You are on page 1of 7

See discussions, stats, and author profiles for this publication at: https://www.researchgate.

net/publication/344179383

Online distribution service transformer health assessment using real-time grid


energy monitor

Conference Paper · August 2020

CITATIONS READS

0 4

5 authors, including:

Quynh Tran Doan Van Binh


Università degli Studi di Palermo Vietnam Academy of Science and Technology - Institute of energy science, Hanoi, Vie…
18 PUBLICATIONS   57 CITATIONS    11 PUBLICATIONS   8 CITATIONS   

SEE PROFILE SEE PROFILE

Ninh Nguyen Quang


Vietnam Academy of Science and Technology
24 PUBLICATIONS   103 CITATIONS   

SEE PROFILE

Some of the authors of this publication are also working on these related projects:

Greening the power systems with solar power for GreenHouse Gas emission reduction in Vietnam: Italy - VIETNAM - research cooperation 2017 – 2019 View project

“Design and installation of a grid connected microgrid 100 kWp PV plant and study solutions for developing solar generation in Vietnam to 2030, taking into account
greenhouse gas emission reduction View project

All content following this page was uploaded by Quynh Tran on 09 September 2020.

The user has requested enhancement of the downloaded file.


Online distribution service transformer health
assessment using real-time grid energy monitor
Quynh Thi Tu Tran1,2 Kevin Davies Leon Roose Binh V Doan, Ninh Q Nguyen
1Hawaii 2Institute of Energy Science
Natural Energy Institute Hawaii Natural Energy Institute Hawaii Natural Energy Institute
University of Hawaii at Manoa University of Hawaii at Manoa University of Hawaii at Manoa Vietnam Academy of Science and
Honolulu, Hawaii, USA Honolulu, Hawaii, USA Honolulu, Hawaii, USA Technology
tranttq@hawaii.edu kdavies@hawaii.edu lroose@hawaii.edu Hanoi, Vietnam

Abstract— This paper presents an online method of monitoring the various transformer component failures is ranked in Table
the health of distribution service transformers using data from a I, where the importance is rough, qualitative product of severity
real-time data acquisition and analysis system located at the
and probability of failure.
transformer. The results can be provided to utility operators in
near real-time, enabling them to assess the status of the distribution
transformer and take actions to avoid or mitigate costly TABLE I. RANKING OF TRANSFORMER COMPONENTS FAILURES
transformer failures and service outages. After thorough review of Component Failure modes Importance
potential non-invasive monitoring techniques, top oil temperature,
Insulation Water accumulation in the oil/paper
vibration, transformer loading, and power factor were chosen as
1 Thermal degradation oil/paper High
monitored parameters, and the overall health index is determined
Aging of oil/paper
using fuzzy logic. A case study is implemented on a specific Short circuit between turns/strands
distribution transformer to show the advantages of the proposed Short-circuit to ground
method. Winding
Open circuit
Conductor tilting, conductor
I. INTRODUCTION 2 bending, clamping system failure, High
Distribution transformers are one of the most expensive axial instability, and lead
deformation
and important components of the distribution power system [1].
Winding bulk movement
Failures in distribution service transformers occur due to a Buckling
number of reasons including operation stresses, oil leakage, Short circuit within capacitance
Bushing
thermal stress due to overload, harmonics and unbalanced 3
graded layers
High
loading [2]. If one or more of these conditions consistently Bushing thermal expansion
Failure due to porcelain damage
exceed the design and operational limits, the service life of the
On-load tap Open contacts
transformers may be reduced or failure may occur. Therefore, it changer Coking of contacts
is important to develop the effective methods to monitor the 4 Low
Drive mechanism
health and the overall condition distribution transformer so that
utilities can proactively mitigate failures and degradation, Failure of Internal rupture
5 tank Low
increase the reliability of electric service and reduce the cost of Leakage
utility operations and maintenance. Core Short circuit core laminations
Many causes of failure of the transformer have been 6
Multiple grounding
Medium
reported in literature. Different investigations and test analyses Ungrounded core
are conducted to identify the root causes and to identify the Core deformation
Operational errors, lack of
preventive measures to avoid these breakdowns. Paper [3], the 7 Other maintenance, and protection system Low
authors focus on components: core, winding, bushing, tank, trips/failures
cooling fins and the transformer dielectric materials. The
authors propose a top-down approach using a three-level model Table 1 indicates that nearly all the failure modes are internal
to the transformer and may be difficult to detect in-situ and non-
consisting of failure mechanism, failure linkages and failure
invasively. Ideally, there would be a set of electrical,
mode. In [4-7], the authors aggregated the data of component
mechanical, thermal or other detectable signatures for the most
failures to classify and identify causes of failures of power important failure modes.
transformers. They proposed a method to calculate a health
index and recommend specific maintenance to electric utilities. Currently, most monitoring programs are implemented on
Based on the component failures addressed in the above papers, large (MVA and greater) power transformers [8-10]. The
the most critical components on transformer health appear to be application to distribution transformer has largely been
the insulation, winding, bushing and on-load tap changer. Other neglected considered. In [11], the remote condition monitoring
components such as cooling system, core, operation errors also techniques are proposed and results are discussed. The
mentioned but appear to be less significant. The importance of parameters such as temperature, oil level, transformer loading,
humming sound were applied to determine the condition of a

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on September 09,2020 at 23:43:43 UTC from IEEE Xplore. Restrictions apply.
distribution transformer. However, the results shown were on stages. Figure 1 shows the block diagram of real time energy
the very small time frame to accurately assess the status of the monitoring device architecture.
transformer. In [12], a wireless monitoring technology using
internet of things (IoT) was designed to assess the state of
distribution transformers in real-time. The system can provide
high reliability and the fast results storing data on server every
90 second. The observed parameters in the paper are limited to
common measurement data such as ambient temperature,
voltage and current.
Table 1 implies that the parameters used to monitor power
transformer typically require expensive sensors, which must
continue to operate properly over the full lifetime of the
transformer and survive in oil environments. These sensors are
also difficult to install or replace without opening the
transformer tank; therefore, they have not been previously
considered for use on distribution service transformers. Any Fig. 1. Block diagram of real time energy monitoring device
potential method that relies instead on inexpensive and easy to architecture
install sensors is of immediate interest
The online monitoring program uses data from smart meters
With the development of science and technology, smart and present results through a user interface (UI). This UI also
meters are innovating to support for energy consumption
provides a means to input various system parameters such as
management. The Hawaii Natural Energy Institute is developing
transformer ratings. The model of online monitoring program is
a low cost, high fidelity real-time energy monitoring device for
use on distribution service transformer. In this paper, we propose described in Figure 2. The decision-making steps are given in a
an online monitoring distribution transformer health method that flow chart on Figure 3.
can be implemented on such a system. This approach (and
challenge) is to determine the most appropriate online sensors
and algorithms to employ for low cost, accurate and in-situ
transformer health monitoring and assessment. Top oil
temperature and vibration, transformer loading and power factor
were chosen as desired parameters. Top oil temperature and
vibration are estimated using dynamic models and real-time data
from the energy monitoring device. Top oil temperature is an
indicator of transformer stress—thus related to degradation over
time—and expected vibration could be compared to sensors on
the tank surface as a means to detect transformer degraded
transformer health. Both methods could be implemented at low
capital and maintenance costs relative to traditional invasive
approaches. Fuzzy logic is used to combine various indicators of
transformer health into an overall health index. Algorithms are Fig. 2. Scheme diagram of the online monitoring program
implemented centrally at the host computer to analyze the
operational characteristics of distribution service transformer
and predict the need for maintenance, reconfiguration, upgrade
or replacement.
II. ONLINE MONITORING SYSTEM DESCRIPTION
A real-time data acquisition and analysis device has been
developed by the Hawaii Natural Energy Institute from 2017 for
high-resolution data acquisition at distribution service
transformers, with the potential to control distributed energy
resources. This monitoring device can be installed on
distribution service transformer or other key locations near the
edge of the grid to record real-time voltage, current, active power
and reactive power and power factor every second. The device Fig. 3. Flowchart of real time transformer health monitoring process
includes a hybrid computing architecture consisting of an FPGA,
64-bit quad-core processor, a timing-optimized GPS receiver, There are several different techniques are used to process
super capacitor to record grid outage. Data from real time energy data to assess transformer health such as fuzzy logic, neural
monitoring device is streamed continuously to other device by network, and health index calculation. While health index
using a flexible messaging with data backup with multiple calculation strongly depends on experience parameters, it is
difficult to implement and cannot provide precise results [13].
Artificial neural network is a good tool for this type of problem

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on September 09,2020 at 23:43:43 UTC from IEEE Xplore. Restrictions apply.
but they requires a large amount of historical transformer failure 1. Top oil temperature estimation
data and can not apply to small sample data [14]. Therefore, in
this paper, fuzzy logic is chosen to solve the problem. In this paper, top oil temperature estimated from a dynamic
thermal model that depends on transformer loading, which is
Fuzzy logic is applied by assigning membership functions indicated by measured voltage and current. The maximum
which describe the extent to which a value of each input permissible top oil temperature is 95°C [15].
parameters(includings vibration, power factor, transformer The top oil thermal model can be described by the following
loading and top oil temperature) – indicate “good”, “average”, deferential equation:
“poor” and “very poor” transformer health. Trapezoidal and
triangular curves are applied as “expert rules” over the d Δ θ oil (θ oil − θ am b )1/ n (1)
respective ranges to describe these associations. Overall q fe + q cu = C oil . +
transformer health or health index (HI) is the final output of the dt R oilR
fuzzy inference system. Specific details are presented in the where the left hand side is the rate of heat generation in the
section III.
core (qFe) and windings (qCu). One the right hand side, the first
The expert rules are constructed following the important term is the rate of change of internal energy and the second is
order of parameters to produce the health index for distribution the rate of heat transfer to the environment. θoil is the top oil
transformer. The top-oil temperature is selected as the most temperature (°C) and θamb is the ambient temperature (°C). n is
important parameter in contributing to the health of transformer, oil exponent, the value of n =0.8 or 1 depending on the type of
followed by transformer loading, vibration and power factor. cooling system using in transformer.
Figure 4 provides a block diagram of the assessment process
using fuzzy logic system. The algorithms were implemented and At rated load, the change of internal energy approaches zero,
executed in Matlab. then:
(θ oil , R − θ am b )1/ n
q fe + q cu , R = (2)
R oilR
Or
(θ oil , R − θ amb )1/ n Δ θ oil , R 1/ n
R oilR = = (3)
q fe + q cu , R q fe + q cu , R
where Δθoil,R is the top oil rise under rated conditions, °C
Equation (1) and (3) are now employed to the following
equation:
Δ θ oil , R 1/ n d Δ θ oil
( q fe + q cu ). = C oil . R oilR + Δ θ oil 1/ n (4)
q fe + q cu , R dt
Or
Δ θ oil , R 1/ n d Δ θ oil
( q fe + q cu ). = τ TO + Δ θ oil 1/ n (5)
Fig. 4. The model of the assessment process q fe + q cu , R dt
The limits of health index and the corresponding Set R is the ratio of qcu to qFe at rated load, qFe is the fixed iron
recommended actions are illustrated in Table II: loss; K is the load factor, which is the ratio of the load current
to the rated load current; τTO is the top oil time constant, hour
TABLE II. LIMITS OF HEALTH INDEX Equation (5) is convert to equation (6):
n
Health index Condition Recommended action  K 2 R + 1 d Δθ oil (6)
range  R + 1  .Δθ oil , R = τ TO . dt + Δθ oil
 
7.5 ≤ HI ≤ 10 Good Normal maintenance
5 ≤ HI ≤ 7.5 Fair Increase diagnostic testing The value of the oil time constant varies with the top-oil
temperature rise and power losses. The top-oil time constant (in
3.0 ≤ HI ≤ 5.0 Poor Start planning change request seconds) at the rated apparent power (kVA) is given by the
HI ≤ 3.0 Very poor Immediately assess risk following equation [16]:

 Δθoil ,U Δθoil ,i 
III. DATA ACQUISITION  − 
 Δθ oil , R Δθoil , R  (4)
Top oil temperature, vibration, transformer loading and τ TO = τ TO , R 1/ n 1/ n
power factor are calculated or estimated locally (at the  Δθ oil ,U   Δθ oil ,i 
transformer) and send for final health assessment at the central   −  
serve.  Δθoil , R   Δθ oil , R 

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on September 09,2020 at 23:43:43 UTC from IEEE Xplore. Restrictions apply.
where: and thus the transmission of vibration from the core to the
Δθoil,i is the initial top-oil rise over ambient temperature for t = surface of transformer tank. Equation (8) is the complete
0, °C estimation model of vibration on the transformer tank.
Δθoil,U is the ultimate top-oil rise over ambient temperature for
load L, °C νtank = (α+βθtop oil).i2 + = (γ+δθtop oil).u2 (8)
The top-oil time constant at rated kVA is given by the where α, β, γ, and δ are parameters whose values depend
following: on transformer geometry which must be measured. The limits
3600.C.Δθ oil , R of vibration are shown in Table IV.
τ TO , R = (5)
Ptotallosses , R TABLE IV. LIMITS OF VIBRATION
Where: Limit of vibration Condition
C is the thermal capacity of the transformer, W-h/°C
Ptotalloses,R is the total loss of transformer at rated load, W 0 ≤ νtank ≤ 48dB Good
τTO,R is the time constant with initial top-oil temperature 48dB ≤ νtank ≤ 60dB Need caution
rise of 0 °C, second
The thermal capacity for an oil natural (ON) cooled νtank ≥ 60dB Poor
transformer (without forced oil circulation) is given by the
following equation [8]:
C = 0.1323*weight of core and coil assembly in kilograms+ 3. Transformer loading
0.0882*weight of tank and fittings in kilograms+ 0.3513*liters Transformer loading (TL) is determined based on the ratio
of oil (6) of the load to the actual nameplate power of the distribution
The algorithms for top oil temperature estimation are transformer. If it is larger than the nameplate rating, the risk of
implemented in Matlab. The limits of top oil temperature are unexpected is evaluated. This failure can damage the thermal
shown in Table III. insulation of distribution transformer or break down the
transformer coil. Therefore, in this paper, transformer loading
TABLE III. LIMITS OF TOP OIL TEMPERATURE is chosen as an important parameter to evaluate the transformer
health. The limits of transformer loading are shown in Table V.
Limit of top oil temperature Condition
Good TABLE V. LIMITS OF TRANSFORMER LOADING
θTO ≤ 95
95 ≤ θTO ≤ 105 Need caution Limit of transformer loading Condition

105 ≤ θTO ≤ 115 Poor TL ≤ 95% Good

θTO ≥ 115 Very poor 95 ≤ TL ≤ 150 Need caution


TL ≥ 150 Poor

2. Vibration estimation
4. Power factor
Vibration is chosen as one of the important parameters to
assess the status of transformer. Vibration is generated Power factor is determined by the ratio of working power
absorbed by load to the apparent power. This parameter is used
primarily at the winding and the core of transformer by
to demonstrate the electricity efficiency of distribution
electrodynamic forces and magnetic forces, respectively. The
transformer as well as evaluate the status of operating
electrodynamic forces result from the interaction between the transformer. The limits of power factor are described in Table
winding current and the leakage flux which are proportional VI.
with the current squared. The magnetic forces are created by the
alternating flux in the core of transformer which is proportional TABLE VI. LIMITS OF POWER FACTOR
to the voltage squared [17]. The authors of [18] proposed a
method to calculate the tank vibration for transformer following Limit of power factor Condition
the relation between cores and winding vibration at the 0.85 ≤ PF ≤ 1.0 Good
corresponding operating frequency. The relationship is
described in following equation: 0 ≤ PF ≤ 0.85 Fair
PF ≥ 1.0 Poor
νtank = t1.i2 + t2.u2 (7)
where t1 and t2 are transmission coefficients IV. SIMULATION AND RESULTS
The authors of [9] also mention that the variation of oil A case study is carried out on a simulated 50kVA-
temperature will effect on the vibration at the tank of 7.2/0.208kV distribution transformer to show the effectiveness
transformer. Changes in oil temperature affect the oil viscosity of proposed model. The electrical conditions on the transformer

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on September 09,2020 at 23:43:43 UTC from IEEE Xplore. Restrictions apply.
were measured by a real-time energy monitoring device in
November and December of 2019. The nameplate parameters
of transformer are described in Table VII.

TABLE VII. FEATURE PARAMETERS OF 50KVA DISTRIBUTION


TRANSFORMER

Quantity Transformer/winding
Top oil temperature rise at rated load 65oC
Oil weight (kg) 105
Winding weight (kg) 70
Fig. 7. The top oil temperature estimation of distribution transformer
Fe weight (kg) 80
The general trend in temperature curve are expected. At
Tank weight (kg) 165 peak hours of the day, the temperature increase from 43 oC to
around 60oC and decrease gradually when the loading reduces.
Load loss (W) 870
The temperature is lower than its limit. Unfortunately, there is
No load loss (W) 130 no temperature sensor available to validate the calculated
profile, it would need to be installed inside the transformer tank.
The profile of apparent power and voltage sent from smart The results of transformer tank surface vibration simulation
meter are shown in Figures 5 and 6, respectively. are shown in Figure 8. The vibration is typically below 30dB
but it increases to 40dB or even 48dB on some peak demand
days.

Fig. 5. The profile of apparent power

Fig. 8. The vibration estimation of distribution transformer


The transformer loading and power factor profile are
provided in Figure 9 and Figure 10, respectively.

Fig. 6. The voltage profile of transformer


Figure 7 shows the variation in top oil temperature
indicated by the model.

Fig. 9. The transformer loading of distribution transformer

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on September 09,2020 at 23:43:43 UTC from IEEE Xplore. Restrictions apply.
[2] K. Chitnavis and N. R. Bhasme, "Review of critical analysis for life
estimation of power transformer," in 2017 4th International Conference
on Power, Control & Embedded Systems (ICPCES), 2017, pp. 1-6.
[3] R. M. Rampersad, S. Bahadoorsingh, and C. Sharma, "Multifactorial
Frameworks Modelling Linkages of Power Transformer Failure Modes,"
in 2018 IEEE Electrical Insulation Conference (EIC), 2018, pp. 398-402.
[4] R. Murugan and R. Ramasamy, "Understanding the power transformer
component failures for health index-based maintenance planning in
electric utilities," Engineering Failure Analysis, vol. 96, pp. 274-288,
2019/02/01/ 2019.
[5] A. Naderian, S. Cress, R. Piercy, F. Wang, and J. Service, "An Approach
to Determine the Health Index of Power Transformers," in Conference
Record of the 2008 IEEE International Symposium on Electrical
Insulation, 2008, pp. 192-196.
[6] W. Wattakapaiboon and N. Pattanadech, "The new developed Health
Fig. 10. The power factor of distribution transformer Index for transformer condition assessment," in 2016 International
Conference on Condition Monitoring and Diagnosis (CMD), 2016, pp.
The health index of the distribution transformer is the final 32-35.
output of the system after applying the fuzzy logic. The results [7] J. Haema and R. Phadungthin, "Condition assessment of the health index
for power transformer," in 2012 Power Engineering and Automation
are shown in Figure 11. Conference, 2012, pp. 1-4.
[8] L. Rosero-Z, A. Pavas, and I. C. Durán, "Analysis of Maintenance in
Transformers Based on a Fuzzy Logic Method," in 2018 IEEE PES
Transmission & Distribution Conference and Exhibition - Latin America
(T&D-LA), 2018, pp. 1-5.
[9] M. Arshad and S. M. Islam, "A Novel Fuzzy Logic Technique for Power
Transformer Asset Management," in Conference Record of the 2006 IEEE
Industry Applications Conference Forty-First IAS Annual Meeting, 2006,
pp. 276-286.
[10] Z. Xuewei and L. Hanshan, "Research on transformer fault diagnosis
method and calculation model by using fuzzy data fusion in multi-sensor
detection system," Optik, vol. 176, pp. 716-723, 2019/01/01/ 2019.
[11] A. A. Nelson, G. C. Jaiswal, M. S. Ballal, and D. R. Tutakne, "Remote
condition monitoring system for distribution transformer," in 2014
Eighteenth National Power Systems Conference (NPSC), 2014, pp. 1-5.
[12] D. Srivastava and M. M. Tripathi, "Transformer Health Monitoring
Fig. 11. The health index of distribution transformer System Using Internet of Things," in 2018 2nd IEEE International
Conference on Power Electronics, Intelligent Control and Energy
The HI varies diurnally between 6.7 to 8. The results Systems (ICPEICES), 2018, pp. 903-908.
indicate that transformer maintenance is not required but [13] I. V. Davidenko and T. S. Kuzina, "Analysis of the modern methods of
diagnostic testing should be applied to identify potential the power transformers health index calculation," in 2017 IEEE
Conference of Russian Young Researchers in Electrical and Electronic
damages. In this case, the top oil temperature and transformer Engineering (EIConRus), 2017, pp. 1491-1495.
loading are the most important input variables. The top oil [14] H.-C. Sun, Y.-C. Huang, and C.-M. Huang, "Fault Diagnosis of Power
temperature is quite stable and does not change much. It can be Transformers Using Computational Intelligence: A Review," Energy
observed that the value of HI varies proportionally with the Procedia, vol. 14, pp. 1226-1231, 2012/01/01/ 2012.
[15] "IEEE Guide for Loading Mineral-Oil-Immersed Transformers -
fluctuation of transformer loading. Therefore, a plan is needed Corrigendum 1," IEEE Std C57.91-1995/Cor 1-2002, pp. 1-16, 2003.
to adjust the load of the transformer to avoid overload and [16] "IEEE Guide for Loading Mineral-Oil-Immersed Transformers and Step-
potential break down. Voltage Regulators," IEEE Std C57.91-2011 (Revision of IEEE Std
C57.91-1995), pp. 1-123, 2012.
V. CONCLUSION [17] Z. Witczak Pawel and M. Swiatkowski, "Magnetic forces applied to the
tank walls of a large power transformer," COMPEL - The international
In this paper, an online method of monitoring distribution journal for computation and mathematics in electrical and electronic
transformer health were presented that uses advanced real-time engineering, vol. 35, pp. 2087-2094, 2016.
grid energy monitor system. The architecture of the smart meter [18] B. Garcia, J. C. Burgos, and A. M. Alonso, "Transformer tank vibration
and structure of the monitoring program based on the online modeling as a method of detecting winding deformations-part I:
theoretical foundation," IEEE Transactions on Power Delivery, vol. 21,
data collected from have introduced. A case study on 50kVA pp. 157-163, 2006.
distribution transformer was performed to illustrate the
effectiveness of the proposed model. The proposed non-
invasive methods can be implemented at low cost and with
minimal disruption. In further work, this monitoring model will
be applied on the larger transformers along with real sensors in
different scenarios to analyze the sensitivity of parameters and
provide the overall validation.
REFERENCES
[1] Transformers: Basics, Maintenance, and Diagnostics. Denver, Colorado,
USA: Reclamation Bureau, 2005.

XXX-X-XXXX-XXXX-X/XX/$XX.00 ©20XX IEEE

Authorized licensed use limited to: UNIV OF HAWAII LIBRARY. Downloaded on September 09,2020 at 23:43:43 UTC from IEEE Xplore. Restrictions apply.
View publication stats

You might also like