Well control

IADC/SPE 62770 Well Control of an Influx from a Fracture Breathing Formation Paul R. Ashley, SPE, Woodside Energy Ltd.

Copyright 2000, IADC/SPE Asia Pacific Drilling Technology This paper was prepared for presentation at the 2000 IADC/SPE Asia Pacific Drilling Technology held in Kuala Lumpur, Malaysia, 11–13 September 2000. This paper was selected for presentation by an IADC/SPE Program Committee following review of information contained in an abstract submitted by the author(s). Contents of the paper, as presented, have not been reviewed by the International Association of Drilling Contractors or the Society of Petroleum Engineers and are subject to correction by the author(s). The material, as presented, does not necessarily reflect any position of the IADC or SPE, their officers, or members. Papers presented at the IADC/SPE meetings are subject to publication review by Editorial Committees of the IADC and SPE. Electronic reproduction, distribution, or storage of any part of this paper for commercial purposes without the written consent of the Society of Petroleum Engineers is prohibited. Permission to reproduce in print is restricted to an abstract of not more than 300 words; illustrations may not be copied. The abstract must contain conspicuous acknowledgment of where and by whom the paper was presented. Write Librarian, SPE, P.O. Box 833836, Richardson, TX 75083-3836, U.S.A., fax 01-972-952-9435.

Abstract Hydrocarbon influxes into the wellbore while drilling are typically caused by over-pressured formations, where mud weight is insufficient to control the pore pressure of the formation drilled. Formation permeability and over-pressure determine the intensity of the kick. Well kill procedures in such situations are standard industry wide. Shut in drillpipe pressure is used to determine the pore pressure of the formation, and the driller’s or weight and wait method is used to remove the hydrocarbon influx from the well. Hydrocarbon influxes associated with fracture breathing are much less common, and methods to deal with such influxes are not taught within the drilling industry. This paper describes two case studies: the first, Bard–1, an exploration well in the Timor Sea, offshore Australia, describes a well control situation that was initially interpreted as over-pressure. The well could only be killed by pumping cement. Investigation showed that the influx mechanism was a fracture breathing and hydrocarbon swap out mechanism. The high mud weights used in the attempt to kill the well actually exacerbated the situation. The second case study, Jura-1 which was located nearby Bard-1, describes design and procedures used to manage a similar influx mechanism. These were implemented successfully enabling the well to achieve its objectives. Introduction Bard-1 was drilled in October 1998. The well is located in the Timor Sea, offshore northwest Australia. The 9 5/8” casing was set at 1460m and a formation integrity test (FIT) performed, limited to 1.41sg. While drilling ahead at 2164m mud flow increased at surface followed by significant gas readings. The well was shut in and conventional (wait and

weight) kill operations commenced. After raising the mudweight to 1.31sg and then 1.36sg the well appeared to be killed, was opened and found to be static. However, after circulating bottoms up, gas peaks were registered and the well return flow increased. Despite several further increases in mud weight to a maximum of 1.48sg, the ingress of hydrocarbons into the wellbore could not be prevented. At various stages of the attempted kill the shut in drillpipe pressure (SIDPP) was reduced to zero, suggesting the well was overbalanced by the current mud weight. Approximately 70bbl of oil with associated gas were recovered to surface during well kill operations. Efforts continued for several days before the well was abandoned with a series of barite pills and cement plugs. The subsequent investigation concluded that the ingress method was a fracture breathing/fluid exchange mechanism instead of over-pressure. The variations in bottom hole pressure due to circulation were causing charging of the fracture system near TD, which then returned hydrocarbons and mud when circulation ceased. This swap out mechanism appears to have been exacerbated by the high mud weights used to attempt to kill the well. Jura-1 was drilled on the same structure approx. 45 km from Bard–1 in July 1999. The well was designed to enable management of a “Bard event”. Casing design and wellsite procedures are described in this paper. Casing was set close to the potentially fractured formations in order to permit a low mud weight to be used and thereby minimise overbalance. Special procedures to shut down and start up the mud pumps were implemented to reduce bottom hole pressure fluctuations. Jura–1 also encountered a fracture breathing formation, with return mudflow of up to 15bbl when circulation ceased during drillpipe connections. During these flow back periods, the “flow signature” was plotted i.e. volume vs. time. Comparison of “flow signatures” helped to identify that the well remained over balanced despite flow being excessive by normal well control standards. Jura-1 was drilled to TD successfully. Bard-1 Well Design and Objectives Well Location. Bard-1 was an exploration well in the Bonaparte Basin of the Timor Sea, offshore northwest Australia. It was located 444km from the port of Darwin in block ZOCA 95-19. This block was then part of the Australian/Indonesian Zone of Cooperation. The primary

2

PAUL R. ASHLEY

objective of Bard-1 was to test the presence of gas in the Plover Formation on the western flank of the Troubadour Structure. Figure 1 shows Bard’s location and the adjacent wells. The nearest wells to Bard-1 are Troubador-1 (Burmah Oil, 1974) 11km northeast, Sunset-1 (Shell, 1997) 14km northwest, and Sunrise-1 (Burmah Oil 1975) 23km northeast. Well Stratigraphy. The following is a summary of the lithologies at Bard-1 by stratigraphic unit. Refer to figure 2 for prognosed formations and depths. Tertiary. The tertiary section comprises the Barracouta, Oliver, Cartier, Prion, Hibernia and Johnson Formations. These are predominantly bioclastic calcarenite and calcilutite. Cretaceous. The Bathurst Island Group comprises the Cretaceous sequences exposed on Bathurst and Melville islands and in the Darwin area. They comprise argillaceous calcilutite, grading to Marl and Calcareous claystone. The Jamieson formation consists of partly calcareous claystone. It is a major lithological change from the carbonates above. The Jamieson clays are reactive to water based drilling fluids, and are typically over-pressured (see “pore pressure prediction” below). The Darwin Formation Radiolarite comprises argillaceous calcilutite, with hard calcareous claystone. Trace amounts of radiolarians are present. The Echuca Shoals Formation consists of hard, slightly glauconitic claystone. The Flamingo Group consists predominantly of claystone and minor siltstone. Sandstone may also be enveloped in the upper part of the interval. Middle-late Jurassic. The Plover Formation, the well objective, consists of an interbedded sequence of quartzoze sandstone, siltstone and claystone. Pore Pressure prediction. Matrix associated over-pressure was modeled for Bard-1 using Shell’s propriety DISCO pore pressure prediction program. This models pore pressure due to undercompaction, which is the typical cause of over-pressure in the region. The Bard model was run on the Sunrise-1, Troubadour-1 offset wells, and the results from this application did not contradict the actual and inferred pore pressure data from these wells. Figure 2 shows the predictive DISCO model for the Bard-1 location. A maximum pore pressure (1.30 – 1.31sg) was predicted for the interval 1850 to 1955 mRT within the Cretaceous Jamieson Formation. The Darwin Radiolarite, which has been associated with a number of kicks across the Timor Sea, had a predicted pore pressure of 1.12sg. It was assumed that the proximity of the Darwin Formation to the normally pressured Plover Formation allowed pressure relief of the Radiolarite into the adjacent porous reservoir below. This theory formed the basis for the modeling of lower pressures in the Radiolarite at Bard-1. Faults. Bard-1 was drilled to penetrate the Bard Horst. A cross section of the seismic line is shown diagrammatically in figure 3. In the area of Bard-1, the major structural events at Top NKA marker (Darwin Formation) are interpreted to penetrate top reservoir. The definition of the bounding faults (Fault A)

IADC/SPE 62770

to the south of Bard-1 is clear on the 2D data set. A smaller antithetic fault (Fault B) is apparent to the north of the main bounding fault. This antithetic fault dissects the well trajectory approximately 200m into the Jamieson Formation between 1870 and 1950 m RT. Between Faults A & B there is considerable offset, including pop-up and pop-down structures associated with the NKA event. Post-well interpretation also implies that Bard-1, although situated to the north of Fault B, penetrates the Darwin Radiolarite at a discrete inflection point (Fault C) which could be interpreted as a minor antithetic to Fault B. Well design. The objective of Bard-1 was to fully evaluate the primary Plover target while retaining a simple, safe and cost effective design. The proposed drilling plan drew upon knowledge gained from recent wells drilled in the immediate area i.e. Sunset West-1, Sunrise-2, and Sunset-1. A reduced casing design with a single 9 5/8” casing string set within the Cretaceous Bathurst Island Group was chosen for the well. An FIT limited to 1.41sg was planned after drilling out the casing shoe. This value was chosen based on knowledge of regional fracture strength, and was sufficient for 30bbl swabbed kick tolerance from TD. A test to leak off was not planned to minimise damage to the formation at the shoe. It was deemed unnecessary as local pore pressure knowledge was thought to be adequate. 8½” hole was to be drilled to TD using a high specification KCl/PHPA/Glycol mud system. An initial mud weight (MW) of 1.31sg was chosen to drill the section based on the maximum predicted pore pressure. It was planned to gradually reduce this mudweight after the well was drilled past the region of maximum pressure. This reduction was to be conducted in a controlled manner with continuous shale shaker monitoring in order that any signs of borehole instability were recognised early and remedial action taken. The minimum mud weight planned was 1.18sg corresponding to the predicted pressure within the Darwin Formation plus a 200psi overbalance. The reduced mudweight, (if possible), was intended to minimise the risk of both differential sticking and losses through the normally pressured Plover objective. Log of events Bard-1 was drilled in October 1998 by the Ocean Epoch semisubmersible. The 9 5/8” casing was set at the planned depth of 1460m in the Bathurst Island Group. After running and testing BOP’s, the casing shoe was drilled out and an FIT performed to 1.41sg. The 8 ½” hole was drilled using the planned mud weight of 1.31sg to 1900m, then mud weight was gradually reduced as planned, with no bore-hole problems. The well penetrated the Darwin Radiolarite, Echuca Shoals, and possibly the Flamingo Formation. While drilling with 1.25sg mud at 2164m flow increased at surface followed by significant gas readings. The following table logs events over the next 6 days: Increased flow from well – flow check, well static. Day 1 11:15h Resumed circulation – increased flow and gas to 7%. Shut in, SIDPP and SICP 0psi.

IADC/SPE 62770

12:45h

14:45h

18:45h

22:00h 22:30h

Day 2 00:00h08:00

01:15h

03:15h

06:45h

08:30

11:30h

WELL CONTROL OF AN INFLUX FROM A FRACTURE BREATHING FORMATION

Circulated well through choke as a precaution. Mud weight (MW) = 1.25sg. Shut in to check pressures after 45min circulation. SIDPP = 0 psi, SICP = 120psi. Circulated for 45 min, then shut in. SIDPP = 170 psi, SICP = 180psi. 2.25h spent circulating 1.31sg kill mud into well. SIDPP = 0 psi, SICP = 160psi. Circulated for another 1.5h to clear apparent annular gas. Shut in well and monitored for 45min. SIDPP = 110 psi, SICP = 150psi. Bled off 6.9bbl in stages until pressures zero. Monitored well on trip tank – static. Circulated choke/kill lines and riser to 1.31sg mud. Opened well. Initial minor flow of 2bbl over 30min, then well was observed static for 10min. Commenced circulating well conventionally with 1.36sg mud (increase in mud weight for trip margin) After 70% annulus volume had been displaced, the shaker gas alarm was triggered and a 20bbl pit gain measured. SIDPP = 0 psi, SICP = 80psi Continued to circulate. Approx. 1.5 times well bore volume over choke with 1.36sg mud until return mud weight was 1.36sg. Shut in. SIDPP = 0 psi, SICP = 135psi. Surmised that MW was correct, but hydrocarbons were still present in the annulus. Continued to circulate approx. 0.75 times well bore volume over choke with 1.36sg. Shut in. SIDPP = 90 psi, SICP = 135psi. Assumed trapped pressure, so bled off 3.9bbl. Now SIDPP = 20 psi, SICP = 60psi. Circulated approx. 1.0 times well bore volume over choke with 1.36sg mud. (Note that throughout circulation, gas returns were observed from the mud gas separator, and an oil scum was observed in the mud leg of the separator). Shut in. SIDPP = 100 psi, SICP = 130psi. Monitored pressures for 1h, no change. Circulated approx. 1.25 times well bore volume over choke with 1.36sg mud, while holding an additional 100psi back pressure at choke. Shut in. SIDPP = 200 psi, SICP = 260psi. Assuming trapped pressure, 16.5bbl mud was bled off in stages over a period of 1.25h. SICP and SIDPP fell fairly equally apparently supporting the theory. With SIDPP = 0 psi and SICP = 25psi the well was left open at the choke. 2.5bbl fluid bled back over one hour. Circulated choke/kill lines and riser to 1.36sg mud. Opened well. There was an initial minor flow of 1.4bbl over 15min, then well was observed

3

static for 10min. Commenced circulating well conventionally with 1.36sg mud. After 65% annulus volume had been displaced, a gas peak of 17% and a 10bbl pit gain was measured. SIDPP = 0 psi, SICP = 80psi At this stage of the attempted well kill the SIDPP had again been reduced to zero, suggesting that the mud weight was sufficient to overbalance the formation pressure. However the well was still subject to static gains of hydrocarbons during bleed off periods, and the amount bled back was far in excess of the storage volume expected due to trapped pressure. The influx mechanism was believed at this stage to be from a high pressure/low permeability formation. This type of mechanism had been encountered in the Darwin and Echuca Shoals formation in a number of wells in the Timor Sea on previous occasions, though none was drilled by Woodside. The forward plan was to attempt to kill the well using additional mud weight. The following sequence of events led to the eventual abandonment of the well: 14:30 Circulated approx. 1.5 times well bore volume over choke with 1.36sg mud, while holding an additional 75psi back pressure at choke. Shut in. SIDPP = 60 psi, SICP = 200psi. 17:00h Circulated approx. 3 times well bore volume over choke with 1.38sg mud. Shut in. SIDPP = 100 psi, SICP = 150psi Bled off 8.1bbl mud, SIDPP = 30 psi, SICP = 60psi. 21:45h Weighed up mud in pits to 1.42sg. Note that the FIT at the casing shoe was limited to 1.41sg, however offset data suggested a leak off strength of up to 1.5sg 22:30h Commenced circulating over choke with 1.42sg mud. Completed 2 circulation’s with 1.42 sg mud. Day3 Returns were still heavily gas cut. Circulated well to 1.44sg mud, gas cut returns and residual casing pressure. Circulated well to 1.48sg mud. During circulation, pit losses of 30bbl/h were noted at 4bbl/min circulation rate. Losses reduced to zero with 3 bbl/min pump rate. The losses were assumed to be below the casing shoe, so a 50bbl lost circulation material (LCM) pill was spotted across and 300m below the shoe. The LCM pill was allowed to soak for 2h. Day4 SIDPP = 60 to 90 psi, SICP = 160 to 170psi. Circulated LCM pill from wellbore. Max 17% gas was measured from above LCM pill, reducing to 1% with pill at surface. Continued to circulate 1.48sg mud over choke (approx. 2 well bore volumes) Circulated second 50bbl LCM pill and spotted across to 300m below shoe. Shut in well for 2.5h to allow pill to soak. SIDPP = 30 psi. Meanwhile 13:30h

4

Day 5-8.

PAUL R. ASHLEY

squeezed 7bbl mud down kill line. Final SICP = 190psi. Circulated LCM pill from wellbore, and attempted to increase circulation rate to 4 bbl/min. Losses of 20bbl/h were induced, which reduced to zero with 3.5bbl/min circulation rate. Note that the LCM pill circulated from the well was heavily contaminated with oil. Continued to circulate well over choke with 1.48sg mud (approx. 5 wellbore volumes). At this stage it was decided to abandon the well by pumping a cement plug via the drillstring, cementing the BHA in place. This was due to the concern that any further increase in mud weight would cause additional, potentially severe losses. A cement plug was set from 2164m to approx. 1850m, then allowed to set for 10h. The drillpipe was severed at 1824m, and the well was circulated at 3 bbl/min to condition the mud. However, despite circulating for 8h, gas levels could not be reduced below 5%. A 50bbl barite pill was spotted and allowed to settle. The drillpipe was then perforated at 1730m, and circulation established. After two circulation’s gas levels were reduced to 0.3%. The well abandonment was completed with a second barite pill and a cement plug inside the casing.

Discussion of influx mechanism Behaviour of the influx at surface, where gas was continuously evolving from the mud, formed gas pockets only after being shut-in and showed relatively high wetness ratios, strongly suggests that the influx was oil. Corresponding increases in mud oil content confirmed this (see figure 5). The formations penetrated below the Jamieson shale are thought likely to have fractured lithology, where storage capacity and permeability of the rock are provided by the fracture system. This is consistent with known characteristics of the formations penetrated, i.e. Darwin Radiolarite and Echuca Shoals. Post well analysis shows that Bard-1 penetrated an inflection point on the seismic line that may represent anti-thetic faulting (shown as fault C in figure 3). The following two theories are postulated, over-pressure and swap-out mechanisms: Over-pressure. The first theory is that the well penetrated a zone of high over-pressure that could not be controlled, despite increasing mud weight to 1.48sg. The formation penetrated appears to have low permeability. It may be a siltstone or fracture system in the Flamingo Formation. Swap-out Mechanism. The second theory is that hydrocarbons entered the wellbore via a swap out mechanism. This postulates that pressure fluctuations associated with drilling caused an opening and closing (or “breathing”) of oil filled rock fractures in the Flamingo Group, that resulted in

IADC/SPE 62770

mud and oil being exchanged with the wellbore, and that subsequent hydrostatic pressure increases magnified these exchanges. The fracture system may have extended from the Flamingo Formation up into higher formations of the Echuca Shoals, Darwin Radiolarite or Jamieson. Over-pressure. Data from the early part of the well kill suggests that some degree of over-pressure was present that caused the initial increase in well flow. SIDPP of 170 psi and SICP of 180 psi were recorded at 14:15h on day 1, and mud weight was increased to 1.31sg on that basis. Subsequently the well was bled off, opened up, and found to be static. There was some anomalous data, however, initial SIDPP was 0psi, and the higher pressure was not recorded until after one hour of circulation over the choke. This may have been due to low permeability. Subsequently there is body of evidence to suggest that a low-permeability, high-pressure formation was not the mechanism at work. This includes: SIDPP does not decrease with increased mud weight. Figure 5 shows that there was no decreasing trend in SIDPP after a mud weight increase by 700psi at TD. Even if the final mud weight of 1.48sg was underbalanced, it is expected that the SIDPP would have decreased as weight increased. The well was bled off to zero after the mud weight had been increased to 1.31sg and showed only very slight flow, reducing to static. This would not be possible if the well were still underbalanced unless the over-pressured zone was very tight. The well was again bled down and opened with 1.36sg mud in the hole. SIDPP was bled off to 0psi on multiple occasions with no subsequent increase. This suggests overbalance on bottomhole pressure. There was a significant volume of hydrocarbon influx without associated pit gains. Figure 4 shows that the percentage of oil in the mud increased to 9% over the duration of well control operations. The active mud system was approximately 1000bbl with initial 2% oil in mud due to glycol content. The increase equates to approximately 70bbl of oil influx, however there were no net pit gains over the period. Higher pore pressures in the Darwin Radiolarite and the Flamingo Group can be modeled in DISCO only if the stratigraphic thickness assumptions are changed. Pore pressures equivalent to 1.48sg are possible if the sequence from the Flamingo to Jamieson inclusive, is increased in thickness by 500 m. There is no seismic evidence to support an increase of this proportion either at the Bard location or at locations immediately adjacent to the Bard Horst. Swap-Out. Data supporting a swap out mechanism includes: There was a hydrocarbon influx without pit gain. Approximately 70 bbl of oil entered the mud system, with a corresponding loss of mud from the wellbore. SIDPP = 0psi was recorded on multiple occasions. This suggests overbalance on bottom-hole pressure, and supports the swap-out theory.

IADC/SPE 62770

WELL CONTROL OF AN INFLUX FROM A FRACTURE BREATHING FORMATION

On a number of occasions the surface pressures were bled off, as they were believed to be “trapped pressure”. During bleed off SIDPP and SICP fell roughly equal amounts, which was consistent with trapped pressure. However, the volume of fluid returned when bleeding off the casing pressure was much more than expected for the pressure decrease. Using compressibility data from the 9 5/8” casing pressure test, approx. 0.5bbl would be expected to bleed off for a 100psi pressure drop. Instead, 10bbl or more was bled off to achieve the same pressure drop. This strongly suggests that bleed back is causing open fractures to close, and that the volume compressed is far in excess of the wellbore volume. The cause of the trapped pressure is thought to have been additional bottom hole pressure due to wellbore and choke line friction, and applied back pressure at the choke creating supercharging of fractures. The influx of hydrocarbons did not respond to increasing mud weight and continued to increase steadily until placement of a cement plug and the pumping of the barite pills (see figure 4). Increased mud weight would not be expected to prevent swap out and may increase the rate at which it occurs, by causing additional fracture opening. After the mud weight was raised above 1.44sg to 1.48sg, losses were experienced when circulating at rates of 4 bbl/min, though they could be controlled if pumping was reduced to 3 bbl/min. At this time losses were being followed by gains which suggests fractures/or a fracture zone were being flexed in and out. During this time, SIDPP and SICP were generally higher, possibly due to increased charging of the flexed zones and shutting in before they could bleed back. Post Bard-1 Conclusions and Recommendations The mechanism was concluded to be a fluid swap-out, where mud displaces hydrocarbon from fractures. The fractures are likely to be localised in the Flamingo Group, and Echuca Shoals Formations. Fluid exchange was thought to be associated with a “breathing” mechanism where fractures open and close as overbalance fluctuates with bottom hole pressure. The following recommendations were suggested to better manage well control on future wells in the area: • Consideration should be given to placing a casing seat into the Jamieson Formation in order to isolate a return fault path should one be intersected. Such a design would offer the contingency of higher mud weights than those used on Bard-1, should the pore-pressure prediction be wrong. • Leak off tests should be conducted on exploration wells whenever possible. The exception should be where the casing shoe is set in a brittle formation type. • A drill-ahead strategy in case of re-occurrence should be developed in order to manage any influx safely. • Locations which show seismic evidence of structural complexity at the Darwin Radiolarite marker, and consequent likelihood of fracture zones, should be avoided.

5

Jura-1 Well Design and Objectives Location. Jura-1 was an exploration well located offshore in the Northern Bonaparte Basin, appoximately 45km from Bard1 (figure 1). The primary objective of Jura-1 was to test the presence of hydrocarbons in the Flamingo and Plover Formations. Jura-1 was located in an area of good seismic data quality, and was sited to avoid potential fracture zones associated with the Jura prospect bounding faults, a learning point from Bard-1. Well Stratigraphy. The predicted stratigraphy is similar to that encountered in Bard-1. For lithology description refer to Bard-1. Predicted formation tops are shown in figure 6. Pore Pressure prediction. The formation pressure profile was again modeled using the ‘DISCO’ programme and is shown in figure 6. A maximum pore pressure of 1.25 – 1.31sg was predicted within the Jamieson Formation. The range was dependent on the degree of development of sands in the Flamingo Formation. The pore pressure in permeable formations, with the exception of the Darwin/Echuca Shoals Formations, was expected to be normal, in line with observations in previous wells. The pore pressure at the top of the Darwin/Echuca Shoals Formations was predicted between 1.13 -1.27 sg. The lower pressure was in the case of normally pressured sands in the deeper Flamingo Formation, the higher in case of shales. Well design. The objective of Jura-1 was to evaluate the hydrocarbon potential of the Flamingo and Plover sands whilst allowing for a possible re-occurrence of the problems encountered whilst drilling Bard–1. 13 3/8” casing was planned to be set in the Turnstone Formation, then the shoe drilled out and an FIT to 1.40sg performed. This gave adequate kick tolerance in the unlikely case of an overpressured kick from the Jamieson formation. 12¼” hole was planned to be drilled to 2250mRT in the lower Jamieson formation, using a KCl/PHPA/Glycol mud system with initial mud weight of 1.2sg. This was deliberately underbalanced to the DISCO prediction. A leak off test was planned after setting 9 5/8” casing. Providing the Jamieson Formation remained stable in 12 ¼” hole, it was planned to reduce mud weight to 1.15sg in 8 ½” hole, so that the Darwin Radiolarite and Echuca Shoals formations would be penetrated with minimum overbalance. This was expected to minimise any fluid exchange tendency in these formations, as observed at Bard–1, should hydrocarbon bearing fractures be present. A much improved kick tolerance compared with Bard-1 was also expected, by setting the shoe in a claystone close to the Darwin Radiolorite Formation. Additional precautions taken included installation of a kick detection system in the mud return flow line, and a Pressure While Drilling (PWD) module was included in the BHA.

6

PAUL R. ASHLEY

Special Drilling procedures. Specific procedures for drilling the Darwin Radiolarite were developed and implemented. They adopted the following principles: • Reduction of equivalent circulating density (ECD) by using a reduced circulation rate. • Monitoring ECD using the PWD tool. • Performing a slow, staged mud pump shut down and start up procedure over 5 minutes in order to reduce fracture closure gradually, enabling any fluid swap out to enter the well bore in a progressive and controlled manner and then be circulated out as oil or gas cut mud. • To reduce ECD while circulating over the choke, mud return flow was to be diverted through both the choke and the kill line. • If a fracture breathing mechanism was encountered, “flow signatures” to be monitored, by plotting bleed back volumes vs. time. • If hydrocarbons were encountered, then every time the pumps were shut down, bottoms up was to be circulated over the choke and hydrocarbon content reduced to a acceptable level before opening the BOP. • Before drilling into the Darwin, the drill string was configured with at least three IBOP valves below the rotary table. These were spaced out one drillpipe-stand apart and were intended to enable the drill string to be stripped out if required • A decision tree was developed to assist wellsite personnel in the case that a “Bard style” hydrocarbon swap out mechanism was encountered. This is shown in Appendix 1. Well Construction. Jura-1 was drilled in July 1999 by the Ocean Epoch semi-submersible. The 13.3/8” casing was set at the planned depth of 1695m in the Turnstone Formation. After running and testing BOP’s, the casing shoe was drilled out and an FIT performed to 1.40 sg. The 12¼” hole was drilled with 1.2sg KCL/PHPA/Glycol mud as planned to 2303m. There were no signs of hole instability. After setting 9.5/8” casing, the shoe was drilled out and a leak off test performed to 1.6sg equivalent mud weight. The 8½” hole was drilled using the planned mud weight of 1.15sg. Top Darwin Formation was identified using FEWD and the special drilling procedures were implemented. This required a staged shut down and start up of the mud pumps at the first drillpipe connection, then circulating bottoms up over the choke following the connection. No hydrocarbons were returned, and conventional drilling practice was continued with a reduced circulation rate of 400gpm. At 2431m, in the Echuca Shoals Formation, pump rate was increased to 600gpm, which resulted in an increase in bottom hole ECD from 1.2sg to 1.25sg, measured by the PWD sub. This increase in pressure caused downhole losses at a rate of 250bbl/h. When flow rate was reduced to 400gpm, the losses ceased. At the next connection (2462m in the Echuca Shoals Formation), the well was flow-checked and a 12bbl gain measured in a period of 8.5min. The well was shut in and pressures monitored, SIDPP=0psi and SICP=60psi. Due to the

IADC/SPE 62770

zero DP pressure, it was thought that a Bard-1 mechanism had been encountered again. It appeared that the fracture/loss zone was returning drilling fluid to the wellbore when the pumps were stopped due to the decrease in ECD. The casing pressure was bled off via the choke to the trip tank, returning 8.6bbl in 30min. The well was then shut in, SIDPP=0psi, SICP=30psi. Bottoms up were then circulated through the choke and kill lines, however, unlike Bard-1, no hydrocarbons were returned. The well was shut in again, with SIDPP=90psi and SICP=75psi. The pressure bled off via the choke to the trip tank with 7.4bbl returned in 20min. The well was then opened to monitor flow. There was a return of 1.4bbl in 10min. The well was now near static, and the evidence suggested that the well was returning drilling fluid without hydrocarbons from the fractured zone. The next stand was drilled down to 2490m (Flamingo Formation). The kick detection system was used to monitor for any increase in mud return flow; none was recorded. The well was flow-checked at the connection and returned 13.2bbl in 8min. The well was shut in, SIDPP=60psi, SICP=40psi. This pressure was bled off via the choke to the trip tank, 1.8bbl was returned in 10min. The “flow signature” (i.e. volume returned vs. time) was recorded. As a precaution, bottoms up was again circulated through the choke and kill lines with no hydrocarbons returned (maximum gas was 0.07%). The well was flow-checked, and 2.9bbl were bled back via the choke in 10min. The well was opened to monitor flow, and there was a return of 1.8bbl over 10min. The “flow signature” was the same as previously recorded. At this stage it was decided to drill to TD, monitoring for increased flow using the kick detection system, and performing extended flow checks on connections. The next stand was drilled to 2520m, and the well flow-checked, returning 13bbl in 25min. The “flow signature” was again similar to the previous one (see figure 7). The flow back was interpreted to be the same fracture breathing mechanism closing and returning mud after being opened from increased bottom hole pressure due to circulating. The well was drilled to TD at 2598m. Conclusion The flow back from the 81/2” hole on Jura-1 appeared to be driven by a similar mechanism to Bard-1, i.e. a fluid return mechanism, driven by localised charging of a fractured formation open to the wellbore. The charging appears to have been caused by increased bottom hole pressure from circulating. The essential difference on Jura-1 was that the formation did not contain hydrocarbons. The decision to use the low mud weight of 1.15sg appears to have been justified in reducing the amount of pressure differential across the fractured zone. Some of the important factors that enabled Jura-1 to be drilled to TD were: Use of the “flow signature” to compare flow back on connections and confirm the mechanism was behaving in the same way. A change in flow back would have been interpreted as potential hydrocarbon influx, and circulated over the choke.

IADC/SPE 62770

WELL CONTROL OF AN INFLUX FROM A FRACTURE BREATHING FORMATION

It will be noted that this methodology of deliberately allowing the well to flow is contradictory to normal well control practice! The kick detection system was invaluable. It helped to confirm that the well was overbalanced, despite return flow on connections. It enabled early detection of an influx or losses with an accuracy of less than 2bbl. The PWD sub was a useful tool to effectively monitor bottom hole pressure and manage ECD. Nomenclature FIT= Formation Intake Test sg= Specific Gravity SIDPP= Shut In Drill Pipe Pressure SICP= Shut in Casing Pressure BOP= Blow Out Preventor TD= Target Depth ZOCA= Zone of Cooperation “A” MW= Mud Weight ECD= Equivalent Circulating Density EMW= Equivalent Mud Weight LCM= Lost Circulation Material BHA= Bottom Hole Assembly Gpm= Gallons per minute Acknowledgements Thanks to Phil Scott, Gary Jones, and Rebecca Jones for their work on the Bard-1 investigation team, and their support in writing this paper. Thanks to Nigel Walters for preparation of the Jura-1 drilling programme, and to Vince Tilley for preparing the Jura-1 special drilling procedures.

7

127°30’ E

128°20’ E

2000m 9°10’ S

-9°10’ S

1000m m 2000

F EO ZON

ION RAT E P O CO

LETI INTERNATIONAL BOUNDARY NT/P55

AREA C

Sunrise 2

Loxton Shoals 1

200m Sunset West 1

ZOCA 96-20

Sunrise 1 Sunset 1

ZOCA 95-19 ZOCA 95-19

NT/RL 2 NT/P 56

Jura-1

Bard-1

Troubadour 1

ZOCA 94-07 Inset Map

-10°00’ S

OR TIM

Kelp 1

ZOCA 95-18

AC / P8

N. T.

BROWSE BASIN

ZOCA 98-24

ZOCA 95-17 127°30’ E

BONAPARTE BASIN BARD - 1 & JURA - 1 : LOCATION MAP

W. A.

Sikatan 1(ST)

. N.T

Kelp Deep 1/DW1

ZOC BONAPARTE BASIN A. W.

VULCAN SUB - BASIN

Figure 1

10°00’ S

Kupang

PETREL SUBBASIN

DARWIN SHELF

Darwin

NT/P 47 W.A.

0

25km

N.T.

128°20’ E

Ref. Dwg. 400-14235

Predicted Depth

Geological Time

mRT

Unit

Lithology

Lithological Description

Pressure Profile

Casing Depths ( mRT )

Pressure ( S.G. Equivalent )

124

1.0

Calcilutite

1.1

30” / 13-3/8” @ 157m

1.2

1.3

1.4

Legend

Alaria Formation

Pore Pressure Range (S.G.)

Calcarenite

Mud Weight Range (S.G.) Losses experienced in Troubadour-1 to 1200m

461

Calcarenite

500

Calcareous Claystones Oliver Formation

Calcarenite with minor Dolomite Calcarenite Calcilutite, Marl & Calcareous Claystone

917

Cartier Formation 1087

Calcarenite and Calcilutite interbedded with Calcareous Siltstone, Claystone and thin Sandstone

1000

Prion Formation 1258

Hibernia Formation 1377

Johnson Formation

1462

Turnstone Formation

Argillaceous Calcilutite with occ. Chert Calcilutite with trace Claystone

9-5/8” @ 1545m

1500

Argillaceous Calcilutite, Marl, Calcareous Claystone

1642

Jamieson/ Lower Wangarlu Fm.

Calcareous Claystone & Claystone 2000

2149 2179 2189 2234

Darwin Formation Echuca Shoals Fm

Flamingo Group

Plover Formation

2492

Argillaceous Calcilutite (radiolarite) Claystone & Siltstone Interbedded Sandstone, Claystone & Siltstone Claystone w/minor Siltstone & Sandstone Sandstone 2500

Figure 2

BARD - 1 PREDICTED SECTION AND PRESSURE PROFILE

Ref. Dwg. 400-14377

NNW

SSE

Bard - 1 Late Cretaceous + Tertiary Carbonates Fault A

Fault B

Jamieson Formation Shales Fault C

Darwin Radiolarite Flamingo Group

Plover Formation

Bard Horst 0

1

2 km Ref. Dwg. 800-22117

Figure 3

CROSS - SECTION OF THE BARD HORST

%OIL & %Gas vs Time 18

16

14

% gas

%Oil / % Gas

12

10

8

% oil

6

4

2

gas peaks

0 0

1

2

3

4 Time (Days)

Figure 4

5

6

7

SICP-SIDPP & Mud Weight vs Time 800

700

Mud Weight (as equivalent bottom hole pressure)

Pressure (psi)

600

500

400

300

200

100 SICP - SIDPP 0 0.00

1.00

2.00 Time (Days)

Figure 5

3.00

4.00

PREDICTED FORMATION TOPS

Approx Depth mRT

GROSS LITHOLOGY

Lithological Description

Pressure Profile

Casing Depths ( mRT )

1.0

387

Barracouta Formation

1.4

Calcilutite

Calcarenite

627

Pressure ( S.G. Equivalent ) 1.1 1.2 1.3

30” / 20” @ 421m ± 5m

500

Calcarenite

Losses experienced in Troubadour-1 to 1200m

Calcareous Claystones fault

Calcarenite with minor Dolomite Oliver Formation

Calcarenite Calcilutite, Marl & Calcareous Claystone 1063

Prion Formation

1000

Legend

Calcarenite and Calcilutite interbedded with Calcareous Siltstone, Claystone and thin Sandstone

Pore Pressure - Flamingo Fm Developed (S.G.) Pore Pressure - Flamingo Fm Underdeveloped (S.G.) Mud Weight Range (S.G.)

1285 fault Hibernia Formation

1411

Argillaceous Calcilutite with occ. Chert

Johnson Formation

1500

1510

Calcilutite with trace Claystone

13-3/8” @ 1550m

Turnstone Fm

Argillaceous Calcilutite, Marl, Calcareous Claystone 1875

2000 Calcareous Claystone & Claystone Jamieson Fm

Darwin Formation

2348

Argillaceous Calcilutite (radiolarite) Claystone & Siltstone

9-5/8” @ 2250m

2381

Interbedded Sandstone, Claystone & Siltstone

Flamingo Group

2504 Laminaria Fm Plover Formation

Figure 6

2542

Claystone w/minor Siltstone & Sandstone Sandstone

2500

2592

JURA - 1 PREDICTED SECTION AND PRESSURE PROFILE

Ref. Dwg. 400-14314

Jura-1 Static Flowback Signatures 14.0

12.0

Volume (bbl)

10.0

8.0

2462.5m 2519.6m 2547.9m 2598.0m

6.0

4.0

2.0

0.0 0

10

20

30 Time (minutes)

Figure 7

40

50

60