By Ray T Oskarsen, John W Wright, John Wright Company; Ole B. Rygg, Thomas Selbekk, Add Wellflow Dynamics; and Mike Allcorn, Alert Well Control
This article summarizes the innovative use of a relief well as a cost-effective and safe alternative for a non-blowout well control intervention.
The incident was initiated by a catastrophic failure of a kelly-saver sub during tripping, which resulted in the dropping of the drillstring. While attempting to spear the fish, the well began to flow. Subsequently, 60 bbl was gained and a stabilized shut-in pressure of 3,000 psig was recorded after closing the blind rams. The influx appeared to be mostly oil in an oil-based mud without migration.
Conventional control options consisted of attempting to bullhead the influx or bring in a snubbing unit to fish the string under pressure and circulate the influx out. Bullheading was ruled out as impractical due to a weak shoe, a strung out influx and a long open-hole section. Snubbing would require the importation of an expensive unit and crew. Quick economic calculations indicated a relief well could be drilled at a lower cost.
Using a combination of wireline electromagnetic and MWD-magnetostatic ranging, we successfully located and intersected the problem well. The intervention strategy was to displace the hydrocarbons using a modified driller’s method, equivalent to having the bit off-bottom. In this case, mud was circulated down the relief well annulus with returns up and out through the choke manifold of the problem well.
Table 1: Several considerations should be made
when deciding whether to drill a relief well.
In December 2006, a well was spud in eastern Syria, close to the Iraqi border. At 3,838 m, during a trip to change the bit, the drillstring parted at the kelly-saver sub and dropped 28 m. To complicate matters, the well took an influx of 60 bbl while the operator was waiting for the appropriate fishing equipment. The rig crew responded by closing the blind/shear rams, which resulted in a stabilized shut-in pressure of 3,000 psig.
The two primary options considered for forward intervention included bullheading heavy mud down the annulus to push the hydrocarbons back into the reservoir or bringing in a snubbing unit to recover the fish under pressure and circulate the well dead.
Bullheading calculations indicated the formation would fracture below the shoe, leaving hydrocarbons in the open hole, and in the worst case cause a broach to surface behind the casing. There were no snubbing units in the country, and import/export costs of the spread would be high.
The alternative was drilling a relief well with an existing low-cost rig, which had limited risk and a high probability of success. Hence, the operator decided to drill a relief well that would intersect above the formation that was predicted to be the origin of the influx.
Several issues complicated this relief well operation: No gyro-surveying tools were available in the country, the choke operator and pump operator were located on separate rigs, and the target well contained a combination of oil, gas and several slugs of mud with different densities. Ultimately, the combination of electromagnetic ranging and passive-magnetics analysis facilitated an efficient well-intersection process, and the hydrocarbons were circulated out according to the drill pipe pressure-decline schedule, which was developed using a transient multiphase model.
Bullheading is typically the first intervention method attempted due to its low cost, but it is also the method that will exert the highest pressures on the surface equipment, casing and open-hole formation. In this case, the open hole was long, the annulus was narrow and filled with viscous oil-base mud, and the influx was strung out. The resulting bullhead pressures on the formation indicated the chance of success was very low.
Relief wells are considered by many in the industry as a last resort that should be considered only when all other options fail. However, Table 1 shows circumstances when the best intervention for a well control situation may be a relief well.
This 2006 incident was not the first time the operator had encountered a well control situation in this area. In 1995, a much more disastrous incident took place at a nearby location. The blowout in 1995 occurred after losing returns during the final angle-building phase of a well that had been planned to be a horizontal well. Despite several unsuccessful attempts at pumping and spotting diesel and lost-circulation material pills to halt the fluid losses, the well pressure exceeded the rupture limits of the last casing string, and a broach to surface occurred from the shoe of the surface string. The blowout preventer stack remained intact, with the drillstring suspended 276 m off the bottom of a 1,563-m open-hole section. The well was flowing oil and gas at extremely high rates.
Two attempts to kill the well from surface through the drill pipe failed, and the rig was skidded off the location. Although two direct intersecting relief wells were required based on the hydraulic calculations, the blowout severity, the kill program complexity and the intervention time mandated that a third relief well be drilled to provide reservoir flooding capabilities and backup options. Two of the three relief wells, which were planned to intersect the blowout well, intersected their respective targets two days apart, and, following a dynamic kill, the blowout was brought to static conditions.
Eleven years later, the same operator spud a well a few kilometers from the location of the 1995 blowout. A 12 ¼-in. hole was drilled to 207 m, where 9 5/8-in. casing was run and cemented. Further drilling was carried out in 8 ½-in. hole to 2,241 m and 7-in. casing was cemented at 2,231 m. No drilling complications were encountered in this part of the well.
Following usual practice in this field, the operator took no leak-off formation data after setting the 7-in. casing. The large set of data available on the shoe strength indicated that the formation strength was 14.5 ppg.
Figure 1: The relief well (schematic not to scale) intersected
the target well at 2,828-m MD for successful kill intervention.
When the 6-in. hole was drilled with 9.5-ppg mud to 2,952 m, gas levels up to 30% were observed, indicating that the top of a hydrocarbon-bearing sand had been penetrated. The sand, which in this text is referred to as Sand A, has a fracture system but a very low permeability matrix. Sand A is also a relatively high-pressured sand, where taking kicks is common. The nature of the fracture system and the fairly tight matrix cause the formation to influx at low volumes; hence, it is common to drill this section underbalanced and flow the sand to surface prior to drilling through the deeper but lower-pressured reservoirs, which were the well objectives.
After the well entered Sand A, the mud weight was increased to 9.9 ppg. The first influx of 10 bbl, which was taken at 3,113 m, was circulated out via the rig’s choke manifold system. Drilling continued to 3,338 m in rotary mode while allowing the sand to flow dynamically. The operator decided to completely deplete the sand before picking up directional equipment to drill further; a total of 2,947 bbl was produced from Sand A through the drill pipe casing annulus, first through the rig’s choke manifold and later through a mobile production test unit.
Once the sand was sufficiently depleted, the well was displaced in stages to 11.4-ppg oil-based mud (OBM), and drilling continued in directional mode. At 3,464 m, flow rate increased, so the well was closed in and Sand A was depleted further by 150 bbl; total depletion since initial kick was 3,298 bbl. Subsequently, the mud weight was increased to 11.7 ppg before entering the target reservoir, after which the well was drilled to 3,838 m.
LEAD-UP TO WELL CONTROL INCIDENT
At 3,838 m, during a trip to change the bit, a total gain of 24 bbl was recorded and circulated out. The bit was subsequently pulled out and changed. When running back in hole, the hole appeared to be tight. During washing and reaming to bottom, the drillstring twisted off at the kelly-saver sub and the string dropped 28 m. The drillstring was greatly worn, and the outer diameter of the tool joint was reduced to a point where the assigned overshot was not usable and, to complicate matters, neither a suitable overshot or spear was available in the country.
While waiting for the appropriate fishing equipment to be mobilized, the rig crew attempted to fish the drillstring with a 5 7/8-in. overshot dressed with a 4 7/8-in. grapple, but the fish slipped off with 50,000 lbs. Following a further gain of 60 bbl at 45 BPH, the well was closed in using blind/shear rams. Initial attempts were made to control the well used dynamic lubrication to open the choke and circulate from the kill line over the choke, keeping the pit volume constant. The casing pressure subsequently reduced to 10 psi with pit level constant. However, when the blind/shear rams were opened, the well was venting gas below the bell nipple, so the well was once again closed in using the blind/shear rams. Casing pressure increased rapidly to 1,900 psi. During attempts to reduce the casing pressure (by circulating via the kill line and taking returns through the choke line while trying to keep the pit volume constant), the well took a second kick of 180 bbl, probably of oil.
Although several sands below Sand A had been penetrated, Sand A appeared to be the flowing formation because of its high pore pressure. A 10- to 30-BPM crossflow was likely occurring from Sand A down to one of the lower high-permeability formations.
Figure 2: Cell phones were used to communicate between
the relief well rig (foreground) and the problem well rig (background).
Following the increase in wellhead pressure to 1,900 psi, the rig crew carried out the static volumetric lubrication method – bleeding off lighter gas/mud while replacing it with heavier mud – to reduce the surface pressure. A total of 27 bbl of 15-ppg mud was lubricated. Because the annular cross section was narrow and OBM mud in hole was highly viscous, the kill mud would not swap with the original mud.
Available options identified for forward intervention included:
- Continuing volumetric lubrication with heavy mud (15 ppg), which would have had slow progress. Getting mud to swap with hydrocarbons would have been difficult.
- Bullheading heavy mud, which could have broken the 7-in. shoe and, in the worst case, led to a broach to surface.
- Stripping in coiled tubing and circulating into the drillstring, in which case success would depend on drillstring/annulus communication.
- Installing a snubbing unit and fish drillstring under pressure, except that no snubbing unit was available in the country.
- Relief well intersection, which had limited risk and a high probability of success
With the lessons learned from the blowout in 1995 and excessive cost of a snubbing spread, the operator was reluctant to attempt a surface intervention. Combined with the fact that the relief well option had a high probability of success and could be drilled with one of the many affordable rigs in the area, it was decided to attempt a relief well intervention.
RELIEF WELL STRATEGY
The objective of the relief well was to intersect above Sand A and displace the hydrocarbons using a modified drillers method, equivalent to having the bit off-bottom (Figure 1). In this case, mud would be circulated down the relief well with returns up and out through the choke manifold of the target well. Once the target well had been brought to static conditions, fishing operations could commence.
In late January 2007, a relief well was spud (Figure 2) with the following project milestones:
- Set + cement 13 3/8-in. casing.
- Set + cement 9 5/8-in. casing.
- Locate target well using electromagnetic ranging.
- Cross-by and triangulate relative position of target well.
- Set and cement 7-in. liner.
- Drill 6-in. hole to intersection point.
- Assess hydraulic communication with target well.
- Kill target well annulus.
- Kill target well drillstring if necessary.
- Fish out drillstring in target well.
RELIEF WELL CHALLENGES
Several issues complicated this relief well operation: No gyro-surveying tools were available in the country; the target well contained a combination of oil, gas and several slugs of mud with different densities; and the choke operator and pump operator were located on separate rigs. The planned method of communication between the two rigs was to use hand-held radios, which turned out to be unreliable. Instead, communication was accomplished using cell phones.
Since all relief wells are inherently different, the planning phase should always include dynamic modeling of the kill hydraulics (Rygg and Gilhuus, 1990).
After mobilizing a hydraulic-modeling team, we performed extensive dynamic simulations to determine the source of the kick, the pressures and potential fluid gradients in the well after the influx, and the feasibility of intersecting above Sand A. From the hydraulic modeling, we decided to plan for pumping a 12.5 ppg OBM at 2 BPM after intersecting the target well. This approach was expected to yield fluid pressures that would exceed the pore pressure gradients in the open-hole section of the target well while not exceeding the shoe fracture gradient.
Figure 3 shows the predicted pressure development for a number of initial scenarios and the actual pressure development when the job was performed.
Gas bleed-off early in the operation caused the rapid pressure drop in the initial phase. As mud started filling the annulus, the pressure decline followed the same trend as the simulated curves for our Scenario 1. Stabilization of pressure at zero psig occurred as expected. Volumes pumped when mud was observed in returns at surface indicate communication between the annulus and drill pipe at the top. After pumping operations were completed, we observed 154 bbl less fluid volume at surface than pumped in the problem well. This indicated a large volume of gas in the well at the start of the operations, mud losses into the formation, or a combination.
For embargo reasons, no north-seeking gyros were available in the country. North-seeking gyros are normally used to get a north-bearing azimuth for locating a target well because magnetic interference from the target well’s casing/drillstring corrupts the measurement while drilling (MWD) readings. The MWD readings were, however, instrumental in estimating high side-to-target direction using the target drillstring magnetic error vector (Jones et al, 1985).
This process matched the electromagnetic-ranging target direction reasonably well when the error field had a magnitude greater than 1 or 2 mT and the inclination was over 2º.
The electromagnetic ranging tool is run on wireline and consists of a sensor to measure the magnetic field and an electrode that injects current into the surrounding formations. If the current is picked up on the target well casing/drillstring, it creates a magnetic field that can be measured using the ranging-tool sensor. The sensor data indicate the direction and distance to the target well (Kuckes et al, 1984).The amount of current that will be picked up on the target casing/drillstring, and consequently the signal strength measured on the sensors, depends on many factors, such as distance to the target well, formation type and mud properties.
Figure 3 shows the predicted pressure development for several initial scenarios
and the actual pressure development when the job was performed. Gas bleed-off caused
rapid pressure loss early in the operation, but pressure behaved nearly
as expected as mud filled the annulus of the target well.
We expected the signal strength to be lower because of the oil and OBM in the target wellbore, so we decided to drill the relief well with a water-based mud before the bypass section. Once the proximity between the two wells was less than 2 m, we changed the mud system back to an invert OBM system as a contingency for premature communication with the target wellbore. The OBM electrodes worked reasonably well; we observed only slight problems with loss of signal if the electrode was electrically isolated from the borehole wall. This was fixed by adjustments in the electrode lengths.
The intersection was accomplished by 12 electromagnetic ranging runs in conjunction with high side reading from the MWD used for steering short distances between ranging runs.
INTERSECTION AND KILL
We suspended drilling approximately 2 m from the intersection point to wait for dawn so we could conduct the kill operation during daylight hours. At 0400 hrs, we held a morning meeting to prepare for the kill operation and restart drilling. At 0545 hrs, the relief well intercepted the target at 2,827.8 m MD, as indicated by increasing casing pressure in target well and losses in the relief well. We closed the pipe rams on the relief well and commenced pumping 12.5-ppg mud down the annulus at 2 BPM as planned.
The target well mud/gas separator was not large enough to handle the gas rates circulating at 2 BPM, and the flow was redirected to the flare-pits. After pumping approximately 375 bbl of 12.5-ppg mud, we observed mud returns that matched the total volume of the target well including the drill pipe. At 0800 hrs, the target well was hydrostatically dead, but circulation continued with relatively high pumping pressure on the relief well. At 1350 hrs, pumping was stopped and a flow check indicated no flow. A pressure of 350 psi was recorded on the relief well, which was subsequently bled down without any pressure buildup. The relief well took 45 days from spud until target well was killed.
- An intersection well (relief well technique) is a reliable and field-proven method for well intervention in the case of problem wells that cannot be re-entered from the surface safely using conventional methods.
- The combination of electromagnetic ranging and passive-magnetics analysis worked as planned and facilitated an efficient well intersection process.
- The relief well located the target well at 2,160 m MD from 21 ± 5 m away and successfully intersected the target well at 2,828 m MD/2,760 m TVD; 12 electromagnetic ranging runs were used to successfully locate and home in on the target well. No gyro surveys where required.
- The hydrocarbons in the target well were circulated out according to the modified drillers method kill plan developed from the hydraulic simulation results; the transient multiphase modeling was invaluable when planning the kill intervention.
This article is based on a presentation at the IADC Well Control Middle East Conference & Exhibition, 2-3 December 2008, Muscat, Oman.
Since this case study was written and presented, John Wright Company has been acquired by Boots & Coots.
Rygg, O.B. and Gilhuus, T. Use of a Dynamic Two-Phase Pipe Flow Simulator in Blowout Kill Planning. Paper SPE 20433 presented at the 1990 SPE Annual Technical Conference and Exhibition, New Orleans, 23-26 October. SPE-20433-MS DOI: 10.2118/20433-MS
Jones, D.L., Hoehn, G.L., and Kuckes, A.F. Improved Magnetic Model for Determination of Range and Direction to a Blowout Well. Paper SPE 14388 presented at the 1985 SPE Annual Technical Conference and Exhibition, Las Vegas, 22-25 September. SPE-14388-MS DOI: 10.2118/14388-MS
Kuckes, A.F. Lautzenhiser, T., Nekut, A.K., and Sigal, R. 1984. An Electromagnetic Survey Method for Directionally Drilling a Relief Well Into a Blown out Oil or Gas Well. SPEJ 24 (3): 269-274. SPE-10946-PA DOI: 10.2118/10946-PA