Gulf Oil Spill Velocimetry-Based Flow Rate Estimate

  1. russ_watters

    Staff: Mentor

    In the thread in P&WA on the issue of the oil spill, I expressed my skepticism that the rate of the spill could be accurately measured via analysis of underwater video clips. That skepticism was strengthened by the recent statement by Prof Steven Wereley, the originator of the first such estimate, that his earlier estimate was "consderably" low, despite his previous claim of a 20% margin for error. So I'd like to gain a better understanding of his method and if it is reasonable to believe it can be accurate in this case.

    For background information....
    Here is the wiki on particle image velocimetry:
    Here is the Wereley's 1998 patent on the subject:
    Here is Wereley's 2006 patent on the subject:
  2. jcsd
  3. Well here is the problem so far I have only seen clips from one of the places the oil is leaking and it looks like a ton of oil is coming out. However I heard from cnn that there are multiple places the oil is leaking from. If I could see all the places the oil is coming from I would just add about how much oil is coming out at any moment and then just do the math to figure out how much should have came out factoring for that the oil probably came out faster as more pressure when the oilfield first started to leak. I mean im not totally sure the people in charge of this even know the main places it's leaking. I heard someone suggest droping an atomic bomb down there to try and seal it up however at that pressure im not sure what an atom bomb will do to seal anything. I saw a video on how to extract oil from water with straw. It might be a good idea to buy some straw for your drinking water in the far future depending on how bad this gets.
  4. russ_watters

    Staff: Mentor

    The primary source of skepticism about this method could be accurately applied to the gulf oil spill is the quality of the video. The video is low resolution and presumably shot at a pretty high shutter speed due to the depth, which presumably would make it very difficult to lock-on to particles and track them to measure their velocity.

    Here's the primary claim from the first patent:
    This says that the method uses depth of field effects of photography to ensure you can focus on a thin plane of particles injected into the fluid stream and a strobe light operating on the nanosecond range to illuminate discrete particles. Neither of those conditions are available in this video and they seem to me to be pretty important and specific requirements.

    The 2003 patent appears to me to be a microscopic version of the same thing.

    From a company specializing in the technique:
  5. If I could get my hands on the size of the oilfield I could probably just figure it out by factoring for the size of the pipe the oil travels out of with all pressures and stuff...
  6. mheslep

    mheslep 3,401
    Gold Member

    From the 2006 patent:
    So the patented method is not applicable, as it requires active controlled illumination.
  7. Perhaps you could see he would be willing to address these concerns here? You never know, and his info is available at the Purdue U site . Divining what his methodology is from, a 12 and 4 year old patent respectively seems to be more amusing than scientific.

    How does this square with a similar estimate using pencil and paper by Chiang? In addition, when we know that surface estimates are meaningless in the presence of deep plumes, and the use of dispersants at such a depth?

    On the technical side, mheslep is simply wrong, in that the method requires this laser pulse. It is PREFERABLE, and necessary to achieve the desired sub-pixel accuracy. Note then, that Wereley's figures differ from controlled PIV confidence in assemblymen by orders of magnitude.

    Clearly this was not achieved in the absence of markers added, or a controlled pulse or binocular setup. That reduces accuracy, but does not necessarily invalidate the estimate as offered, nor the conclusion that surface estimates are grossly misleading by comparison. Perhaps if BP or the USCG would allow the addition of a small number of markers, this could be achieved. One wonders why this has yet to occur given that we are hours away from this having been an issue for a month.
    Last edited: May 19, 2010
  8. I'm new here and unfamiliar with the rules on specificity to subject matter being discussed so please excuse me for introducing a tangential question about the oil leak. I recently asked this question in Earth forum but it seems I should have asked it here.
    According to news media sources the vast majority of the oil is either slowly settling to the bottom, billowing around at various depths, or slowly surfacing. Nobody seems to know.
    I've been thinking that the potential impact of the leak would depend on the answer.
    I've been wondering how much the density of the crude is affected by the pressures at that depth. These long chain hydrocarbons are subject to compressibility so density should increase. But, enough to keep the stuff down there?
    My tech college math skills are thirty years unused and not in the engineering field, so basically I'm a layman.

    came across this but I'm unsure of how to apply it, or even whether it's pertinent to the question:

    Can anyone tell how the pressure is affecting the density of the stuff?
  9. Q_Goest

    Q_Goest 2,989
    Science Advisor
    Homework Helper
    Gold Member

    Hi gwangi,
    Interesting question. Petroleum is of course, a mixture of various hydrocarbons, and I have no idea what hydrocarbons are being released but I suspect it is fair to say they are relatively heavy with carbon chains in the neighborhood of 15 to 25 carbons. Let's assume the carbon chains follow the general rule CnH2n+2 and look at a couple of different hydrocarbons at 2500 psia and 40 F. There are reports of tar balls washing ashore, so I'll include an even heavier hydrocarbon. Here's what the density looks like:
    C16H34: 34 lb/ft3
    C24H50: 50 lb/ft3
    C30H62: 52 lb/ft3

    From this, it looks like these hydrocarbons are less dense than seawater, so they should come to the surface. Why is there a layer of these hydrocarbons reported deep under the surface? I don't know...
  10. The temperatures and introduction of dispersants have been implicated, but it is still uncertain. There seems to be the range from "sheen" to "tar" and the mayonnaise-like slick in between. Hopefully congress will demand a public analysis of what BP is siphoning so we can have a better idea of how to answer these questions. The disaster has occurred, we must at least learn from this.
  11. mheslep

    mheslep 3,401
    Gold Member

    Apparently naphthenes are the majority share of petroleum; they follow CnH2(n+1-g) with g the number of H rings, and would thus be a little less dense - C24H48 and so on.
  12. mheslep

    mheslep 3,401
    Gold Member

  13. russ_watters

    Staff: Mentor

    From my post in the politics forum, here are the differences I see between the measurement methods he's patented and what is available to him with the gulf spill:

    1. Low-speed photography, using a standard camera.
    2. Poor light source (no laser or strobe light).
    3. Opaque fluid.
    4. No pre-selected, suspended particles.
    5. No specialized depth-of-field focusing.
    6. Uncertain opening size (even if he got the size right, breaking-off a pipe can change the geometry of the outlet).
    7. Unstable flow.
    8. Uncertain and inconsistent mixture of liquid and gas.
    9. Uncertain camera angle.
  14. russ_watters

    Staff: Mentor

    I'm trying to understand/assess the method he used to measure the flow rate. That is the purpose of this thread. A link to his homepage is not helpful.
    Do you have a source for that?
    Do you have a source for that?

    What I'm searching for myself is for any reference that says it is possible to do the measurements without such features, not personal opinions of laypeople that it should still be possible.
  15. There are at least 8 cameras in place, and there is a binocular method. Your reasons certainly explain why a method that usually has significant sub-pixel resolution yielded a result with more leeway than accuracy, but it does not invalidate the method. The presentation by the media has been flawed, and one can argue that Wereley is to blame for that in part if that is pleasing.

    I still wonder, with so many ROVs and so much access, why is the US government, BP, and other not using test particles in the effluent and reaching averages? This would seem to be critical in any attempt to back-fill, given the need to understand the pressures involved.
  16. russ_watters

    Staff: Mentor

    What method? I've found no reference anywhere to a method existing to do what is being suggested Wereley did. Do you have one?
    What method? Flawed how? Can you explain exactly what he's doing and if there is any precedent for the method?
  17. I'm afraid that without you having read the relevant book, and piracy being illegal, it would be difficult. For the flaws in the media presentation, your own posts have been more than adequate in that regard, in the 'politics' thread as you say it.
  18. The method does not image single particles nor track the movement of single particles.

    The field of view is segmented and Auto-correlation is performed on Fourier transforms of "snapshots" of assemblies of particles at different times to extract flow fields - so the particles are used as markers for the bulk fluid flow.

    In this case, I assume, the frames are used instead of laser flash illumination.

    Auto-correlation in Fourier space has a similar effect to confocal imaging, it decreases the depth of focus of the conventional image (preferentially weights the contribution of particles that are exatly at the plane of focus).
  19. mheslep

    mheslep 3,401
    Gold Member

    Regarding the eight cameras, I'd not heard that, but Wereley's stated he made his estimate from the same single grainy video we have all seen.

    Without an operable valve upstream of the leak, and a pump that can over come the leak pressure, inserting any tracers into the flow must be nearly impossible.
  20. russ_watters

    Staff: Mentor

    I've found a copy of a ppt presentation Wereley gave on the subject, on the US House website.

    Of note:
    -He used a single, 15fps internet video for his analysis.
    -He matched features and counted pixels, apparently manually, from one frame to the next.
    -He scaled based on the 21" shaft size (not the 9" riser size).

    Based on his ppt presentation, his method - as I speculated before - bears only superficial resemblance to PIV and really isn't any more sophisticated than any armchair observer could do. There ws no special skill involved in this estimate.

    Also, another potential source of error I didn't think of before but can see in his ppt presentation:
    10. Buoyancy of oil and gas provides an upward component of velocity immediately after leaving the pipe.
  21. russ_watters

    Staff: Mentor

    Welcome to PF.
    Understood, but I would say this is a difference between the typical method of PIV and what Wereley did. He manually identified features and tracked them across frames of video.
Know someone interested in this topic? Share this thead via email, Google+, Twitter, or Facebook

Have something to add?