Case histories bring reality to well control training
WELL CONTROLCase histories bring reality to well control trainingLouisiana State University, with support from the Minerals Management Service, used two actual underground blowouts as student teaching / learning toolsJohn Rogers Smith, Adam T. Bourgoyne, Jr., Sherif M. Waly and Eileen B. Hoff, Louisiana State University, Baton Rouge, Louisiana
Three principal conclusions developed in this presentation are:
Introduction The ability to successfully prevent blowouts is widely recognized as a critically important element of any drilling operation. Well control training is used to give rig-site personnel the practical and theoretical knowledge needed to develop this ability. However, many preventive / monitoring practices taught in this training can seem arbitrary, mundane and unnecessarily exacting to trainees. The case histories described here demonstrate clearly why careful control and monitoring practices are important for avoiding blowouts, particularly in deep, high-pressure wells. The Minerals Management Service supported the acquisition of these case histories, and their adaptation for training purposes, as part of a project on prevention / control of underground blowouts. Prevention / control of such blowouts is complicated by several factors. By definition, an underground blowout is an uncontrolled flow from one subsurface zone to another. Consequently, there is no opportunity to observe the flow directly, and diagnosing what is happening in the subsurface is difficult. Even identifying the occurrence of an underground blowout can be difficult. The consequences are also uncertain, ranging from insignificant to catastrophic. Usually, the most serious risk is of the blowout broaching to the seafloor or ground surface and causing a crater, with consequent damage and danger at the surface. The industry lacks systematic procedures for analyzing / controlling underground blowouts. In addition, most conventional well control training courses allocate less than 5% of course resources to the subject. Some well control manuals and texts, such as Murchison,1 Abel,2 and Kelly, Bourgoyne and Holden,3 give some guidelines or example approaches for specific situations, but not a general method. Therefore, the industry needs additional training resources to address the prevention and control of underground blowouts. Such a need was the justification for using records of previous experiences to create case histories for use as training exercises, as previously proposed by Bourgoyne and Kelly.4 Overview. A case history of a deep, underground blowout offshore Texas has previously been used as an effective, interactive, group learning exercise. 5 A later section of this article will describe briefly how that case history is being adapted to be an individualized, interactive, programed learning exercise. Two additional case histories are described herein. The focus will be on an underground blowout that occurred during a trip in a deep, high-pressure gas well. The important factual events are presented chronologically, and resultant key learnings are emphasized. Another case history of a "near miss" reinforces some of these learnings. These two experiences are the conceptual basis for a training simulation exercise. Although a complete history match of a major well control event would be too time consuming for most training, specific situations can be re-created. These require the trainee to diagnose the situation with the same kind of information that would exist in reality. Using a simulator allows the trainee to actually implement decisions rather than just think about their likely results. The exercise and its results from use by industry professionals are described. Case history applications. Multiple case histories have been provided by industry sources for the underground blowout project, and several have been selected as appropriate for training purposes. In addition, case histories are also described in some existing well control literature, by Grace,6 Muchison2 and Abel. 3 In general, the examples in literature provide enough information to make and support a key point, but not to develop a full training exercise. The following two case histories are the basis for the key learnings and simulation exercise described herein. C.H. No. 1: Underground Blowout In Deep Gas Well The primary well control event reviewed in this article is an underground blowout that occurred during a trip in a deep gas well. A diagram of the well at the time it was shut in is shown in Fig. 1.
Chronological description. A deep, offshore gas well had been drilled to a TD below 22,000 ft MD and 21,000 ft TVD. The objective sand was reached and found to be gas productive. Several conventional cores were taken and recovered. A 12.5-lb/gal mud was being used, which provided a 550-psi overbalance. A 9.625-in. liner was set and cemented just below 13,300 ft and tied back to surface with 10.75-in. pipe. The leak-off test at the liner shoe was equivalent to 13.7 lb/gal. A 60-ft core of the objective sand was tripped out of the hole. Previous trips had caused no problems, and the overbalance should have provided more-than-adequate trip margin. However, careful monitoring of the trip tank indicated that the hole had taken 2.5 bbl less than expected. This was not considered a problem, because it was a relatively small error for such a long trip, and previous, successful trips had experienced even larger discrepancies. There was no flow from the well after the trip out, and it was considered successful. A new bit was picked up, and the trip began uneventfully. The trip was interrupted at the 9.625-in. casing shoe to slip and cut the drill line. There was still no indication of flow from the well. The trip was continued to 18,400 ft, where pit level measurements indicated a gain was occurring. The well was checked for flow and observed to be "flowing slightly." This was concluded to be thermal expansion of mud, and the trip was continued. Tripping continued to about 19,300 ft, but pit level continued to increase in excess of pipe displacement. The trip was again interrupted, and the well was shut in. The shut-in drill pipe pressure was 0 psig, but shut-in casing pressure was 100 psig. The cause of the casing pressure was concluded, by rig personnel, to be "U-tube effect from out-of-balance mud." Consequently, the trip was again continued. Continued pit gain. In reality, a pit gain of at least 55 bbl had been recorded over the previous 1.5 hr. A quick calculation shows that this size gas influx would cause 450 to 600-psi loss of hydrostatic head, depending on where it was in the annulus. Actual total pit gain was probably larger, given that the kick had probably occurred much earlier. A larger gain could easily cause the 650-psi loss of hydrostatic head necessary to cause the 100-psig, shut-in casing pressure. If so, the well was truly underbalanced when the trip continued, and influx from the formation was almost certainly occurring. After tripping to 19,750 ft, the crew concluded that they should circulate to get the mud back in balance. Although the continuing pit level gain showed that the well was still flowing at this point, it was not shut in or circulated on the choke. When circulation began, the total pit gain was at least 73 bbl more than calculated pipe displacement. At this point, shut-in casing pressure (SCIP) would probably have been about 300 psig. If the well had been shut in, it would have been obvious that mud imbalance was not the problem. An SICP of at least 830 psig could be contained without losing returns, based on the leak-off test. Therefore, it is likely that the well could have been killed conventionally at this time. Circulation continued for about 30 min more. During this time, the pit level indicator being observed by the operator’s representative malfunctioned and showed no gain. The mud logging crew observed that the pit gain was continuing, but did not advise the operator’s representative. Afterwards, a review concluded that a "lack of clear communication" contributed to the "improper" actions taken by the crew. Another 108 bbl of gain were taken before the pits ran over. Problem becomes obvious. Circulation continued another 30 min before rig personnel concluded that the well was really flowing, and it was shut in. The loss of mud from the pits means that the total pit gain when the well was finally shut in is unknown, but it was substantially more than the 181 bbl indicated by mudlogging records at the time the pits overflowed. The SICP was 1,150 psig. If all or most of the influx was below the 9.625-in. casing shoe, this pressure would result in lost returns and would initiate an underground blowout. The drillstring was almost 3,000 ft off bottom. Stripping was begun to get the bit closer to TD to allow for attempting a conventional well kill. After stripping eight stands into the hole, SICP had reached 3,400 psig. The cause for this increase is not certain, but gas migration, pressure buildup after initial shut in and failure to bleed off pipe displacement would all cause shut-in pressure to increase. The upper stripper began leaking at this time, allowing an additional 40-bbl gain. After shutting in to repair the stripper, the SICP was 4,100 psig. Given these excessive pressures, lost circulation leading to an underground blowout was almost certainly in progress. An additional complication was that the drillstring became stuck during the stripper repair with the bit still more than 2,000 ft above TD. Kill preparations. The seriousness of the situation was finally obvious. It was clear that conventional control methods were unlikely to succeed, and preparations were begun to perform an off-bottom kill. The primary focus of this case history is prevention, but the following summary of kill preparations and operations is given for completeness. Engineering calculations indicated that a dynamic kill could be achieved by pumping several thousand barrels of 13.5-lb/gal mud through the drillstring and into the loss zone at a rate of about 17 bbl/min. This would require additional mud, pumps, personnel and other resources to be delivered to the rig. In the meantime, the risk of a surface blowout could be reduced by minimizing surface pressures. Surface pressures were reduced by intermittently bullheading mud into the annulus and pumping mud into the drillstring to keep it at least partially full. Gas that migrated to the top of the annulus was bled off and replaced with mud during periods when bullheading was interrupted. Noise logs were run in the drill pipe to confirm that flow in the annulus was occurring. The outer-casing annulus pressures were monitored for changes to verify that conditions were not becoming worse. After five days of preparation and rig-up, the kill operation was ready to begin, and shut-in casing pressure had been reduced to 1,055 psig. Kill operation. The final kill plan was to pump 4,000 bbl of 13.5-lb/gal, water-based mud down the drillstring at a rate of 17 bbl/min and a pressure of at least 6,500 psig. The mud would exit the bit and return up the annulus with the gas flow to the loss zone just below the 9.625-in. shoe. This rate and density was calculated to raise pressure at the bit enough to prevent further gas influx. Then, 1,000 bbl of 15.5-lb/gal mud would be pumped and left to fill the annulus between the bit and the shoe. This would ensure a hydrostatic kill of the well even if gas was left in the well below the bit. Three workboats were tied into the rig mud pits to provide a total of 5,600 bbl of 13.5-lb/gal mud. Another workboat held 2,000 bbl of 15.5-lb/gal mud. Three turbine-driven pump skids provided 4,000 hydraulic hp for the large-volume, high-pressure pumping job. Engineers had not only designed the procedure, but had also predicted the expected pressure-vs.-rate response throughout the job and provided criteria for determining whether the job was succeeding and should be continued. The kill operation began by pumping the 13.5-lb/gal mud at 3 bbl/min and staging up to 17 bbl/min to allow verification that hydraulic predictions were correct. Initial pump pressure at 17 bbl/min was 6,000 psig. As mud began to fill the annulus, this pressure increased to 6,900 psig. A steady-state condition of 6,800 psig at 16.3 bbl/min was achieved after pumping about 700 bbl. This combination indicated that a dynamic kill had been achieved. Circulation continued for another 1,000 bbl to help remove some remaining gas from the open hole and annulus. Then the planned 1,000 bbl of 15.5-lb/gal, was pumped at a final rate / pressure of 14 bpm and 5,350 psig. Success of the kill operation was confirmed by running noise and temperature logs to verify that downhole flow had ceased. C.H. No. 1 Analysis And Key Learnings There are several important learnings that can be drawn from this experience. These relate to the causes of kicks, detection, reaction to unconfirmed kick indicators and control of severe well control problems. Key learning No. 1. The actual cause of the initial gas influx into this well is not known. The operator concluded that one, or both, of the following could have caused the initial kick that proved so hard to detect conclusively.
The key learning is that: A small volume of gas influx can expand enough to displace mud from the well and cause it to become underbalanced, especially in a deep well. Key learnings No. 2. Swabbed-in kicks, or other kicks taken during a temporary underbalance, can be difficult to detect. In this case, the "kick" was almost certainly taken during the trip out of the hole, as explained above. It was only detected after tripping over 18,000 ft back in the hole. Key learnings are:
Key learnings No. 3. Reaction to an unconfirmed indication of a kick is important. In this case, tripping was continued, and then circulation initiated, without careful monitoring or evaluation of observed flow, pit-level and pressure indications. The initial indication of a possible kick was apparently a steady pit gain in excess of what was expected due to pipe displacement while tripping in the hole. The minor flow noted at 18,400 ft confirmed the possibility that a kick had been taken. The increasingly strong indicators that a kick had been taken continued to be discounted until after a very large influx had occurred. The pit level increased rapidly after enough mud was displaced to cause the well to be underbalanced. Pit level increased even more rapidly when circulating, because gas was being brought to surface and expanding even faster, simultaneous with new influx from the formation. Key learnings are:
Key learning No. 4. Delay in reacting properly to the initial gas kick and the subsequent large kick caused a major underground blowout. A gas formation with 120 ft of 40-mD sand and a 13,400-psig reservoir pressure was flowing uncontrolled into another permeable zone almost 8,000 ft shallower An effective kill procedure required density, volume and rate high enough to overcome this high-rate, underground flow and ultimately regain hydrostatic control, despite having over 2,000 ft of open hole below the bit. The key learning is that well control for underground blowouts and off-bottom conditions requires special procedures not typically addressed in conventional training. Nevertheless, even a severe loss of control can be corrected with a properly designed and executed operation. C.H. No. 2: Near Miss Due To Small Swabbed Kick Near misses can also provide a basis for case history-based learning exercises. This case history is based on one of the authors’ personal experience with a small kick that was apparently swabbed-in during a trip out of a deep gas well. This kick did not result in an underground blowout. However, the well had previously experienced lost returns, and underground blowouts had been experienced in previous wells in the area. Consequently, although it was successfully controlled, it is considered to have been a near miss. The significance of this case is that it developed in a similar manner to the previous case, but the cause of the kick is somewhat better documented and understood. The kick was identified just as the crew was beginning to run a production liner, and the kick can be concluded to have been caused by the trip out of the hole. A summary of the experience is described in the following paragraphs. Fig. 2 is a wellbore sketch indicating the general configuration of this well during the trip out of the hole.
The well had been drilled to the objective TD below 18,000 ft. After correcting lost returns experienced at TD, it was logged, and a cement plug was set below 17,500 ft to isolate the lost-circulation zone near TD from shallower, potentially productive intervals in the open hole. The cement plug was dressed off, the well was circulated clean with 18.5-lb/gal mud and a trip out was made to run a production liner. Fill-up volumes were monitored throughout the trip, which was judged to be routine except for two factors. First, the trip was made somewhat faster than most previous trips. Second, a 4-bbl "gain" had been noted while laying down drill collars. The gain was believed to have been caused by floor-washing water spilling into the trip tank. The well was checked and found not to be flowing. Preparations were then made to run the 7-in. production liner, and there was no additional pit gain during this time. However, excess displacement while running the liner was noted almost immediately. Running was halted temporarily, and flowchecks were made on two occasions in the first few stands. No flow was observed, and running continued. "Auto-fill" float equipment was being used to minimize surge pressures that could have re-initiated lost returns and was thought to be a possible cause of the fill-up discrepancies. The remainder of the liner, and 15 stands of drill pipe were run, as the well was observed to finally be flowing. Having all of the liner in the hole and drill pipe opposite the BOP stack allowed the well to be shut in. Shut-in drill pipe and casing pressures were equal at just above 900 psig. The decision was made to strip the liner in the hole to enable circulation closer to the likely kick zones. The float equipment was activated so that it would prevent flow up the drill pipe, and stripping commenced. Ten stands were stripped before the annular preventer element failed. The well was shut in on the pipe rams, and the element was replaced. Stripping then continued to about 17,000 ft, where the liner became stuck. The well was killed conventionally at that depth, and the liner was cemented successfully. A maximum pressure of about 1,400 psig was encountered during stripping. Gas and saltwater-cut mud were circulated out during the kill, but in smaller volumes than expected. A large amount of mud had been lost during the stripping operations; apparently, this was equivalent to bullheading a significant fraction of the kick fluids back into the formations in the open hole. Analysis / Key Learnings From Swabbed Kick The precise cause of this kick has never been determined. Most likely, it resulted from a small, swabbed-in kick at the beginning of the trip. This conclusion is based on: the trip being faster than previous trips, occurrence of balling and swabbing on previous trips, "gain" observed while laying down drill collars and the common occurrence of incorrect hole fill-ups observed on the first few stands pulled. Although this last possibility is not documented for this trip, it was experienced on earlier trips; it has been simulated and demonstrated as a feasible explanation for the actual sequence of events. The seemingly-late reaction to the kick indicators was still quick enough to prevent the excessive pit gain and excessive shut-in pressures experienced in the previous case history. The rig personnel’s acknowledgment that a kick had occurred and readiness to initiate stripping before flow became excessive were important to their success. The low formation permeability and combination of water and gas production made their challenge easier. The key learnings are similar to the previous case history:
Simulation-Based Learning Exercises A simulator training exercise has been developed using a situation analogous to the second case history described above. It is intended to demonstrate how a small, swabbed-in kick can be very difficult to detect, but can develop into a blowout if ignored for too long. The scenario is also representative of the kick that caused the problem in the first case history. The exercise is implemented on stand-alone, PC-based, well-control, training / simulation software.
How the training works. The training exercise begins by advising students that they have just arrived on the rig for a crew change. A trip from TD has just begun, and the third stand is being pulled. The floorhand monitoring the continuous-fill trip tank has advised that the hole has taken at least two barrels less mud to fill than calculated. The supervisor they are relieving has just requested that the trip be postponed and the well watched for flow. The students must then perform the flowcheck and decide what to do next. They should not be told that the well had kicked. They should have the same uncertainty about the meaning of a small volume discrepancy that they would have in a real situation. There are a number of alternative actions the students might take after observing that there is no flow. Some of the possible reactions are:
There are additional, various reactions that might be taken as the students observe the results of their decision. In any case, careful observation will eventually demonstrate that a kick has been taken. Training objective. A major objective of this exercise is to allow the student to reach the conclusion that there is a kick in the well independently of confirmation by the instructor. This shows that the presence of a small kick is almost undetectable initially. As the kick migrates, it will expand. The expansion is very slow while the kick is migrating in the lower section of a deep well, and the resulting pit gain is too slow to detect. As the kick reaches the upper portion of the well, the gradual expansion will have resulted in a larger kick volume. The hydrostatic pressure at the kick will also be changing proportionately faster, so the kick will be expanding more rapidly – even if the migration velocity is the same. The more-rapid expansion, loss of hydrostatic head due to the kick volume resulting in a formation influx, or the combination of both will eventually result in detectable flow and significant pit gain at surface. If identified quickly enough, the well can still be shut in and controlled. In the case of the "near miss," control was regained fairly quickly. In the "deep underground blowout," it was misjudged for too long, and the underground blowout occurred essentially as soon as the well was shut in. However, this was still less dangerous than a surface blowout. Results of training. The training exercise has given similar results. It has been used in training more than 20 industry professionals. Their first reaction is usually to ask the instructor what to do. Because they are in a well control training course, they expect to have to shut the well in. Asked what they would do in the field if this were a real case, the reaction is usually to carefully check for flow. After identifying that there is no flow, there will typically be one of two reactions. One is to trip back to bottom to circulate bottoms up and check whether a kick was taken. The other is to continue tripping out of the hole. In either case, essentially nothing happens to indicate whether a kick is present until the gas has reached the upper few-thousand feet of the well. The simulation is of a 3-bbl, swabbed-in kick in a 10,000-ft well. This was selected to simulate the effect of an even-smaller kick that had migrated to 10,000 ft in a deeper well. The 10,000-ft depth was chosen to reduce the total amount of simulation time. A wellbore diagram for the simulation is shown in Fig. 3. In actual case histories, the flow was not detectable until after more than 10 hr of migration.
The simulation demonstrates this effect with a much shorter period of apparent inaction. Even so, one student team had an experience similar to the deep underground blowout. Apparently, the lack of any "action" during the early part of the simulation caused them to become distracted. They missed the initial indications that the well was flowing and that a pit gain was occurring. The pits eventually overflowed, causing the simulator screen to flash red, and the simulator was "frozen" to prevent the simulation from blowing out. Fig. 4 is an example of the subsurface pressures and pit gain vs. time recorded for a sequence of events like this that ultimately simulates a blowout. Note that once, the well becomes underbalanced, it unloads more quickly and wellbore pressures drop rapidly. The reduced pressure causes an increased rate of flow into the well. In the simulation, several hours are required for the gas to migrate to a depth of about 2,200 ft, where it has expanded enough to cause a 10-bbl-total pit gain. Within 10 min, the kick has migrated to about 1,300 ft and expanded enough that the well becomes underbalanced, and the formation begins to flow. Within another 10 min, the gas reaches the surface.
Key learnings. The most common reaction by industry participants is to use precautions commonly taken in the field. They trip back to bottom and begin circulating bottoms up. Again, there is very little "action" initially. However, the 69 min required to circulate bottoms up only requires 7 min at X10 on the simulator. So, the kick reaches the upper portion of the well fairly quickly. A gradually-increasing pit gain warns an observant team that there may indeed be a kick in the well. Depending on how rapidly the team detects this, a flow can usually be observed if the pumps are stopped. If so, the well is shut in, and conventional well control is begun. Given that the well already contains kill-weight mud, the driller’s method can be used to remove the kick and re-fill the well with mud. This approach is usually successful and fairly uneventful. The casing pressures and pit gain observed during this procedure re-confirm that a kick had been taken. Fig. 5 is an example of the pressures and pit gain vs. time recorded in a simulation of this sequence of events.
Three key learnings can be experienced using this exercise:
Given that many kicks occur on trips – and that some result in serious surface or underground blowouts – this is an important learning exercise to supplement more-routine exercises, where the kick is taken while drilling and is more easily detected. Other Training Exercises The original training exercise developed for this project was based on a deep, underground flow offshore Texas. It has been used as a successful group learning exercise for almost 100 industry professionals and rig personnel, and more than 50 petroleum engineering students. The Petroleum Engineering and Industrial Engineering departments at LSU have also initiated a cooperative effort to adapt the original, interactive group-training exercise as a programed learning exercise for individuals. This exercise is being implemented on PC software. It would allow an individual to make decisions by selecting from a predefined list of alternatives and then to be advised of the result. Wellbore sketches are used to clarify well conditions resulting from previous decisions. Key learnings are emphasized as they become evident in the results of the student’s decisions, or as results of the actual decisions made in the field are explained. Another new training module has just been started to simulate the "deep underground flow." It is intended to re-create major decision points and actions that resulted in the underground blowout and which ultimately brought it back under control. Re-creating the entire case history in a single simulation would be time consuming, even at the X10 maximum speed of the simulator. However, specific point-in-time situations that occurred during the case history allow simulation of subsequent events, as well as alternative courses of action that might have been taken. Each such situation then becomes a separate learning exercise within the context of the actual events that occurred. This allows key lessons from the case history to be learned in a "hands-on," experiential manner by students operating the simulator. Note: This article was prepared from
the paper, "Case histories bring reality to well control training," presented by
the authors at the IADC Well Control Conference of the Americas, Houston, Texas, Aug. 25–26,
1999. Acknowledgment The authors appreciate the support and encouragement from the Minerals Management Service for this project. They also thank LSU for its support and permission to publish. This work would not have been possible without the information provided by operators and their drilling personnel; it is greatly appreciated. Colin Leach is also acknowledged for sharing his understanding of swabbed-in kicks. Literature Cited
 The authorsJohn Rogers Smith, a Campanile Professor and assistant professor in PE at Louisiana State University, holds a BS in EE from the University of Texas and MS and PhD degrees in PE from LSU. He previously worked for Amoco Production Co. He is currently an SPE distinguished lecturer on deep drilling and a member of AADE, ASME and SPE. Adam T. (Ted) Bourgoyne, Jr. received BS and MS Degrees in PE at LSU and a PhD in PE at the University of Texas at Austin. He began his career as a senior systems engineer with Conoco. In 1971 he joined LSU as an Assistant Professor. Since then, he has worked in the undergraduate, graduate, and continuing education programs of LSU’s PE Department. He served as Chairman of the Department from 1977 to 1983 and as Acting Dean of Engineering from 1985 to 1987. Until his retirement from LSU in December 1999, he was the Campanile Professor of the Craft & Hawkins Department of PE and Dean of the College of Engineering. He has been active in blowout prevention and guided development of a research / training well facility at LSU. Sherif M. Waly earned a BS degree in systems / biomedical engineering in 1981 and MS degrees in biomedical and industrial engineering in 1988. He received a PhD from the University of Miami in 1994. He joined the Industrial and Manufacturing Systems Engineering department of LSU in 1995. He was previously a research associate and instructor at the University of Miami. His research areas include industrial human factors, occupational biomechanics and safety engineering. Mr. Waly has authored / co-authored more than 70 articles / papers. Eileen Hoff is a graduate student at Louisiana State University where she is completing work on her dissertation in engineering science. Her current research focuses are on ergonomic aspects of oil production / safety training. She has BS and MS degrees in industrial engineering.
|
ase
histories can be used in training to demonstrate the importance of effective well control.
This article focuses on the case history of an underground blowout that occurred during a
trip in the hole on a deep gas well. It reviews the chronological events and emphasizes the
key learnings. Another case history of a near miss due to a small swabbed kick reinforces
some of these learnings. Also described is a case-history-based training simulation which is
designed for students to experience these learnings by making the operational decisions
themselves that might lead to, or prevent, an underground blowout. 
