You are on page 1of 13

QUESTIONS AND ANSWERS

TRANSPORTATION
RISK ASSESSMENT
WEBINAR
Table of contents
1. WHY WAS THE RELEASE ANGLE CHOSEN AS 10 DEGREES FROM THE HORIZONTAL
FOR THE PIPELINE SECTION FROM POINT A TO POINT B? 2
2. WHAT IS THE THOUGHT BEHIND 10% ANGLE AS A SIMULATION OF CRATER? 5
3. WHAT IS THE BEST DATA FOR BURIED PE PIPELINES? 5
4. ARE THERE UNDERGROUND PIPELINE MODES OR HOW TO CORRECT PARAMETERS TO
SIMULATE UNDERGROUND PIPELINE LEAKAGE? 7
5. IS THERE AN EQUIVALENT TO EGIG FOR HDPE PIPELINES? 7
6. BURIED PIPELINES HAVE DIFFERENT METHODOLOGIES FOR MODELLING
CONSEQUENSE AND THERE IS GUIDANCE IN SUCH TEXTS AS OGP ON HOW TO
MODEL THESE AND PROPORTION THE RISK. WHAT DOES SAFETI ENCOMPASS WHEN
YOU SPECIFY THAT THE PIPELINE IS BURIED? 7
7. HOW CAN PHAST AND SAFETI SIMULATE THE CRATER FORMATION IN CASE OF
UNDERGROUND PIPELINE RUPTURE? 7
8. WHY DOES THE GAS RELEASE AT 0 FT ABOVE GROUND LEAD TO OVERESTIMATION
OF CLOUD DISPERSION AS CONSEQUENSE OF GROUND EFFECTS? 7
9. HOW CAN YOU ESTIMATE THE INVENTORY OF GAS/OIL/RELEASED/SPILLED TAKING
INTO ACCOUNT THE EFFECTIVENESS OF LEAK DETECTION SYSTEM? 8
10. THE PIPELINE WALL THICKNESS IS PRINCIPALLY DESIGNED ON THE BASIS OF INNER
PRESSURE. THE WALL THICKNESS OF PIPELINES OBSERVED BY EGIG ARE DESIGNED
ON THE BASIS OF INNER PRESSURE. HOW CAN YOU ESTIMATE THE FREQUENCY
REDUCTION ON THE BASIS OF WALL THICKNESS INCREASING? 9
11. HOW CAN YOU USE THE OGP HOOK-UP CORRELATIONS FOR IGNITION LIKELIHOOD
EVALUATION INSTEAD OF “PURPLE BOOK” CORRELATION INTO SAFETI? 10
12. HOW CAN YOU EVALUATE THE OIL SPILL INVENTORY IN OIL PIPELINE WHERE THE
SPILLAGE DEPENDS ON STATIC HEAD AS WELL AS INNER RESIDUAL PRESSURE? 12
13. IS PHAST AND SAFETI ABLE TO SIMULATE TRANSIENT RELEASE CONDITIONS? 12
14. WHAT IS THE DIFFERENCE BETWEEN PHAST AND SAFETI? 12
15. CCPS TANK TRUCK TRA CASE STUDY WHAT YOU ARE SHOWING IS DONE IN SAFETI
7.2. IS SAFETI 7.2 RELEASED? 12
16. HAS “LONG PIPELINE” MODEL BEEN IMPROVED? CANNOT MODEL 1" HOLE IN 5" LINE
USING CURRENT MODEL BECAUSE LOWER LIMIT OF APERTURE RATIO = 0.2. 12
17. PHYSICAL LOCATION OF THE PIPE? UNDERGROUND OR ABOVEROUND? 13
18. HOW DO YOU DEFINE THE IGNITION PROBABILITY ALONG THE ROUTE? 13
19. IS THERE ANY BASIS FOR USING A 20% PRESSURE DROP WHEN DISCRETIZING THE
PIPE LINE FOR FAILURE FREQUENCY BASED CALCULATIONS? 13
1. WHY WAS THE RELEASE ANGLE CHOSEN AS 10 DEGREES
FROM THE HORIZONTAL FOR THE PIPELINE SECTION FROM
POINT A TO POINT B?

This angle is taken directly from the CCPS case study, page 258 of CCPS. (1995). Guidelines for
Chemical Transportation Risk Analysis.

Legacy DNV used internal guidance which recommended 45° from horizontal impingement with a 0.25
momentum reduction factor to simulate crater impingement. This was validated against hydrogen
pipeline experiments as described in this paper and showed good correlation for flowrate and radiation
metrics. This approach has not been automated in Safeti 7.2, but the factors to control these inputs are
made available in the dialog (see Figure 2 and Figure 3) used to define release scenarios. So that you
can follow guidance such as OGP’s pipeline release recommendations:
Figure 1 - OGP 434-7 buried pipeline release recommendations
Figure 2 - Release angle definition fields in Section Breach dialog
Figure 3 - Velocity modification fields in Section Breach dialog

Since DNV’s merger with GL the Phast and Safeti modelling team have access to legacy GL tools and
models. We are currently working on implementing a proprietary crater model based on experiments
conducted at our Spadeadam test facility into Phast and Safeti which will account for crater effects in a
less generic way. We will discuss this in a future monthly status note. Please email
software.support@dnvgl.com if you wish to sign up for the Safeti status note. The paper describing this
method is: A model for the initial stages following the ruptureof a natural gas transmission pipeline,
Cleaver R. P., Halford, A.R., Process Safety and Environmental Protection 95, 2015 pp202–214.

2. WHAT IS THE THOUGHT BEHIND 10% ANGLE AS A


SIMULATION OF CRATER?
See question 1.

3. WHAT IS THE BEST DATA FOR BURIED PE PIPELINES?


Criteria for inclusion in the EGIG data set are:
Figure 4 - EGIG inclusion criteria

And therefore polyethylene piping is not included. However, a range of coatings are present in operation
of the EGIG pipeline source dataset with a recent increase in Polyethylene coatings, and therefore,
recent trends for such pipeline could feasibly be sourced from this reference.
I am not familiar with the best source of failure rates for PE pipelines (non-coated).

Figure 5 - EGIG pipeline coating trend


4. ARE THERE UNDERGROUND PIPELINE MODES OR HOW TO
CORRECT PARAMETERS TO SIMULATE UNDERGROUND PIPELINE
LEAKAGE?
See question 1.

5. IS THERE AN EQUIVALENT TO EGIG FOR HDPE PIPELINES?


See question 3.

6. BURIED PIPELINES HAVE DIFFERENT METHODOLOGIES FOR


MODELLING CONSEQUENSE AND THERE IS GUIDANCE IN SUCH
TEXTS AS OGP ON HOW TO MODEL THESE AND PROPORTION THE
RISK. WHAT DOES SAFETI ENCOMPASS WHEN YOU SPECIFY
THAT THE PIPELINE IS BURIED?
See question 1.

7. HOW CAN PHAST AND SAFETI SIMULATE THE CRATER


FORMATION IN CASE OF UNDERGROUND PIPELINE RUPTURE?
See question 1.

8. WHY DOES THE GAS RELEASE AT 0 FT ABOVE GROUND LEAD


TO OVERESTIMATION OF CLOUD DISPERSION AS CONSEQUENSE
OF GROUND EFFECTS?
We did not look at any dispersion results during the webinar. Please contact me at
Colin.Hickey@dnvgl.com so we can discuss your case.

With Phast and Safeti’s current modelling capabilities, for a case which is used to simulate a below
ground release it is typical to set the height to ground level and to use a release angle and an
impingement factor. In addition, a grounded release has a drag factor applied from interaction with the
ground. In the jet phase of dispersion, momentum is a major factor in the rate of dispersion. With
reduced momentum the rate of entrainment of the cloud will be less than a non-impinged and/or an
elevated release. The difference between impinged non-impinged, grounded and elevated releases can
be seen in these comparison runs:
The cases are:

1m elevation, 45° impinged


1m elevation, 45° (not impinged)
0m elevation, 45° impinged
0m elevation, 45° (not impinged)

As you can see, the impinged cases extend the furthest (red and blue). The non-impinged cases travel
less distance (yellow and green). This is due to the momentum having been factored by 0.25 for the
impinged cases and therefore entraining less air.

If you compare the evident two pairs of impinged or non-impinged cases, those at 1m elevation travel a
small amount further. This is due to the extra drag factor reducing momentum and decreasing the rate
of entrainment.

9. HOW CAN YOU ESTIMATE THE INVENTORY OF


GAS/OIL/RELEASED/SPILLED TAKING INTO ACCOUNT THE
EFFECTIVENESS OF LEAK DETECTION SYSTEM?
Phast and Safeti consider time varying flow of gases and superheated materials from the pipeline
breaches. This takes account of the pressure wave travelling through the pipeline, contribution of flow
from upstream and downstream branches and interaction of valves. Valves can be assigned a location
along the pipeline and can be given closure properties. The types of valves available in Phast and Safeti
for use on the pipeline are: closure, where a closure time is assigned (this could be used in conjunction
with info from studies into closure times or general guidance e.g. HSE CR99206 (section 4.6)), excess
flow valves, where the valve will close when flowrate reaches a set level at the valves location on the
pipeline, and finally it is also possible to assign non-return valves (resulting in no reverse flow from
downstream branches.)

The method behind these calculations are in the modelling theory documentation installed with Phast
and Safeti. Please let me know if you require more details and want a copy of these documents.
10. THE PIPELINE WALL THICKNESS IS PRINCIPALLY
DESIGNED ON THE BASIS OF INNER PRESSURE. THE WALL
THICKNESS OF PIPELINES OBSERVED BY EGIG ARE DESIGNED
ON THE BASIS OF INNER PRESSURE. HOW CAN YOU ESTIMATE
THE FREQUENCY REDUCTION ON THE BASIS OF WALL
THICKNESS INCREASING?
This highlights a very good point. The factors influencing the likelihood of loss of containment from a
pipeline are numerous, and include things such as transported material properties which may give rise to
depletion of the pipe wall (flowrate, pressure, temperature, corrosivity), maintenance strategy, location
of the pipeline, etc.

EGIG provide primary failure frequencies for their entire data set, and then secondary frequencies for
subsets of the data set. The secondary frequencies are described in EGIG 2015 as follows:

Figure 6 - EGIG secondary failure frequencies description

As you can see, despite mention of the influence of design (including pressure) as a parameter, no
secondary failure frequency results are provided in terms of operating pressure. And therefore we cannot
perform any deeper analysis to discount the effect of increased pressure from increased pipe wall
thickness and therefore “clean” the data for generic use.

This is the cost of using generic industry data and as with all engineering problems we must have a good
understanding of the reference information when we use it on projects. In this way, the influence of
increased pipe wall thickness is used in the case study to account for the increased thinning life and
increased resistance to external impact, but is used at the cost of not accounting for the fact that
increased operating pressure may have increased the rate of failure for pipelines with increased wall
thickness. In such cases, if the underlying data leads to more conservative results used inappropriately
it may be acceptable to use the data for comparison and decision making purposes. That is to say,
applying the failure rates of high pressure large wall thickness failure rates to lower pressure pipelines
may overestimate the failure rate and therefore can be used indicatively.

11. HOW CAN YOU USE THE OGP HOOK-UP CORRELATIONS FOR
IGNITION LIKELIHOOD EVALUATION INSTEAD OF “PURPLE
BOOK” CORRELATION INTO SAFETI?
For immediate ignition control you can choose one of the purple book options, or you can define the
immediate ignition probability directly.

Figure 7 - immediate ignition probability control

For delayed ignition there are several options:

Ignition sources on the map


In Safeti you can draw ignition sources on the map. This will not achieve the intent described in the OGP
reports as ignition of the cloud will be based on the location and strength of ignition sources, (in addition
to exposure time of the cloud) and not just the release properties or genetal properties of the plant.

Define a delayed ignition probability


With a 3D Explosions license, Safeti offers you more capabilities and allows you to assign a delayed
ignition probability directly to the release case. This allows you to run the discharge calculations, observe
the size of the release and then use the OGP 434-6 reference tables to determine the ignition probability
(Figure 8). The OGP report refers to UKOOA ignition probabilities, which assume an immediate ignition
probability of 0.001. This can also be inserted as described above.
Figure 8 - specify minimum probability of delayed ignition in Safeti with 3D Explosions

For info, we have built a UKOOA ignition model into Safeti Offshore, and it is possible to simply choose
which ignition model is used in the offshore QRA study and it will be used for all cases. This same
technology will also be included in Safeti in future. (Figure 9)

Figure 9 - Safeti Offshore explosion and risk parameters


12. HOW CAN YOU EVALUATE THE OIL SPILL INVENTORY IN OIL
PIPELINE WHERE THE SPILLAGE DEPENDS ON STATIC HEAD AS
WELL AS INNER RESIDUAL PRESSURE?
There is no commercially available model known to us. For this reason we sponsored a PhD at University
College London several years ago to create a sub-cooled liquid pipeline breach model. The thesis has
been published and the development of the theory into a mathematical model is in our future roadmap. I
cannot currently estimate when this will be released. Popularity of a feature determines priority, so
please let me know if this is of interest to you.

I can make the thesis available to you upon request. Please email me at Colin.Hickey@dnvgl.com to
request a copy.

13. IS PHAST AND SAFETI ABLE TO SIMULATE TRANSIENT


RELEASE CONDITIONS?
Yes, Phast and Safeti consider time varying flow for gases and liquids from vessels and pipelines.

14. WHAT IS THE DIFFERENCE BETWEEN PHAST AND SAFETI?


Risk = Consequence x Likelihood.

Safeti performs risk calculations (QRA) and as such contains Phast.

Phast is all of the consequence models from Safeti offered in a consequence only package with no
consideration of likelihood factors:

• Release frequency
• Weather condition or direction probability
• Ignition probability

15. CCPS TANK TRUCK TRA CASE STUDY WHAT YOU ARE
SHOWING IS DONE IN SAFETI 7.2. IS SAFETI 7.2 RELEASED?
Safeti 7.2 is currently under development. It will be released at the end of the summer 2015. One of the
new features in Safeti 7.2 is the new pipeline risk model, with automatic failure case generation as
demonstrated in the webinar. We will announce to all customers when Safeti 7.2 is released. We also
highlight new features and give updates on project status and more accurate estimates of release date in
our monthly status notes. Please contact software.support@dnvgl.com if you would like to sign up for
the Phast and Safeti monthly status note.

16. HAS “LONG PIPELINE” MODEL BEEN IMPROVED? CANNOT


MODEL 1" HOLE IN 5" LINE USING CURRENT MODEL BECAUSE
LOWER LIMIT OF APERTURE RATIO = 0.2.
For the automatic section breaches, it is now possible to model releases less than 0.2 of the cross
sectional area of the pipe. We have performed sensitivity analysis and reviewed validation for breaches
for gas and superheated liquid pipelines and have extended the limits for the pipeline risk model in Safeti
7.2 to allow for holes down to the lowest orifice size found in the rest of the software: 0.01mm.

It is not possible to create location specific breaches (the non-automated failure cases) in Safeti, nor in
Phast. The method will be extended to these model entry points in the next version of Phast and Safeti
(7.3 or an update e.g. 7.2x).

17. PHYSICAL LOCATION OF THE PIPE? UNDERGROUND OR


ABOVEROUND?
In the case study we looked at simulation of a underground pipeline, this was achieved using the method
described under Question 1.

18. HOW DO YOU DEFINE THE IGNITION PROBABILITY ALONG


THE ROUTE?
See question 10.

19. IS THERE ANY BASIS FOR USING A 20% PRESSURE DROP


WHEN DISCRETIZING THE PIPE LINE FOR FAILURE FREQUENCY
BASED CALCULATIONS?
We performed sensitivity analysis and found that this is the border of seeing significant change in
discharge results. Note that we then check for differences in the discharge results too and rationalise
again where possible. This strikes a balance between accuracy and computational efficiency.

The parameters are available for you to edit, should you want to perform your own sensitivity analysis.

Figure 10 - Pipeline discretisation parameters, available for edit

You might also like