Underbalanced plug milling with coiled tubing: An engineered approach

A 60-well field study provides a means of optimizing common operations and minimizing nitrogen use, risks of becoming stuck and time on location—all adding up to more cost-effective jobs.

Luis Castro, BJ Services Company

Multistage fracture stimulation treatments are increasingly being used to access hydrocarbon reserves. Techniques such as “plug and perforate” allow the operators to systematically fracture multiple reservoir zones in short periods of time, minimizing overall job duration. Composite bridge plugs or composite fracturing plugs (CFPs) are commonly used to isolate the recently fractured zone from the next. To bring the well to production quickly, these isolation plugs must be removed from the wellbore as soon as possible after the job. Milling with coiled tubing (CT) and a downhole positive-displacement motor (PDM) has been the method of choice for this task for multiple reasons. When compared with workover rigs, coiled tubing can generally mobilize on location, rig up, and run in and out of the well more quickly. CT can also operate under live well conditions. These are all desirable traits, especially when dealing with underpressured or water-sensitive reservoirs requiring the use of nitrogen to maintain circulation.

In normally pressured to overpressured reservoirs, CFP milling operations with CT are routinely performed using single-phase fluids as the power media to the PDM. However, in reservoirs where bottomhole pressure (BHP) is lower than hydrostatic pressure, lower-density fluids are required to maintain circulation and ensure a safe and viable plug milling operation.

CHALLENGES OF USING NITRIFIED FLUIDS

Using low-density fluids, such as a mix of water and nitrogen, decreases the hydrostatic load on the reservoir, thereby lowering the risk of circulation loss, the risk of becoming stuck, and the possibility of fluids invading the reservoir, where they may cause damage.

However, milling CFPs with CT and nitrified fluids has challenges as well. Nitrogen introduces the additional variable of compressibility. Simulation models can account for nitrogen’s compressibility during the job design stage, but handling the dynamic changes of the nitrified system in the field can be complex due to all the variables involved: CT string volume, annular volume, pressure drops, PDM configuration, bottomhole pressure and temperature, etc.

Additionally, nitrogen flowing through a PDM can negatively affect motor performance. Downhole motors typically have elastomer internal parts that are prone to elastomer swelling and a subsequent decrease in performance after prolonged exposure to gases.

Operationally, the impact of compressibility is much larger. During plug milling jobs, CT and PDM operators mainly rely on pump pressure and CT weight as indicators of motor performance and job progress. When a plug is being milled, a decrease in CT weight signals the presence of an obstruction to be removed or milled; an increase in pump pressure implies that the PDM is being placed under load while milling on the obstruction. Unlike milling with non-compressible fluid, where almost immediate pressure increases are observed at surface, milling with compressible fluids adds an attenuated delay to the pressure surge. If this effect is not carefully accounted for in the field, it could lead to a motor stall; the load is too high for the motor, so rotation stops. The BHA then must be picked up to decrease the load. Motor stalls increase the fatigue cycles on the CT string, increase the operational time on location, increase the volume of nitrogen used, and could result in the job failing to meet economic goals.

FIELD STUDY

A field study was performed in the Piceance Basin of western and northwestern Colorado. The Williams Fork Formation is enclosed in the basin and dates from the Late Cretaceous, ranging in depth from 6,000 ft to 10,000 ft. Reservoirs are typically underpressured, with pressure gradients around 0.3 psi/ft. Most wells in the area are completed with 4½-in. casing to a maximum depth of 10,000 ft. After the fracture treatment is completed and the CFPs are removed, production tubing is run into the well and the well is brought on to sales. Typically, each well is fractured in three to five stages, although up to nine stages have been fractured in a single well.

Composite frac plugs and milling BHA. The CFPs milled in this campaign were capable of allowing gas to flow from beneath the plug before or after fracturing. The CFP comprises composite parts and upper and lower metal slips. After stimulation operations are complete, it allows for rapid mill-out with a PDM and a junk bit. The composite parts and slips are circulated out of the wellbore after being milled. The CFPs used in these operations were 4½-in. flow-through CFPs with pressure differential ratings of 8,000 psi.

The top of the BHA is a 2⅞-in.-OD CT external connector. A 2⅞-in.-OD dual backpressure valve is used to isolate pressure and hydrocarbons from the wellbore up the CT. Since milling plugs in low-BHP wells often presents the risk of getting stuck, jars are always run on the BHA. The jars are of 2⅞-in. OD, with 20,000 lb of impact (without accelerators). A hydraulic disconnect, attached below the jars, has a 2⅞-in. OD and releases the remaining tool string when a ⅝-in. steel ball is pumped down the CT and through the hydraulic disconnect.

Often, when reaching TD of a low-BHP well after completing a plug milling operation, it is beneficial to lift the fluid that remains in the wellbore with nitrogen. As described earlier, a PDM’s integrity is compromised when nitrogen is in contact with the elastomers inside it. Therefore, to preserve the PDM, a 2⅞-in.-OD dual-circulation sub is incorporated on the BHA, operated by a ½-in. steel ball. For effective removal of the CFPs, a 2⅞-in.-OD mud motor is used. The PDM operates between 25 and 120 gal/min. (0.6–2.8 bbl/min.), giving it a speed of 110–490 rpm. When milling plugs, the motor experiences torque as weight is applied to the CFP. The motor has a maximum torque of 830 ft-lb and can withstand 4,720 lb of WOB. A 3.603-in.-OD, three-bladed junk mill is used, since this is 93% of the casing’s drift ID (3⅞ in.). This allows plug parts to be circulated around the mill and still cover a large surface area of the plug.

Variables and logic. To mitigate the effect of compressible fluids and optimize the CFP milling process in the Piceance Basin, a number of study variables were identified. These variables were chosen based on the fact that using less nitrogen could reduce the effect of compressibility and speed up the milling process. However, enough nitrogen had to be introduced to ensure proper returns to surface and maintain circulation. The variables are 1) fluid and nitrogen rates; 2) rate of penetration (ROP); and 3) wellhead pressure (WHP).

Among the factors that were considered constant are: CT size (1¾-in. OD), average CT length (16,000 ft), CT material (100-grade steel), CT string makeup (tapered), CT unit (60,000-lb pull injector), fluid and nitrogen pumps, CT connector, milling bottomhole assembly (check valves, jars, PDM and mill), type of CFP, casing size, casing material and surface flowback equipment. Additionally, all operations were performed within the same field and by the same core personnel, bringing a high level of consistency to the study.

OPTIMIZING FLUID AND NITROGEN RATES

To optimize the fluid and nitrogen rates, an iterative method was used as follows:

1. Design the operation via proprietary CT simulation software to obtain initial fluid and nitrogen rates based on pre-job bottomhole temperature and pressure provided by the customer

2. Monitor the field data (WHP and pump pressure) in real time, and recalculate optimal rates via proprietary CT simulation software by back-calculating the BHP from the known surface variables

3. Test new rates and evaluate performance

4. Continuously monitor changes in pump pressure and WHP.

Using a customer-provided estimated BHP of 2,700 psi, initial flowrate calculations through proprietary CT simulation software yielded expected rates of 1.5 bbl/min. for the fluid (water plus 0.1% friction reducer) and 250 cu ft/min. of nitrogen. At these rates, pump pressure was expected to be 3,462 psi, WHP to be 161 psi, volumetric rate through the PDM to be 1.8 bbl/min., and gas volume factor to be 30% (i.e., 30% of the fluid placed through the PDM would be gaseous).

In the first field trial, actual conditions proved to be slightly different. Pump pressure was slightly lower and WHP was slightly higher, as were the volumetric rate and the gas volume factor.

After adjusting bottomhole parameters in the CT simulation software and recalculating the optimal rates, slightly higher fluid rates could be used while avoiding circulation loss by a safe margin. Fluid rates were adjusted for both water and nitrogen. Higher rates were conducive to slightly higher pump pressures that were well within safe operating ranges. These higher rates also yielded a higher volumetric rate through the PDM (more than 2 bbl/min.), while keeping the gas volume factor fairly constant. At the same pressure drop through the PDM (which is controlled by the weight set on the plug), the higher volumetric rate yielded 360 rpm from the motor.

Results. In general, higher fluid rates with a constant nitrogen rate were found to speed up the milling process, as lower gas-to-liquid ratios in the treating fluid reduce compressibility. Similarly, higher calculated PDM rotational speeds were obtained after the optimization procedure as higher volumetric flowrates were pumped through the PDM.

The optimization process proved to be quick and efficient, as better operating flowrates were found and implemented immediately. The consistent nature of the work allowed the crew to readily implement more refined “initial rates” based on the experience from previous wells. After a few operations, further iterations of these “initial rates” became the standard operating flowrates in the field. Both the CT provider and the customer benefit from such knowledge, since it allows CFP milling jobs to commence with a well-defined operational envelope and with the efficiencies inherent rather than optimizing flowrates through trial and error on location.

Key lessons learned. Compressible fluids require extra time to stabilize (i.e., to reach steady state) when changes in the flowrates are introduced to the system. Therefore, it is essential to allow enough time after a flowrate change for the CT and backside to react to the redistribution of fluids before making further changes and before making judgments about the surface pressure responses to such changes. An estimated response time can be calculated using CT simulation packages.

In many instances, BHP estimations—necessary for the job design stage—can be obtained from the fracturing treatments performed on location before the milling operation.

OPTIMIZING ROP WHILE MILLING

The method to optimize the ROP while milling was more empirical and required much more field observation by the CT operator as well as post-job analysis. The procedure was as follows:

1. Monitor ROP while milling the plug

2. Monitor CT weight constantly to minimize stalls

3. Analyze ROP data post-job to identify behavior trends.

To analyze the changes in ROP while milling, depth-versus- time data generated from the data acquisition system on the CT cabin was plotted for each plug milled during the campaign.

Milling operations may involve PDM stalls that require pulling the CT off the plug and then running back in to resume the milling operation. The time involved in overcoming the PDM stall and resuming the operation was not included as “effective milling” time, since no actual removal was being performed at the time.

Results. In most cases, ROP was found to be highly variable near the top of the plug. This can be attributed to the presence of leftover fracturing sand on top of the plug. Since sand is expected, a less-aggressive approach to the top of the plug will speed up the milling process, because the sand can be “washed” as opposed to “milled.” This fact aided in minimizing the number of stalls caused by aggressively running the PDM and mill into the top of the plug.

An interesting result from analyzing the depth-versus-time plots is that milling was fast through the top of the plug—from the upper section of composite material down to the top of the metal slips. ROP for this area was as high as 20 ft/hr. Further analysis of the depth-versus-time plots showed that milling ROP decreased considerably in two distinct areas. If a schematic of a CFP is superimposed to scale on the plot, the locations of the low-ROP zones coincide with the locations of the CFP slips, Fig. 1. This trend was observed in the vast majority of the depth-versus-time plots.

Fig. 1. A typical time-versus-depth plot for the field study shows that ROP is reduced at two depths, 8,582.0 ft and 8583.1 ft. Superimposing a scale schematic of a CFP, the low-ROP zones are seen to coincide with the locations of the CFP slips.

This finding has great importance for milling operations. In the field, many PDM stalls occur due to a false perception that the PDM and the mill are not making adequate progress through the plug. As a result, the CT operator may try to increase progress by setting more weight on the mill and the CFP. The increased load on the PDM may lead to a stall and lost time to recover. Instead, if field operators are aware that ROP drops significantly at certain locations in the plug, they understand that apparent stagnation may just mean that a harder obstruction is being removed and that further weight on bit may induce a PDM stall. The consequences of this knowledge are twofold: It allows quicker mill-outs that use less nitrogen with better fatigue life management of the CT string, which translates into better asset management for the CT provider and more cost-effective jobs for the operator.

From a statistical standpoint, analysis of more than 60 CFP mill-outs showed that, in 50% of the cases, the bottom slips took longer to mill than the top slips; in 25% of the cases, the top slips took longer to mill; and in 17% of the cases, they took about the same time. Additionally, in 96% of cases, the top CFP milled much more quickly than the rest of the plugs in the same wellbore.

To identify noticeable changes or efficiencies in the process introduced by the optimization of rates and ROP, an analysis of milling time was also performed, comparing the first 25% of the campaign to the latter 75%. During the first 25% of the work, 52% of the time spent working on a CFP was dedicated to actual milling and the remaining 48% to stalls, pulling off the CFP, and running back onto it. In contrast, during the latter 75% of the project, as much as 74% of time was spent actually milling and only 26% on stalls and moving the CT. It is estimated that increased efficiencies from the optimization saved the operator as much as 43% of operating time per well.

Key lessons learned. Aggressive CT run-in speeds when approaching the top of the CFP can lead to premature PDM stalls, since it is likely to find sand collected on top of the CFP. Additionally, ROP decreases dramatically when milling through the slips on the CFP.

OPTIMIZING CHANGES IN WHP

Wellhead pressure optimization was more challenging than the other changes, as WHP takes the longest to stabilize after a change has been introduced to the system. When a flowrate is altered, such change has to first affect all the surface equipment, the CT string, the BHA and the CT-casing annulus before finally affecting the WHP. Compressible fluids exacerbate the delay, Fig. 2. However, ensuring that the milling system is as stable as possible before setting the mill on the CFP can help in the optimization process. WHP was carefully monitored before reaching a CFP, during the milling operation, and after the CFP had been milled.

Fig. 2. Pressure responses to a motor stall are shown for various locations in the CT intervention system. The yellow line represents BJ’s proprietary TeleCoil downhole monitoring tool. (The tool also records annular pressure--—the purple line). The tool sends real-time data to surface via a transmission medium that runs inside the CT string. The delayed response at the CT surface gauge and at the wellhead can be seen.

Results. WHP monitoring yielded straightforward results. As a general trend, the WHP before tagging the top of the CFP was fairly constant. This can plausibly be attributed to the stabilization of the fluid system prior to reaching the CFP. Since no meaningful changes are introduced to the wellbore flow dynamics, no noticeable change is experienced by the WHP. After commencing the milling operations, the WHP gradually dropped. This can be attributed to the additional weight of the solid particles (sand and debris milled from the plug) in the CT-casing annulus fluid column. Further milling into the CFP amplifies the effect.

WHP generally stabilized at a lower value than the initial WHP immediately after the plug was removed. This effect can be explained in the intermediate term by the transitional presence of sand and debris in the backside fluid column. In the longer term, the effect may remain due to the increase in the size of the fluid column and additional exposure of previously isolated intervals: As the CT is run in hole to the next plug or toward the bottom of the well, the height of the fluid column being moved in the backside is increased, regardless of the type of fluid that occupies the space (gas, liquid or a mix of both). The plug below acted as a check valve for the previously isolated perforated interval that was then exposed to the fluid column on the backside. This perforated interval, with some exceptions, has a similar pressure gradient to those above.

Key lessons learned. An isolated producing interval may have a lower pressure gradient than those above it. In these cases, removing the isolating plug triggers a circulation loss that, depending on formation conditions, can be temporary or of longer term. Because the success of milling in underbalanced conditions hinges on the ability to maintain returns, it was found in the field that carrying out milling operations with a slightly higher WHP than designed could be beneficial in combating sudden circulation losses.

Higher WHPs are achieved by running slightly higher nitrogen rates to reduce the hydrostatic column in the annulus. Field experience established the increase increment at about 30–40% of the design levels. A rule of thumb used during this study was 250 psi over the running WHP. This increment gives the CT operator more time to react to sudden decreases in WHP (an indicator of possible circulation loss).

As an example, when milling operations are performed at a 500-psi WHP and it starts to drop after removing a plug, the CT operator may not notice the change immediately or may not have enough time to react (e.g., pull the CT out of the hole, increase the nitrogen rate, etc.). When running the same job at 650 psi or higher, the change is more obvious sooner, providing more time to react. Avoiding circulation losses would also be beneficial for preserving the integrity of the PDM, because there is no need to pump extra nitrogen through the PDM to regain circulation and the overall operational time is minimized.

OTHER FINDINGS

Other operational efficiencies observed during the course of the campaign include the use of a larger crane when performing milling operations in multi-well pads. Spotting the CT unit for the first well in such a way that it can remain there for the other wells can make a great difference in rig-up time between wells. All that is necessary is the use of a crane large enough to swing the entire surface assembly (injector head, lubricator, BOP, flow-tee, etc.) to the next well, Fig. 3.

Fig. 3. Using a large crane on a multi-well pad can save time rigging the CT equipment down and up between wells.

Communication between the CT provider, the customer and third parties is crucial. As an example, flowback providers during this campaign were kept informed of the CT operation. In turn, they were able to rig up the flowback iron so that it would not interfere with the CT equipment in the current well or subsequent wells in the same pad. This saved precious rig-up and rig-down time, streamlining operations.

Finally, it is common in multi-well pads to find plugs set at similar depths in two or three wells. Prior knowledge of this allows the CT provider to plan for removing sections from the end of the CT between wells, in order to minimize CT fatigue and maintain string integrity.

ACKNOWLEDGMENT

This article was prepared in part from SPE 121736 presented at the SPE/ICoTA Coiled Tubing and Well Intervention Conference and Exhibition held in The Woodlands, Texas, March 31–April 1, 2009.

THE AUTHORS
	Luis Castro is a coiled tubing Region Engineer for BJ Services Company in Houston. He has a master’s degree in management from Hamline University in St. Paul, Minnesota, and a chemical engineering degree from Simón Bolivar University in Caracas, Venezuela. In his 10 years as an oilfield engineer, he has written numerous technical papers and articles. He can be reached at luis.castro@bjservices.com.