Wednesday, February 4, 2015

Examining PHMSA's Faulty PIR formula

(Note: I wrote these notes up thinking about the Spectra/Algonquin AIM pipeline, but I've been doing this PIR research for some time -- BH)

I have something to contribute here concerning
the abuses of the Potential Impact Radius (PIR) formula
used by PHMSA.

An a recent conference call, Richard Kuprewicz spoke eloquently
about how the PIR was never intended to be a prediction
of the Actual Impact Radius
(AIR), but was only a tool developed
to help assess the calculation of High Consequence Areas (HCA).

What I am working on is this:

Since the existing PIR formula is clearly wrong,
What is a better formula for predicting
the actual impact radius?


I examined 5 major gas pipeline accidents
where I was able to get these metrics:

1) diameter of the pipe
2) Max allowable operating pressure (MAOP)
3) Actual impact radius

In every case the predicted PIR was short of the actual impact radius,
sometimes by a long shot.  Sometimes by a factor of 2x predicted.

I grabbed this stock diagram plotting the existing (broken) formula's prediction as a function of diameter and pressure:


Here's my version showing what I found in the data. The arrow going up shows the actual radius of impact vs. predicted down on the line.

The longer the arrow, the farther off the actual (measured) was from predicted.

So you can see that Carlsbad and Sissonville, while the AIR
was found to be over the PIR, it was not by much. ~20%.

But as you can see, Appomattox, San Bruno,
and Cleburne were significantly over, by ~200% or more.

I'm not sure how to explain this. I'm normally good at curve fitting.
I think there may be 2 missing terms.




Bottom Line for the AIM:

In my opinion, to properly assess the risk of a potential castrophic incident at Indian Point in relation to the new 42" pipeline, we must use an accurate PIR formula tuned to real-world data. This should be a top priority of PHMSA.

A good estimate for the Actual Impact Radius for AIM, a 42" 800 psig MAOP pipeline in a dense urban area could easily be 2,500 ft., whereas the calcuated PIR would be about 825 ft. using the existing formula.


I say this because I feel the most important factors
are "pipeline diameter" and also "dense urban area".
This last term is absent in the present PIR formula.

The existing PIR formula is
0.69 * SQRT(Max Allowable Operating Pressure) * Diameter

Sometimes they write that to make it look like D is a squared term.
I think it should be!

Remember, the area of a circle is Pi * R SQUARED,.

Where is the squared term for R (or D) in the PIR formula?
Which grows exponentially as the radius of the pipeline grows?
 
Also, IMO San Bruno may have gotten bonus points (feet added) for being  a "dense urban area". Sissonville and Carlsbad maybe should get demerits (feet deducted) due to terrain issues: speculative-- 

"Dense Urban Area" should be reflected in the PIR, although I'm not sure I can say how without a larger dataset.

Here is a map I derived from one generate by someone in the SAPE group:

The inner yellow band here is the predicted PIR.
The black line is from Cleburne TX (36" vs 42").
The outer pink is a better indication (IMO) for a potential impact radius.

Take it to 2,500 ft in those neighborhoods to the south.



 
--
--
May you, and all beings
be happy and free from suffering :)
-- ancient Buddhist Prayer (Metta)

Don't forget to sign the
Pledge to Resist
the Constitution Pipeline:

3 comments:

Robert Dickerman said...

Bill,

I have calculated the number of LNG tanker trucks worth of LNG that would be equivalent to the amount of gas released in a worse-case event here in Northfield, MA. The calculation includes the following factors, amongst others:

1. cross-sectional area of pipe (using r**2, which you have alluded to)
2. length of pipe to nearest Main Line Valves (both upstream and downstream)

I don't know what a good model for AIR or PIR would be, but I would guess that it would involve both of the above factors.

If you are interested, I can send a copy of my handwritten calculations.

Bob Dickerman

Bill Huston said...

Feel free to send it along, but LNG is quite different from gas in pipeline, since it is 600x more dense then methane @ 1 ATM. I've calculated blast potential of LNG tanker ships, but using a different method, i.e., looking at the total joules in that gas volume, then converting to kilotons of TNT.

Bill Huston said...

Feel free to send it along, but LNG is quite different from gas in pipeline, since it is 600x more dense then methane @ 1 ATM. I've calculated blast potential of LNG tanker ships, but using a different method, i.e., looking at the total joules in that gas volume, then converting to kilotons of TNT.