + 38 𝑙𝑜𝑔10 (𝑑) + 𝑘𝑛 (3)
𝑔𝑚
where:
0 𝑓𝑜𝑟 𝑛 ≤ 8
𝑘𝑛 = {
5 𝑓𝑜𝑟 𝑛 > 8
However, the formula used in Reference [3] is different, namely:
580
𝑅𝐼 = 51.488223 − + 38 𝑙𝑜𝑔10 (𝑑) + 𝑘𝑛 (4)
𝑔𝑚
To obtain this formula, a different statistical approach to fit the RI performance data was used.
This approach is described in [4]. There is a difference of 4.48 dB between the two formulas to
compute the same quantity, using the same method (ERPI). The results given by equations (3)
and (4) are for ANSI measurements (bandwidth of 5 kHz and measuring frequency of 1 MHz). To
convert them to CISPR specifications (bandwidth of 9 kHz and measuring frequency of 0.5 MHz),

3.1 dB should be added to the computed values. This CISPR correction, is indicated in [2] and [3]
for this method. However, many other corrections, are not indicated:
Altitude Correction:
+𝐴/300 (5)
𝐴 is the altitude above the sea, in meter.

Temperature Correction:
+40(1 − 𝛿𝑟 ) (6)
273+25 𝑝
𝛿𝑟 is the relative air density (𝛿𝑟 = . ).
273+𝑇 760
Radio Frequency Correction:

1.25
+20 𝑙𝑜𝑔10( ) (7)
1 + 𝑓2
𝑓 is the radio frequency, in MHz.
Operating Frequency Correction:
𝑓𝑜𝑝𝑝
+20 𝑙𝑜𝑔10 (0.5 (1 + )) (8)
60
𝑓𝑜𝑝𝑝 is the operating frequency in Hz.

In order to take those corrections into account, SESEnviroPlus adds corrections (5) - (8) to the
result given by equation (3). It is noticed that the three approaches (SESEnviroPlus, equation (3)
and equation (4)) will provide different results. The new User-Defined method makes it possible
to use SESEnviroPlus to obtain the results for all three approaches.
Error! Reference source not found. shows the commands needed to compute RI using the
formula in equation (3). Under the GENERAL-SETTING command, the method is defined as
being semi-empirical and that it is used for AC circuits. The VALIDITY-RANGE command
specifies that the method is used only for Heavy Rain weather conditions. Under the
CORRECTIONS command, the CISPR correction of 3.1 dB is added. The 0 before 3.1, indicates
that the correction is a constant. When, the correction is a function, the value 0 is replaced by the
Function ID that will used to compute the applied correction. The coefficients of equation (3)
are defined under the command COEFFICIENTS. As for the corrections, the value 0 before each
coefficient indicates that the related coefficient is a constant:
 The Reference 𝑪𝟎 is specified as 𝟓𝟓. 𝟗𝟔𝟖𝟐𝟐𝟑.
 The coefficient 𝒌𝒅𝒊𝒗 of the function 𝒇𝒈𝒎 is −𝟓𝟖𝟎 . 𝑔𝑚 is the average maximum
gradient.
 The coefficient 𝒌𝒍𝒐𝒈 of the function 𝒇𝒅 is 𝟑𝟖. 𝑑 is the diameter of the conductors.
 The coefficient 𝑪𝑪𝒐𝒏𝒔𝒕 is a function of the number of sub-conductor in the bundle, 𝑛. It

is given by 𝑘𝑛 in equation (3). In this case, the Function ID of 𝑘𝑛 is 3. In the upcoming

user interface, a description of each function (and the corresponding formula) will be
provided, to facilitate the selection of those functions.
USER-DEFINED-METHOD, RADIO-INTERFERENCE, UD-EPRI (Eq. 3)

GENERAL-SETTING
METHOD-TYPE,SEMI-EMPIRICAL
SOURCE-TYPE,AC
COEFFICIENTS
REFERENCE,0,55.968223
AVERAGE-MAXIMUM-GRADIENT,0,0.,0,0.,0,0.,0,-580.
CONDUCTOR-DIAMETER,0,0.,0,38.,0,0.,0,0.
CONSTANT,3,0.
VALIDITY-RANGE
WEATHER,HEAVY-RAIN
CORRECTIONS
RADIO-INTERFERENCE
CISPR,0,3.1
Figure 3: Commands to Compute RI, using formula (3).
3.2 Example
In this section, the RI will be computed using the built-in EPRI method of SESEnviroPlus and the
EPRI methods in equations (3) and (4), specified as User-Defined methods. As mentioned
before, the built-in method is based on equation (3), to which the CISPR correction and
corrections (5)- (8) are added. For the methods in equation (3) and (4), only the CISPR correction
is added. The AC line for which the calculations were made is a 250 kV, 3-phase circuit where:
𝐻1 = 18.2 m, 𝐻2 = 22.4 m, and 𝑆 = 4.9 m. See Figure 4 for the geometrical details.
𝑆
𝐻1 𝐻2
Figure 4: AC circuit configuration.
In Command Mode, Figure 5, the Built-in EPRI method of SESEnviroPlus is specified by simply
giving its name: EPRI. The User-Defined methods UD-EPRI (Eq. 3) and UD-EPRI (Eq. 4)
are defined explicitly, under the command USER-DEFINED-METHOD, by appropriate
coefficients and correction.

RADIO-NOISE,0.5,1,0
METHOD,EPRI
METHOD,USER-DEFINED
GENERAL-SETTING
SOURCE-TYPE,AC
DEFAULT-WEATHER,HEAVY-RAIN
COEFFICIENTS
CONSTANT,3,0.
VALIDITY-RANGE
WEATHER,HEAVY-RAIN
CORRECTIONS
RADIO-INTERFERENCE
CISPR,0,3.1
GENERAL-SETTING
SOURCE-TYPE,AC
DEFAULT-WEATHER,HEAVY-RAIN
COEFFICIENTS
CONSTANT,3,0.
VALIDITY-RANGE
WEATHER,HEAVY-RAIN
CORRECTIONS
RADIO-INTERFERENCE
CISPR,0,3.1
Figure 5: The commands used to define the EPRI method given by equations (3) and (4).
The computed RI is presented in Figure 6:

 Build-in Method: EPRI/GE in red curve,
 Method in equation (3): UD-EPRI (Eq. 3) in blue curve.
 Method in equation (4): UD-EPRI (Eq. 4) in black curve.
As mentioned, the difference between UD-EPRI (Eq. 3) and UD-EPRI (Eq. 4) is 𝟒. 𝟒𝟖 𝐝𝐁. Even if
the formula used in the Built-in Method is also given by equation (3), there are 𝟏. 𝟑𝟑 𝐝𝐁 difference
between EPRI/GE and UD-EPRI (Eq. 3). This difference is due to the added corrections (5) to (8)
for the Built-in SESEnviroPlus method.
4 Conclusion
To make SESEnviroPlus more useful and powerful, a new internal architecture of the code is
provided by using a general formulation for all corona methods. The example, shown in section
3, shows how the user-defined approach, using the new formulation can be useful to have a full
control over used corona methods.

EPRI Radio Interference

80
75
70
RI (dB)
65 EPRI/GE
UD-EPRI (Eq. 3)
60
UD-EPRI (Eq. 4)
55
50
-50
-45
-40
-35
-30
-25
-20
-15
-10
-5
0
5
10
15
20
25
30
35
40
45
50
X (in meter)
Figure 6: Radio Interference computed using Built-in EPRI method of SESEnviroPlus, equation (3) and
equation (4). The results are in dB above 1 𝜇V/m.
5 References
[1] P. S. Maruvada, Corona Performance of High-Voltage Transmission Lines, England: Research

Studies Press Ltd, 2000.
[2] CIGRE Working Grop 36.01, "Interferences Produced by Corona Effect of Electric Systems.," Cigre,
1966.
[3] S. D. S. a. V. L. C. R. G. Olsen, "Comparison of Several Methods for Calculating Power Line

Electromagnetic Interference Levels and Calibration With Long Term Data.," IEEE Transaction on
Power Delivery, vol. 7, no. 2, pp. 903-913, April 1992.
[4] SES & Technologies Ltd, "SESEnviroPlus - Electromagnetic Environment Study (How To...
Engineering Guide)," SES & Technologies Ltd, Montreal, 2012.

Part IV: Recently Published Technical Articles
IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 51, NO. 6, NOVEMBER/DECEMBER 2015 5023
Using Fall-of-Potential Measurements to Improve

Deep Soil Resistivity Estimates
Robert D. Southey, Member, IEEE, Majid Siahrang, Simon Fortin, Member, IEEE, and
Farid P. Dawalibi, Senior Member, IEEE
Abstract—When measuring the ground impedance of an elec-

trically isolated grounding system, it sometimes happens that the
test electrodes are placed at considerably greater distances from
the installation under test than the maximum electrode spacing
used during the soil resistivity measurements carried out during
the predesign phase of the grounding system. As a result, ground
impedance measurements carried out with the fall-of-potential
method may contain valuable supplemental information about Fig. 1. Arrangement of test electrodes in the Wenner method.
deep soil strata that can improve grounding system performance
predictions made during the design phase with computer mod-
eling software and explain discrepancies encountered between
predicted and measured ground impedance values. These data ferent depths. The method typically used by grounding design
can be also used during subsequent grounding design work as- engineers to measure the resistivity of the soil as a function of
sociated with facility expansion. It is shown in this paper how, depth is the Wenner four-electrode method [5], [6], although
in the absence of interfering metallic infrastructure, such ground other four-electrode methods can be used as well, as will be
impedance measurements can be converted into apparent soil
resistivity values corresponding to deeper soil layers, with an ex- discussed in the following. In this method, as shown in Fig. 1,
ample showing how dramatic improvements in grounding system four electrodes are equally spaced along a straight line, and
performance predictions can be obtained. while a current is forced to flow through the soil between
Index Terms—Fall-of-potential (FOP) test, ground impedance
the two outer electrodes (i.e., current electrodes C1 and C2),
measurement, power system grounding, soil resistivity data inter- the voltage difference between the two inner electrodes (i.e.,
pretation, soil resistivity measurement, touch voltage. potential electrodes P1 and P2) is measured. As the spacing be-
tween the electrodes is increased from small to large values, the
I. I NTRODUCTION ratio of the measured voltage between the potential electrodes
to the circulating current between the current electrodes is
H AVING adequate information regarding the soil structure

is of capital importance in the design and performance
prediction of grounding systems. Ground impedance, touch and
recorded: this is the so-called “apparent resistance.” This set of
apparent resistance versus electrode spacing data is interpreted
most efficiently using numerical inversion methods to find an
step voltages, and ground potential rise (GPR) very much depend equivalent horizontally layered soil structure that best fits the
upon the electrical resistivity of soil, not only in the immediate measurement data [7].
vicinity of the grid but also at depths significantly greater than The apparent resistance obtained at each electrode spacing is
the maximum dimension of the grounding system [1]. a function of the resistivity values of a range of soil depths,
For grounding design purposes, the soil is typically mod- with the weight of deeper layers increasing with electrode
eled with a horizontally layered isotropic electrical medium. spacing. Data from larger electrode spacings, therefore, contain
Modeling the soil with a multilayer structure makes it possible more information about the resistivity of deeper soil layers.
to account for variations in moisture content, concentration of It has been shown to be necessary to perform soil resistivity
mineral salts, grain size, compactness, temperature, and other measurements along traverses whose lengths are on the order of
factors affecting resistivity, as a function of depth [2]–[4]. several times the maximum dimension of the grounding system
Data obtained from soil resistivity measurements are inter- under study, in order to obtain good accuracy in grounding
preted to estimate the electrical resistivity of the soil at dif- system performance predictions [1], [6].
As a measure to verify that the design calculations correctly
Manuscript received December 31, 2014; revised February 20, 2015; predict the performance of the grounding system, a fall-of-
accepted February 23, 2015. Date of publication April 30, 2015; date of current
version November 18, 2015. Paper 2014-PSEC-0886.R1, approved for publica- potential (FOP) test is often required after the grounding system
tion in the IEEE T RANSACTIONS ON I NDUSTRY A PPLICATIONS by the Power is installed [6]. This test, under ideal conditions, measures the
Systems Engineering Committee of the IEEE Industry Applications Society. ground impedance (or resistance, when the reactive part is
This work was supported by Safe Engineering Services & Technologies Ltd.
The authors are with Safe Engineering Services & technologies ltd., Laval, small, as is the case for smaller grounding systems or high-
QC H7L 6E8, Canada (e-mail: robert.southey@sestech.com; majid.siahrang@ resistivity soils) of the installed grounding system. In this
sestech.com; simon.fortin@sestech.com; farid.dawalibi@sestech.com). method, a test current is caused to flow through the soil, from
Color versions of one or more of the figures in this paper are available online
at http://ieeexplore.ieee.org. the grounding system under test to a current return electrode,
Digital Object Identifier 10.1109/TIA.2015.2428679 located at a remote distance from the grid. At the same time, the
0093-9994 © 2015 IEEE. Personal use is permitted, but republication/redistribution requires IEEE permission.
See http://www.ieee.org/publications_standards/publications/rights/index.html for more information.
5024 IEEE TRANSACTIONS ON INDUSTRY APPLICATIONS, VOL. 51, NO. 6, NOVEMBER/DECEMBER 2015
Fig. 3. Extracting an extra Wenner data point from the FOP test data.
Fig. 2. Two ground impedance measurements yield one equivalent soil resis-
tivity measurement.
resulting potential rise of the grounding system with respect to

a series of test points along a straight line running between the
grounding system and the current return electrode is measured.
As recommended by standards and guidelines [6], the FOP
Fig. 4. General four-electrode method.
test electrodes are usually placed at a distance from the ground-
ing system under test of several times the largest dimension
of the installation under test. On the other hand, although a enough for the grounding grid to be considered equivalent to a
similar recommendation is made to carry out soil resistivity relatively small current injection electrode, as will be shown. In
measurements to electrode spacings that are on the same order the next two sections of this paper, we introduce two approaches
or greater than the largest dimension of the installation under to incorporate the data obtained from FOP tests into the soil
test, this recommendation is often not heeded, in the authors’ interpretation process.
experience. Grounding systems are therefore often designed
based on inadequate estimations of the soil structure. It is
A. Wenner Method
therefore often the case that the length of the FOP test traverse
is considerably greater than the maximum electrode spacing Since the position of each current injection electrode remains
used during the soil resistivity measurements carried out prior fixed throughout the FOP testing, it is only possible to generate
to the design of the grounding system. As a result, ground one Wenner data point. This requires the availability of test data
impedance measurements carried out with the FOP method (or an approximation thereof) corresponding to the potential
may contain valuable supplemental information about deep soil electrode located at 1/3 and 2/3 of the total FOP path length,
strata that can be used to verify grounding system performance as measured from the center of the grounding system under test
predictions made during the design phase with computer mod- to the current return electrode.
eling software and explain discrepancies encountered between With this approach, the difference between the apparent
predicted and measured ground impedance values. The more resistances read at points 1/3 and 2/3 of the FOP path length
extensive knowledge of the soil structure can be also used for is used as an extra Wenner data point, with an “a” spacing
future grounding studies associated with facility expansion. equal to 1/3 of the FOP path length. Since the length of the
In this paper, we show that, under certain conditions, the data FOP path is often considerably greater than the length of the
obtained from the FOP test can be incorporated into the origi- soil resistivity measurement traverse, this additional Wenner
nally measured soil resistivity data to improve estimated deep data point extracted from the FOP measurement can be helpful
soil resistivity and thus better estimate grounding system per- to better estimate the resistivity of the soil at greater depths,
formance. To this end, two different approaches are presented, the which can be detected based only on the original soil resistivity
performance of which is demonstrated with numerical simula- measurement data (see Fig. 3).
tions of an example study. All computer simulations have been
carried out using the CDEGS suite of software packages [8].
B. General Method
Even more information can be extracted from the FOP data if
II. M ETHODOLOGY
a more flexible soil resistivity interpretation method is used. As
As shown in Fig. 2, as long as the position of the current shown in Fig. 4, soil resistivity measurements need not be made
return electrode remains fixed, as is usually the case, the appar- with equally spaced electrodes: the potential electrodes can be
ent resistances obtained from two FOP measurements, shown located at arbitrary positions along a straight line joining the
on the left of the figure, can be combined to infer a corre- current electrodes [9]. This method will be called the “General
sponding apparent resistance for a four-electrode soil resistivity method” in this paper. It will be noted that the Wenner method
measurement, by subtraction of the two apparent resistances. is a special case of the General method, in which the electrode
Thus, a series of FOP measurements can be converted into a spacings (i.e., Se1, Si, and Se2) are equal.
series of four-electrode soil resistivity measurements made at As shown in Fig. 2, any number of pairs of data points
large electrode spacings. obtained from an FOP test can be used to generate a set of soil
It is of course necessary that the distance between the resistivity measurements corresponding to the General method.
grounding grid under test and the other electrodes be large For example, as shown in Fig. 5, the differences between
SOUTHEY et al.: USING FOP MEASUREMENTS TO IMPROVE DEEP SOIL RESISTIVITY ESTIMATES 5025
TABLE I
A CTUAL S OIL S TRUCTURE AT S ITE U NDER S TUDY
Fig. 5. Extracting extra soil data for the General method from the differences
between apparent resistances measured at adjacent locations of the potential
electrode in the FOP test. TABLE II
S OIL R ESISTIVITY M EASUREMENT D ATA
A. Case Description
The site under study is square, with a length of 30 m (98.42 ft).
The actual soil structure at the site is horizontally layered, with
Fig. 6. Earth potentials as a function of distance from center of 30-m square
grounding grid versus those from 1-m vertical electrode, for 1-A injection, in
the characteristics presented in Table I.
100-Ω · m soil. Let us suppose that the limiting grounding design specifica-
tions are as follows:
• ground resistance not to exceed 1.0 Ω;
the apparent resistances measured at adjacent locations of the • touch voltage not to exceed the IEEE Standard 80 limit.
potential electrode can be used to extract a corresponding four-
electrode measurement data set. It is worth mentioning that the For the purposes of this example, let it be also assumed that
FOP data obtained from points close to the grounding grid (i.e., it has been established that the 1-Ω ground resistance of this
on the order of one half grid diameter away from the edge of the grounding system is large compared with the ground impedance
grid) should not be used, as the soil potentials at these locations of the system to which it is being connected, resulting in a GPR
may still be influenced by the proximity of the edge of the during fault conditions that is insensitive to small changes in
grounding system (see Fig. 6, for example, which compares the ground resistance of the grounding system under study. Let
earth potentials from a point source with those from a square us assume that this GPR is 2.0 kV (symmetrical, RMS).
grounding grid in a uniform soil).
Note that, to be able to use the FOP test data to improve the B. Available Data From Soil Resistivity Measurements
soil resistivity estimations, the following additional conditions
Let us suppose that, prior to the construction of the instal-
should be satisfied.
lation, soil resistivity measurements were carried out with the
• The grounding grid under study should preferably not Wenner four-electrode method, at “a” spacings varying from
be connected to any external grounds at the time of 0.1 m (0.33 ft) to 30 m (92.48 ft), during the design phase of the
the testing (e.g., construction power neutral, distribution project. Table II shows the Wenner four-electrode measurement
feeder neutrals, and transmission line shield wires). data, in the form of apparent resistances, corresponding to the
• No extensive grounded metallic infrastructure should be three-layer soil structure presented in Table I for the electrode
located close enough to the measurement traverse to sig- spacings listed. These data points are also shown in Fig. 7, in
nificantly distort earth potentials. This condition applies the form of apparent resistivity values, as orange circles.
equally well even if conventional resistivity measure- The data listed in Table II are processed with a numerical
ments are carried out. data inversion algorithm to obtain an equivalent horizontally
layered soil that fits the measurement results. Fig. 7 shows
the results of the soil resistivity interpretation process. The
III. C ASE S TUDY
“computed curve” shown in the figure represents the apparent
This section presents a case study illustrating how the meth- soil resistivity values that would have been measured for the
ods presented earlier can improve soil structure estimates and interpreted soil structure.
thereby grounding performance predictions for an example As it is observed, the computed curve of the apparent soil re-
grounding system. sistivity matches the measured data quite well, which indicates
TABLE III
G RID P ERFORMANCE IN I NITIAL AND A CTUAL S OIL S TRUCTURES
TABLE IV
T OUCH V OLTAGE S AFETY L IMIT
Fig. 7. Soil interpretation results based on the Wenner data obtained from the
soil resistivity measurements only.
Fig. 8. Grounding system design.
that the deduced soil structure reflects the characteristics of the

soil at each measurement location. The reported RMS error
is the root-mean-square error between the value of apparent
resistivity obtained from the measurement data points and from
the interpreted soil structure (i.e., from the computed curve). Fig. 9. Contour plot of touch voltage in the actual three-layer soil. Shaded
areas exceed the IEEE limit of 936.5 V. Grid GPR is constant at 2 kV.
Note that the soil resistivity data, due to a limited maximum
electrode spacing, barely detect the presence of a third deeper
layer, with lower resistivity: only a very slight drop in apparent soil is quite a bit larger than the value computed with the two-
resistivity at the largest electrode spacing begins to suggest the layer soil. In both cases, the grid GPR is approximately 2.0 kV.
presence of such a layer, but this drop is too slight and supported If we use the system data shown in Table IV for the IEEE
by too few data points for the inversion process to resolve. Standard 80 safety calculation, we obtain a tolerable touch
voltage limit of 936.5 V (here, we apply the dc offset decrement
factor of 1.052 to reduce the touch voltage limit so that we can
C. Grid Design and Its Performance
compare it with the symmetrical RMS value of the computed
Fig. 8 shows a grounding design for the site, consisting of a touch voltage).
4 × 4 mesh grid buried at a depth of 0.5 m (1.6 ft), along with Fig. 9 shows a contour plot of the touch voltage occurring
6-m (20 ft) ground rods installed along the perimeter of the grid when the actual three-layer soil is considered: shaded areas
on 7.5-m centers. represent locations exceeding the IEEE safety limit. Clearly, the
Table III presents the computed ground resistance and max- grounding system is underdesigned, as a result of inadequate
imum touch voltage (symmetrical, RMS) of the grid for the soil resistivity data.
actual three-layer soil structure, on one hand, and the initial Why this result? Neglecting the third low-resistivity bottom
two-layer soil structure obtained from the limited soil resistivity layer in the initial two-layer soil structure resulted in a higher
measurement data, on the other. As shown in Table III, although apparent ground resistance. When the low-resistivity bottom
the ground resistance of the grid in the actual three-layer soil layer of the true soil structure is considered, the ground re-
is smaller than the value predicted with the two-layer soil, the sistance drops, drawing more of the available fault current out
maximum touch voltage resulting from the actual three-layer of the grounding grid, through the surrounding soil, into the
Fig. 10. Locations of test electrodes in the FOP test.
TABLE V
FOP T EST D ATA (A CTUAL T HREE -L AYER S OIL )
Fig. 12. Soil resistivity inversion results after the inclusion of the additional
Wenner data point from the FOP test.
Fig. 13. Extracting additional soil data for the General method from the
FOP test.
Fig. 11. Extracting an additional Wenner data point from the FOP test. Accordingly, the ground resistance and the maximum touch
voltage computed for the grid become 0.87 Ω and 1111 V,
low-resistivity deep soil, thereby increasing earth potential respectively. As expected, the modified soil structure results in a
gradients and therefore touch voltages, an effect not seen with significantly better prediction of the performance of the ground-
the initial two-layer soil. ing system, as compared with the soil structure interpreted with
only soil resistivity measurement data limited to a maximum
D. FOP Test “a” spacing of 30 m.
After the installation of the designed grid, an FOP test

F. Improving Soil Structure Interpretation With Multiple
is carried out, along the path shown in Fig. 10. The return
Additional Data Points Extracted From FOP Test
electrode is located 180 m (590.5 ft) away from the center of
the grounding system, and the apparent ground resistance is The differences between the apparent resistances measured
measured along the FOP path at intervals of 15 m (49.21 ft), at adjacent potential electrode locations shown in Fig. 13 can
starting 30 m (98.42 ft) away from the center of the grid and be used to generate multiple data points, as if measured with
ending 15 m (49.21 ft) before the current return electrode. the General four-electrode soil resistivity measurement method
Table V lists the apparent resistance data obtained from a illustrated in Fig. 4.
computer simulation of the FOP test, with the grid modeled in The apparent soil resistances generated this way from Table V,
the actual three-layer soil structure listed in Table I. as well as the corresponding electrode spacings (see Fig. 4 for
nomenclature), are presented in Table VI.
Fig. 14 shows the interpreted soil structure, which is com-
E. Improving Soil Structure Interpretation With Extra Wenner
puted using the original soil resistivity data in Table II along
Data Point Extracted From FOP Test
with the eight additional soil resistivity data points obtained
As shown in Fig. 11, the difference between the apparent from the FOP test. In this figure, the computed curve is plotted
resistances measured at 60 m (197 ft) and 120 m (394 ft) can be as a function of the distance between C1 and P1 electrodes
added to the soil resistivity measurement data (see Table II) as according to Figs. 1, 2, and 4. As shown in Fig. 14, the inter-
a supplementary Wenner data point with an “a” spacing equal preted three-layer soil structure is in excellent agreement with
to 60 m (197 ft), which is one third of the FOP path length. the actual soil structure and therefore is expected to accurately
The soil interpretation results after the inclusion of the predict grounding grid performance.
additional data point obtained from the FOP test are shown Indeed, with the refined soil structure, the computed grid
in Fig. 12. This figure shows that the new interpreted soil resistance is 0.86 Ω, and the maximum computed touch voltage
structure, with the benefit of the additional Wenner data point, is 1015 V, both of which are very close to the values listed in
becomes significantly closer to the actual soil structure. Table III for the actual three-layer soil.
TABLE VI • As shown in the case study presented in this paper, de-

D ATA O BTAINED F ROM THE FOP T EST TO BE U SED IN G ENERAL
M ETHOD S OIL R ESISTIVITY D ATA I NVERSION pending on the actual soil structure, even a maximum soil
resistivity test electrode spacing equal to the maximum
length of the grounding system may not be adequate to
sound deeper soil layers that have a significant influence
on grounding performance.
• The ground resistance value alone, such as the one ex-
tracted from the FOP test, is not a reliable indicator
of the adequacy of a grounding system design. Indeed,
measuring a smaller ground resistance than predicted by
the design calculations can be a sign that actual touch
(and step) voltages will be higher than those predicted
by the design calculations. On the other hand, a complete
measured FOP curve contains valuable information about
the resistivity of soil at greater depths, which can be used
to help verify design calculations and thus improve the
accuracy of grounding performance predictions.
ACKNOWLEDGMENT
The authors would like to thank the constructive comments
from the Research and Development Team involved in the
development and maintenance of the RESAP and HIFREQ
modules of the CDEGS software package.
R EFERENCES
[1] R. D. Southey and F. P. Dawalibi, “Improving the reliability of power
systems with more accurate grounding system resistance estimates,” in
Fig. 14. Soil interpretation results after inclusion of eight extra soil resistivity Proc. Int. Conf. Power Syst. Technol., Oct. 2002, vol. 1, pp. 98–105.
data points obtained from the FOP test. [2] F. P. Dawalibi, J. Ma, and R. D. Southey, “Behaviour of grounding systems
in multilayer soils: A parametric study,” IEEE Trans. Power Del., vol. 9,
no. 1, pp. 334–342, Jan. 1994.
[3] H. Lee et al., “Efficient grounding designs in layered soils,” IEEE Trans.
IV. C ONCLUSION Power Del., vol. 13, no. 3, pp. 745–751, Jul. 1998.
[4] F. P. Dawalibi and N. Barbeito, “Measurement and computation of the
This paper has presented the idea of extracting information performance of grounding systems buried in multilayer soils,” IEEE Trans.
Power Del., vol. 6, no. 4, pp. 1483–1490, Oct. 1991.
from FOP tests to improve estimates of deep layer soil electrical [5] IEEE Guide for Safety in AC Substation Grounding, IEEE Std. 80-2000,
resistivity. Two approaches have been described to incorporate Jan. 2000.
the FOP data into the soil resistivity interpretation process. [6] IEEE Guide for Measuring Earth Resistivity, Ground Impedance, and
Earth Surface Potentials of a Ground System, IEEE Std. 81-2012,
Through numerical simulations applied to a case study, the Dec. 2012.
performance of the proposed methods has been examined. The [7] F. P. Dawalibi and C. J. Blattner “Earth resistivity measurement interpre-
results obtained from the simulations have led to the following tation techniques,” IEEE Trans. Power App. Syst., vol. PAS-103, no. 2,
pp. 374–382, Feb. 1984.
conclusions. [8] F. P. Dawalibi and F. Donoso, “Integrated analysis software for grounding,
• Both of the proposed methods for extracting data from EMF, and EMI,” IEEE Comput. Appl. Power, vol. 6, no. 2, pp. 19–24,
Apr. 1993.
the FOP tests are quite effective in improving estimates [9] J. He, R. Zeng, and B. Zhang, Methodology and technology for Power
of deeper soil layer resistivity values. In particular, con- System Grounding. New York, NY, USA: Wiley, 2012, pp. 86–87.
verting the data obtained from FOP tests to soil resistivity
data corresponding to the unequally spaced four-electrode
Robert D. Southey (M’87) received the B.Eng.
General method can result in particularly good estimates (Hons.) degree in electrical engineering from McGill
of the actual soil structure. University, Montreal, QC, Canada, in 1986.
• A two-layer soil model, which may be consistent with Since then, he has been a Researcher, a Project
Manager, and a Department Director with Safe En-
soil resistivity measurements made to short or moderate gineering Services & technologies ltd., Laval, QC,
electrode spacings, such as those that are often used for Canada, specializing in grounding-related measure-
computer simulations of grounding system performance, ment techniques, grounding design, stray voltage
studies, electrical safety studies, and ac interference
can result in poor predictions of grounding system perfor- analysis and mitigation design. He has authored or
mance. Multilayered soil structures, such as those that are coauthored over 50 technical papers and articles on
frequently obtained in practice from soil resistivity mea- the aforementioned subjects, several of which are referenced in ANSI/IEEE
Standard 80–2000, and well over 100 technical reports.
surements made to greater electrode spacings, provide a Mr. Southey is a member of NACE International. He is also a Registered
more reliable basis for grounding studies. Engineer in the Provinces of Quebec, Manitoba, and Alberta in Canada.
Majid Siahrang, received the B.Eng. degree from Farid P. Dawalibi (M’72–SM’82) was born in 1947.
Bu-Ali Sina University, Hamedan, Iran, in 2003, the He received the B.Eng. degree from St. Joseph’s Uni-
M.Sc. degree from Sharif University of Technology, versity, Beirut, Lebanon, affiliated with the Univer-
Tehran, Iran, in 2005, and the Ph.D. degree from sity of Lyon, and the M.Sc. and Ph.D. degrees from
the École Polytechnique de Montréal, Montreal, the École Polytechnique de Montréal, Montreal, QC,
QC, Canada, in 2011, all in electrical engineering Canada.
(power). From 1971 to 1976, he was a Consulting Engi-
From 2005 to 2007, he was a Research En- neer with Shawinigan Engineering Company, Mon-
gineer with the Power Research Institute (NRI), treal. He has worked on numerous projects involving
Tehran. Since 2011, he has been a Research Scientist power system analysis and design, railway electri-
with Safe Engineering Services & technologies ltd., fication studies, and specialized computer software
Laval, QC, Canada. His research interests include power system analysis; power code development. In 1976, he joined Montel-Sprecher & Schuh, Montreal,
system grounding; ac interference analysis; and electromagnetic modeling of a manufacturer of high-voltage equipment, as Manager of Technical Services
electric machines, power transformers, and superconducting power equipment. and was involved in power system design, equipment selection, and testing
for systems ranging from a few to several hundred kilovolts. In 1978, he
founded Safe Engineering Services & technologies ltd., Laval, QC, Canada,
Simon Fortin (M’06) received the B.Sc. degree in a company that specializes in soil effects on power networks. Since then, he has
physics from Laval University, Quebec, QC, Canada, been responsible for the engineering activities of the company, including the
in 1985 and the Ph.D. degree in physics from The development of computer software related to power system applications. He
University of British Columbia, Vancouver, BC, has authored over 300 papers on power system grounding, lightning, inductive
Canada, in 1991. interference, and electromagnetic field analysis. He has also written several
His area of specialization has been in the the- research reports for the Canadian Electricity Association and the Electric Power
oretical aspects of high-energy particle physics. Research Institute.
In 1992–1993, he was a Research Assistant with
the Department of Nuclear Physics, University of
Montreal, Montreal, QC, Canada, again specializing
in particle physics. Since 1994, he has been with Safe
Engineering Services & technologies ltd., Laval, QC, Canada, which he joined
as a Research Scientist. He has authored or coauthored about 40 papers on light-
ning, grounding, and electromagnetic fields compatibility-related problems. His
research interests include the computation of electromagnetic fields at various
frequencies and transient phenomena.
Analysis of Grounding Systems in Horizontal Multilayer Soils Containing Finite
Heterogeneities
Simon Fortin Nina Mitskevitch Farid P. Dawalibi

Member, IEEE Safe Engineering Services & Senior Member, IEEE
Safe Engineering Services & technologies ltd. Safe Engineering Services &
technologies ltd. Laval, Quebec H7L6E8, Canada technologies ltd.t
Laval, Quebec H7L6E8, Canada info@sestech.com Laval, Quebec H7L6E8, Canada
info@sestech.com info@sestech.com


Abstract – This paper presents a theoretical model for the analysis However, all of those techniques for analyzing
of grounding systems located in multilayer soils in which arbitrary heterogeneous soils are restricted to finite volumes embedded
heterogeneities (finite volumes) are embedded. The presence of the in a uniform soil. As shown in [11], finite volumes embedded
heterogeneities is handled through a Boundary Element Method while in a more realistic multilayer soil can also be analyzed by
that of the horizontal soil layers is handled analytically, using Image representing the soil layers by very large finite volumes of
Theory (or any other equivalent technique). Numerical results are
presented for several cases, and are shown to agree with limiting
appropriate shape. In most cases, however, such an approach
cases. The results also clearly show that the presence of finite requires an extremely large amount of computer resources in
heterogeneities in multilayer soils can have a very strong impact on order to represent the effect of the soil layers accurately.
the safety performance of a grounding system. This paper presents a technique for analyzing grounding
systems buried in horizontally layered soils that include
Index Terms – grounding/earthing, layered soils, heterogeneous heterogeneous volumes of any shape. The method does not
soil volumes, earth potential and ground potential rise. require the horizontal layers of the multilayer soil to be
represented as large finite volumes. In fact, the presence of the
horizontal layers can be accounted for using any existing
I. INTRODUCTION technique (for example, image theory) applicable for the case
It is well known that the electrical characteristics of the soil where the heterogeneous volumes are absent.
at the location where a grounding grid is buried have a very The method uses the same amount of computer resources
strong impact on the performance of the grounding grid. Most as for finite volumes embedded in a uniform soil and is
soil resistivity measurements reveal the presence of therefore applicable for the same types of scenarios [4], [5]. It
horizontally layered soils, with uniform characteristics in each allows for the analysis of grounding grids in complex soil
layer. Accordingly, the analysis of the performance of environments that is as realistic as possible, resulting in more
grounding grids in this type of soil has received a lot of accurate and more economical grid designs.
attention from researchers [1]-[3]. The presence of local The following section describes the proposed method for
heterogeneities (i.e., regions of finite volume where the the analysis of grounding systems buried in layered soils with
resistivity differs from that of the surrounding soil) can also heterogeneous soil volumes. The next section presents
influence the performance of a grounding grid in several examples of calculations carried out using this method for
scenarios. Examples of finite volumes include scenarios where simple systems, and compares them to results obtained using
the grounding system is located in the vicinity of various existing methods. The last section discusses the computation
natural or artificial soil elevations or depressions such as results for a more complex and realistic case.
backfill layers, cliffs, open mines, water reservoirs (ponds,
lakes or sea shore), hydroelectric dams, etc. [4], [5]. II. DESCRIPTION OF THE APPROACH
There are only a few special types of heterogeneous soils A. Summary of Theoretical Model
that allow exact analytical closed form solutions. These
include soil types where the heterogeneity is of hemispherical The analysis of grounding systems buried in a soil
[6], cylindrical [7], or hemi-spheroidal [8] shape. The analysis including heterogeneous volumes is based on a low-frequency
of grounding systems that includes finite heterogeneities of version of the moment method [12] combined with an
arbitrary shape can be carried out only using numerical application of the Boundary Element Method [13].
methods, involving a discretization of the boundary of the The conductors of the grounding grid are assumed to be
heterogeneous volumes. The theoretical basis and the cylinders of length considerably larger than their radius. These
computation algorithms for heterogeneous soil volumes of conductors are subdivided into ܰ௖ conductors segments.
arbitrary shape have been developed during the past several Segment ݅is assumed to discharge a current ‫ܫ‬௜ into the soil,
years [4], [5], [9]-[11]. with the current being uniformly distributed over the length ‫ܮ‬௜
of the segment.
Similarly, the surface of each soil volume is subdivided soil interface. Cases where a finite volume either crosses or
into a total of ܰ௉ small elements (patches), with patch ݅ coincides partly with a soil layer interface require a special
assumed to carry a uniform charge ܳ௜ distributed over the area treatment. This will be discussed in a future publication.
‫ܣ‬௜ of the patch. Consider, then, a finite volume of uniform resistivity ߩ௏
The solution proceeds by adjusting the strength of the embedded in a layer of resistivity ߩ௘ of a multilayer native
current and charges until the following conditions are satisfied: soil, as shown in Fig.1. The general task at hand is to find a
1. The potential at the midpoint of each conductor solution of Poisson’s equation that is valid both inside and
segment is an unknown constant ܸ଴ (the outside the volume and that satisfies all boundary conditions
equipotentiality condition). (continuity of the potential and of the normal component of the
2. The total current leaking out of all conductor current) in the native soil.
ே಴
segments is a known constant ‫ܫ‬଴ , i.e. ∑௜ୀଵ ‫ܫ‬௜ =
‫ܫ‬଴.
3. The potential and normal component of the
current density is continuous at a single point (the
centroid of the patch) across each soil volume
patch.
Other conditions are possible. In particular, the
equipotentiality condition can be replaced by an equation
relating the mid-point potentials of pairs of segments joined at
a node to the longitudinal current flowing in the conductors to Fig. 1. Soil volume located in the horizontal multilayer soil model
account for the voltage drop due to the impedance of the
conductors. This modification does not change the nature of In principle, this could require four different solutions of
the approach used to account for the soil layering and will not Poisson’s equation: one for each combination of the location
be discussed further. of the source and observation point (i.e., inside and outside the
The potential to be used when imposing the above volume). As will be shown shortly, however, the solution for a
conditions is due to the current discharged by all conductor given location of the source can be taken to be the same for
segments and to the charges on all soil volume patches: observation points located both inside and outside the volume.
ே಴ ೔ ூ ேು ೔ ொ
This has the advantage of automatically satisfying the potential
Ԧ) = ∑௜ୀଵ
ܸ(‫ݎ‬ ∫ ‫ݎ( ܩ‬ Ԧ௦) ݀‫ܮ‬௜ + ∑௜ୀଵ
Ԧ, ‫ݎ‬ ∫ ‫ݎ( ܩ‬
Ԧ, ‫ݎ‬
Ԧ௦) ݀‫ܣ‬௜ (1) boundary condition at the surface of the volume.
௅ ௅ ூ
೔ ೔ ஺ ஺ ொ ೔ ೔
For a point source located outside the volume, the Green’s
where ‫ݎ‬ Ԧ is the location of the calculation point and the function can be taken as the known solution applicable in the
integrals are computed over the length of conductor segment ݅ native soil, in the absence of the volume. Call this solution ‫ܩ‬ே .
(for the first integral) and over the area of soil volume patch ݅ In the present work, this solution is obtained using the well-
⃗(‫ݎ‬ known Image Theory, but any technique that yields a valid
(for the second integral). The current density ‫ܬ‬ Ԧ) needed for
solution for this function could be used.
the third condition can be calculated through ‫ܬ‬ ⃗(‫ݎ‬Ԧ) = ‫ܧ‬ሬ
⃗ (‫ݎ‬
Ԧ)/ Note that this solution is valid for observation points located
ߩ(‫ݎ‬Ԧ), with ߩ(‫ݎ‬ Ԧ) being the soil resistivity at point ‫ݎ‬ Ԧ and the both inside and outside the volume. This is obvious for
electric field ‫ܧ‬ ሬ
⃗ (‫ݎ‬
Ԧ) being obtained by analytical or numerical observation points located outside the volume, since function
differentiation of (1). ‫ܩ‬ே was selected. This is also true for observation points
The Green’s functions ‫ܩ‬ூ(‫ݎ‬ Ԧ, ‫ݎ‬
Ԧ௦) and ‫ܩ‬ொ (‫ݎ‬Ԧ, ‫ݎ‬
Ԧ௦) appearing located inside the volume, since ‫ܩ‬ே actually satisfies
in (1) give the potential at point ‫ݎ‬ Ԧ due to a point source of Laplace’s equation everywhere in the volume, as it must.
current (charge) located at point ‫ݎ‬ Ԧ௦. They should be selected in The solution for a point source located inside the volume
such a way that Poisson’s equation is satisfied everywhere and must be different, since the local resistivity around the point
that the boundary conditions of continuity of the potential and source is the resistivity of the volume, which is normally
normal component of the current density are satisfied at the different than that of the layer of the native soil in which it is
layer interfaces of the multilayer soil. Their determination is embedded. A solution ‫ܩ‬௏ of Poisson’s equation that is suitable
the object of the next section. for a point source located inside the volume is then:
B. Determination of Green’s Functions ఘೇ
‫ܩ‬௏ = ‫ܩ‬ே (2)
ఘ೐
This section completes the formalism by providing explicit
choices for the Green’s functions introduced in the previous For observation points located in the native soil outside the
section. For simplicity, the discussion will be limited to the volume, this solution obviously satisfies the boundary
case where a single volume is present; the generalization to conditions since it is just a multiple of the solution for a point
multiple volumes is straightforward and will not be discussed. source located in the native soil. Inside the volume, this
It will also be assumed that the volume is located in its entirety solution has the correct singular behavior around the point
within one layer of the multilayer soil, with the exception that source. This last condition simply means that if we were to
part of the boundary of the volume may coincide with the air- consider the total current ‫ ܫ‬passing through a sphere with a
very small radius ܽ centered on that point source, then when Combining equations (4) and (5) gives:
using ‫ܩ‬௏ as Green’s function, this current would be related to ଵ ଵ ఙ ଵ ଵ
the electric field by ‫ߩ = ܧ‬௏ ‫ܫ‬/4ߨܽଶ, as required. ‫ܧ‬଴ ቀ − ቁ= ቀ + ቁ (7)
ఘೇ ఘ೐ ଶఢబ ఘೇ ఘ೐
Together, functions ‫ܩ‬ே and ‫ܩ‬௏ define the Green’s
function ‫ܩ‬ூ in (1) for a point source of current. The which is the final form of the boundary equation to be satisfied
on each patch.
corresponding function ‫ܩ‬ொ to be used for the point charges
located along the surface of the volume should also be a
III. VALIDATION
multiple of ‫ܩ‬ே . The proportionality factor is actually arbitrary,
since the sources along the surface of the volume are not real This section presents some examples that verify the
and therefore do not have to yield a definite total current. A validity of the results obtained with the method described in
convenient and dimensionally correct proportionality factor is this paper. The calculations were carried out with the
1/ߩ௘߳଴, i.e. MultigroundZ software package [14], which includes an
ଵ
implementation of this method.
‫ܩ‬ொ = ‫ܩ‬ே (3) All examples assume that a current of 1,000 A is injected
ఘ೐ఢబ
into the modeled grounding system.
With this choice of Green’s function, the potential
boundary condition at the surface of the soil volume is A. Validation Using Metallic Volumes
automatically satisfied. Only the boundary condition on the When the resistivity inside a finite volume is very low
normal component of the current density needs to be satisfied compared to the resistivity outside the volume, the volume
numerically. This can be done in much the same way as should behave essentially like a conductor. To verify this
described in [9]. For completeness, the necessary equations are equivalence, results computed for a model including low-
summarily described in the following paragraphs. resistivity finite volumes should be compared to those obtained
Consider an arbitrary patch on the boundary of the finite when using a model where the finite volumes have been
volume. Arbitrarily define the positive direction of the normal replaced by solid conductors. From potential theory, it is
on that patch to be the outward normal to the volume at that actually sufficient to represent the equipotential surface of the
point. Let ‫ܧ‬௏ be the electric field caused by all sources solid conductors in this last model, since in both cases the
(including the charges on all the patches) along the direction entire volume occupied by the conductor would be
of the normal to that patch and just inside the volume and let equipotential. This surface, in turn, can be approximated by a
‫ܧ‬௘ be the corresponding electric field just outside the volume dense metallic cage of thin wires, for which a solution can be
(in the medium of resistivity ߩ௘). The boundary condition to readily obtained.
be imposed for the patch is then:
ாೇ ா೐
= (4)
ఘೇ ఘ೐
Note that the contribution of all sources except that of the

charges on the patch under consideration is continuous at the
surface of the patch. The contribution of the charges on the
patch itself is discontinuous, due to the presence of a 1/‫ݎ‬ଶ
term in the Green’s function for the point source ‫ܩ‬ொ .
In principle, it would be possible to compute the terms
appearing in the above equation numerically. However, it is
possible to compute part of it analytically, namely the field due
to the 1/‫ݎ‬ଶ term. Extracting this term explicitly gives:
ఙ
‫ܧ‬௏ = ‫ܧ‬଴ −
ଶఢబ
ఙ (5)
‫ܧ‬௘ = ‫ܧ‬଴ +
ଶఢబ
where ߪ is the charge density per unit area on the patch and
‫ܧ‬଴ is the continuous electric field due to all other sources
(conductor segments and patches) and to the continuous part
of the field generated by the charge on the patch itself. The
latter (which is due, for example, to the images of the patch
across the soil layer interfaces when the Green’s function is
computed using Image Theory) should be computed using a Fig. 2. Top and side views of the two-layer soil structure with low
resistivity volumes
reduced Green’s function, with the singular part removed:
ଵ This technique makes it possible to validate the behavior
‫ܩ‬෨ொ = ‫ܩ‬ொ − (6)
ସగఢబ௥మ of the approach for low resistivity soil volumes embedded in
multilayer soils by using a conventional multilayer soil model model are shown as dotted curves. As expected, the two
where all soil volumes are replaced by conductor segments approaches yield very similar results.
(metallic cages). In this section, the soil potentials are
B. Comparison with Multilayer Soil Model
computed for a two-layer soil model that includes two volumes
with a very low resistivity of 10-3 Ω-m. The top and side views Another limiting case that can be used for validation
of the soil structure layout along with the observation profiles purposes is that of a multilayer soil where one of the layers is
at which the potential is to be computed are shown in Fig. 2. replaced by a soil volume. When the soil volume has the same
The two soil volumes cover the same zone of 20m x 20m thickness as the layer it replaces and extends sufficiently far
in the XY plane. They are placed in different layers of the beyond the edge of the grid, the soil volume model should
multilayer soil and are separated by a vertical distance of 2.5m. behave in essentially the same way as the infinite layer.
An energized 10m long conductor is embedded in the middle For example, consider a 49-mesh, 100m_x_100m grid
of Volume 1. buried at 0.5m below grade in the three-layer soil structure
For validation purposes, the two volumes are replaced by whose parameters are shown in Table_I. For comparison, the
dense metallic cages made of conductor segments with a 0.5m top layer of this soil model is replaced by a 1000m x 1000m
mesh size. A perspective view of the metallic cage model is soil volume of the same thickness (1.5m). A schematic side
shown in Fig. 3. Note that for Volume 1 the energization is view of such a soil structure is shown in Fig.5.
applied directly on the metallic cage.
TABLE I
MULTILAYER SOIL STRUCTURE
Layer Resistivity (Ω-m) Thickness (m)
Top Layer 1000 1.5
Central Layer 500 2.0
Bottom Layer 70 Infinite
Fig. 5. Side view of the three-layer soil structure with the top soil layer
Fig. 3. Perspective view of metallic cages used to simulate low resistivity formed by the large-extent soil volume
volumes
The computation results for both cases are presented in
Table_II. The results for the ground potential rise (GPR) of the
grid and for the maximum values of the touch voltage and of
the soil potential at the surface of the earth are all in very good
agreement between the two models.
TABLE II
COMPARISON OF RESULTS FOR TWO SOIL STRUCTURES
Three-Layer Two-Layer Soil with
Soil Structure
Soil Soil Volume
Grid GPR (V) 1029.12 1031.93
Max. Soil Potential (V) 1009.12 1012.04
Max. Touch Voltage (V) 750.18 750.92
Fig. 4. Distribution of soil potential for Profile 1 and Profile 2 for IV. PRACTICAL EXAMPLE
heterogeneous volumes (solid curve) and metallic cages (dotted curve) As a practical example of the use of soil volumes in
multilayer soils, consider a 52m_x_52m, 36-mesh grid
The distribution of the soil potential along Observation
installed inside a 2.6m thick backfill soil layer with a resistivity
Profiles 1 and 2 is shown in Fig. 4. Results for the soil volumes
of 80 Ω-m. The system configuration is shown in Fig. 7. The
are shown as solid curves while those for the metallic cage
backfill material is installed above grade and is held in place the three layer soil model is much lower than that obtained for
by retaining walls with a slope of 22.5 with respect to the earth the heterogeneous volume soil model, which is expected since
surface. The soil structure beneath the backfill layer represents the 68.6m_x_68.6m backfill material is represented with an
a horizontal two-layer soil structure. The resistivity of the 5.3m infinite horizontal soil layer in that model.
thick top layer is 250 Ω-m and the resistivity of the bottom soil The step voltages distribution for the two soil models is
layer is 12,000 Ω-m. Since all conductors and soil features shown in Fig.7. For ease of comparisons, the curves show the
should be modeled below ground in the software package used step voltage expressed as a percentage of the system GPR
for the simulation, this scenario was modeled by specifying a (which is different for each soil model). The step voltage
trapezoidal soil volume (representing the elevated backfill pattern for the two soil models are very similar at points
layer) in a layer with a large resistivity (100,000 Ω-m), located above the grid. However, there are significant
representing the air. differences at the edge of the grid and of the backfill area. The
most interesting aspect in this case is the presence of secondary
peaks that occur at the edge of the elevated backfill layer when
the person steps down. Those peaks are only present for the
heterogeneous volume soil model where the observation points
strictly follow the shape of the elevated backfill layer.
Fig. 6. Top and side views of system configuration and soil model with Fig. 7. Step voltages plotted as percentage of system GPR for
finite heterogeneity used to represent the elevated backfill layer heterogeneous volume (red curve) and three-layer (blue curve) soil models
V. CONCLUSIONS
TABLE III
GROUND RESISTANCE FOR THE ELEVATED BACKFILL LAYER SCENARIOS A theoretical model for the analysis of grounding systems
Soil Structure Resistance (Ω) located in multilayer soils in which arbitrary heterogeneities
Three layer soil model with elevated backfill (finite volumes) are embedded was presented. In the presented
layer modeled as a trapezoidal soil volume in a 9.7 approach, the presence of the heterogeneities is handled
high resistivity (air) top soil layer through a Boundary Element Method while that of the
Three layer soil model with elevated backfill
18.5
horizontal soil layers is handled analytically, using Image
layer modeled as the top layer of the soil Theory (or any other equivalent technique). This results in
significant savings of computer resources compared to existing
The closest limiting case solution for this scenario is a techniques that use finite volumes to represent the soil layers.
horizontal three-layer soil where the backfill layer is modeled The model was validated by comparing the computation
as the top layer of the soil structure. The computation results results obtained with this model to relevant limiting case
are compared for these two soil models assuming that a current solutions. A practical example of a grounding system
of 1000 A is discharged by the grid. Table_III shows the grid embedded in an elevated backfill layer was presented and
resistance for the two soil models. The ground resistance for compared to a model where the backfill layer is replaced by a
soil layer of infinite extent. It was shown that safety parameters
could be underestimated when using this approximation.
This model should allow for the analysis of grounding
systems buried in complex soil environments in a manner that
is as realistic as possible, resulting in more accurate and more
economical grounding grid designs.
ACKNOWLEDGMENTS
The authors would like to thank Safe Engineering Services
and technologies ltd. for the financial support and facilities
provided during this research effort.
REFERENCES
[1] T. Takahashi and T. Kawase, "Calculations of earth resistance for a deep-
driven rod in a multilayer earth structure," IEEE Trans. Power Delivery,
vol. 6, pp. 608-614, Apr.1991.
[2] J. Ma, F.P. Dawalibi, and R,D. Southey, “On the equivalence of uniform
and two-layer soils to multilayer soils in the analysis of grounding
systems,” IEE Proceedings Generation, Transmission and Distribution,
vol. 143, no.1, pp.49-55, Jan. 1996.
[3] P.J. Lagace, J.L. Houle, Y. Gervais, and D. Mukhedkar, ”Evaluation of
the voltage distribution around toroidal HVDC ground electrodes in n-
layer soils,” IEEE Trans. Power Delivery, vol. 3, pp. 1573-1579,
October 1988.
[4] J. Liu, F. P. Dawalibi, N. Mitskevitch, M-A Joyal and S. Tee, "Realistic
and Accurate Model for Analyzing Substation Grounding Systems
Buried in Various Backfill Material," in Proc. The 6th IEEE PES Asia-
Pacific Power and Energy Engineering Conference 2014, Hong Kong,
2014.
[5] F. P. Dawalibi, N. Mitskevitch and G. Allard, "Grounding Analysis of
Large Hydroelectric Generating Complex Using Soil Structure
Containing Heterogeneous Volumes," in Proc. The 17th Conference of
the Electric Power Supply Industry (CEPSI), Macau, 2008.
[6] J. Ma, F.P. Dawalibi, and W.K. Daily, “Analysis of Grounding Systems
in Soils with Hemispherical Layering,” IEEE Trans. Power Delivery,
Vol. 8, No. 4, pp. 1773-1781, October 1993.
[7] J. Ma and F.P. Dawalibi, “Analysis of Grounding Systems in Soils with
Cylindrical Soil Volumes,” IEEE/PES Trans. Power Delivery, Vol. 15,
No. 3, pp. 913-918, July 2000.
[8] A. Hajiaboli, S. Fortin, F.P. Dawalibi and P. Zhao, “Analysis of
Grounding Systems in the Vicinity of Hemi-Spheroidal
Heterogeneities”, in this issue.
[9] J. Ma and F. P. Dawalibi, "Analysis of grounding systems in soils with
finite volumes of different resistivities," IEEE Trans. Power Delivery,
vol. 17, No. 2, pp. 596-602, Apr. 2002.H.Hongisto and M. Oksama,
“Constraining of the zero total surface charge in galvanic modeling”,
Geophysical Prospecting, vol. 48, pp.647-658, 1998.
[10] H.Hongisto and M. Oksama, “Constraining of the zero total surface
charge in galvanic modeling”, Geophysical Prospecting, vol. 48, pp.647-
658, 1998.
[11] M.B. Kostic and G.H. Shirkoohi, “Numerical analysis of a class of
foundation grounding systems surrounded by two-layer soil,” IEEE
Trans. Power Delivery, vol.8, pp.1080-1087, July 1993.
[12] R. F. Harrington (1968). Field Computation by Moment Methods. Latest
printing by IEEE Press in 1993, ISBN 0780310144N.
[13] N. Morita, “The boundary-element method,” Analysis methods for
Electromagnetic Wave Problems, E. Yamashita, Ed. Norwood,
MA:Artech, pp.33-78, 1990.
[14] Safe Engineering Services and Technologies web site
(www.sestech.com)
Analysis of Grounding Systems in the Vicinity of Hemi-Spheroidal Heterogeneities
Amir Hajiaboli, Simon Fortin, Farid Paul Dawalibi, Peter Zhao and Adrian Ngoly
Laval, Canada
Abstract -- This paper presents a method of analysis of An example using a hemispheroidal soil model to model an
grounding systems located inside or near a hemi-oblate HVDC sea electrode has been fully explored in [7].
spheroidal soil heterogeneity. This type of soil is particularly The mathematical background necessary for the calculation
useful for modeling grounding grids close to certain types of finite
of the potential due to current discharged in a soil with a hemi-
inhomogeneities, such as lakes or some types of backfill material.
The developed analytical framework is based on the moment spheroidal geometry is discussed first. Results obtained using
method and involves solving Poisson’s equation in an oblate the hemi-spheroidal soil model are compared with those
spheroidal coordinate system. Computation results obtained obtained using other published methods (e.g. hemi-spherical
using this modeling approach for several electrodes and hemi- soils, finite volume and finite element models) to evaluate the
spheroidal geometries are compared with those obtained using accuracy and precision of the proposed approach. The studies
other numerical and analytical techniques. In all cases a good
and comparisons carried out in this work are based on the
agreement has been achieved.
CDEGS suite of Safe Engineering Services and Technologies
Index Terms—Grounding Systems, Green’s Function, Oblate Software packages [8].
Spheroidal Coordinate, Poisson’s Equation, Associated
Legendre’s Functions, Moment Method. II. DESCRIPTION OF THE PROBLEM
The problem addressed in this paper is the calculation of the
I. INTRODUCTION performance of a grounding grid buried in the vicinity of a
It is well known that the properties of the soil can have a medium consisting of a finite volume of soil in the shape of a
significant effect on the electrical performance of grounding hemi-spheroid having a uniform resistivity 𝜌2 embedded into
systems. The more accurately the properties of the soil are a uniform medium with a resistivity 𝜌1 . As shown in Fig. 1,
represented in computational work, the more accurate and the flat side of the hemi-spheroid coincides with the air-soil
realistic the computed results are. Motivated by this interface. It is characterized by two parameters: the radius R of
requirement, several research projects have been carried out to the circular trace of the hemi-spheroid on the air-soil interface
improve modelling techniques in the vicinity of different soil and the depth D of the lowest point of the hemi-spheroid in the
structures. A part of this effort is related to advanced numerical ground.
techniques (e.g. Finite Volumes or Boundary Element
Methods [1], [2]) which can model arbitrary soil
heterogeneities. However, analytical or semi-analytical
techniques have also been utilized to accommodate different
soil geometries including horizontally and vertically layered
soils as well as cylindrical, hemi-spherical, and inclined soil
structures [3]-[6]. Even though the geometrical shape of the
boundaries in these analytical models is limited, they are the
preferred methods in many applications due to their accuracy
and ease of use. Furthermore, the analytical models can be
used for validation purposes to establish the modeling
framework used for more general numerical techniques.
In this paper, the mathematical formulation for modeling
grounding systems in the vicinity of hemi-spheroidal
heterogeneities is presented. This type of soil model can be
useful when modeling grounding grid structures located in the
vicinity of certain types of finite inhomogeneities whose shape
approximates the geometrical shape of a hemi-spheroid, such
as lakes or some types of backfill materials. Theoretically, the
range of resistivities can vary from very low to very high
values, which makes the hemi-spheroid an ideal option for Fig. 1. Top and side view of a hemi-spheroid with a grid located inside the
hemi-spheroid.
modeling extremely low resistivity water in the sea or a lake.
The analysis is restricted to the case of an oblate hemi- a hemi-spheroidal soil is verified to have the correct symmetry.
spheroid, for which the condition 𝐷 < 𝑅 is satisfied. Limiting A vertical electrode is located inside a hemi-spheroid with
cases of this type of geometry include a thin circular plate radius of 20 m and depth of 10 m. The resistivity inside the
located at the surface of the earth (when 𝐷 → 0) or a hemi- hemi-spheroid is 1000 Ωm. while that of the native soil is 100
sphere (when 𝐷 → 𝑅). Grounding system conductors can be Ωm. The hemi-spheroid is centered at the origin of the
located anywhere in the vicinity of the hemi-spheroid, i.e. coordinate system, and the electrode lies along the z-axis, from
inside, outside, or on the interface of the hemi-spheroid. 5 m to 8 m in the ground.
Fig. 2 shows the potential at the surface of the earth due to
III. DESCRIPTION OF THE ANALYSIS current leaking out of that electrode. The figure shows that the
The analysis is based on the Method of Moments (MoM) computed potential exhibits the expected symmetry.
[9] in oblate spheroidal coordinates. The application of the
Method of Moments for this particular geometry requires the
development of the Green's function for a point source in hemi-
spheroidal coordinates. The development of the Green’s
function can be simplified by considering the equivalent
problem of two point sources located symmetrically with
respect to the air-soil interface along with a full spheroid
instead of a hemi-spheroid. This setup automatically satisfies
the interface boundary conditions for the continuity of the
potential and of the normal component of the current density
at the air-soil interface. Appendices I and II discuss the
procedure involved in deriving the analytical solution of the
Green’s function for a point source in oblate spheroidal
coordinates and give some details about the numerical
computation of this function. Fig. 2. Scalar potential in dB at the surface of the earth due to a vertical
electrode. 0 dB corresponds to 1 V
To calculate the current distribution in the grounding grid,
the grid conductors (assumed to be of cylindrical shape) are B. Comparison with a Uniform Soil Model
subdivided into smaller segments so that each segment resides In this section, a hemi-spheroidal soil model where the
in only one soil region. The leakage current over each segment hemi-spheroidal volume and the native soil have the same
is assumed to be uniform. The potential due to current leaking resistivity is compared against a uniform soil model. Fig. 3
out of a segment is obtained by integrating the Green’s shows the grid inside the hemi-spheroidal soil. The uniform
function for a point source over the line segment. The total soil resistivity inside and outside the hemi-spheroid is 500 Ωm.
potential at any point is then obtained as a linear superposition The hemi-spheroidal soil volume has a depth of 10 m and a
of the contributions of the individual segments. Finally, by radius of 30 m. The grid dimensions are 20 m×20 m and it is
imposing the equipotentiality condition for all conductor located at a depth of 0.5 m.
segments using the Method of Moments, the current leaking The computation method used for the uniform soil approach
from the segments is determined. More details about the is also based on the moment method. The Green’s function
Method of Moment can be found in [9]. used in this case is that for a point source in an unlimited
After finding the current distribution in the network, the 𝜌2 𝐼
uniform soil (V = ⃗−𝑟⃗ |
).
potential at any point in space can be calculated by adding the 4𝜋|𝑟 𝑠
potential contribution from each conductor segment. This Fig. 4 shows the scalar potential at the surface of the earth
potential can then be used to find various parameters related to for these two models. As expected, the two results agree very
safety around grounding grids, such as the touch voltage (the well with each other.
potential difference between earth and the grounding grid) and
the step voltage (the potential difference between two points
on the earth surface) [10].
IV. NUMERICAL RESULTS

This section presents some numerical results obtained with
the presented approach, and compares them to results obtained
using other techniques, whenever possible.
A. Symmetry of Potential
As a first check of the approach, the potential calculated in
Fig. 3. The Grid inside the spheroidal soil.
Fig. 4. Potential at the earth surface for hemi-spheroidal and uniform (a) Potential at earth surface for hemi-spheroidal and hemi-spherical soil
soil models. models; grid inside the hemi-spheroid.
C. Comparison with an Hemi-spherical Soil Model

In this section, the behavior of the presented approach for
hemi-spheroids approaching the shape of a hemi-sphere is
investigated. A 20 m×20 m grid is located at a depth of 0.5 m
in the vicinity of a hemi-spheroid with a radius of 30 m and
depth of 29.9 m. The resistivity inside the hemi-spheroid is
1000 Ωm while that outside the hemi-spheroid is 100 Ωm.
Three different locations for the grid are considered: (a) totally
inside the hemi-spheroid, (b) crossing the hemi-spheroid or (c)
completely outside the hemi-spheroid. These three scenarios
are shown in Fig. 5.
Fig. 6 shows the scalar potential obtained along x-
coordinate at the surface of the earth and compares them with
those obtained in a hemi-spherical soil with a radius of 30 m,
(b) Potential at earth surface for hemi-spheroidal and hemi-spherical soil
using the technique described in [4]. As expected, the results models; grid crossing the hemi-spheroid.
are essentially identical.
(a) Inside (b) Crossing
(c) Potential at earth surface for hemi-spheroidal and hemi-spherical soil

models; grid outside the hemi-spheroid.
Fig. 6. Comparison between hemi-spherical and hemi-spheroidal soil
models.
(c) Outside
D. Comparison with a Finite Volume Soil Model
Fig. 5. Location of grid with respect to hemi-spheroid. In this section, calculation results obtained using the
approach described in this paper are compared to results
obtained using a Boundary Element Method known as the
Finite Volume model [1]. This method allows for modeling soil models.
any number of soil volumes of trapezoidal shape, each with its E. Comparison with Finite Element Model
own resistivity. With this model, an approximate hemi-
In this section, the results obtained using the hemi-
spheroidal soil model can be obtained using four trapezoidal
spheroidal soil model proposed in this paper are compared
soil volumes, as shown in Fig. 7. Although the cross-sectional with those obtained using a rotationally symmetric
view in the XY plane at z = 0 is a square with a side length of implementation of Finite Element Method (FEM). The hemi-
40 m, this should represent a fair approximation of a hemi- spheroid has a radius of 20 m and depth of 5 m. The resistivity
spheroidal volume with a depth of 10 m and radius of 20 m. inside the hemi-spheroid is 1000 Ωm while that outside the
The resistivity of the soil inside and outside the hemi-spheroid hemi-spheroid is 100 Ωm. The conductor network is a ring of
is 1000 Ωm and 100 Ωm respectively. The grid dimensions are radius 2 m which is located at a depth of 0.5 m and centered in
20 m×20 m, and it is located at a depth of 0.5 m. the middle of the hemi-spheroid. The wire radius of the ring is
The results are shown in Fig. 8, and differ by 2 to 3 % which 0.01 m and it is kept at a potential of 100 V.
is a fair agreement. The finite element model is obtained by triangular
discretization of the trace of the system in a 𝜌 − 𝑍 plane of a
cylindrical coordinate system centered on the hemi-spheroid.
To minimize the number of triangular elements required for
the discretization, the computational domain is restricted to a
circle of 100 m in this plane. The air-soil interface and the axis
of symmetry are implemented using a Neumann’s boundary
condition while Dirichlet’s boundary condition is used at the
(a) Side view boundary of the computational domain. Fig. 9 shows the finite
element mesh together with the boundary condition considered
for setting up the finite element method. To replicate the
Dirichlet’s boundary condition at the computational domain
boundary, a dense spherical conductor mesh with radius of
100 m has been added in the hemi-spheroidal model. This
external conductor grid is kept at 0 V.
(b) Top view
Fig. 7. Finite volume resembling a spheroid.
Fig. 9. The finite element mesh, the red color shows the Dirichlet’s boundary
condition at the boundary of the computational domain (V = 0 V) and at the
surface of the ring (V = 100 V), the blue color shows the Neumann’s
boundary condition at the air-soil interface and along the axis of symmetry.
Fig. 8. Potential at earth surface for hemi-spheroidal and finite volume

Fig. 11. Scalar potential for a metallic plate calculated based on hemi-
spheroidal soil model.
G. Comparisons against Known Analytical Results

In this section, the resistance of a low resistivity hemi-
Fig. 10. Potential at the earth surface for hemi-spheroidal and finite element spheroidal soil calculated based on the model described in this
soil models. 0 dB corresponds to 1 V. paper is compared against a value obtained analytically. The
resistivity inside the soil volume is set to a low value (0.001
Fig. 10 shows a comparison between the potential at earth
Ωm) in order to maintain a quasi-equipotential surface on the
level computed using the FEM and the hemi-spheroidal
models. The results for the FEM are obtained using a very fine boundary of the hemi-spheroid. The resistivity outside the
mesh with a maximum element edge of 0.1 m. The two models volume (native soil) is 100 Ωm. Using the expression given in
agree very well all along the observation profile. [12] for the potential due to a charged ellipsoid, the resistance
𝑅𝑇 to remote ground of an equipotential hemi-spheroid can be
F. Comparison with a Metallic Plate derived as:
In the limit where the depth of the hemi-spheroid soil is very 𝜌 𝜋 𝐷
𝑅𝑇 = ( − tan−1 ) (2)
small compared to its radius and the resistivity inside the hemi- 2𝜋√𝑅 2 −𝐷2 2 √𝑅 2 −𝐷2
spheroid is much smaller than that outside of the hemi-
spheroid, a hemi-spheroidal soil should behave in much the Here, 𝑅 is the radius of the hemi-spheroid, 𝐷 is its depth,
same way as a thin metallic plate located at the surface of a and ⍴ is the resistivity of the soil outside the low resistivity
uniform earth. hemi-spheroid. The computed grid resistances based on the
To verify that those results can be reproduced, a metallic model described in this paper for a low resistive spheroid for
plate is approximated by a hemi-spheroidal soil with a radius various depths and radii are given in Table I.
of 500 m and a depth of 2 cm. The resistivity inside the hemi-
spheroid is 0.001 Ωm and that outside the hemi-spheroid is TABLE I: CALCULATED RESISTANCE VERSUS THEORETICAL VALUE.
1000 Ωm. A current of 1 Amp is injected in a horizontal Computed Theoretical
conductor located inside the hemi-spheroid. A 6 m long R (m) D (m) Resistance Resistance
conductor buried at a depth of 1 cm, at a symmetrical position (Ω) (Ω)
with respect to the center of the hemi-spheroid is used for this 10 0.1 2.487 2.484
purpose. Since the resistivity of the spheroid is very small, it is 15 10 1.197 1.197
expected to be essentially an equipotential volume and it is not 70 69.5 0.2281 0.2279
expected that the actual position and characteristics of this
energization conductor will have a strong impact on the results. V. CONCLUSIONS
The potential at the surface of the earth above and around This paper presents a technique for analyzing grounding
the hemi-spheroid is shown in Fig. 11. The potential in the flat structures located close to a hemi-oblate spheroidal soil.
area is 0.5 V. This agrees very well with the computed value Detailed formulas for the Green’s function due to a point
of the voltage on a plate given by [11]: current source located in the vicinity of such soil were given.
𝜌
Results obtained using this approach were presented, and
𝑉=𝐼 (1) compared with those obtained using other numerical and
4𝑎
where 𝜌 is the resistivity of the earth and 𝑎 is the radius of analytical techniques such as hemi-spherical soil model, finite
the plate. volume and finite element methods. A good agreement
between the results obtained using this technique and the
results obtained using other numerical techniques has been
achieved in all the benchmark cases.
This type of soil model should prove useful, for instance,
when modeling grounding grid structures in the vicinity of a
lake or the sea, or to represent certain backfill materials.
APPENDIX I. SOLUTION OF POISSON’S EQUATION IN OBLATE The boundary conditions on the surface of the spheroid are
SPHEROIDAL COORDINATES the continuity of the potential and the continuity of the normal
This section presents a derivation of the Green’s function component of the current density on the surface of the
solution for Poisson’s equation in oblate spheroidal spheroid:
coordinates [12].
𝑉1 = 𝑉2
A. Oblate Spheroidal Coordinates 1 𝜕𝑉1 1 𝜕𝑉2 | (7)
=
𝜌1 𝜕𝜁 𝜌2 𝜕𝜁 𝜁=𝜁
A spheroidal coordinate system can be obtained by rotating 0
a plane containing confocal ellipses and hyperbolas around the Using separation of variables and the addition theorem to
1
minor axis of the ellipses. Fig. 12 shows this coordinate system expand |𝑟⃗−𝑟⃗ |, the potential inside and outside the spheroid can
𝑠
together with the variables used for defining the parameters of be written as shown in (8) and (9):
the system. Assuming that the minor axis of the ellipses lies in
the z direction of a Cartesian coordinate system, the ∞ 𝑛
relationship of this coordinate system to the Cartesian 𝑉1 = 𝜌1 𝐼 ∑ ∑ (𝐴 𝑃𝑛𝑚 (𝜂) + 𝐵 𝑄𝑛𝑚 (𝜂) )(𝐴′ 𝑃𝑛𝑚 (𝑗𝜁)
coordinate system is described by equations (3) and (4): 𝑛=0 𝑚=0
+ 𝐵′𝑄𝑛𝑚 (𝑗𝜁)) (C′mn cos(𝑚φ)
+ D′mn sin(𝑚φ) ) (8)
∞ 𝑛
𝑉2 = 𝜌2 𝐼 ∑ ∑ (𝐴𝑚𝑛 𝑃𝑛𝑚 (𝜂) + 𝐵𝑚𝑛 𝑄𝑛𝑚 (𝜂))(𝐴′ 𝑚𝑛 𝑃𝑛𝑚 (𝑗𝜁)

𝑛=0 𝑚=0
+ 𝐵′ 𝑚𝑛 𝑄𝑛𝑚 (𝑗𝜁)) (𝐶𝑚𝑛 cos(𝑚φ)
+ 𝐷𝑚𝑛 sin(𝑚φ) ) +
∞ 𝑛
𝜌2 𝐼 ∑ ∑ 𝑀𝑚𝑛 𝑃𝑛𝑚 (𝑗𝜁)𝑃𝑛𝑚 (𝜂) cos(𝑚(𝜑 − 𝜑0 )) (𝜁 < 𝜁𝑠 )

𝑛=0 𝑚=0
∞ 𝑛 (9)
Fig. 12. Oblate coordinate system and relevant parameters for defining the 𝜌2 𝐼 ∑ ∑ 𝑁𝑚𝑛 𝑄𝑛𝑚 (𝑗𝜁)𝑃𝑛𝑚 (𝜂) cos(𝑚(𝜑 − 𝜑0 )) (𝜁 > 𝜁𝑠 )
system. { 𝑛=0 𝑚=0
𝑥 2 +𝑦 2 𝑧2 where, 𝑃𝑛𝑚 (𝜂) and 𝑄𝑛𝑚 (𝜂) are the associated Legendre’s
+ = 1/4 (3)
𝑐 2 (𝜁 2 +1) 𝑐2𝜁2 functions of the first and second kind with real argument and
𝑐
𝑥 =
2
√(1 + 𝜁 2 )(1 − 𝜂 2 ) cos𝜑 𝑃𝑛𝑚 (𝑗𝜁) and 𝑄𝑛𝑚 (𝑗𝜁) are the associated Legendre’s function
𝑐 of first and second kinds with imaginary arguments. Here, 𝜁𝑠
𝑦 = √(1 + 𝜁 2 )(1 − 𝜂 2 ) sin𝜑 (4)
2 is the value taken by the 𝜁 coordinate at the location of the
𝑐
𝑧 = 𝜂𝜁 source 𝑟⃗𝑠 . 𝑀𝑚𝑛 and 𝑁𝑚𝑛 are the coefficients calculated
{ 2 1
where 𝑐 = 2√𝑎1 2 − 𝑎2 2 is the focal distance, 𝜁 ∈ [0, ∞), based on the expansion of |𝑟⃗−𝑟⃗ | using the addition theorem
𝑠
𝜂 ∈ [−1,1] and 𝜑 ∈ [0, 2𝜋]. and are given below [12].
B. The Green’s Function 2𝑛+1 (𝑛−𝑚)! 2
0 )(−1)𝑚
𝑀𝑚𝑛 = 𝑗(2 − 𝛿𝑚 [ ] 𝑄𝑛𝑚 (𝑗𝜁𝑠 )𝑃𝑛𝑚 (𝜂𝑠 )
The equations solving Poisson's equation for a point source {
2𝜋𝑐 (𝑛+𝑚)!
(10)
2
of current inside a spheroid with a surface 𝜁 = 𝜁0 = 𝑐𝑠𝑡 can 0 )(−1)𝑚 2𝑛+1 [(𝑛−𝑚)!] 𝑃 𝑚 (𝑗𝜁 )𝑃 𝑚 (𝜂 )
𝑁𝑚𝑛 = 𝑗(2 − 𝛿𝑚 (𝑛+𝑚)! 2𝜋𝑐𝑛 𝑠 𝑛 𝑠
be developed as follows:
0 0
Where 𝛿𝑚 = 1 for m=0 and 𝛿𝑚 = 0 for m≠0. Applying
∇2 𝑉 = 𝜌2 𝐼𝛿(𝑟⃗ − 𝑟⃗𝑠 ) (𝑟 < 𝜁0 )
{ 2 2 (5) the boundary conditions (7) on (8) and (9), the unknown
∇ 𝑉1 = 0 (𝑟 > 𝜁0 ) coefficients can be obtained as follows:
where 𝑟⃗𝑠 and 𝑟⃗ are the source and observation point vectors 𝑚
(𝜌2 − 𝜌1 )𝑄𝑛𝑚 (𝑗𝜁0 )𝑄′ 𝑛 (𝑗𝜁0 )𝑁𝑚𝑛 cos(𝑚𝜑𝑠 )
from the origin, 𝑉2 is the potential inside the spheroid and 𝑉1 𝐶𝑚𝑛 = 𝑚 𝑚 (11. 𝑎)
𝜌1 𝑄𝑛𝑚 (𝑗𝜁0 )𝑃′ 𝑛 (𝑗𝜁0 ) − 𝜌2 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′𝑛 (𝑗𝜁0 )
is the potential outside the spheroid.
Assuming that V2 can be separated into two terms, one 𝑚 𝑚
[𝑃 ′ 𝑛 (𝑗𝜁0 )𝑄𝑛𝑚 (𝑗𝜁0 )˗𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′𝑛 (𝑗𝜁0 )] 𝜌2 𝑁𝑚𝑛 cos(𝑚𝜑𝑠 )
specialized solution (𝑉2′ ) due to the point current source and 𝐶 ′ 𝑚𝑛 = 𝑚 𝑚 (11. 𝑏)
one general solution ( 𝑉2′′ ) due to the boundary harmonics 𝜌1 𝑄𝑛𝑚 (𝑗𝜁0 )𝑃′ 𝑛 (𝑗𝜁0 ) − 𝜌2 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′𝑛 (𝑗𝜁0 )
contribution, leads to the following simplified formulation 𝑚
(𝜌2 − 𝜌1 )𝑄𝑛𝑚 (𝑗𝜁0 )𝑄′𝑛 (𝑗𝜁0 )𝑁𝑚𝑛 sin(𝑚𝜑𝑠 )
below for 𝑉2 : 𝐷𝑚𝑛 = 𝑚 𝑚 (11. 𝑐)
𝜌2 𝐼 𝜌1 𝑄𝑛𝑚 (𝑗𝜁0 )𝑃′ 𝑛 (𝑗𝜁0 ) − 𝜌2 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′𝑛 (𝑗𝜁0 )
𝑉2 = 𝑉2′ + 𝑉2′′ = ⃗−𝑟⃗ |
+ 𝑉2′′ (6)
4𝜋|𝑟 𝑠
REFERENCES
𝑚
[𝑃𝑛′𝑚 (𝑗𝜁0 ) 𝑄𝑛𝑚 (𝑗𝜁0 ) − 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′𝑛 (𝑗𝜁0 )]𝜌2 𝑁𝑚𝑛 sin(𝑚𝜑𝑠 ) [1] J. Ma and F. P. Dawalibi, "Analysis of Grounding Systems in Soils with
𝐷′ 𝑚𝑛 = 𝑚 𝑚 (11. 𝑑)
𝜌1 𝑄𝑛𝑚 (𝑗𝜁0 )𝑃′𝑛 (𝑗𝜁0 ) − 𝜌2 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′ 𝑛 (𝑗𝜁0 ) Finite Volumes of Different Resistivities," IEEE Transactions on Power
Delivery, vol. 17, no. 2, pp. 596-602, Apr. 2002.
where 𝜑𝑠 is the value taken by the 𝜑 coordinate at the [2] I. Colominas, F. Navarrina, and M. Casteleiro, “Analysis of transferred
location of the source 𝑟⃗𝑠 . earth potentials in grounding systems: A BEM numerical approach,”
Similarly, the coefficients for the case where the point IEEE Transaction on Power Deliver., vol. 20, no. 1, pp. 339–345, Jan.
source is located outside the spheroid are: 2005.
[3] F. P. Dawalibi, J. Ma, and R. D. Southey, “Behaviour of grounding
𝑚
systems in multilayer soils: a parametric analysis,” IEEE Transactions
(𝜌1 − 𝜌2 )𝑃𝑛𝑚 (𝑗𝜁0 )𝑃′𝑛 (𝑗𝜁0 )𝑀𝑚𝑛 cos(𝑚𝜑𝑠 ) on Power Delivery, vol. 9, no. 1, pp. 334-342, Jan. 1994.
𝐶𝑚𝑛 = 𝑚 𝑚 (12. a) [4] J. Ma, F. P. Dawalibi, and W. K. Daily, "Analysis of Grounding Systems
𝜌2 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′ 𝑛 (𝑗𝜁0 ) − 𝜌1 𝑃′𝑛 (𝑗𝜁0 )𝑄𝑛𝑚 (𝑗𝜁0 )
in Soils with Hemispherical Layering," IEEE Transactions on Power
𝑚 𝑚
Delivery, vol. 8, no. 4, pp. 1773-1781, Oct. 1993.
[𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′ 𝑛 (𝑗𝜁0 )˗𝑃′𝑛 (𝑗𝜁0 )𝑄𝑛𝑚 (𝑗𝜁0 )]𝜌1 𝑀𝑚𝑛 cos(𝑚𝜑𝑠 ) [5] J. Ma and F.P. Dawalibi, "Analysis of Grounding Systems in Soils with
𝐶 ′ 𝑚𝑛 = 𝑚 𝑚 (12. 𝑏) Cylindrical Soil Volumes," IEEE Transactions on Power Delivery, vol.
𝜌2 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′𝑛 (𝑗𝜁0 ) − 𝜌1 𝑃′ 𝑛 (𝑗𝜁0 )𝑄𝑛𝑚 (𝑗𝜁0 )
15, no. 3, pp. 913-918, July 2000.
𝑚
[6] F. Freschi, M. Mitolo and M. Tartaglia, “An Effective Semianalytical
(𝜌1 − 𝜌2 )𝑃𝑛𝑚 (𝑗𝜁0 )𝑃′𝑛 (𝑗𝜁0 )𝑀𝑚𝑛 sin(𝑚𝜑𝑠 ) Method for Simulating Grounding Grids,” IEEE Transaction On
𝐷𝑚𝑛 = 𝑚 𝑚 (12. 𝑐) Industry Applications, Vol. 49, No. 1, Jan. 2013
𝜌2 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′𝑛 (𝑗𝜁0 ) − 𝜌1 𝑃′ 𝑛 (𝑗𝜁0 )𝑄𝑛𝑚 (𝑗𝜁0 )
[7] A. Hajiaboli, S. Fortin, F. Dawalibi, “Numerical Techniques for
𝑚 𝑚 Modeling HVDC Sea Electrode”, In this issue
[𝑄′ 𝑛 (𝑗𝜁0 )𝑃𝑛𝑚 (𝑗𝜁0 )˗𝑃′ 𝑛 (𝑗𝜁0 )𝑄𝑛𝑚 (𝑗𝜁0 )]𝜌1 𝑀𝑚𝑛 sin(𝑚𝜑𝑠 ) [8] Safe Engineering Services and Technologies web site
𝐷′ 𝑚𝑛 = 𝑚 𝑚 (12. 𝑑)
𝜌2 𝑃𝑛𝑚 (𝑗𝜁0 )𝑄′ 𝑛 (𝑗𝜁0 ) − 𝜌1 𝑃′ 𝑛 (𝑗𝜁0 )𝑄𝑛𝑚 (𝑗𝜁0 ) (www.sestech.com)
[9] R. F. Harrington, Field Computation by Moment Methods, Wiley-IEEE
Numerical computation of (11) and (12) involves calculation Press, April 1993.
of associated Legendre’s functions of first and second kind, [10] F. P. Dawalibi, R. D. Southey, R.S. Baishiki, “Validity of Conventional
which is discussed in more detail in Appendix II. Approaches For Calculating Body Current Resulting From Electric
Shocks,” IEEE Transactions on Power Delivery, vol. 5, no. 2, Apr. 1990.
[11] IEEE Guide for Safety in AC Substation Grounding, IEEE Standard 80-
APPENDIX II. NUMERICAL COMPUTATION OF THE ASSOCIATED 2000, The Institute of Electrical and Electronics Engineers, Inc., January
LEGENDRE’S FUNCTIONS 2000.
[12] W. R. Smythe, Static and Dynamic Electricity, McGraw-Hill, 2nd edition
The numerical computation of associated Legendre’s
1950.
function of first and second kind has been carried out using a [13] S. Chang and J. M. Jin, Computation of Special Functions, Wiley, 1996.
combination of techniques described in [13] and [14]. [14] A. Gil and J. Segura, "A code to evaluate Prolate and Oblate Spheroidal
The technique described in [13] is based on a direct forward Harmonics," Computer Physics Communications, vol. 108, pp. 267-278,
recursive calculation for 𝑃𝑛𝑚 (𝑧). For 𝑄𝑛𝑚 (𝑧), if |𝑧| < 1 a 1998.
direct forward recursion has been used, and if |𝑧| > 1 a
backward recursion has been used. The technique is similarly
applied to both real and imaginary arguments of the associated
Legendre’s functions.
In the technique described in [14], the associated
Legendre’s function with imaginary argument is first
transformed to a real function according to the following
transformation:
𝑗𝜋𝑛
𝑅𝑛𝑚 (𝑋) = exp (− ) 𝑃𝑛𝑚 (𝑗𝑋)
2
{ 𝑗𝜋𝑛
(13)
𝑇𝑛𝑚 (𝑋) = 𝑗 exp ( ) 𝑄𝑛𝑚 (𝑗𝑋)
2
Then, a forward recursion is applied for the computation of

𝑅𝑛𝑚 (𝑋) and a backward recursion is used to compute
𝑇𝑛𝑚 (𝑋)for all real values of X. To evaluate the values of 𝑇𝑛𝑚
𝑚
and 𝑇𝑛−1 needed for starting the backward recursion, the
Wronskian relating the R’s and T’s together with the
𝑚
evaluation of the continued fraction of 𝑇𝑛𝑚 /𝑇𝑛−1 is used [14].
Finally, the inverse of the transformation shown in (13) is
applied to obtain 𝑃𝑛𝑚 (𝑗𝑋) and 𝑄𝑛𝑚 (𝑗𝑋) . This technique is
computationally more efficient, especially for higher order
associated Legendre’s functions with imaginary arguments.
Numerical Techniques for the Analysis of HVDC Sea Electrodes
Amir Hajiaboli, Simon Fortin and Farid Paul Dawalibi

Laval, Canada
Abstract — This paper presents an analysis of HVDC sea installation on the shore line or infinite seabed resistivity [1].
electrodes using different numerical techniques. The large In recent years, in parallel with advances in computer
volume of sea water that must be included in the analysis is architecture, usage of more sophisticated computer aided
represented using infinite (Inclined Layer model) and finite design techniques for analyzing grounding systems has
models (Finite Volume and Hemi-Spheroidal models). The
increased substantially and has led to an unprecedented growth
comparison between the results obtained using these models are
based on the potential rise of the electrode, the sea potential rise in this area [8-11].
in the vicinity of the electrode and on the computational burden Due to the very low resistivity of sea water, leakage current
for utilizing each model. The analysis shows that the results from the HVDC electrode can propagate very far from the
obtained using finite models can predict the performance of the electrode within the sea water before passing through the
electrode more accurately and realistically. seabed. This means that the calculated electrical performance
of the electrode, e.g. the maximum Ground Potential Rise
Index Terms—High Voltage Direct Current (HVDC), Sea (GPR) and electric field observed in a model, can be
Electrode, Finite Volume, Spheroidal Soil, Inclined Soil, Ground substantially affected by the extent of sea water that can be
Potential Rise (GPR), Computational Burden included in the model. This sets a hard compromise between
the achievable accuracy and the available computational
I. INTRODUCTION resources.
Over the past several years, High Voltage Direct Current In this paper, the performance of an HVDC sea electrode
(HVDC) links have been increasingly used to interconnect is analyzed using three different numerical techniques to
distant or non-synchronous systems. Due to their advantages account for the sea water and the seabed. Two of these
over HVAC links, it is predicted that this type of transmission techniques are based on a finite volume of sea water [8, 9] and
lines will be utilized even more often in the coming decades. the other is based on an infinite inclined sea model.
One major advantage of HVDC lines is the possibility of The structure of the paper is as follows. First, the problem
using sea water as the return path (monopolar operation) which considered in this paper is briefly described. The three
reduces the installation cost over HVAC lines substantially, techniques used to model the water and seabed are then
especially for transmission over long distances. However, for presented. These three modeling techniques are then compared
HVDC transmission over land, monopolar operations are using a parametric analysis on the extent of water volume,
typically restricted to short periods of time during maintenance resistivity of the seabed and the distance of the electrode to the
activities on bipolar lines because the continuous injection of shore.
DC current in the ground may result in interference issues such
as corrosion of nearby buried metallic structures such as II. GENERAL PROBLEM
pipelines. Fig. 1 shows a schematic view of the general problem
One major concern is the heating of the environment in the considered in this paper. An electrode is located in the sea, at
vicinity of HVDC electrode, especially in monopolar a distance 𝐷𝑔 from the shoreline. The resistivity of the sea
operations. To address this issue, as suggested by CIGRE1998
water is denoted as 𝜌𝑠𝑒𝑎 and is kept at a low value of 0.2 Ωm
guide for HVDC electrodes, installing the electrode in close
in the analysis. This is a typical value and can vary depending
vicinity to the sea or a lake is preferable, for better heat
on the salinity and on the temperature of the water. The
dissipation. However, the electric field in the vicinity of this
resistivity of the soil on shore and in the seabed is assumed to
type of electrodes can be a serious concern for marine life.
be the same, and is denoted by 𝜌𝑛𝑎𝑡𝑖𝑣𝑒 . The value of seabed
Furthermore, corrosion effects on buried metallic structures
resistivity depends on the type of soil beneath the water. While
near the shore can be significant due to the continuous
the shallow earth crust has a relatively low resistivity, the
injection of current and the low resistivity of the shallow earth.
resistivity of deeper layers of the crust (at depths ranging from
All these issues have to be addressed properly before HVDC
1 km to 30 km), can be significantly higher (~ 14000 Ωm) [7].
links can be effectively installed and exploited [1-4].
To account for this variability of the seabed resistivity, the
Several techniques have been utilized to investigate the
analysis was carried for resistivity values ranging from a few
performance of HVDC sea electrodes [5-7]. These techniques
hundreds Ωm up to several thousands Ωm.
are usually based on simplifying assumptions such as electrode
The electrode consists of a 10 m square grid and is located electrode on the air-water interface and the edge of the hemi-
0.5 m below the sea surface. Although the size of this electrode spheroid.
is about one fifth to one tenth that of a typical sea electrode, it
nonetheless illustrates realistically the performance of a typical
sea electrode and is well suited for comparing the results
between the three models used in this paper.
Fig. 1. The general structure of the problem.
(a) Top View

III. METHODOLOGY
Three different computational techniques are used in this
paper, differing by how the boundary separating the native soil
and the sea water is defined and how its effects are computed:
 Hemi-Spheroidal Model: the sea water is enclosed
in a hemi-spheroidal surface.
 Inclined Layer Model: the sea water occupies a
semi-infinite wedge.
(b) Side View
 Finite Volume Model: the sea water occupies a finite Fig. 2. Hemi-spheroidal representation of the sea.
wedge-like region.
The Green’s function for this case is obtained from a solution
In all cases, the current distribution in the grid conductors of Poisson’s equation in spheroidal coordinates, and involves
is obtained by an application of the moment method [12]. First, Associated Legendre’s functions of real and imaginary
the conductors are subdivided so that each segment reside in arguments [9, 13, 14].
only one region of homogeneous resistivity. A linearly varying
current distribution is assumed over each segment, Inclined Layer Model: In this model, the water and native soil
corresponding to a uniform leakage current distribution. are separated by an inclined surface where 𝜑 = 𝑐𝑠𝑡 in a
Imposing the boundary condition 𝛥𝜙 = 𝑍𝐼 between pairs of cylindrical coordinate system whose symmetry axis lies in the
connected segments using the method of moments, the current air soil interface (Fig. 3).
in all conductor segments is determined. In the above equation,
𝜙 is the potential of the segments (calculated by integrating the
Green’s function for the problem over the length of the
segment), 𝐼 is the current flowing in the segment and 𝑍 is the
total impedance per unit length of the segment.
Hemi-Spheroidal Model: Fig. 2 shows a schematic view

of a system with hemi-spheroidal geometry. The flat side of
the hemi-spheroid coincides with the interface with the air. It
is characterized by two parameters: the radius 𝑅 of the circular Fig. 3. Schematics of Inclined Layer Model for the sea.
trace of the hemi-spheroid on the air interface and the depth 𝐷
of the lowest point of the hemi-spheroid in the ground. In all For this geometry, the distance 𝐷𝑔 is defined as the shortest
the cases considered in our analysis we have assumed that D = distance between the projection of the center of the electrode
10 km. In this geometry, the distance 𝐷𝑔 is defined as the on the air-water interface and the axis of symmetry of the
shortest distance between the projection of the center of the cylindrical coordinate system.
In general, the Green’s function for this case is obtained in modeling techniques considered in this paper. The plots
terms of Bessel functions of imaginary order and arguments. present several curves, corresponding to different values
A limiting case of this model can be obtained using Image assumed for the seabed resistivity.
Theory when the resistivity of the seabed is infinite and the In all cases, the GPR of the sea electrode decreases as the
inclination angle is an exact divisor of 180⁰ [1]. volume of water increases, as could be expected. In the
Inclined Layer Model (Fig. 5a), the electrode GPR is not
Finite Volume Model: Fig. 4 shows a schematic view of strongly affected by the value of the seabed resistivity, as long
the model used to represent the sea using the Finite Volume as the seabed resistivity is sufficiently large compared to the
approach. The volume of sea water is parameterized by its resistivity of the water. However, in the Finite Volume and
length 𝐿, its width 𝑊, and by the largest depth 𝐻 reached by Hemi-Spheroidal models (Fig. 5b and Fig. 5c), the electrode
the water. In all the cases considered in the analysis, the width GPR is strongly dependent on the seabed resistivity and also
and length are assumed to be equal and H was set to 10 km.The on the size of the lateral side of the volume or on the diameter
distance 𝐷𝑔 for this model is defined as the shortest distance of the hemi-spheroid. Also, the variation of the electrode GPR
between the projection of the center of the electrode on the air- as a function of the volume of water is less pronounced for
water interface and the edge of the shallow part of the water lower seabed resistivities for those two models. Furthermore,
volume. for low seabed resistivity and large water volume, the electrode
The technique used to calculate the effect of the volume of GPR calculated using the Hemi-Spheroidal and Finite Volume
water is considerably different for this method than for the models tends towards the GPR value obtained when the sea
other two methods. It is based on discretizing the surface of the extends to infinite depth. This behavior is not seen in the
boundary into small patches, each carrying an unknown Inclined Layer Model for an electrode located at 1 km from the
amount of charge. The charges are then adjusted until the shore.
boundary conditions at the surface of the volumes are satisfied
[5]. This adjustment is carried out by solving the patch
boundary condition equations together with the moment
method equations for the conductor segments, often resulting
in very large matrix systems. Moreover, the boundaries of the
water volume must be subdivided adequately in order to reach
good calculation accuracy.
(a)
Fig. 4. Schematics of Finite Volume Model
IV. PARAMETRIC ANALYSIS FOR DIFFERENT MODELS

In this section, a series of parametric analyses are carried
out to investigate the effect of various parameters on the GPR
of the electrode. These analyses are based on three different
aspects of the model 1) the effect of the extent of water, 2) the
effect of the distance of the electrode to the shore, 3) the seabed
resistivity. In all cases, a current of 1000 A is injected into the
electrode.
(b)
A. Effect of the volume of water
Fig. 5 shows the variation of the GPR of the sea electrode
when increasing the volume of sea water for the different
(c)
Fig. 5: Parameteric analysis of the electrode GPR versus volume of water for Fig. 6. GPR versus seabed resistivity for different models. The electrode is
differerent analysis techniques. (a) Inclined Layer Model; (b) Finite Volume at 0.5 meter below the sea surface and is located 1000 m away from the shore
Model; (c) Hemispheroidal Volume Model. The electrode is 1000 m from the in all the models. In the Inclined Layer model, the inclination angle is 1⁰. In
shore and is at a depth of 0.5 m in all models. the Hemi-Spheroidal model, the diameter is set to 200 km and in the Finite
Volume model the lateral side of the volume is set to 200 km.
B. Effect of the seabed resistivity C. Effect of the electrode distance to the shore
The effect of the resistivity of the seabed on the GPR of the Fig. 7 shows the effect of the distance of the electrode to
electrode can be seen more clearly in Fig. 6, which shows the the shore in the different models. To facilitate comparisons, the
value of the GPR of the electrode for different values of the curves for each model are normalized to the value obtained
seabed resistivity for the three models. For all models, when when the electrode is located at 1 km from the shore in that
the resistivity of the seabed approaches that of the sea water model.
the GPR of the electrode tends to the value (11.85 V) that Once more, the behavior of the Inclined Layer model is
would obtain for an electrode located in an infinitely deep sea. quite different than that of the other two models. The GPR of
The GPR increases from this value as the seabed resistivity the electrode in the Inclined Layer model decreases sharply
increases, but in a very different manner for the different when the electrode distance to shore increases up to about 2
models. km. For distances to shore greater than 2 km, the GPR is
For the Inclined Layer model, the GPR of the electrode is essentially constant. For the models including a finite amount
nearly independent of the seabed resistivity, as long as it is of sea water (Hemi-Spheroidal and Finite Volume) the
sufficiently large. This is due to the fact that the sea water variation in the electrode GPR as a function of distance of the
extends to infinity in this model, meaning that the current electrode to the shore is negligible for suffienctly large values
discharged by the electrode has a path to infinity through a low of the seabed resistivity.
resistivity medium (the sea water) with a cross-sectional area
that is expanding as the distance from the source is increasing.
As a result, the resistance of this “through-water” path is finite,
and much lower than that of the path through the seabed when
the resistivity of the latter is sufficiently large. Therefore,
current flow is essentially confined to the water in this limit,
and the GPR of the electrode is function mainly of the
properties (resistivity and geometry) of the water.
In contrast with the results for the Inclined Layer model,
the results for the GPR of the electrode for the Hemi-
Spheroidal and Finite Volume models show a linear increase
as a function of seabed resistivity, when the latter is
sufficiently large. This is because any current discharged by
the electrode must pass through the seabed in order to reach
infinity (remote earth), causing a potential rise proportional to
the seabed resistivity. Fig. 7. Normalized electrode GPR (to the GPR at 1000 m from the shore) for
different values of electrode distance to shore. The electrode is at 0.5 meter
below the sea surface and the seabed resistivity is 10,000 Ω-m. In the
Inclined Layer model, the seabed inclination angle is set to 1⁰. In the Hemi-
Spheroidal model the diameter is set to 200 km and in the Finite Volume
model the lateral side of the volume is set to 200 km.
V. POTENTIAL AND ELECTRIC FIELD DISTRIBUTION
CLOSE TO THE ELECTRODE
This section shows the potential and electric field
distribution calculated at the surface of the water, in a 40 m by
40 m region centered on the electrode. In all cases, the
electrode is at 0.5 meter below the sea surface and is located
1 km away from the shore.
The resistivity of the seabed is 10,000 Ohm-m. The
inclination angle is 1⁰ for the Inclined Layer model. For the
Hemi-Spheroidal model, the radius of the hemi-spheroid is (a) Inclined Layer model (b) Finite Volume model
200 km and its depth is 10 km. For the Finite Volume model,
the size (length and width) of the volume is 200 km and its
height is 10 km.
Fig. 8 shows the scalar potential distibution in the vicinity
of the electrode as calculated using those different water
volume models. Fig. 9 shows the electric field distribution
calculated at the surface of the water above the grid. Despite
the differences in the computed scalar potential for the
different water volume models used in this paper, the electric
field has a similar magnitude and distribution in all the water (c ) Hemi-Spheroidal model
volume models. This can be explained by the fact that the
electric field, being the gradient of the potential, is less Fig. 9. Magnitude of Ex for Inclined Layer, Finite Volume and Hemi-
Spheroidal models, in dB. The dark square in the middle of the figure
sensitive than the potential to the detailed properties of the represents the HVDC electrode. 0 dB corresponds to 1 V/m.
seabed boundary, especially when the observation points are
far from the boundary.
VI. PERFORMANCE ANALYSIS
Tables I shows the computation performance for the

different models considered in this paper. The simulations
were carried out on a computer with Intel i7-4770 CPU
(3.4GHz clock speed) and memory of 24 GByte. While the
values given in Table I are only valid for this particular
computer architecture, they provide an estimate of the
computation time for the models used in this paper based on
the implementation available in the CDEGS software package
(a) Inclined Layer model (b) Finite Volume model
[15].
For the particular system being studied, the Inclined Layer
model is the fastest both in terms of GPR calculation and for
the calculation of the scalar potential and electric field at
observation points. The Hemi-Spheroidal model, when used
only for GPR computation, is fast relative to the Finite Volume
model but the computation of the scalar potential and electric
field at observation points in the Hemi-Spheroidal model takes
longer than in the Finite Volume model. This is because the
Green’s function which is used for the computation of point
(c) Hemi-Spheroidal model sources of current in the Hemi-Spheroidal model requires the
Fig. 8. Magnitude of the scalar potential in Inclined Layer, Finite Volume calculation of associated Legendre’s functions of high order
and Hemi-Spheroidal models, in dB.The dark square in the middle of the [9], which is a time consuming task. In the Finite Volume
figure represents the HVDC electrode. 0 dB corresponds to 1 V. model, most of the computation time is spent building and
inverting the matrix giving the charge distribution at the
surface of the patches defining the water volume, since the
number of patches needed to represent this volume accurately
is very large. Once this charge distribution is obtained, the
computation of the GPR and of the scalar potential and electric
field at observation points is rather fast.
Table I: Computation Time for Different Models 6. E. Uhlmann, Power transmission by direct current, Chapter 15,
Computation Computation Time for Springer-Verlag, 1975
Model Time for only GPR and 1681 7. E. W. Kimbark, Direct Current Transmission, Chapter 9, Wiley
GPR (s) Observation Points (s) Interscience, 1971.
Inclined Layer 3.34 10.30 8. J. Ma and F. P. Dawalibi, "Analysis of Grounding Systems in Soils with
Hemi-Spheroidal 73 1456 Finite Volumes of Different Resistivities," IEEE Transactions on
PWRD, Vol. 17, No. 2, pp. 596-602, April 2002.
Finite Volume 1517 1539 9. A. Hajiaboli, S. Fortin, and F. P. Dawalibi, P. Zhao and A. Ngoly
“Analysis of Grounding Systems in the Vicinity of Hemi-Spheroidal
VII. CONCLUSION Heterogeneities,” in this issue.
10. F. Freschi, M. Mitolo and M. Tartaglia, “An Effective Semianalytical
In this paper, three numerical techniques for modeling Method for Simulating Grounding Grids,” IEEE Transaction On
HVDC sea electrodes were presented and compared. Two of Industry Applications, Vol. 49, No. 1, Jan. 2013
the models (the Finite Volume and Hemi-Spheroidal models) 11. H. Rosenberg, “Electric power, HVDC from land to off-shore structures
consider a finite extent of water volume while one model (the utilizing sea-electrodes for return current, concerns and precautions with
regard to environment and corrosion,” in International Gas Union
Inclined Layer model) considers an infinite extent of sea water. Research Conference 2014, September 17-19, Copenhagen, Denmark.
A parametric analysis on the seabed resistivity, water volume 12. R. F. Harrington, Field Computation by Moment Methods, Wiley-IEEE
extent and also on the installation distance of the electrode to Press April 1993.
the shore was carried out for each model. 13. W. R. Smythe, Static and Dynamic Electricity, McGraw-Hill, 2nd
The results reveal a basic difference between models edition, 1950.
14. Shanjie Zhang, Jian-Ming Jin, Computation of Special Functions, John
assuming a finite amount of water and a model assuming an Wiley, 1996.
infinite amount of water. Essentially, in the case where the 15. Safe Engineering Services and Technologies web site
resistivity of the water is much smaller than that of the seabed, (www.sestech.com)
the behavior of the electrode is nearly independent of the
resistivity of the seabed for the model with an infinite amount
of water while it depends strongly on the resistivity of the
seabed when a large but finite amount of water is used in the
model. Moreover, the location of the electrode with respect to
the shore can have a strong impact on the behavior of the
electrode for a model assuming an infinite amount of water but
has a very weak effect when a finite amount of water is used
in the other models.
The main conclusion is that models including an infinite
amount of sea water (such as the Inclined Layer model
considered in this paper, or models based on image theory)
should be avoided when modeling sea electrodes. Instead,
models including a large but finite amount of water (such as
the Hemi-Spheroidal model and the Finite Volume model
considered in this paper) should be used to represent the sea
volume more realistically and to achieve better accuracy in the
computed results.
REFERENCES
1. P. Girdinio, P. Molfino, M. Nervi, M. Rossi, A. Bertani and S.
Malgarotti, “Technical and compatibility issues in the design of HVDC
sea electrodes ,” in IEEE Symposium on Electromagnetic Compatibility
(EMC EUROPE), Rome, Sept 17-21, 2012.
2. J. Ma, F. P. Dawalibi and W. Ruan, "Design Considerations of HVDC
Grounding Electrodes," in IEEE/PES T&D 2005 Asia Pacific, Dalian,
China, August 14 - 18, 2005.
3. W. Ruan, J. Liu, J. Ma, F. P. Dawalibi, and R. D. Southey, "Performance
of HVDC Ground Electrode in Various Soil Structures," in Proceedings
of the IEEE-PES/CSEE International Conference on Power System
Technology, PowerCon 2002, Kunming, China, October 13-17, 2002,
Vol. 2, pp. 962-968.
4. J. Liu, F. P. Dawalibi, J. Ma, and R. D. Southey, "HVDC Advanced
Analysis Methods for Grounding Design and DC Interference Mitigation
Techniques," EMC 2002, Beijing, May 21-24, 2002.
5. S. Rusck, “H.V.D.C. Power Transmission: Problems Relating to Earth
Return”, Direct Current, pp. 290-300, Nov. 1962.

Conference Proceedings 2016

Uploaded by

Document Information

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

Conference Proceedings 2016

Uploaded by

Copyright:

Available Formats

June 21 to 24, 2016

Copyright © 2016 SES ltd. All rights reserved. Page i

Farid Paul Dawalibi

Group photo of the first Users’ Group Meeting in 1991.

Page ii Copyright © 2016 SES ltd. All rights reserved.

26th Annual CDEGS Users’ Group Conference

Boulder, Colorado, USA

Dear CDEGS Users:

Copyright © 2016 SES ltd. All rights reserved. Page iii

1 INFLUENCE OF GROUND CONSOLIDATION ON LOCAL

2.1 Project Requirements

3 Soil Characteristics and Ground Consolidation

3.1 Natural Soil Conditions

Layer Resistivity (Ωm) Thickness (m)

3.2 Engineered Soil Characteristics

Page 1-2 Copyright © 2016 SES ltd. All rights reserved.

Figure 1: Foundation Piles Model vs Rods Model.

Figure 2: Consolidated Ground and Soil Volume.

Table 2: Achieved Earth Resistance Comparison

Model Earth Resistance (Ω)

With 13.5 m Piles 0.368

With 8 m Rods 0.384

Copyright © 2016 SES ltd. All rights reserved. Page 1-3

Table 3: Equivalent Soil Model

Layer Resistivity (Ωm) Thickness (m)

4 Influence on the Earth Impedance and Electrode System

4.1 Overall System Earth Impedance

4.2 Substation Earth Mat

Page 1-4 Copyright © 2016 SES ltd. All rights reserved.

Figure 3: Substation Earth Mat.

4.3 Overall 400kV Earthing System

Figure 4: Overall Earthing System.

4.4 Overall Earth Electrode

Figure 5: Overall Electrode and DNO Network.

4.5 Summary of Earth Impedance Results

Model Earth Impedance (Ω)

Substation Earth Mat in Isolation (Fig.3) | 0.384 |

Overall Earthing System (Fig.4) | 0.147 |

Overall Electrode and DNO Network (Fig.5) | 0.143 |

Value Required to Ensure Safe Post-Construction Earth

Page 1-6 Copyright © 2016 SES ltd. All rights reserved.

5 Effect of the Man-Made Ground on the Surface Potentials

5.1 Investigations into the Earth Mat in Isolation

21.95 % (508 V) 46.31 % (1072 V) The effect of the piles is

30.86 % (714 V) 44.28 % (1025 V) Smaller contours due to

28.34 % (656 V) 44.36 % (1027 V) The effect of the piles is

MINIMUM VALUE : 40.33281

Touch Voltage (% Ref. GPR) [Wors] Potential Profile (% reference GPR)

Figure 6: Touch Voltages. Figure 7: Surface Potentials.

Copyright © 2016 SES ltd. All rights reserved. Page 1-7

Natural Soil with Piles

Touch Voltage (% Ref. GPR) [Wors] Potential Profile (% reference GPR)

Figure 8: Touch Voltages. Figure 9: Surface Potentials.

190 55.55 200 3

Touch Voltage (% Ref. GPR) [Wors] Potential Profile (% reference GPR)

Figure 10: Touch Voltages. Figure 11: Surface Potentials.

Backfill Material and Piles

190 42.50 200 3

Figure 12: Touch Voltages. Figure 13: Surface Potentials.

Page 1-8 Copyright © 2016 SES ltd. All rights reserved.

Ground Conditions Touch Voltages Comparison Transferred Voltage Comments

12.86 % (298 V) 40.22 % (931 V)

8.46 % (196V) 40.72 % (943 V)

12.85 % (297) 40.38 % (935 V)