Fix the pumps

Monday, April 02, 2007

Oh. My. God.

Corrections have been made to the post below. They can be found here. Note that conclusions remain true, it is merely the methodology which was partially in error.

This one is aimed at the engineers in the audience, but everyone has to read it.

We know the following information about the welded pipe that runs from the drive units to the pump units, conveying the high pressure hydraulic oil to spin the pump impellers:

Max working pressure = 3000 psi (actually can go higher, see Ms. Garzinos's memo)
Diameter of high pressure line = 3" (see the Floodgates Operating Manual)
Thickness of high pressure line = Schedule 80 (nominally 0.3") (See the original specs, or the job sites. "Schedule 80" is written all over the pipes, which are plainly visible from the levees at Orleans Avenue. I have pictures, which I will post later)
Material of high pressure line = A106, probably grade B (See the original specs), with Maximum Allowable Stress in Tension of 15,000 psi [corrected to 20,000 psi here]
Pipe supplier = MWI (See the original specs)

Engineers in the audience will know to run the numbers through equation (4) in section 104.1 of ASME B31.1 [corrected here]. This is an equation that has been around for decades and is well known to every engineer who has ever specified pipe.

When you run that calculation with just the standard numbers (assuming a standard 12.5% wall deviation, joint efficiency = 1, y = 0.4), you find something absolutely horrifying.

The pipes are too thin to take the normal operating pressures! They're too thin by a lot. When you run the numbers, you find that the maximum burst pressure for 3" Schedule 80 A106B pipe is only 2393 psi [corrected here]. Even if the material is actually A106, Grade C, that still only gets you to 2791 psi. Even when you knock the wall allowance out completely, you still don't get there. When I did this calculation, I actually felt physically sick when I got the result. I've redone it a few times, hoping the numbers were different. They're not. And keep in mind that the pressure can exceed 3000 psi during operations, as also shown in Ms. Garzino's memo.

There are two of these lines on every pump/drive unit pair, each of them about 250 feet long. There is over 17,000 feet of this pipe installed at all three floodgate sites. All of it is unfit for the service it will see. All of it is prone to bursting under normal service [corrected here]. How this could have slipped through so many layers of review is unknown, but I can tell you that if there was a "Fitness for Service" declaration signed, it must be either fraudulent or signed under duress. I bet that the welders working these jobs probably asked for some kind of release from liability, so they wouldn't be blamed when the pipes burst. This is the worst kind of engineering mistake. This is "crash-into-Mars-because-we-didn't-convert-the-units-correctly" stuff.

This is equivalent to driving the sheet piles too shallow on the levees. It is a fundamental flaw in the system that has not been addressed by anyone, and has most likely been covered up to have gone this long. And it could have been remedied so simply: the Corps should have specified Schedule 160 pipe instead of Schedule 80 (by the way, there are questions to be answered as to why a thickness for the pipe was specified, but no diameter. That's very unusual. It's almost like the Corps knew in advance what they were going to get before they issued the specification...).

For reference, I suggest you take a look at this company's website:
Redox, Inc.

This page on their site is particularly instructive.

I have no association with them, other than they happened to have all the information that makes my point - conveniently organized and well written.

This is so incredibly shocking it bends the brain. It appears that the Corps and/or MWI didn't do the simplest of calculations involved in sizing a pipe. Any newly minted engineer who does anything with piping becomes familiar with this equation within five minutes of being handed a job. By the way, that calculation is also called for in the Corps' own engineering manual dealing with piping, EM-1110-1-4008, "Engineering and Design - Liquid Process Piping." It shows up on page 5 [correction: page 16] of chapter 3 (note that the Corps wrote the manual on WordPerfect, so the characters for plus, minus, and equals won't show up properly in Adobe Reader) [correction:

Put simply, all of the pumping systems, as designed, procured, and built, have a strong likelihood of failure when placed under normal operating conditions.

[Corrections to this post are available here]

9 Comments:

  • Matt,

    Though I didn't work the equation, from my experience, you are totally right.

    Could another Class-Action be swung on this data?

    By Anonymous GentillyGirl, at April 02, 2007 8:06 PM  

  • Matt -

    Go to one of the Corps' public meetings and tell them of your calculations. Ask the tough questions and get their answers. Get very public with this information as quickly as possible.

    Lakeview Property Owner

    By Anonymous Anonymous, at April 03, 2007 8:59 PM  

  • I often wonder what fools we may be. We have had an exceptional dry season.Little to no rain, it can't last.

    We will flood again.

    By Anonymous Anonymous, at April 04, 2007 12:09 AM  

  • Matt,
    Please make this public! The Corps is just impossible.

    Well, there is one silver lining... turns out we can thank God that the pumps don't work! :)

    By Anonymous Anonymous, at April 04, 2007 12:28 PM  

  • You are performing an invaluable service for all of us. Please send copies of this latest blog to our congressional delegation. Perhaps if each of us reading this forwarded a link it would get their attention. This level of incompetence must be addressed and corrected BEFORE another levee breach - not after. I really hope Landrieu, Vitter or Jindal will do something with this information.

    By Anonymous Anonymous, at April 04, 2007 12:57 PM  

  • I have written to Bobby Jinal the following:


    Please read the latest post in
    http://fixthepumps.blogspot.com/2007/04/oh-my-god.html

    The author of this blog, Matt McBride, has written about the shoddy work the Army Corps of Engineers is doing in "repairing" the pumps in the Orleans Avenue Canal Floodgate site. I'm sure that this is not an isolated situation.

    The lack of care that is going into this work is criminal: rusting, unpainted pipes, bolts and flanges, piping that runs from the drive units to the pump units are too thin to handle normal operating pressure.... I could go on and on, but I urge you to read Matt's findings. The author is an engineer and knows what he's reporting. In his words, "There is over 17,000 feet of this pipe installed at all three floodgate sites. All of it is unfit for the service it will see. All of it is prone to bursting under normal service. How this could have slipped through so many layers of review is unknown, but I can tell you that if there was a "Fitness for Service" declaration signed, it must be either fraudulent or signed under duress. I bet that the welders working these jobs probably asked for some kind of release from liability, so they wouldn't be blamed when the pipes burst. This is the worst kind of engineering mistake......Put simply, all of the pumping systems, as designed, procured, and built, have a strong likelihood of failure when placed under normal operating conditions."

    You were instrumental post Katrina in getting things done. That is why I am contacting you. Please look into this matter. We cannot have New Orleans flood again due to this kind of criminal behaviour. Thank you."

    We must all take action on this. It's almost too late.

    By Anonymous judyb, at April 05, 2007 11:02 AM  

  • Hey, I hate to throw a monkey wrench into the works, but there is still a CHANCE they didn't completely muck things up.

    The B.31 standards are all for sizing pipe. I deal with them quite a bit.

    B.31.1 is Power/Industrial stations, etc.

    B.31.4 is for liquid hydrocarbons

    B.31.8 is for gas

    B.31.3 is for process fluids. It's by far the most conservative of the bunch. Except for rare instances (like ordering 10 miles of gas pipeline), all (good) engineers always use B.31.3 for all short lengths of pipe. It's the most "lawyer-proof."

    Most companies have their own pre-existing pipe specs. Shell's is Schedule Q (or P, I can't remember off the top of my head). It's surprising the Corps doesn't have their own.

    You the Corps specified (or at least seems to specify... they aren't clear) sizing per B.31.3, not B.31.1.
    http://www.usace.army.mil/publications/eng-manuals/em1110-1-4008/c-3.pdf


    Anyway, I crunched the numbers with B.31.3 and Grade B pipe and the MAWP came out WAY short, but I resized it for X42 and X60 pipe (the next most common in my experience) and X60 pipe does come out to a MAWP of 2994 psig. That's still a little short, but close.

    The real deciding factor is what is the maximum pressure (Total Dead Head) that the hydraulic power pumps can put out. That's what the pipes should be sized on. Any clue on the model number and whatnot?

    Also, what corrosion allowance did you use?

    By Blogger Clay, at April 05, 2007 7:34 PM  

  • I have provided another entry on this topic which incorporates some of the comments I've received. You can find here.

    By Blogger mcbrid35, at April 19, 2007 5:17 PM  

  • Hey Matt,
    Hope to hear a report from you on today's heavy rain. Hearing reports of widespread street flooding. Any evaluations on canal levels today?

    By Anonymous Laureen, at May 04, 2007 2:07 PM  

Post a Comment

<< Home

Go to older posts Go to newer posts