Trends extracted from 1,200 Gulf Coast blowouts during

Transcription

1 Comprehensive analysis of 1,06 blowout reports covering four U.S. Gulf Coast states and OCS waters over 6 years includes database handling procedures and discussion of well control failure causes Trends extracted from 1,00 Gulf Coast blowouts during / WORLD OIL / JUNE 1 67 Fig. 1. Wells drilled in Texas, Offshore Texas and Louisiana and the U.S. Fig.. Number of blowouts in Texas and OCS by year. the U.S. Gulf Coast area during the period 160 through Presently, Texas blowouts make up some two-thirds drilling, tripping out, and circulating or killing. that problems of influx detection, kick handling and Even though the drilling industry continually learns more about how to handle an unstable well, it seems number of future occurrences. the study of past occurrences to reduce probability and the real world, i.e., on the rig. The small changes and stab a string valve, or ) P failure after closure. Formation pressure control is an important concern. Minerals Management Service (MMS), Outer Con- statistical data. A new, recently compiled database con tains extensive data on more than 1,00 blowouts from Cost and loss of life incurred from blowouts warrant well pressure control do not change much with time in improvements that do occur are, first of all, seen through contributed blowout report information, as will be noted. quently, blowouts occurred during activities of actual The analysis indicated also that major causes of loss into a high-pressure zone, or they were associated with formation breakdown. Secondary barriers which failed were either: 1) failure to close the Ps, ) failure to A. L. Podio University of Texas, Austin, Texas PM Skalle, NTNU, Trondheim, Norway; and of pressure control were related to swabbing or drilling 16, as described and analyzed here. Five agencies of the incidents in the database, but range below the average frequency of 0.1 per 100 wells during newwell drilling/completion. Data indicates that, most fre most often after a kick occurred and resulted in a blowout INTRODUCTION. Louisiana Office of Conservation. Mississippi State Oil and Gas Board. Texas Railroad Commission (TRRC) 1. State Oil and Gas Board of Alabama ence in general and extensive training, through required of kick detection and killing procedures. Conversely, large reducing well control loss frequency. Technological We would expect to observe that operating experi from warrant the study of past occurrences to reduce the kick detection and handling may result in loss of control. essarily be accompanied by a corresponding reassessment This could cause an increased risk, which may not nec both directions. Examples of technology changes are drilling operations. Improper procedures with respect to The cost of blowouts and loss of lives incurred there incidence of future problems. blowout schools especially, are slowly paying off and respect to shallow seismic proffling for detecting shallow gas hazards and improving kick detection systems for floaters and land rigs. These in turn would be expected occurrence data for the period from 160 through 16 was gathered from the following five sources: DATABASE DEVELOPMENT! ORGANIZATION A database comprising drilling and blowout () to reduce risk and incidence of kicks and loss of control. developments may, however, override these effects in use of diverterless drilling combined with drilling in deeper water, and pushing surface casing deeper and improvements were made during the last decade with deeper so as to require fewer casing strings. both in the planning phase and during execution of rt WELL CONTROL

2 FIg.. Frequency of blowouts per 100 wells drilled in Texas. FIg.. Frequency of blowouts per million feet drilled in Texas. Table 1. Overall activity, no. of blowouts (), 160 to 1 6 Area Total No. Footage Blowouts during no. of during wells drilled, drilling pen drilling drilled 106 ft loowells 106ft Louisiana* 0, OCS 16 1, Texas 0 7,00, All 1,06 70,7, lncomplete data Table. No. of blowouts by activity/area, 160 to 16 Phase Texas OCS Exploration drlg. Development drlg. 10 drlg. 1 Completion 6 Workover 17 Wireline 1 Production 1 Missing phase data 1 Total 16 tinental Shelf (OCS). The data was organized in an Excel spread sheet, and subsequently analyzed in terms of trends of the characteristics and occurrence of blowouts. The original database, initiated by Hughes, Podio and Sepehrnoori, was used as the starting point, but since data from Alabama and Mississippi are still incomplete at this stage, it was decided not to publish them in this paper. It should also be noted that it was not until 17 that mandatory reports were instituted in all the U.S. states; before then, no systematic method of record keeping existed. Consequently, statistics before 17 are of questionable accuracy. It is hoped that it will be possible to update the blowout database every th year, to observe new trends and changing statistics. From the records analyzed, both before and after 17, we have seen that, often, kicks are reported together with blowouts. Kicks which have simply been circulated out in a routine manner without subsequent problems, have been excluded from the data. Those kicks that have caused extraordinary problems during shut-in or killing, such as circulation loss, stuck pipe and underground communication, are normally included in the Texas RRC database, and often, it was somewhat difficult to deter mine whether a true blowout took place. When in doubt, such occurrences were included and denoted with a very short blowout duration. In addi tion to drilling operations (sulfur drilling and water 6 WORLD OIL / JUNE 1 1 wells are not included), blowouts occurring during completion, workover and production are included and categorized as such. Wells drilled before 160 were excluded due to lack of accuracy and report consistency. Reports about blowouts in wells drilled after 160 show a gradual increase in both con sistency and detail level, although some interpretation ofthe parameters and description of events was necessary; even after 17. Data was organized by geographical location and divided into major groups: 1) identification, e.g., well, operator and contractor identity, type of well, data source and data quality; ) operational characteristics, e.g., depth, operation in progress, wellbore configura tion, event sequence and causes leading to blowout; ) consequences, e.g., extent of losses, occurrence of fire and loss of life; and ) event disposition, e.g., duration, method of control and final status. And since the database contains too much data for one paper, to avoid a voluminous presentation, we decided to defer statistics dealing with what happened after the blowout (blowing fluid type, mode of control, duration, consequences, etc.) to a later analysis. Regarding the method of analysis, although data com pilation is an ongoing process, it was believed that a representative data volume had been collected to permit an analysis of occurrences and characteristics. The anal ysis comprised the grouping of events characteristics and corresponding activities, and developing frequency distributions. Our data shows frequency similar to the drilling operation blowouts published by Daneberger. DRILLING ACTIVITY VS. BLOWOUTS To date, a total 1,06 blowouts have been included in the database, divided according to operational phase and geographical areas, as shown in Tables 1 and. As noted, data for Louisiana is complete only as far as 10. Note that there is a significant difference in the blowout frequency-while-drilling of OCS wells, when considered as a separate group. Figs. i and show, that U.S. drilling activity reached a peak during the 10s, with a corresponding high level of blowouts. Figs. and indicate that overall blowout frequency for Texas is rather stable after 170, and independent of the activity level. It is believed that before 170, due to the absence of regulatory require ments to file accident reports, the number of blowouts

3 Fig.. Frequency of blowouts per 100 wells drilled in the OCS. Fig. 6. Number of blowouts in Texas and OCS by well depth, >1 000 ft Fig. 7. Number of blowouts in Texas and OCS by last-set-casing size, Fig.. Wells drilled in the U.S. distributed into four depth intervals. yields a blowout frequency too low to be trustworthy. Fig. shows the corresponding data for OCS wells. BLOWOUT DEPTH Number of blowouts were compared to actual depth at which the kick occurred, and depth of last casing set, Figs. 6, 7. It is apparent that most kicks/blowouts occur at shal lower depth since most U.S. wells are drilled to shal lower depths, Fig.. It is also observed that blowout frequency is much higher in deeper wells, since drilling lengthjexposure time/formation pressure are higher. These facts are shown in Table. Assuming that depth interval distribution in Texas and the OCS are similar to U.S. depth interval dis tribution, we see that % and % of the blowouts occurred in wells below 1,000 ft, while only between 0.% and.7% (average 1.6% in the period) of the wells were drilled below this depth. As expected, the frequency is highest in the deepest depth categories due to higher pore pressure gradients and difficulties in handling highly compressed gas. In addition, as stated by Wylie and Visram, increased exposure time, longer openhole sections, more tripping time and increased risk of lost circu lation problems increase blowout probability. Fig. shows an increas ing trend in the number of wells drilled below 1,000 ft, which may result, in the future, in an increasing number of blowouts, if equipment and procedure improvements are not Table. Blowouts in TexaslOCS, 160 to 16, in four depth categories, compared to all U.S. wells Depth Texas OCS U.S. wells drilled interval, ft blowouts blowouts No. % No. % % % <, , , >1, implemented. Detailed distribution of blowout numbers as a func tion of depth is shown in Fig. 6. Two peaks are observed: one at,000 to,000 ft and one around 10,000 ft. The first peak correlates with drilling operations undertaken when only surface casing is set. The peak at 10,000 ft correlates with onset of abnormal formation pressure in the Gulf Coast region. This is also evident from Fig., which presents mud weight variation with depth, which shows that normal mud weights ( to 10 ppg) are seldom seen below 10,000 ft. OPERATION IN PROGRESS Operations in progress at time of initial pressure con trol loss have been divided first by the phase of well oper ation (drilling, completion, production, workover and wireline); then, secondly, the events have been further split by the specific activity or operation during which the blowout occurred. This analysis is presented in Fig. 10 and Tables,, 6 and 7. The most risky operational phase corresponds to exploratory drilling in which blowout frequency is three times higher than for development. This trend has been reported also by other studies. Fig. 10 also shows that a significant number of blowouts have occurred during workovers. During drilling, most events occurred during making hole or moving drillstring operations, as seen in Table. Those that correlate with drilling onbottom typically imply drilling into a formation 6 WORLD OIL / JUNE 1 6

4 Fig.. Mud weight vs. depth for blowout wells drilled in Texas Districts 1 and. Fig. 10. Number of blowouts vs. operational phase in progress, Table. No. blowouts () vs. operation/activity in progress during drilling phase (Lou isianaltexas/ocs, ) Operation Drilling Circulation Technical problem Well testing Abandon well Pressure testing Missing operation data Total Activity 0 Tripping out/cnx/wiper Actual drilling (bit on bottom) Circulating Out of hole Tripping in Coring Circulating/killing Tripping Wait on order 6 Fishing Stuck pipe Killing with unknown pressure. The high percentage of blowouts during drilling and tripping indicates that operators/contractors may not be taking enough precautions to compensate for the decreased overbalance margin. The industry must walk a fine line in choosing a fluid that will optimize both pressure control and drilling rate efficiency. It is interesting to note from Table that the largest number ofblowouts during completion corresponds to wait ing on cement. This indicates that more attention must be paid to the problem of gas migration in cementing. Workover blowouts are also more numerous for oper ations that involve running or pulling tubing or drill pipe. This is another indication that not sufficient atten tion is given when minimizing well back pressure with the intent of reducing formation damage. Gas wells characterize the majority of blowouts in producing wells. Typically, these correlate with high welihead pressures, corrosive fluids and failures of pip ing and valving. 70 WORLDOIL/JUNE 1 Table. No. of blowouts () vs. operation/activity in progress during completion phase (TexasIOCS, ) Operation 1 Installing equip Circulation 6 Running well equip. 10 Well testing Perforation Missing operation data Total ActivIty 1 WOC Nipple down P Run csgitbg. Set well plugs Cementing casing Initial production Killing Casing running Cleaning well Gas lifting/initiate prod. BLOWOUT CAUSES Blowouts are the result of failure to control a kick and regain pressure control. They are typically caused either by equipment failure or human error. A blowout occurs for about every 110 kicks, both for development and exploratory wells. In Tables through 10, causes of blowouts have been systematized in terms of the pri mary cause of the loss of pressure control (failure of the primary barrier), and then in terms of the cause of the uncontrolled flow of fluid (blowout), or failure of the secondary barrier. The high incidence of swabbing as the primary bar rier failure is another indication that insufficient atten tion is given to trip margin, keeping the well full and controlling pipe movement speed. Since blowout preventers are the most commonly used secondary barrier it would be expected that they would be operated, maintained and used properly. How ever the data shows that in a total times, these systems did not perform their function and resulted in uncontrolled flow. In addition, a total 0 events occurred when Ps were not installed. Table shows the breakdown of primary and sec ondary failures in terms of operational phase. Table 10 shows details for drilling operations.

5 Table 6. No. blowouts () vs. operationlactivity in progress during the workover phase (TexasIOCS, ) Operation Pulling well equip. Installing equip. Abandon well Perforation Circulation Running equip. Well testing Drilling activity Missing operation data Total Activity 67 Pull tubing Pull/drill out plugs Stuck pipe Pull WL Cleaning well Pull pump rod Pull casing Plugged pipe Run tubing Install P Run WL Nipple down P Pull tubing Set well plugs Killing Pull casing Note also that this type of information was not avail able for a large number of wells (over 0) due to the lack of uniform reporting requirements. This is very unfortunate since it limits the possibility of undertak ing detailed failure-tree analysis and developing reliable failure probability estimates. REDUCING NUMBER OF BLOWOUTS Despite improved regulation and inspection of P equipment and rigs, and improved blowout preven tion training for drillers and weilsite supervisors, no improvement in blowout frequency in Texas was seen during This may be explained in part by the fact that the prevailing contracts were of the footage or turnkey type. One conclusion from this is that improved rig equipment and crew 10 ability to detect and control kicks was offset by con tractor and operator efforts to: 1) maximize revenue and cut expenses through increasing the driffing rate (minimum overbalance); andjor ) decrease non-pro ductive rig time (ignoring or missing drilling breaks, tripping too fast, not stopping to circulate bottoms up to remove trip gas or to check for any minor influxes or more frequently fill the hole). suggested that such trends may Wylie and Visram be combatted by: Eliminating footage contracts in favor of daywork contracts. Alternatively, operators should take a more proactive role in specifying minimum mud density and 7 offer a bonus program for minimizing kicks. Taking the same approach with regard to kicks as for rig and equipment inspections. A kick report and analysis should be submitted to the Conservation Board and made part of the public record for review, analy sis and continuous improvements. This is a view supported by our study of this database. Of the 1,06 blowouts in our database, only about 600 include reports which may be considered sufficiently complete from the standpoint of providing enough infor mation for analysis. Thus, despite the important effort and good work being done by some regulatory agencies in collecting data, a lot of information is lost. This prob lem could be solved if the drilling industry were to develop a standard procedure and report format for kicks, loss of pressure control and blowouts. CONCLUSIONS An extensive database containing blowout data from Texas, OCS and U.S. Gulf of Mexico states has been compiled, containing data from 160 through 16. Data quality is improving, especially since 17, when mandatory blowout reports were instituted in all U.S. states. However, to obtain all details necessary for a better blowout analysis, an industrial standard for data gathering is required. The data shows that blowouts continue to occur at Table. Most frequent primary and secondary barriers that failed in all phases (Lou isianaltexas/ocs, ) Primary barrier Swabbing Drilling breaklunexp. high press. Too low mud weight Formation breakdown/lost circ. Gas cut mud Trapped/expanding gas Wellhead failure Xmas tree failure While cement setting Secondary barrier Failed to close P Rams not seated Unloaded too quickly DC/KeIly/TJ/WL in P P failed after closure P not in place 1 Fracture at casing shoe Failed to stab valve/kelly/tiw Table 7. No. of blowouts () vs. operation in progress during production phase (TexasIOCS, ) Operation Production of gas 6 Production of oil 1 Closed in well 10 1 Missing operation data 6 Total 7 Poor cement Tubing leak Improper fill up Tubing burst Tubing plug failure Uncertain reservoir depth/pressure Water cut mud Flange leak Annular losses Csg. collapse Unknown 0 Casing leakage 1 Annular valve/choke String safety valve failed Form. breakdown/lost circ. Diverter failed after closure String failure Casing valve failed Wellhead seal failed Xmas tree failed String safety valve not insti. SCSSV/storm choke failed Unknown WORLD OIL / JUNE 1 71

6 Table. Distribution of most frequent operation phase failures (TexasIOCS, ). Swabbing Drilling break Formation breakdown Trapped/expanding gas Gas cut mud Too low mud weight Wellhead failure Xmas tree failure While cement sets Failure to close P P failed after closure P not in place Fracture at casing shoe Failed to stab string valve Casing leakage Blowouts Texas OCS 1* Primary barrier Secondary barrier I Explor. drilling Devel. drilling Completion ProductIon Workover 6 WIroline DO Most frequent secondary causes 16 Failed to close P Rams non seated prop. () Unloaded too quickly (16) DCITJ etc. inside P (1) Failed to stab string valve P failed after closure Fracture at casing shoe String safety valve failed P not in place Annular valve failed Missing data 6 P failed after closure Failed to close P Fracture at casing shoe Casing leakage Formation breakdown Annular valve failed Missing data 7 P failed after closure Failed to close P Fractured formation Failed to stab string valve Missing data Failed barrier dist Table 10. No. blowouts () vs. primary/secondary barriers during the drilling phase (TexasIOCS, ) Primary barrier Swabbing Unexpected well pressidrlg. break Formation breakdown about a constant rate and are triggered by two main sons: 1) drilling into a formation we thought we knew, but found that formation pressure was higher than expected, combined with reduction of BHP caused by axial movement ofthe drilistring, causing swabbing; and ) too high mud weight and high equivalent circulation density (ECD) during circulation, or downward move ment of the string, leading to simultaneous loss of circu lation and kicks. Once the first pressure barrier was lost in such man ner, the kicks could have been controlled, if the Ps: had been closed, would have worked correctly, or had been installed as planned. It sounds simple, but occa rea L The author Pal Skalle, currently an associate professor at the Norwegian University of Science and Technology (NUST), Trondheim, received an MS degree from the University of Leoben, Austria (17), and a PhD in petroleum engineering from NUST (1). His main focus has been drilling fluid technology/hydraulics. Mi: Skalle is a member of NPF and SPE. A. L. (Tony) Podio is holder of the H.B. (Burt) Harking Ji: professorship in petroleum engineering at the University of Texas at Austin. His teaching and research interests are pri marily in drilling/produc tion, and include studies of well pressure control in drilling, kick control simulation, multi-phase flow, beam pumping efficiency, single- and multi-phase flow metering, and applications of sonics/ultrasonics. He has authored/co authored numerous papers on methods and equipment for analyzing performance ofpro duction systems, and has lectured extensively in the U.S., Europe and South America. 1 1 sionally, the real world is more corn- 1 plex than the simple rules of well con trol and blowout prevention. 1 The problems of kick detection, handling kicks and losing control do not seem to change much over the years. Improved technology and 10 improvements in procedural abilities 16 appear to have a tendency to be lost 1 when striving to maximize efficiency 6 and revenue, thereby overlooking vital pieces from the fine set of small mdi- cators. The solution to this complex problem may be found in enhanced motivation for risk reduction and bet ter knowledge and understanding of drilling operations. Who knows? ACKNOWLEDGMENTS The authors wish to thank all the entities that provided data and especially Ms. Belinda Wolf of the Texas Railroad Commission for her continued assistance and providing access to all the original files of blowouts in Texas. This article was prepared from paper SPE, of the same title, pre sented by the authors at the IADC/SPE Drilling Conference, Dallas, Texas, March 6, 1. LITERATURE CITED Hughes, V M. P., A. L. Podia and K Sepehrnoari, A computer-assisted analysis oftrends among Gulf Coast blowouts, in Situ, 1(), 10, pp. 01. Daneberger, E. P., Outer Continental Shelf drilling blowsuto , paper OTC 7, Proceedings, th Annual Offshore Technology Conference, Houston, May 6, 1, pp. 1. Th,entieth Century Petroleum Statistics, Slot edition, DeGolver & MscNaugton, Dallas, Texas, 1. Basic Petrslesm Data Book, Vol. XVI, No. 1, American Petroleum Institute, Washington, D.C., August 16. Wylie, W. W. and A. S. Visram, Drilling kick statistics, paper IADC/SPE 1, Prorn ceedings, IAIC/SPE Drilling Conference, Houston, Feb. 7 March, 10, pp Holand, P., Offshore blowsuts, causes and trends. PhD thesis, 600, NTNU, Trond helm, Norway, February 16, 7 WORLD OIL / JUNE 1