shelburn Deepwater Horizon – A Possible History of the Leak

Posted: May 21st, 2010 by: h2

Shelburn came up with another excellent overview/analysis article, buried in the comments section of the The Gulf Deepwater Oil Spill, sheen, other oil layers, and RIT flows thread.

Shelburn is an oil industry guy, ROV was/is his specialty if I remember right. Great stuff, with ROCKMAN, there’s a lot of good information coming out, and good interpretation of what’s going on now. Fairly objective I’d say too, not everything is 100% but the effort is clearly being made to be as clear as possible.

(Update: just published on theoildrum: What caused the Deepwater Horizon disaster? by aeberman)

shelburn on May 20, 2010 – 4:10pm Permalink | Subthread | Comments top


The original survey, performed shortly after the rig sunk, did not locate any oil leaking – at that time.

Comments have been made that the bottom was obscured with mud and that is quite possible, the rig and riser reaching bottom would have stirred up a lot of the very soft mud.

But ROVs don’t depend on video for locating oil leaks, they use sector scan sonar which is standard equipment on any work class oilfield ROV. Even a small oil leak will show up like fireworks on a sonar screen, even if the optical visibility is zero.

I think it is a reasonable assumption that there was an initially a very small area of leakage, literally a pinhole, most likely inside the BOP. It is also possible that the kink in the riser, and presumably a kinked or broken piece of drill pipe inside the riser, was holding back some or most of the flow.

The 21 inch riser has enough volume to hold about 2,000 barrels of oil so a slow leak would take quite a bit of time to fill it, as much as 2 days if you assumed an initial 1,000 bpd leak rate.

Some hours and days after that initial survey other inspections started locating oil flowing from three places.

1 – A video of a small oil flow from the end of a drill pipe got extensive media coverage. This was the smallest leak and was capped in a few days.

2 – A much larger flow was located at the broken end of the 21 inch riser. That outflow is about 600 feet from the BOP but the oil travels through about 4,000 to 5,000 feet of riser to get there. The riser is kinked a couple feet above the BOP and then descends to the seabed for a distance before rising off the seabed in a long loop. Originally the top of that loop was 1,200 feet above the sea bed and has gradually descended. Several days ago it was reported to be 300 feet above the bottom. I don’t know where it is now. After the loop the riser leads back towards the BOP until it reaches the broken end which was partially buried in the mud.

3 – A small leak under some pressure was observed where the riser was kinked at the BOP. An early and poor quality still photos does not seem to show any leakage while videos released in the last few days show a very substantial flow from several leak points, some with pressure behind them.

Any place there is a reduction causing high velocity flow you will have erosion. The speed of that erosion depends of a number of factors. The primary factors are velocity (which is generated by the pressure differential), the material being eroded and any entrained solids like sand in the flow. Other factors can also contribute such as temperature, the corrosive value of the flow, etc. In this case there is almost certainly some amount of sand in the flow which would act just like a sand blaster.

A small orifice will erode more quickly that a large one as more of the flow comes in direct contact with the material, but as long as the stream is running the erosion will continue and the leak area will keep growing.

It is therefore logical that the leak would start small, grow rapidly and over time the growth rate would slow down as the pressure differential and the velocity slow.

The leak at the kink has continued to erode and it has certainly gotten larger. A week or two ago the ratio between the two leaks was described as 85% out of the riser end and 15% at the kink. As time goes on that ratio has been changing so the kink will continue to leak a larger portion of the total oil and gas.

In summary, the leak started small and grew as the leak point(s) eroded. It will continue to grow until it reaches an equilibrium with the amount of oil and gas the formation is capable of producing. I doubt that equilibrium will be reached for many months as there still seems to be a substantial pressure differential between the pressure below the BOP and above the BOP. The last information I saw was several days ago but it seemed that the pressure differential was about 6,000 psi.


From day one I have been intensely curious about what BP would say about the flow rate, so I have specifically looked for BP to make a statement. I have never seen anything where BP actually stated any flow rate or range of flow rates. Every statement I have seen, often attributed to BP by the media, has actually traced back to a statement by the USCG, NOAA, or other government agency making estimates based on the spill size.

The only thing BP has said, as far as I can find, is a couple days after the 5,000 bpd announcement, when pressed, a BP spokesman basically said “that number is as good as any other”. They have also reacted to media statements that the flow rate is 70,000 or more bpd as being inaccurate.

In Congressional testimony BP mentioned a maximum rate of 60,000 bpd but I believe the phrasing of the question (conveniently eliminated by most of the media) was what was the theoretical absolute maximum rate of flow if the well was totally unrestricted – no BOP, no downhole obstructions, etc. I think in later testimony they reduced that to a maximum of 55,000 bpd.

Historically the largest completely open well gushers have all maxed out at about 100,000 bpd and usually dropped off within a few days. The Lakeview gusher (California 1910) was estimated to reach 100,000 bpd, ran unchecked for 18 months, but over that time “only” averaged between 15,000 and 20,000 bpd.

Gulf of Mexico wells have never been capable of delivering the flow rates of the largest onshore wells in the Middle East, Texas or California. The maximum perforated and controlled flow rates have been about 40,000 bpd so it kind of requires a suspension of reality when people outside the industry start talking about flow rates of 70,000 to 100,000 bpd (and higher) through a well that does have restrictions. Even if the well was totally unrestricted, ie: no pressure reduction in the BOP, it is unlikely the well is physically capable of producing over 60,000 bpd.

From the above I doubt that the flow is much above 30,000 bpd, maybe up to 40,000 bpd.

At the same time I think it has been over 5,000 bpd and growing since at least the end of April. The videos shown in the last hour would seem to absolutely confirm this. There is still a substantial flow at the riser where BP says they are recovering 5,000 bpd – plus the leak at the kink.

A low range would seem to be in the area of 10,000 to 15,000 bpd.

My best guesstimate – more than 10,000 bpd; less than 40,000 bpd.

The accuracy of estimates of blowouts has always been problematic. Even in blowouts that have been extensively studied after the fact the minimum and maximum reasonable estimates seems to be about 50% and 200% of the median.

Trying to get an accurate volume from a short piece of video is also full of problems. To me, the statement that it is within 20% shows that the scientist involved does not fully appreciate these problems.

I am not a piping or process engineer and my fluid dynamics education was decades ago but I would love to know if he considered some of the following:

– He did state that this is mixed phase flow and that he had worked with that before but was his science based on flow at 2,250 psi ambient? I know that gases can have substantial changes in properties when under pressure and some liquids also.

– A portion of the gas is probably going into phase change as it combines with the seawater which might increase its real or apparent volume and velocity.

– Did he adjust for the optical distortion of the wide angle lens used by the ROV? As objects move out from the center of the lens they elongate, or if in motion seems to accelerate. If you are doing a pixel based measurement you would need to know the lens distortion and adjust for it. If he was measuring off center someplace where the elongation 1s 25% then that would already be outside his stated error range if he did not adjust for it. The human eye seems to adapt for surprisingly large variations in optic distortions unless a grid is overlaid on the picture.

– Other videos from the opposite side of the riser leak show a piece of dill pipe bent at about 90 degrees partially blocking the bottom of the riser. Not only does this reduce the flow area in the bottom half of the riser but also is positioned so it appears that it diverts the flow towards the side that he used to measure particle velocity. That diversion could accelerate the velocity which seems to be the entire basis of his calculation.

– As subsequent 5 minute videos show the flow is not a steady state but seems to vary, especially the gas flow. If you consider the length and layout of the riser this make sense as the riser acts a bit like a separator (note the gas separation and also the lack of separated gas at the kink) and as it has high places there is a real potential for gas to collect and cause the flow to vary even if it was entering the riser at a steady flow. Longer segments of video, as now available, would be more accurate.

I do believe BP knows much more than they are telling about what the flow rate actually is. There are some real solid technical reasons they need that information.

In designing for the top kill and/or junk shot it is really vital to the decision making. In fact, the approach they are now taking of trying the top kill before the junk shot could be a clue that the flow rate is not as great as they originally thought (we have no idea what their original estimate was) as they have to be able to pump the mud against that flow.

And I think they would be much better off if they had been open and communicated more information from the beginning. The impression that they are hiding information will persist long after the well is killed and (most of) the spill is cleaned up.

A little further downthread shelburn explains how the mud injection system will work, including the difficulties he knows about, and how the stuff gets built and tested and installed. Interesting stuff.

xazp on May 20, 2010 – 7:27pm Permalink | Subthread | Parent | [Parent subthread ] Comments top

Thanks for your comments. In the end of your comment, you talk about the top kill/junk shot procedure, and I was wondering if you (or the TOD staff) might take a stab at some of the issues BP may encounter with the approach. From reading media reports, it’s hard to understand why “dumping mud down a pipe” would take weeks; so I was wondering if you could educate us on some of the preparations/precautions that might be involved with doing this at 5,000 feet. I’d find it educational and it’d probably help me gain an understanding of some of the challenges BP is working through.

shelburn on May 21, 2010 – 1:26am Permalink | Subthread | Parent | [Parent subthread ] Comments top

I think there are some risks to the casing and BOP from this operation but I don’t have the expertise to talk about them in detail. Maybe some of the actual experts could comment.

From the operational end there are some real bottle necks.

I think from what I have read that BP has assembled most if not all of the topside equipment. I heard about mud pumps totaling 30,000 horsepower, 50,000 or more barrels of mud, etc. That seems to work out to over 15,000 TONS of mud.

A lot of the underwater equipment is probably being engineered and built from scratch. The manifold would be an example.

I know the choke and kill lines were damaged and it was announced they had been cut by the ROVs. Just as an example of what has to happen lets take the task of connecting those lines to the hoses delivering the mud.

The choke and kill lines are at least 3 inch diameter and capable of handling 15,000 psi or more. It is difficult for an average person whose hydraulic experience is usually limited to his home plumbing and garden hose to comprehend what 15,000 psi means in real life.

If you go to your local hardware store the largest valve they will probably stock is a 2 inch PVC or brass valve weighing about 2 to 5 pounds. If you live near an oil patch you can get a 600 psi valve weighing about 10 lbs pretty easily.

A 2 inch valve rated at 15,000 psi is a special made item that probably weighs about 300 to 500 pounds and has a lead time of several months. They are attached to the piping in a shop with specialized welding, 100% X-rayed and have to pass a number of other tests.

In this case we are talking about a 3 inch (or larger) fitting that has to be installed by an ROV, clamped to the line and pass a pressure test before it is used. I will wager the each fitting will weigh at least a half ton, it will be machined from a solid block of specially metal, probably titanium or specialty stainless steel. It will have a variety of toothed slips, smooth slips, seals, actuation pistons, etc. all machined to extremely high tolerance. The fitting must grip the pipe tight enough it can withstand a 100 ton load trying to pull the fitting off (due to the internal pressure), seal without leakage against 15,000 psi, and still not damage the pipe.

The normal lead time to engineer, manufacture a prototype, test it and manufacture the final fittings (assuming the prototype passes all the tests) would usually be about six months or longer – as a rush job.

After the fittings are manufactured they will probably perform a SIT (Systems Integration Test) where they model as closely as possible the actual installation procedure with all the components.

In this case I expect there are at least 40 or more people working in shifts around the clock just doing these fittings. Machine shops will be keeping their best people on standby to immediately start work when the engineers have drawings ready. If this is the critical path item BP will have a jet ready to fly them from Houston to the heliport, or if they are too heavy for a chopper a fast crewboat will be at the dock with engines ticking over.

All of this will be supervised by a variety of safety inspectors from BP and in this case certainly the MMS and probably the USCG. A number of engineers from different companies will review the final design trying to find any potential flaw.

I know the public perception right now it that offshore oil people are a bunch of cowboys cutting corners whenever possible but nothing could be further from the truth. When you are working at these depths, pressures and volumes any mistake can have serious results. The industry has learned over the years that safety, redundancy and intricate planning leads to successful operations – and successful operations are profitable operations. Unsafe operations can have disastrous consequences – as we have just seen.

The fitting I have described is probably the most difficult individual item required to go ahead with the top kill or junk shot, but there are several hundred other items and pieces of equipment required. It really is a major undertaking.

shelburn on May 21, 2010 – 1:43am Permalink | Subthread | Parent | [Parent subthread ] Comments top

I forgot to mention that while all the engineering and logistics are going on the ROV companies will have a virtual reality model of the BOP, riser and all the components set up and the ROV pilots who will do the actual installation are probably already practicing and refining the operation. These models are extremely realistic and are very similar to a commercial flight simulator. As they accurately model the ROV and the task they can often identify problems before the actual operation takes place.

shelburn on May 21, 2010 – 11:00am Permalink | Subthread | Parent | [Parent subthread ] Comments top

The 15,000 psi is the working pressure of the BOP, the designs have to have safety factor and the prototype must pass pressure tests well in excess of 15,000 psi.

The last report of the actual well head pressure was “8,000 to 9,000 psi”. That pressure will increase when they pump in the mud. There were reports that during the blowout surge the recorded pressures reached as much as 30,000 psi. I haven’t heard any recent confirmation of that report but if true it will probably figure heavily in the writing of new regulations and may have damaged the BOP.

Hypothetical calculations – based on well 13,000 feet below the BOP and a formation pressure of 13,000 psi.
Well full of gas – pressure on BOP about 13,000 psi
Well full of oil – pressure on BOP about 8,250 psi
Well full of 18.5 lb mud – pressure on BOP 0 psi = no flow

I’m sure that BP and all their consultants are extremely concerned about the possibility of damage to the BOP and possibly increasing the flow and/or making it impossible to take any other remedial action until one of the relief wells is successfully completed. It would not surprise me, in fact you can bet, that there is a contingent that is advising BP that to do nothing is the safest course of action.

newdood on May 21, 2010 – 11:07am Permalink | Subthread | Parent | [Parent subthread ] Comments top

Oh. Damn.

I’ve dealt with valves for natural gas service, but the “high-pressure” valves in the range I’m familiar with were ANSI 600#

The question that now jumps into my head is about the flexible hoses that are supposed to connect the new choke/kill manifold to the BOP stack. I’ve never HEARD of flexible tubing rated for ANYWHERE CLOSE to these pressures. Is that just further evidence of my inexperience?

From what I can put together, it sounds like there are shutoff valves at the kill/choke ports rated for the full 15000 psi pressure, but they know (hope?) that the actual pressure inside the stack is considerably lower.

Maybe the fitting to attach the manifold hose to the stack has to be designed/tested for ONLY 8000 or 10,000 psi.
– With an orifice large enough to pass the “junk”
– And a connection to the BOP that can be handled by an ROV

And you’re STILL gonna tell me they can’t run down to the local Home Depot and pull it off the shelf!!?

shelburn on May 21, 2010 – 11:29am Permalink | Subthread | Parent | [Parent subthread ] Comments top

The Coflexip hoses to hook up the mud lines to the BOP have the same relationship to your garden hose as the several hundred pound 2 inch valve has to a ANSI 600# valve weighing about 10 – 15 pounds.

Fortunately these kinds of Coflexip hoses are used quite a bit around deep wells so they are available.

There is a more than reasonable chance that when they do the top kill the pressures will reach 15,000 psi, if not more. They have to have a substantial pressure over the internal pressure of the well to force enough mud through two 3 inch lines to overcome the flow, which seems to be increasing every day.

Lots of details we don’t know that make a huge difference. Is the leakage going through the annular space? Is the flow entering the drill pipe some distance below the BOP? Is the flow going directly from the liner to the BOP? BP may have an answer to some of these questions due to the inspection of the BOP but I doubt they have any definitive information of the condition of the formation, cement, casing and liner at the bottom of the well.

Comments are closed.