Deepwater Oil Spill - the Congressional Illustrations

One of the considerable problems that arise in writing about the Deepwater Horizon oil spill in the Gulf of Mexico comes in determining what information that is given is actually accurate. I’m going to assume however, that when a company provides information to Congress that it is, to the best of their knowledge, correct. Since the illustrations that accompanied the testimony of BP officials is now available, albeit in draft form, I am going to summarize some of this so that, as discussion and later posts require, it can serve as a useful point to find reference material. This information was presented the Energy and Commerce Committee of the House, on May 24th. (Note that the illustrations have been condensed to fit on the page, clicking on the image should lead to a larger view on a separate page).

UPDATE: 9:00 pm Monday In regard to the metal cutting post just before this one, they have lowered the big shearing cutter down to the riser, and cut off the power cables, so I suspect that the first shear may be sooner rather than later, END UPDATE

There are several points that I want to highlight, and so I will be taking selected parts of the BP slides to record separate pieces of information. The first of these is where the actual pay zones of oil and gas are, relative to the total depth of the well, and what the pressures are within them. Notice that the pressure at the bottom of the well is about a thousand psi lower than I had been using in earlier posts.


There are several different parts of the overall structure for which more information is available in these slides (and my many thanks to several folk that have pointed me to their existence).

Consider first the Blow-Out Preventer, on which I have focused recently. The exact nature of the different rams is not the same as that illustrated by the paper from which I took the illustration.


The various components of the BOP, from the top are:

UAP – Upper Annular Preventer, used in normal drilling operations for well shut-in, rated at 10,000 psi
LAP – Lower Annular Preventer - down rated to lower wellbore retaining pressure to 5,000 psi.
BSR – Blind Shear Rams – cuts through the pipe and seals the well. 2 rams, one on either side of the drill pipe (DP).
CSR – Casing Shear Rams – non sealing, cut drill pipe and casing, but not designed to seal the well.

The three variable bore rams (VBO) would normally be the following with the functions designated:
UPR – Upper Pipe Rams – Ram packers can close on a range of drill pipe, from 3.5” OD to 6 5/8” OD and seal at 15,000 psi well bore pressure.
MPR – Middle Pipe Rams - Ram packers can close on a range of drill pipe, from 3.5” OD to 6 5/8” OD and seal at 15,000 psi well bore pressure., can also be used to hand-off drill pipe to a maximum weight of 600,000 lb.
LPR – Lower Pipe Rams – A Test Ram that can hold up to 15,000 psi.

The lowest ram did not function, I believe, and there was evidence of hydraulic leakage which has raised questions as to whether, or which, of the other rams functioned. However, as a part of the protocol for the positive pressure test, the drill pipe is lifted out of the BOP, and the BSR is closed to seal the well, while fluid is pumped into the well through the kill line, to raise the pressure and ensure that the cement job had held pressure. That test was successfully completed.


Following the positive pressure test the drill pipe is lowered back into the hole, note that it is in three sizes, as it goes down the well, with the upper segment at a diameter of 6 5/8” , and the lowest at a diameter of 3.5” bextending down to 8,367 ft below the rig.


To separate the seawater that is then going to be pumped into the riser a special, heavy mud known as the Lost Circulation Material (LCM) which has a weight of 16 pounds per gallon (ppg) is injected into the well, through the DP, and it circulates from the bottom of the DP back up to the BOP. The choke and kill lines had been filled with seawater, and the kill line pressure gage was still reading 1200 psi. (This takes the time to 16:28 on the 20th April. The green color is 14 ppg mud, the pink the 454 barrels of 16 ppg LCM, and the blue is seawater at 8.6 ppg.


The sea water is pumped into the well and circulated behind the spacing LCM, material. Once it has filled the drill pipe, and started to rise back up the production casing, the pressure rating of the water in the DP is at 2325 psi, and that in the kill line is 1200 psi. The annular seal is closed around the BP, and the pump is switched off. The pressure in the DP (on the rig floor I imagine) is reduced to 1200 psi, to equate with that of the kill line. Once the two are at the same pressure the kill valve at the BOP is opened, but this is where, at around 5 pm the first signs of a problem appear. The fluid in the well, for the first time, exerts less pressure on the fluid in the rock, than the value of that pressure.


The annular that is leaking is, presumably, the annular seal at the top of the BOP. The term “U-tube” I believe refers to the fact that there is not a pressure balance at the point it is measured. At this point there are two separate scenarios as to how fluid starts to migrate into the well, as the pressure in the DP is bled back to zero. It is now 17:52. Note that the oil/gas flow is shown as an orange arrow and in either case has a path to the bottom of the DP.


Pressure in the well now starts to rise, fluid is flowing from the kill line (about 3 – 15 bbls, the kill line was closed. Pressure in the DP was monitored and held steady at 1,400 psi, This was the negative pressure test, at 19:55 the decision to continue pumping seawater was implemented, at 20:58 it was noted that the flow out was greater than the flow in. The flow meter indicates that the well is flowing. Pressure in the DP continues to build, even after the pumps are shut down.


The first indications of this flow was 51 minutes before the explosion occurred. Seawater displacement restarted, at 21:14 and at 21:56 the Emergency Disconnect System (EDS) was activated, which should have led the shear rams to activate, and released the riser from the well. It did not work. The gage results leading up to the explosion were provided.


The activation time for the EDS should be 46 seconds. In this time, after the EDS button is pushed, the blind shear rams close, cutting the drill pipe and sealing the well.
The choke and kill line valves are closed and the lines unlatched
The LMRP is unlatched and disconnects
The sequence is complete and the rig can move away from the well.

There is no evidence that the EDS fully activated. Even if hydraulic power to the BOP had been lost there was a second procedure, the AMF, that should have sensed the loss in power, communications and hydraulics and closed the blind shear rams. But it required all three to have been lost to activate. It should take 37 seconds. There is no evidence that it fully activated.

The subsequent use of ultrasonics and gamma ray examination suggests that the blind shear and VBRs may have functioned, and could have reached the locked (closed?) position.

The testimony also showed a simplified version of how the shoe at the bottom of the casing should act to stop re-entry of fluids into the bottom of the production casing. The flappers stay open as first mud, and then cement is pumped down the well. Once the plug reaches the well, and all the cement has been ejected, then the flappers swing back horizontal, sealing the well. (The testimony pictures has the flapper in the wrong place in the middle figure so I deleted it from this copy).

Deepwater Oil Spill - cutting metal underwater

The decision to cut the riser and drill pipe (DP) from the top of the Blow-Out Preventer (BOP) as a first step in putting the Lower Marine Riser Package (LMRP) over the leaking oil well in the Gulf of Mexico is not quite as easy as it might sound to a layman. This post is going to talk a little about a couple of the problems, as well as an alternate way of doing it, and should end with a possibly slightly amusing anecdote.

The current plan is to carry out the operation in two parts, first the main body of the riser and contained BP will be cut using a large shearing machine, and then a precision cut will be made with a diamond wire cutter to prepare the surface to act as a support and seal for the LMRP. (Illustrations are below the fold). There are a couple of reasons why this is going to be done this way, and one or two concerns that will need to be watched as the operation continues.

Part of the problem lies in the considerable stress that the metal is under, which may make it more difficult to get the precise cut that is needed for the sealing surface. The weight of the collapsed riser on the kinked section of the riser means that the metal is under a considerable stress, and as the cut is being made that stress is going to be released. This is a particular problem if the saw blade is in the cut as that stress releases. As an illustration large diamond wheels are used in granite quarrying as a way of making the first cut in a new layer of rock. As the quarry works down into the deposit, however, the stress in the rock increases. So that when the diamond saw cuts into the layer, the walls will move in slightly. Now if the diamond saw is rotating until it is removed, it will grind away that small movement and there is no problem. But if the saw is removed, and the cut restarted the next day, then it might be necessary to recut the same depth again, because the walls have moved in that far. The much worse thing to do is to leave the diamond saw in the cut overnight – the walls will move in and pinch the blade and you can’t get it back out without destroying it. Did I mention these blades can cost up to $100,000.

So given that it would be a bad idea to break the diamond wire in the cut, or trap it for that matter, the first step in the process is to take some of the load from the riser by cutting off the majority of the fallen pipe. To make this simpler and since the quality of the cut is not important, this will be done with a large shearing machine.

Large shearing machine used to cut through the main section of the riser (Unified Command )

For the second stage in the process the riser and pipe will be cut, at the top of the BOP using a wire saw to cut through both pipes. The diamond wire, which is held under tension in the design, should give a relatively smooth surface that can be used for the sealing surface to the LMRP.

One of the problems in using this tool, which is rather sensitive (in other circumstances we have broken several wires in a smaller model) is dealing with the central DP as the cut is being made, and with the flow of oil and gas past the wire, which can send it into vibration, over the relatively long cut.

Wire sawing the riser and drill pipe (Unified Command )

The two pulleys at the top of the saw are used, as well as guides, as tools to keep the wire under the correct level of tension.

A version of an underwater wire saw (Wachs)

It is relatively slow to cut, (perhaps 10 to 30 sq ft/ hour, depending on power and blade size) and requires patience to keep the applied loads at the right level (the joy of automatic equipment over manual operation). There are still possible risks, due to the remaining load on the metal being cut as it is being made, since the amount of metal left gets less over time. As a result it is likely that, in the same way as a forester will use wedges to stop the cut closing as he chops down a tree, so small metal wedges will be put into the cut in this case. The only difficulty that this imposes is that it will be difficult to access the central BP and difficult to deal with that, once cut through, since it is within the riser and not accessible. Hopefully it will be held in place by the bend of the pipe and riser, until the wire has passed through it.

Over the past twenty years diamond wire sawing has become more common as a way of cutting through a variety of materials. It has the advantage of making a relatively narrow cut , and the ability to cut through objects made of different materials, such as reinforced concrete. The cutting wire used to have small diamond particles embedded in the steel of the wire, but in more modern varieties the diamonds are embedded in a plastic such as Teflon, and set as beads along the wire, allowing better cooling during the cutting process.

Segmented diamond wire – new so that the paint is still over the diamonds (Wachs)

Now my students will tell you that I have a bad habit of slipping commercials for another technology into lectures, and so I am going to indulge, and mention that there is an alternate technique that could be used instead. And that is the use of a high pressure abrasive laden, waterjet cutting stream. These are becoming increasingly used in industrial manufacturing, although in that use the technique usually involves a different way of adding the abrasive to that which would be used in the current problem. In the conventional way of adding abrasive it is mixed into the jet after the water is accelerated, and, in the right circumstances, we have been able to cut through half-inch titanium with an accuracy of 0.001 inches. Those jet systems operate at pressures from 40,000 psi to about 90,000 psi. In the current case the abrasive is added to the water from a pressure vessel, and the technique is known as abrasive slurry jetting (ASJ). It allows equivalent cutting at about a tenth of the jet pressure (i.e. 5 - 10,000 psi). The tool has been developed for cutting off different oilwell parts from the wells under the North Sea. Because the jet cuts away from the nozzle, it can cut, for example, out from the center of a set of drill pipes, one nested inside the other, and separate them, as one of the last stages in decommissioning a well. It has a problem with range, the jet cutting to a shorter distance as the back pressure in the water increases, and for cutting down to 500 ft (the deepest we have tried) it can be improved to cut large risers by adding an air sheath around the outside of the jet.

One alternate way that is used for cutting tanks and other containers with possibly explosive content is to use a chain crawler around the pipe. This way the jet does not have to cut quite as far through the water.

Cutting a pipe using a crawler (NLB )

(And for those who question whether jets would work under this depth, Gulf Oil – as was – used an abrasive waterjet to drill a well down to 15,000 ft back in the early 1970’s but a fire destroyed the equipment and the research was stopped).

To end with the anecdote – which relates to the cleaning of oil rigs for inspection. The problem that arises is that barnacles and a variety of sea creatures and plants grow on these structures, and need to be removed. The historic method of doing this was to give a diver a chipping hammer, and send him down to chip off the offending coating. Now when a diver is in the sea there is not a lot of resistance, and so one has to clamp onto something to develop enough thrust to effectively use the hammer. The only object available is the rig itself, and wrapping one’s legs around a strut works to give the platform. The only problem is that the strut is covered in barnacles – ouch!

So the waterjet industry developed a zero-thrust gun (in other words the cutting jet on one end is matched by a jet of similar power on the other, so the forces are balanced) and the diver does not need to hold onto anything. It was taken out to one of the rigs in the North Sea and given to one of the divers.

He took it, without a word, and disappeared over the side of the rig. He was down underwater using it for about fifteen minutes and came back to the surface. He didn’t say anything, but walked over to the chipping hammer. He put down the lance, picked up the chipping hammer, and walked to the side of the platform. “So long , you son of a . . . .”, he said, and threw it overboard.

It was one of the fastest adaptations of a change in technology around the world that I have seen. Anyway I will talk about using jets for efficient cleaning in another post. Enjoy the day!

Deepwater Oil Spill - using a nuclear weapon

While the likelihood of using a nuclear weapon to shut off flow to the leaking oilwell in the Gulf are absolutely zero, nevertheless, under a different circumstance it did work in the Soviet Union. This video of that event shows it happening. Thanks, Rune.

Deepwater Oil Spill - why Top Kill may have failed

The Top Kill attempts have failed, and the Government has a response.

He (President Obama) said US Energy Secretary Steven Chu was leading a team of "the world's top scientists, engineers and experts" in devising a contingency plan should the "top kill" attempt fail.

But while waiting for that, and for the Lower Marine Riser Package (LMRP) I thought I would spend a few paragraphs discussing why Top Kill may have failed.

In a couple of earlier posts I wrote about how it was necessary to fill the gaps that ran through the Blow-Out Preventer (BOP) either with spheres and triangles or with wire(string would act similarly). To refresh your memory, in the initial simpler analysis I had put up a simple sketch of the BOP and well, to show how the blocking particles were injected.

Simple approximation of the situation

Now unfortunately that diagram missed a significant part out, and that is that there are three sets of pipes leading down into the well. These are the well outer casing, which, surrounded by a layer of cement, holds the BOP in place. Then there is the production casing, which had just been set to the full depth of the well. And there is the drill pipe that, at the time of the incident, extended down 8,367 ft from the platform, or roughly 3,367 ft below the BOP. That drill pipe (DP) had previously been used to locate the production casing at the bottom of the well, and itself now rode inside that production casing. In most normal operations it is closed at the bottom by a drill bit, but (and I’ll come back to this later) it had just finished the cementing of the production casing into position, and once it detached from that and was being pulled from the well, it was an open pipe all the way up to the rig floor. And in that condition it could be used for other things. By pulling mud out of the DP and transferring it to the mud pits (or standoff vessel), the level in the riser would fall and be replaced by seawater flowing in at the top. Unfortunately this also lowered the weight of mud in the well, and that is what caused the oil and gas to flow into the well.


Outside of the DP is the casing and cement segments that make up the outer lining of the well. The diagram presented in Congressional testimony, shows these various pipes, except for the central drill pipe.

Casing and cement down the Deepwater well

BP do not know, but believe that the oil is getting into the well through the cement wall at the bottom of the well, and probably rising up the well through the empty space (annulus) between the production casing and the outer lining of the well. However the oil and gas may have broken through the bottom of the cement plug and be rising up within the production casing, in which it is also rising through the BP once the oil reaches its lower end.

Most normal blowouts occur when the well is being drilled, and mud is flowing down, through the drill bit, and then back up the space (the annulus) between the DP and the rock wall. Thus, when there is a blowout, the oil and gas that flow into the well normally flow up this outer passage to the rig, and give the spectacular fountain of oil. The BOP was invented (by Harry Cameron and Jim Abercrombie) to stop that flow and to protect the crew at the surface. Because the flow is normally up the outside of the drill pipe, the initial BOP designs were rams that pushed seals across the flow path through the BOP, and sealed against the side of the DP.

BOP open allowing flow through the annulus (ASME )

BOP closed against the pipe, sealing the annulus (ASME )

A BOP could have two of these mounted so that one sealed to the production casing in the well, and one to the drill pipe, but if underwater then the production casing is tied back to the Wellhead Collet Connector, and then the only tube running through the BOP will be the DP, to which they will seal.

BOP connection to casing at the seabed (PCCI report for MMS)

The problem that this leaves, in the current situation is that the pipe that runs through these two seals is open at the bottom to the oil flow. So how can the flow through this be stopped?

The answer is to mount a top ram set that has a set of shear cutting blades on it, that will cut through the pipe and seal the full face of the well.

Shear blades to cut through the DP and seal the well (Varco )

The BP should shear, but would be held in place by the grip of the annular sealing rams below.

In this case it seems to be recognized that for some reason this shear event did not totally succeed. Thus the pipe was not totally severed and the two shear plates did not fully move over one another to complete the seal.

Now this is where the problem arises, because, in part, that pipe is still open at its lower end. If the leak is around the outside of the pipe, through a gap that has generated between the pipe and the annular seals, then the use of the junk shot to fill the cracks and gaps could conventionally have worked. But the configuration of the rams on the Deepwater Horizon had changed from the initial simpler configuration to add seals for occasions where the drill pipe was not in place.

Ram layout on the BOP (Times Picayune)

And the "junk" is being injected at the bottom of this stack.

Section through the BOP, showing the anticipated mud flow path (initially from BP)

If the leak is coming up through the remnants of the drill pipe then life is complicated. It can’t all be coming up through an undamaged pipe alone, since it was the far open end of that which was successfully closed at the beginning of the remedial steps, but if it is coming through the pipe and leaking out at the shear rams into the annulus that feeds into the riser, and out to the sea, then putting sealing particles into the bottom of the BOP to seal the cracks, could have sealed some of the leakage around the DP trapped in the shears, but not that flowing through the shears in the remaining pipe section.

The reason that it can’t is that the access to that flow is occurring 3,367 ft below the riser, and there is no easy way to get the sealing particles down that far. If they are mixed with mud and pushed down the well to that level and then released they have a different problem. The hope when they were released into the well was that the flow of the current would be enough to carry them up to the cracks that they could seal. But if they have to be carried down to the zone where the oil remains, then their density may be sufficient, if they get into the flow without enough speed to lift them up into the BOP to cause them to sink to the bottom of the well.

The materials that BP tried included materials that might float on the surface, and might not

Those materials, including fibrous pieces of rope and chunks of rubber, were supposed to force more of the mud down the wellbore, but ultimately it did not work.

Rubber has a specific gravity of 0.91 and rope varies from 0.9 to 1.4. But remember that at that depth any buoyancy from air entrainment would be lost.

In other circumstances it might have worked, If they could have dropped the DP out of the shears perhaps, but they couldn’t and it didn’t. So on to the LMRP.

UPDATE: Thinking about this a little more, I had two more thoughts. The first is that once the LMRP preparation cuts off the riser and the bent drill pipe that the full weight of the pipe below the shears may come onto the section in the shear jaws at the moment, pulling them further out of alignment and increasing the flows. It could also cause the pipe to drop out of the jaws, pulled out by the underlying weight, and hopefully not distorting them too much so that in the best of worlds they could then be cranked shut.

One could also, once the bent riser and pipe had been cut, go in down the pipe bit that extends up, go down past the annular seals with an abrasive jet lance (most of the flow is around the DP as we have established above) and cut it off, right above the shears. Then partially open the shears, drop the pipe out, and close them again. If they move all the way closed, without the obstruction, then the well may be sealed.

(Note because a) this is really a sort of Tech Talk, and b) its a nice day, there won't be the regular Sunday tech talk which would have been on coal mining today - it is postponed, as was the Saturday discussion of state climate changes - hopefully I'll be able to get back to both of those before long).

Deepwater Oil Spill - the end of Top Kill

BP and Admiral Landry just held a Press Conference in which they said that, based on a decision 90 minutes ago, by the “best and brightest minds” that it was time to move on the next option, the Lower Marine Riser Package (LMRP). BP was unable to block sufficient flow out of the well to make the injection of cement possible, and thus to kill the well. They had made, I believe he said three attempts to inject material (the junk shots) without being able to get that material to block the passages through the Blow-Out Preventer (BOP) . (Unfortunately I missed a large part of his opening remarks, and thus have only the question response to go on at present). The volume of mud used did not appear to have changed from earlier reports at some 30,000 barrels.

Mr Suttles said that they had given the technique every chance, but could not get it to start to provide an effective seal. They had, however, determined that the majority of the pressure restrictions to the flow of oil was coming from some resistance within the well itself, and from the BOP. Since the riser above the BOP was not contributing much to the resistance, and thus to control of the oil flow, the next plan is to remove it, using a band saw device (of which pictures will be available) and then to lower the LMRP onto the existing BOP. They intend cutting the surface that the LMRP will sit on flat, so that it will give a good, but not perfect seal. Thus there will be some leakage around the joint, and they will monitor that and use dispersant as appropriate.


The new change should take somewhere between 4 and 7 days to implement. The assembly, which has been constructed, and is not the Top Hat assembly built earlier, to fit on the bottom of a riser. Flow of oil from the LMRP will rise up a 6 7/8 inch drill pipe within the riser (the same size as the one currently fitted to the RIT). The riser will also carry hot water down to the LMRP to protect against the formation of hydrates.

He noted that their inability to stop the well “scares everybody” but is reasonably confident (no success percentage estimates) that this will collect the majority of the oil and gas. Because they do not know the flow path of the oil below the seabed it is difficult to estimate what is actually going on in terms of oil path below the BOP. Thus they are, again, trying something that has never been done before, but expect, based on the RIT, that it will work.

On being asked about the cleanup of the dispersed oil – he pointed out that the reason that the dispersant was used was to break the oil into small droplets. These are small enough to be consumed by the microbes in the sea, and thus there is no plan to do other than let nature take its course. For the oil on the surface, they are getting better at spotting oil pools and sending skimmers to deal with them.

The Admiral drew attention to the article on Hurricanes and the Oil Spill which is available at the Unified Command We site.

The relief well is about half-way through the rock it must drill (about 6,000 ft below sea level) but progress will slow as the well deepens.

The Lower Marine Riser Package (LMRP) option

Deepwater Oil Spill - varying the junk mix and cutting the flow

The video from the ROVs monitoring the Deepwater oil leak from the well in the Gulf of Mexico show that the at around 10:45 am they may have started another junk shot injection, given that the flow from the riser has just increased, and various large particles have been coming past the camera. When they are injecting mud the contrast on the picture gets somewhat worse, and so it is probably better if you check this out for yourself. The riser is continuing to flow mud.

between about 10:45 am and noon 29/5/2010

In the earlier post I wrote about one way of tackling the leaks in the BOP, using spheres and triangles of rubber. But as you can see from the flows from the leak at the top of the riser, when the cracks get quite small the injected particles get swept past, and end up coming out of the end of the riser. So I am going to look at what the flow path size is, and another answer.


So let us run some numbers to see if we can estimate what the largest width that the particle has to bridge will be. That would be if the hole in the BOP is circular, since in any other geometry one of the dimensions will be smaller and catch the particle.

Assuming that, if for the sake of the discussion we accept that the well is leaking at 17,000 barrels a day, this translates into around 500 gallons of fluid a minute. (Divide by 24 to get flow per hour, divide by 60 to get flow per minute, and multiply by 42 to convert to gallons).

The next step uses an Excel table that I have generated over the years to calculate circuit flows. It has the orifice diameter on the left and across the top the pressure driving the flow. I have modified the table to show the diameter that would be required to allow 500 gpm to pass (roughly) and have highlighted where that flow is reached (roughly) for different combinations of flow and pressure. (i.e. the red numbers in the table, for a given driving pressure across the top read to the left to get the hole diameter that will give this flow).

For those interested in generating their own the equation I used for the value in space N13, for example, was

=$L$1*(3.1412*60*($A$13/2)^2*12*12.5*SQRT(N3))/231.

The values are very dependant on the discharge coefficient (L1) and this value can vary for flows through orifices from about 0.6 to 0.9. To get the largest diameter I made the value 0.6 (as the discharge coefficient increases the diameter of the hole needed to pass that flow reduces).



You can see that the largest dimension of the flow channel is just over 0.7 inches. (Which means that the BOP rams functioned over at least the majority of their stroke). The minimum is about half an inch, and if I change the discharge coefficient from 0.6 to 0.85 then the diameter range goes from 0.4 to 0.6 inches.

So from this we know that the maximum gap in the BOP is 0.7 inches in diameter. Now this is good news because it means that it is less than a third of the diameter of the feed line (which has an effective inner diameter of possibly 2.7 inches or so).

So we can continue, as they are, to send particles down through the riser to the BOP. But we also know that the flow path through the riser could be a long thin crack, rather than the round hole we used in the example above. So to address that problem a different particle shape and type is needed. Consider now what happens if we send some wire down the line with a thin rubber coating (so we don’t damage the fittings on the way down) and give it say a diameter of 0.4 inches. This is small enough to get through the pipes, but if the crack is narrower than this the wire, because of its shape, will be pulled across the crack, thus:

Wire fills a longer part of the crack if it is flexible enough to follow it.

Of course the cracks won’t run straight, and thicker wire is stiff, so after a while they will probably introduce wires of differing diameters. But this may be the next step in the process. There is, however a precaution that has to be taken. Wires tend to clump together and can build a blockage in the feed line if they are too long, so the pieces should be kept short, and fed in a little at a time, over a size range, to ensure that they help rather than hinder the bridging of the flow.

At the moment (noon) it looks as though they are still pumping mud, so they may be trying to use a slightly heavier mud in order to get balance, though again they are constrained on how heavy they can make this before they start losing it into the formations.

Deepwater Oil Spill - why Top Kill starts and stops

Out in the Gulf BP are going through the preparations for the next stage in the attempt to top kill the deep water well that has been leaking oil and gas for more than a month. The kill attempt has now been underway for a couple of days, and so this is initially a recap on what I believe has happened to date, and what they are currently planning on doing. I will include in this explanation the two attempts that BP have made so far, and what I expect that they will do next.

UPDATE: (9:19 PM) There have been at least two and possibly as many as four times today that there has been something that looks like an explosion, and debris rains around, and for a while the ROV cameras go and look at something else (right now it is the riser outflow that we haven't seen for a while).

What I believe is going on is that BP is running a series of "junk shots" with the Nat Lab "junk", and after they run one they fill the well with mud to see if it has changed anything. To date, while there are changes (you stop doing this when there aren't) they haven't been enough. But after each time that they fill the well with mud, they switch the pumps off while they go and regroup. That allows the gas and oil to push the mud back out of the way (one of these tries, perhaps, it won't and we will know that it has worked).

But when the oil and gas get through the constriction of the BOP there is a drop in pressure and an expansion, and this gives that little bit of excitement that we see. After a while, not so exciting, though this time one of the cameras went off air, so maybe there was something else going on as well?? OH, and BP has decided to halt, for a while, the second drilling of a relief well, so that they can prepare for plan B. One has to remember that these drilling rigs are at a premium, and there is an opportunity cost with just about every decision.

"The (Developmen Driller II) has temporarily suspended drilling operations in preparation for the possible future deployment of its BOP on top of the (Macondo) BOP," he told UpstreamOnline in an email response.

END of UPDATE.

UPDATE 2: 10:53 pm. It appears that they are pumping mud again, so another test may be under way. (I am judging this on texture and jet structure) . There is one interesting thing to note, and you have to be familiar with the patterns to see it. So I am going to repeat a picture from 3:45 am yesterday (which has been my standard base reference for mud, rather than oil flows.

3:45 pm 5/26/2010

Now look at the current flow:

10:59 pm 5/28/2010

Notice the changes in the flow pattern, and particularly that the crack to the right of the central paint removed line, which had a piece of rubber jammed in it . By adding more NL junk they have just about totally bridged that crack and stopped the flow - which illustrates that what they are trying to do at the BOP is at least partially working - as I explain below. In fact, after watching for a short while, both the central jets that used to carry most of the flow, are now significantly diminished, so both have some form of particle bridging the gap.

I am almost tempted to note that the flow seems at a little less pressure, judged again by jet structure, and which would indicate more bridging in the BOP, but I grow more cautious as the process extends.

And one last point - you should not expect the majority of the cracks at the riser (the ones we are looking at) to get closed until almost the end of the injections since their width is much smaller than the one that they are trying to block in the BOP, but as it grows smaller, so they should come into the range of the NL "junk." END UPDATE 2

To help with the explanation I am going to use some very simple models, which oversimplify the situation, but hopefully help in explaining it. To start with I am going to break the overall system down into a very simple diagram.

Simple approximation of the situation

Basically BP are sending mud through a series of feed lines, which I have simplified for the explanation into one feed pipe that I have called the choke line (grey). It has an internal diameter, at most, of 3 inches. (I say at most because most fittings on these lines have smaller diameters). It feeds into the top of the well, which I have colored brass, and which is the pipe (casing) that feeds from the seabed down to where the oil is emerging from the rock, some 13,000 ft further down.

The casing and choke line sit underneath the Blowout preventer (which is the large assembly at the top of the rig). I have colored this bronze, and simplified the shape, for this explanation, to represent a pipe that has been partially closed by the action of the BOP.

Now, here is the problem. BP want to feed mud through the choke line at such a pressure and flow that it will push the oil and gas flowing up out of the well back down the well. To do that they have to create enough resistance to the flow that the combination of the mud flow and the oil can’t all escape out through the hole in the BOP.

They can do this since, as you increase the flow through a hole (or nozzle) it has to move faster to get through the space in a given time. It takes a driving pressure to get the fluid moving at that speed, and for a given driving pressure the fluid will only move at a certain speed, and so only a fixed volume of fluid can go through the hole.

Thus if BP pump more fluid into the well than this volume, for that given pressure (which is higher than the pressure that the oil was flowing at) then all the flow out of the well will change to mud, and the excess mud that is not flowing out of the hole will be at enough pressure over the oil (and gas) in the well that it pushes it back down the well and back into the rock.

Now that was the first step. The idea was that once the column of mud filled the well, down to where the rock reservoir lies, that the weight of the mud would exert a pressure on the oil in the rock, that was higher than the fluid pressure, and the flow would stop.

The first time they did this, the density of the mud (weight of a cubic foot) was not high enough for the full column to balance the pressure in the rock, and the leakage of mud out of the hole in the top of the BOP was higher than BP had thought, so they were losing too much mud to the Gulf.

So they moved to step two. The first part of this is to try and reduce the size of the hole in the BOP. And for this they used a variety of what everyone is calling a junk shot. It actually isn’t. Given the problem that I am about to outline, they have an ally that folks normally don’t have. As Secretary Chu has pointed out there is the full intellectual strength of the Federal Labs behind this attempt, so where one would normally just chop up tires and similar materials, there is a fascinating mathematical problem in designing plug pieces of the right shape that will fit the constraints, and which will accelerate the blocking of the flow path. So I suspect that some of the shapes that will appear in the flow, will have been specificially designed for the problem.

Not being familiar with their answer I’m going to stick with the spheres and triangles of the more traditional, shall we say old fashioned, way of addressing the problem.

It is one of these problems where the bounding conditions can make life rather difficult. Let me redraw the problem with a different orientation. What we have to do is to block (the term that is being used is bridge) the passage through the BOP. This will stabilize the flow, and will allow the cement plug time to set up. (We’ll talk about cements another time).

So here is the slit: through the BOP – it is going to be a more complex shape, but this allow some explanation of the problems.

Simplified picture of the slot to be sealed

Now to block the slot we have to have some pieces of material (although they are quite large for simplicity I am going to call them all particles) that are big enough to wedge in the slot, but small enough to get through the feed lines to that point.

Now here’s the first catch, we don’t want them to be hard enough that we will damage the passages, nor soft enough that they will bend distort and compress and squeeze through the hole. Rubber turns out to fit the bill, and though there are other materials that could be better, in this initial explanation that’s what I am going to use.

The easy thing to do is to use some spheres, not easy to get, though golf balls are an example. Unfortunately they are a bit too big. The reason is that the feed line through the choke has a maximum inner tube size of probably 2.75 inches. A golf ball is about 1.6 inches, which is more than half this, so that two balls together could block the feed line – a definite no-no. (And don’t say it can’t happen, I’ve seen it with smaller particle ratios than this).

The maximum size that you can get through the line should be about a third of the minimum diameter – say 0.9 inches, stretch a point and make it a maximum ball size of an inch. So we fill the mud with miniature golf balls, pump enough of them down that they end up going through the BOP and wait for the effect. Let me show you, using pearl spheres, how the problem evolves:

Spheres in slot

See all the open space around the spheres, and how much of the slot remains open. We really haven’t made much of a blockage in the slot area of contact, and we have made it hard to push other materials into the slot area itself. However the smallest open area now may be at the maximum diameter of the spheres, which is further back, where the feed pipe is larger. Putting more spheres of the same size down won’t improve the situation much, because they still leave room, around the spheres, for fluid to flow.

There are two ways to go, once the initial building blocks for the bridge have been established. The first might be to use triangular pieces of rubber (as we saw protruding from the crack in the riser. These can fit closer together and fill more of the slot and flow passage.

Ideal case of three rubber strips blocking flow

Here the strips have aligned in the right way and have been driven into the slot, reducing the flow path. But note, as with the spheres that the gaps that are left are now too small for more strips of that size to feed into the slot and do more blocking.

So in either case what has to happen is that there should be a second pass, where smaller particles are used . These couldn’t be used before, because they would push through the slot, but now the slot size is smaller, and so these can start to fill in the gaps. Let me illustrate with the spheres:

Filling the gaps with a smaller second set of spheres.

So now the gaps that remain are even smaller, and so in a third shot, with even smaller gaps the feed particles have to be even smaller.

It can take a number of different slugs of material going through the choke line (and being pushed into the BOP instead of falling down the well, before the slot is sufficiently “bridged” that there isn’t much flow out of the BOP.

So expect that there will be a number of these shots, after each of which mud will be pumped in to see how much progress in filling the holes they have made. Bear again in mind that there is this restriction on how big a piece they can feed in, and just hope that all the gaps in the BOP are small enough that big enough particles can be fed into the lines to block it.

Now, as I said, they are probably using more sophisticated shapes from the National Labs, that will allow the number of shots to be reduced, but the relative sequence still has to be followed, as they build the bridge. Let's see how it goes, and be patient, each shot takes time to set up.

And at the same time, given that they have to balance the weight of the column of mud against the rock pressure, they are using the interval to change the mud weight increasing it each time, to seek that balance. (They don't want it higher than it has to be or it could hydraulically fracture the rock and lose the mud into the crack).

(Oh and if some of you remember the class in school where the teacher filled a jar with big pebbles and asked you if it was full, you said yes, then he/she poured in smaller pebbles, now is it full? Then came sand - now is it full? And then the jar was filled with water - its the same basic idea).

Deepwater Oil Spill - restarting the mud, and an apology

It appears that a significant part of what I have written for the past day has been wrong, and the information that I, in good faith, commented on to several radio stations and newspapers about the status of the Gulf leak was similarly not totally correct.

I have reaped, I guess, the rewards for believing statements about transparency. From the LA Times

The company announced late in the day that it had suspended shooting heavy drilling mud into the blown-out well 5,000 feet underwater around midnight Wednesday so it could bring in more materials. Thursday evening, BP PLC said it had resumed the pumping procedure known as a top kill. Officials said it could be late Friday or the weekend before the company knows if it has cut off the oil that has been flowing for five weeks.

The conclusions that I drew about the pressure drop up through UPDATE 8 on Wednesday, when we were losing visual contact with the leak, because of what I thought was mud falling around the well could have been correct.

However, later comments on the relative flow velocity were based in part on the assumption the BP was still pumping mud. They were not. So that while the velocity observation was correct (since BP has stopped pumping) that is the only consolation I have from believing the Admiral’s statement this morning that there was only a slight problem with pressure balance before the company could start pumping cement into the well. (The LA story that I took this from has been updated so I cannot go back to the original quote). As I say this is more than mildly irritating, and I will try and be more cautious in drawing conclusions in the future. It does, however, through the pictures I posted, allow us now to be able to identify the difference between the mud flow (top picture) and the oil/gas flows that are in the remainder, with very little, if any, mud. (and at 8:40 pm on Thursday it is still oil and gas).

There was, however, some additional material available in the article:

Although incident commander Adm. Thad Allen of the U.S. Coast Guard said Thursday morning that BP had temporarily stopped the flow of oil, the company's chief operating officer later said petroleum was still flowing. "Once the well has stopped flowing then we would pump cement down into the hole to fully seal it," Doug Suttles said. "We might finish this in the next 24 hours, or it might take longer." Engineers next plan to inject heavier "bridging material" above the mud to prepare to put a cement seal on the well.

Well I see that the monitoring ROV is back in position to watch the leaks as BP, perhaps, is about to restart pumping mud – if they really are.

I would stop commenting on this, but sadly it is too important a subject not to continue. My apologies if I have, inadvertently misled you.

UPDATE: 9:52 pm the camera is focusing on the cracks in the riser, and it seems that they may be injecting rubber pieces one of which is now stuck in one of the cracks in the riser. (Not very securely it seems)

Piece of "junk" (?) in the riser crack, as BP apparently work to reduce the size of the path through the BOP.

Note that this piece has had to pass through the BOP, and it is sealing the BOP path which is more critical to success. It could also be a piece of the rubber from the annulus seal that broke loose and got caught in the riser. Without knowledge of what BP is trying it is hard to decide, but the flow looks to be still gas and oil without mud, and I would expect that BP would have to use mud as the carrier if they were injecting material into the flow, so this could just be a piece of seal that got caught. If you can't tell where it is, it is in the crack to the immediate right of the center line (without the paint) on the riser. (The view has changed)

UPDATE 2: Mud is clearly visible in the change in the look of the flows out of the riser. But at the moment it does not appear to be under the pressure of the flows on Wednesday. (This could be because it is being pumped in at a lower pressure, or it could be that they have sealed some of the leaks in the BOP and that is cutting back the driving pressure at the riser).

Leak shot at 10:25 pm Central

The problem we saw on Wednesday night with mud being heavier than oil and thus settling more readily and obscuring the view, is also evident.

UPDATE 3: 12:18 AM So it appears that BP have injected "rubber strips" into the flow, and that some of these have lodged in the BOP, reducing the flow channel, while one made it through and is trapped in one of the leaks in the bent portion of the riser.

Now what may happen is that they will slowly increase the mud flow/pressure to a) find out how much the leak rate has been reduced and b) to make sure that the restrictions in the flow channel are stable, and won't blow out. (If they do then they will have to repeat the process). Once they have a sure reduction in leakage then they will re-generate the higher pressures that overcome the pressure in the reservoir and start forcing the oil and gas back down the well, as the mud begins to fill the pipe.

The mud seems to have a slightly different texture from last time, so they could have increased the mud weight so that when the column of mud is re-established that this time it weighs a little more and overcomes the slight pressure imbalance that they were left with the first time they tried this.

Now is a good time for caution and, though the fill time may be reduced because of the smaller leak rate into the Gulf, they may still pump at a relatively only slightly higher pressure that that in the reservoir, to slowly sweep down the well, getting into the necessary channels, and giving time for the oil and gas to be pressed back into the rock that it came from.

UPDATE 4: 9:30 AM The latest report from the Gulf

Hayward told CNN BP engineers had injected a "junk shot" of heavier blocking materials into the failed blowout preventer of the ruptured wellhead, and would also pump in more drilling "mud"- all part of the top kill procedure being attempted.

"We have some indications of partial bridging which is good news," he said.

"I think it's probably 48 hours before we have a conclusive view," he added.

Admiral Allen also noted at that time that the leak had been stopped, but that they were not sure that they could sustain the halt in flow. However at 8:10 am, Sterling925 who wa watching and commenting on The Oil Drum saw some sort of event occur around the BOP.

Chaotic images - looks like an explosion!

09:14 et 5/28/2010

and from SteinarN

It looks like A LOT of gas is coming up from the seabed around the BOP. Considering the large water pressure and the possibly large area this gas is emanating from it ought to be a large flow. This indicate the integrity of the well is not good?

Unfortunately I did not see any of this and haven't been able to see the BOP apart from one short short since, though in that shot it did not appear to have any problem. The PBS viewed ROV at the moment is working with a chain, while the ROV that was monitoring the plume is now staring out into the ocean.

The CNN shot however shows that we are back with oil and gas apparently coming out of the leaks at the top of the riser, and no different to the conditions before they started pumping mud into the well last evening. So the second filling of the well has apparently all been washed out, and they will try again later. The comment from BP was that this might take another couple of days.

UPDATE 5: 10:24 AM Well I am not sure that the CNN feed was actually live and there are other stories catching their attention at the moment, but there is a Youtube recording of what took place (h/t Jessica in Pensacola).

UPDATE 6: 11:09 AM The feed has gone back to the riser, and we are back to the oil and gas flows that we were saw at the beginning. Not quite the same shapes as earlier, so perhaps the block in the BOP was partially effective, but BP have now apparently filled the well twice and failed to get enough weight into the mud to hold the driving pressure from the rock. They could try again with a higher density mud, I am presuming that the second shot had a higher weight than the first, and that while the first left a small pressure imbalance, that the second was closer, but as yet no banana. (Though the Admiral did say that they had stabilized the flow). My presumption is that they will mix up another batch and try again - though whether they will try another junk shot is not clear.

Flow at 11:09 am

The way in which you try and stop leaks is that you put the big stuff in first. If you can get enough of that to stick, it still leaves large flow channels, and so the second shot uses smaller pieces that fit into the gaps. Then you try smaller shots etc until you get as good a seal as you can. Doing this to plug water flows into tunnels can take several shots to get a total seal, working with sequentially smaller sizes of particles.

Deepwater Oil Spill - preparing for cement

Admiral Allen, who is the Government head of the effort to cap the flowing well in the Gulf, and to oversee the cleanup operation, commented this morning that the well had reached a point where the internal pressure difference between the mud pumped in and the reservoir pressure was very low. However, with the relatively high volume of leakage that was passing through the BOP, the plan now included a try at blocking some of that leakage path by injecting debris (for which likely read rubber strips and small spheres) in the hope that these will lodge in the flow path within the BOP and reduce the leakage of fluid.

The leakage rate is significant (I calculated earlier that it was around 17,000 bd, which lies within the newly reported range of 12,000 to 19,000 bd, and may have been higher than BP were actually anticipating. (Though the leak may also have increased a little as the mud was injected at higher pressures). The operation has already used all the mud on one of the supply boats, and has moved to the second (there is a third standing by so they won't run out). The concern, however is now with the volume of cement that will be required for the seal.

The high volume that is leaking would require that additional amount to the volume needed for the seal itself, and that may be closer to the available capacity of the system that they have in place, or the supplies that they have on site to achieve the seal. If that is the case, one can understand the desire to at least partially plug the leaks in the BOP, and to wait until the mud column fully balances the pressure in the oil reservoir before starting this phase of the operation.

Until this point in the operation the volume of cement required to create an effective plug has not been seen as an issue.

Unfortunately as I write the feed from the leak has moved to look at other things, but the last glance I had seemed to suggest that they have been able to reduce the flow somewhat, though I guess not enough. Just to remind you of the problem, this is what the internal flow path through the BOP looks like:

Section through the BOP, showing the anticipated mud flow path (initially from BP)

Now they have to get relatively large particles (larger than the smallest diameter of the flow path) through the feed lines, which have a size of 3-inches, and those of us who pump particles in fluids much prefer that the particle diameter be no more than 30% of the pipe diameter - which is to say in this case an inch. And so if the orifice in the BOP is larger than that, then there is a problem in working out how to get something in there that will be effective, remembering that it has to work through all the flow convolutions of the supply line that has just been installed for the mud.

Deepwater Oil Spill - comparison of flows

I am putting up these pictures to show why I believe that the injection pressure of mud into the well has dropped, indicating that BP have filled the well, and are now holding pressure to see if there are any problems. I would assume, if none develop, that they will inject cement to seal the top of the well, sometime today. And I am slightly modifying the order since, at 6:35 they showed the ROV and the leak to give you some sense of how close the camera is to it.

ROV and plume 6:35 am

3:45 pm yesterday

5:32 am this morning

Notice how the flow was longer and straighter, indicating that it was at higher pressure (velocity) and that now it blows out at much closer distance, meaning it doesn't have the same pressure (velocity). There is a small caveat, and that is that I am assuming that there hasn't been any significant erosion of the surface of the cracks between the two shots, and that may be a possible change, though not enough to cause the reduced throw distance of the central jet.

And here is the flow at 6:30 am

Deepwater Oil Spill - permissions and concerns about Top Kill

Rather than stick continuous updates on the previous post, I am going to add them as updates to this separate post on the progress of the Deepwater Top Kill attempt

By 1 pm Louisiana time, permission had been obtained from Admiral Landry to go ahead with the top kill procedure. The team is therefore waiting for final approval from the top folks at BP, before going ahead.

There have to be concerns about the increasing number of leaks at riser near the BOP, which is lowering the flow out of the end of the riser (and which were visible on a CNN feed earlier this morning). Leaks of a fluid that is carrying abrasive particles can get larger very quickly, and can threaten the integrity of the BOP. This now, therefore, becomes a driver to accelerate the process, given that if the flow is allowed to continue that the BOP may be further eroded and weakened and may collapse on the one hand, and that even if it were not to, that the resistance to the pressure of the top kill process will decline as it erodes.

We are at a point where endless discussion and the need for permit after permit begins to have a negative impact on success.

UPDATE - Well so much for impatience, the process has been started, and we should know fairly quickly whether it works.

UPDATE 2: There appears to be mud coming from the leaks in the riser at the top of the BOP, showing they have started pumping at 2 pm

UPDATE 3: Well BP said that they would provide a live feed, they just didn't say which one - grin - so the feeds are now showing things such as parts of the BOP, where they were looking for leaks initially, but where nothing much is now happening, But it does seem from the occasional glance of the leaks near the top of the BOP that they are increasing the pressure on the feed to the well at 2:30 pm. Hmm, now CNN are showing the top leak, and if you look at the one at the back, it noticeably got bigger in the last couple of minutes.

(Oh, and for those watching CNN you don't need 2000 ft of cement to seal this well, they can get by with a whole lot less (less than 40 ft) but may go for an absolute assurance that nothing will ever get out of that well again.)

UPDATE 4: (3 pm) BP are planning on injecting 50 barrels of mud a minute. Let us assume that the leaks are still allowing 5000 bd of flow, which is equivalent to 150 gal/min. 50 barrels at 42 gallons per barrel is 2,100 gallons. Let us assume, for discussion that the amount that escaping fluid through the BOP doubles during the injection. Then the flow that will go down the well is 2,100 – 300 = 1800 gallons.

There are 231 cu inches in a gallon. So they will be injecting 416,000 cu inches/minute. While the well has various diameters I will assume a diameter of 16 inches as an average. This gives a cross-sectional area of 200 sq inches (roughly).

Then the speed at which mud will travel down the well to fill it is 416,000/ 200 = 2,080 inches per minute, or 172 ft per minute. The underground part of the well is 13,000 ft deep.

At 172 ft/minute, it will take 13,000 / 172 = 75 minutes to fill the well with mud.

So that is about as fast as the operation can determine success. Though, due to pausing between steps to ensure that there aren't too many problems, it will likely be somewhat longer. And it depends on how much worse the leaks at the top of the BOP get.

UPDATE 5: There is nothing really exciting about watching a mud fountain play into the ocean for an hour or more, and so I thought I would stick my neck out a little and predict how this might end. (Nothing dramatic being visible, means that so far it is working).

Once they consider that the well is full of mud they will likely back off the pump pressure that is injecting mud into the well. They will monitor the pressure in the well as they drop the inflow pressure down, and watch to see, as the pressure falls, if there is any increase in pressure from down hole. If they can take the pressure of the pumps all the way down to ambient (which is under 5,000 ft of sea water, and is therefore around 2,400 psi) and stop injecting fluid, and nothing changes on the pressure gages, and there is no flow out of the well, then the well will have been brought under control. As the pressure in the pumps starts to fall, so the mud fountains at the leaks will start to diminish.

UPDATE 6: Well I notice that the BP Press Conference keeps getting postponed, which may be good or bad news. If the Top Kill had immediately failed (unacceptable erosion of the BOP) then there would likely have been a Press call.

If it had immediately worked then there might have been another, but that they are delaying it means that it may be working but a little slower than they had hoped.

UPDATE 7: Ah, in the Press Conference Doug Suttles said that they are only injecting mud at the rate of 20 barrels a minute. (7,000 barrels over 6 hours). This is less than half the anticipated flow (50 barrels) and they may have dropped the injection flow rate to keep pressures in theleak above the BOP at an acceptable level. That does increase the time it will take to fill the well significantly (by several hours, depending on the leak rate). Though it also shows that those estimates that the well was leaking at 100,000 barrels a day were fantasy.

It would take 87.5% of the mud injected being lost to leaks, for it to take 22 hours to fill the well, and that would indicate that the leakage rate was 25,000 bd.

I was watching CNN and Campbell Brown talked over what the biggest problem has been so we'll have to wait until either a reporter or the transcript lets us know.

UPDATE 8: Well it may be my old eyes, but I think that the flow from the leaks is definitely less than it was, leading me to suggest that the well is now full of mud, and they are reducing the flow to see if the well behaves. (8:47 pm) I haven't been watching for a little while however, but did have these other thoughts before I got distracted (its called dinner if I recall).

The leaks at the tops of the riser do appear to be getting bigger. However they look like crack leaks, which seem to get longer preferentially to wider, with less overall increase in flow. I would expect, since BP has more than enough pump capacity, that they will just increase the flow to balance any increase in leakage. They can monitor this through the pressure gages, and so can set the flow rate to give a certain progress down the well, or well pressure. I suspect they are more concerned about well pressure, and will just keep that at a steady value until they are sure that they have filled the well. (Probably indicated by a change in pressure levels as they start to try injecting mud into the formation rather than pushing the oil and gas back.. The mud will coat the walls and make it more difficult to inject and thus they should see a pressure increase). (8;47 pm)

UPDATE 8.5: Oh, and if you were wondering why you can't see the camera feeds any longer. Oil and gas are lighter than water and float off upwards, mud is heavier and so the plume is dispersing and settling back downwards, around the well. Good job we have pressure gages. (9:07 pm)

UPDATE 9: Well the picture is clear again (wonder if they used an ROV backwash to remove the mud?) and I have been sat with an earlier picture of the leak flows, relative to the current picture and I still think, from the lowering of the points at which the flows bulge out, that the jets aren't at as high a pressure as earlier. This could, however, be because the cameras have moved, or changed lenses, but the pictures do have several common features. And the heights of the plumes from both side jets are significantly lower. (In fact by 9:45 the back one appears to have disappeared, while the main central flow is noticeably less powerful).

UPDATE 10: Looking at the flow now, relative to the earlier pictures it does definitely seem to be reduced. So I would suspect that they have reduced the amount they are pumping in to balance against the leaks, while they monitor the pressure in the well and see if there are any problems develop. If there aren't any (and their chance is reducing) then the plan was to inject cement and create a plug. No-one has mentioned how much of that they have available.

Incidentally, in my calculation above, if it took 22 hours to fill the well at 20 barrels a minute, then one could assume, knowing the volume of the well, that the rest was leakage. It looks as though it took about 7 hours to fill the well, which would mean (at that assumed pumping rate) that about 60% of the fluid injected was going out through the leaks, and this comes out at about 12 barrels/min or 17,000 bd. (5 am)