P R O D U C T I O N

O P E R A T I O N S

PRODUCTION ENHANCEMENT OF PROLIFIC, EXTENDED-REACH GAS-LIFT OIL WELLS

Most oil wells producing from the Glauconite YY pool of the Lake Newell field in southern Alberta, Canada, have very high flow capacities. Wellbore operations are complicated by the slant-well configurations, with surface angles of 45 increasing up to 75 bottomhole and horizontal displacements in excess of 6,600 ft. After discovery of the Countess upper Mannville YY pool in 1989, a marine three-dimensional-seismic program, shot in 1991, showed that the reservoir extended 1.25 miles underneath the manmade Lake Newell. The reservoir was developed with 14 producers and one injector. Eleven of the 15 wells were slant drilled from a pad location where drilling begins at an angle of up to 45 at surface (Fig. 1). The original oil in place was estimated to be 15 million bbl, with ultimate recovery estimated at 8.8 million bbl. Primary production began in January 1990, and water injection was implemented in July 1993. Because of the reservoirs high permeability, most wells in the reservoir have high productivity indices. Any pressure drop within the system has a significant impact on productivity. All wells flowed initially, but shortly after the initiation of water injection, water cuts increased and artificial lift was installed. Gas lift was selected because of the availability of compression capacity, infrequent workovers, low operating costs, exceptional well inflow capability, lack of wellbore restrictions for production logging and pressure surveys, and low risk of a potential oil spill in an environmentally sensitive area. This article is a synopsis of paper SPE 48935, Significant Production Enhancement of Extended-Reach, Prolific Gas-Lift Oil WellsCase History of Systematic Problem Resolution, by D. Hahn, SPE, D. Yu, SPE, M. Tiss, SPE, R. Dunn, SPE, and D. Murphy, PanCanadian Resources, prepared for the 1998 SPE Annual Technical Conference and Exhibition, New Orleans, 2730 September.
62

OPTIMIZATION OPPORTUNITY

To evaluate waterflood performance, the reservoir was divided into three areas on the basis of structure and net oil pay. Pressure was maintained in Areas 1 and 2, but increasing water cuts of 70 to 90% resulted in steeply declining oil-production rates. The reserves-life indices (remaining reserves divided by current production rate) of these two areas were in excess of 15 years compared with the desired 4 to 7 years. Cementsqueeze operations were performed on the wells without success. A review of the producing wells in Areas 1 and 2 indicated that gas-lift optimization was necessary to increase drawdown and oil production and to improve the oil-recovery rate. The Area 3 reserves-life index was estimated at less than 2 years. Therefore, optimization efforts were focused on wells in Areas 1 and 2. A study of the pressure drops in the surface system determined that increasing the size of pipeline at the pad site would reduce pressure drops and increase production. However, the most effective measure would be to improve the downhole

artificial-lift system. The system review also determined that adequate capacity existed at the testing and battery facilities to handle increased well production from wellbore optimization.
OPTIMIZATION ATTEMPTS

A flowing-pressure-gradient survey was performed on the most prolific well in the field in September 1996. Subsequent tubing-flow-performance analysis could not match actual data with theoretical calculations, indicating that the production-string design and gas-lift performance were not optimized. Theoretically, for an efficient gas-lift installation with 2.875-in. tubing, fluid production should have increased from 717 to 1,500 B/D of liquid (BLPD). This increase could be accomplished by replacing several gas-lift valves with valves that had different operating pressures. A coiled-tubing-deployed system replaced the three existing valves successfully in November 1996. The well was placed back on production with a minimal increase in fluid production to 850

Fig. 1Slant-well schematic: true vertical depth (TVD) is 3,380 ft, and measured depth (MD) is 7,200 ft.
MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

BLPD. A subsequent flowing-pressure-gradient survey in January 1997 still showed excessive pressure drop in the tubulars. For the next 6 months, significant effort was expended to obtain a reasonable explanation for the differences between actual and calculated tubing performance. Advice was solicited from an international expert on gas lift who also was unable to model the actual performance with various nodal analysis packages. Therefore, it was determined that some other unexplained phenomenon was contributing to the problem. The following production-impairment mechanisms were considered. Production/injection measurement equipment. Hole in the tubing near surface. Nonrepresentative flowing-pressure gradient caused by production interference from flow past the gauges. Phase separation and stratification of fluids (water, oil, and gas) in the tubing. All the metering was verified and deemed to be providing accurate data. A hole in the tubing near the surface was ruled out because the flowing-pressure gradient indicated a definite gradient shift at the gas-injection point. Wireline gauge rings were run to confirm that no tubular restrictions existed. Review of the flowingpressure-gradient results, with and without pressure gauges in the tubing (i.e., lubricator and sump), demonstrated that the gauges were not interfering with production; therefore, the gradients were deemed to be accurate. Recent research and experiments on horizontal- and deviated-well flow characteristics indicate that phase separation in tubulars can be an issue because higher-specificgravity fluids will move at slower velocities or even reverse the flow along the bottom of the tubular. Because these slant wells are a special application of an extended-reach deviated well, it was postulated that the effective flowing diameter of the tubing was possibly smaller because of possible reverse flow of the heavier liquid phase on the low side of the tubing. This effect would increase pressure drops along the tubing. The tubing was replaced with 3.5-in. tubing to obtain at least 1,450 BLPD; however, only 1,130 BLPD was achieved, indicating that the problem was probably of a different nature and still not understood.
EMULSIONS AND DEMULSIFIERS

emulsions in the surface progressing-cavity transfer pumps. The tight emulsions result in significant pressure drops in the surface flowlines. The pressure-drop/emulsion problem was being addressed through the continuous injection of demulsifier upstream of the transfer pump. The design of a typical gas-lift mandrel introduces the lift gas into the tubing flow stream countercurrent to the liquid stream. This action can create significant turbulence in this area and cause significant shearing/agitation of the liquid phases. These turbulent conditions could be the catalyst that promotes severe emulsification of the fluids. Tests of emulsion samples taken at the wellhead revealed that they were very viscous and stable. The viscous emulsified flow regime created excessive pressure drops within the wellbore that impeded production. Surrounding wells are also prone to emulsions. Several weeks after the August 1997 installation of larger tubing, a demulsifier was introduced into the injection-gas stream. After 2 days, the well responded with a very strong production surge. The estimated rate was in excess of 2,830 BLPD. The production spike would last for 2 to 3 hours then revert to its normal rate for 6 to 7 hours. This cycle repeated itself two to three times every day. During these highrate surges, the surface piping at the wellhead vibrated vigorously and operational problems were encountered with the separation and gas-processing equipment. The demulsifier was a two-component blend of active ingredients in hydrocarbon carriers. The dry lift gas probably absorbed the hydrocarbon carrier, causing the resulting thicker demulsifier active ingredient to remain at the top of the annulus fluid. Project economics dictated the installation of a chemical capillary string. The dedicated chemical-injection string allows introduction of chemicals where produced fluids enter the tubing string, letting activation take place before the fluids reach the more turbulent region of lift-gas injection.
RESULTS

measured pressures and pressures calculated with the Hagedorn-Brown correlation. Because the tight produced emulsions in the tubulars impaired gas-lift performance, a second well was upgraded in a similar manner. Production increased from 380 to 1,775 BLPD, an incremental increase of 560 BOPD. The current rate is in close agreement with the theoretical predictions. Subsequent to the introduction of this chemical, no evidence of paraffin deposition within the tubulars has been seen. Dewaxing-related operating costs have been reduced, and flow efficiencies improved, probably because of increased flowing temperatures. Also, with the downhole injection of demulsifier, the use of the chemical for surface treatment at the testing/transfer facility has been reduced significantly. Since July 1997, the oil-production rates from the Countess YY pool increased 1,475 B/D, from 1,825 to 3,300 B/D (even higher than the previous peak rate of 3,000 B/D in early 1994 shortly after the waterflood was initiated). The perseverance in resolving the technical issues surrounding the poor gas-lift performance of these wells has improved cash flow and profitability of this pool significantly.
SYSTEMATIC PROBLEMRESOLUTION CYCLE

While trying to reconcile the underachieving gas-lift performance, discussions held with the property team determined that the Countess YY crude oil tends to form tight
MARCH 1999

The initial chemical-injection rate of 5.3 gal/D was reduced to 4 gal/D after several days. The production stabilized at 3,000 BLPD. High wellhead pressures were caused by surface piping restrictions that were rectified in May 1998. A flowing-gradient survey was run after the tubulars were upgraded and demulsifier was being injected through the capillary string. The gradient survey demonstrated excellent agreement between

The resolution of inadequate well-production performance occurred after several iterations that followed a modified Shewhart cycle. The four steps of this cycle can be summarized as follows. 1. Plan: diagnose the problem, collect data, determine changes, and develop an action plan. 2. Do: execute the action plan to carry out change. 3. Check: observe the results. 4. Act: analyze the results. What was learned? Do side effects or benefits still exist? Was the plan successful? Repeat cycle if unsatisfactory. During the entire process of arriving at the most satisfactory solution for the issue of obtaining production rates near the theoretical predictions, the multidisciplinary team followed the systematic pattern for continuous improvement. This cycle was repeated at least four times before the best solution emerged. Please read the full-length paper for additional detail, illustrations, and references. The paper from which the synopsis has been taken has not been peer reviewed.
63

P R O D U C T I O N

O P E R A T I O N S

METERING MULTIPHASE FLOW IN THE GULF OF MEXICO

Meters for measuring multiphase flow are unique tools that allow measurement of produced fluids (oil, water, and gas) without separation into individual phases. Two distinct and fundamental approaches to multiphase-flow measurement exist. The first includes no fluid separation, and the second uses partial (hybrid) measurement. No single multiphase-flow-measurementsystem design or technology resolves all multiphase-flow-measurement issues satisfactorily. Each approach has benefits as well as shortcomings. Even with these limitations, worldwide use has increased. Approximately 50 multiphase-flow meters were in service in 1995; this number increased to approximately 150 in 1997. Subsea application was the major reason for the original growth of the technology and is expected to be the dominant driver in the future. Most commercial multiphase-flow-measurement systems have no level- or pressurecontrol equipment to maintain but do include process-variable transmitters, which generally are off the shelf with standard calibration procedures. However, most systems require some form of electrical power and, in some cases, a controlled environment (such as a control room) for a flow computer. Because new electronic systems have been problematical when installed in the field, Conoco was compelled to test at the field level.
MPFM 1900VI DEVELOPMENT

between Conoco Inc., Norske Conoco A/S, and Fluenta A/S was reached in 1991 on a program that spanned 2 years and consisted of four development activities. The work resulted in Fluentas MPFM 1900VI multiphase-flow meter. Previous testing of water-cut meters at Conocos Grand Isle shore base in 1988 indicated that live fluids (at bubblepoint) affected the accuracy performance of many instruments, causing them not to meet the manufacturers accuracy specifications. This bubblepoint-fluid condition occurs when pressure drops below the last separation pressure, which occurs as fluid flows through pipes and fittings. Conocos multiphase-flow test and validation flow loop outside Lafayette, Louisiana, was commissioned in March 1993. Its original objective was to validate meter performance with produced fluids by use of multiphase (liquid/liquid, liquid/gas, and liquid/liquid/gas), clamp-on ultrasonic, water-cut, and any other technology with potential operational and economic improvements. Testing of Fluentas MPFM 900V , the forerunner of the MPFM 1900VI, began in April 1993. The project was to end by January 1994 with the tests of the MPFM 1900VI. Because of unplanned events, however, the flow-loop-test phase did not end until February 1997. The JIP test program was planned to be completed after the conclusion of field-level testing in the U.S. Gulf of Mexico.
USE OF MULTIPHA SE-FLOWMEA SUREMENT TECHNOLOGY

daily production volumes from the platform would be increased by 1 to 2%. It was further assumed that this increase in production would come about because the meter would allow production adjustments with faster feedback from measurements, decreasing the apparent decline rate; production tests could be performed more often and for shorter periods of time; and response to unplanned production-rate changes could be faster.
METER DEVELOPMENT

The field-level performance-evaluation test begun in late 1997 was the final step of a development program started in 1992. A joint-industry-project (JIP) agreement

This article is a synopsis of paper SPE 49118, Application of the First Multiphase-Flow Meter in the Gulf of Mexico, by Edward G. Stokes, SPE, Conoco Inc.; Dennis T. Perry, PetroTraces Inc.; Marshall H. Mitchell, Conoco Inc.; and Martin Halvorsen, Fluenta A/S, prepared for the 1998 Annual Technical Conference and Exhibition, New Orleans, 2730 September.
64

The economic incentive to use multiphaseflow technology is derived from the initial savings of weight and space on conventional platforms, providing instantaneous flowrate information, reduced maintenance, and more efficient detection of problems associated with declining production. In a subsea application, multiphase-flow technology becomes an enabling technology, allowing measurements in an environment where separators are unproved. A business case was developed for this final field test that assumed that average

The meter measures oil, gas, and water flow rates without physical separation of the well stream. The nonintrusive, real-time, fullbore instrument requires no bypass line and no invasive mixing device. The meter determines fluid slip automatically and calculates volume flow rates at actual and customer-supplied standard conditions. Fluid slip is the relative velocity between liquid and gas phases in a multiphase system where gas tends to flow faster than liquid. The measurement system includes a capacitance sensor, an inductive sensor, a gamma densitometer, a venturi meter, and a system computer. The capacitance sensor is used to measure the permittivity of the mixture and the gas velocity in oil-continuous multiphase-flow situations. If the flow becomes water continuous, the system flow computer automatically selects the inductive-sensor signals to calculate the conductivity and gas velocity of the mixture. The gamma densitometer is used to measure the density of the flow stream. The flow computer performs the analysis on the data and the data are brought safely to the computer by cables through safety barriers. Principle of Operation. Measurement of the flow is divided into two parts, fluid fractions and velocities. Oil, water, and gas flow rates are calculated on the basis of the measured fractions and velocities. The permittivity and density are different for each of the three components of an oil/gas/water mixture. If these permittivities and densities are known and the total permittivity and density of the mixture are measured accurately, the fractions of each of the three comMARCH 1999

P R O D U C T I O N

O P E R A T I O N S

ponents can be determined. For water-continuous mixtures, the inductive sensor is used to calculate the fractions. The principle is basically the same, except that conductivity (not permittivity) is being measured. Studies have shown that generating gas/liquid flowing conditions without slip is virtually impossible. Even if no-slip conditions could be generated artificially, slip would reoccur a very short distance downstream of the mixing device (typically 5 to 10 diameters). The strategy for this system was to find and develop mathematical models that give dependable velocity measurements under all slip conditions. This system determines the velocities of the large and small gas bubbles and the liquid. The capacitance and inductive sensors contain a number of electrode configurations that are used to measure the velocities of the large gas bubbles through crosscorrelation. The velocity of the small gas bubbles and the liquid is found from the differential pressure across the venturi meter. When the two velocity components are determined, they are combined with information from the fraction measurements to calculate the individual flow rates of oil, gas, and water.
FIELD TESTING

FIELD OBSERVATIONS

Separator-Discharge Sampling. The separator-sediment/-water sampler was not used because a representative sample could not be taken during the dump cycle. Because of the nature of on/off fluid flow, a dump cycle would start with mostly water and end with mostly oil. Although the dump rate was reasonably consistent, sampling the dump cycle yielded nonrepeatable results. Instead, liquid samples were taken manually at the inlet to the multiphase-flow meter. Fluid Cloud-Point Problem. One well was found to flow at close to its cloud point. After several series of tests where various wells were flowed through the metering system, the indicated water cut for all wells became consistently lower than the manual water-cut samples. Inspection and cleaning of the meter corrected the problem and returned the meter performance. The capacitance unit had a paraffin deposit covering the liner surface, which had to be cleaned to enable proper operation. Safety. To use a gamma densitometer in the field, a licensing procedure had to be developed that included the following. 1. Permission from regulatory agencies to proceed with the usage. 2. Removal from previous location. 3. Transportation to field. 4. Installation and leak check. 5. Field training and maintenance. 6. Contingency planning and documentation of Items 1 through 5. 7. Roll-up plan for removal of radioactive source after use. Observations. Multiphase-flow meters proved to be more robust than anticipated, with no mechanical failures during the test period. It was very difficult to acquire all the required data consistently and accurately over different shifts and rotating personnel. However, less than 10% of the data was eliminated for poor quality. The only maintenance problems were with software early in the program and paraffin buildup when testing close to the fluid cloud point. The field performance of the metering system appeared to be similar to the flowloop performance. Measurement repeatability was demonstrated, except for the consistent accumulation of paraffin found throughout the test period for one well. The piping and electrical installations were simple and straightforward, making the system easy to move at low cost.

Testing indicated that the duration of the well test does not affect the relative performance of the meter compared with the separator. Projected 24-hour production rates from the separator and multiphase-flowmeter system were affected in some wells by the length of the well test. Many wells do not flow at the average daily rate constantly; instead, they appear to cycle. At some point in each well, the multiphase-flow meter was subjected to instantaneous flow conditions below the meters specified liquid-rate minimum. This was caused by surging, where a liquid slug is followed by an extremely high gas fraction with very low liquid rates. This slugging occurred at regular intervals. The meter was useful in establishing optimum gas-lift rates through the relatively instantaneous nature of its data calculation and display. Issues for Future Use. Future use of any new-technology multiphase-flow-measurement equipment depends on the following. The absolute accuracy of multiphaseflow-measurement systems must improve to a maximum of 5% at all conditions of flow for gas, oil, and water. How the conflict that arises (because the method of determining a separators hydraulic efficiency in the field often is not specified) when comparing typical performance between a separator and a new-technology multiphase-flow meter is resolved . How multiphase-flow-measurement accuracy is proved after repair, recalibration, or system change (i.e., level or pressure). How one determines whether a measurement-system performance change might have occurred during use (such as the waxbuildup problem experienced during the field test). The U.S. Minerals Management Service (MMS) custody-transfer requirements are 2% for sales of gas and 0.25% for sales of liquids. Currently, multiphase-flow-measurement technology cannot meet these stringent requirements. The question is whether multiphase-flow technology can be improved sufficiently to close this gap and be approved for fiscal measurement by the MMS.

Initial plans called for testing of all wells for 24 hours at a set gas-lift rate to establish a base condition. This test would be followed by another 24-hour test at the same operating conditions of choke size, pressure, and gas-lift rate to evaluate repeatability. After the first two series of long-duration tests, test duration was determined by field-operations personnel. Adjustments also were made to observe, in real time, each wells response to conditional changes (i.e., choke diameter, gas-lift rate, or processsystem settings). Meter Installation. The meter system was installed with upward vertical flow. Physical installation included a drip pan to contain spills, block valves to allow work to be done on the meter without depressuring the whole platform, and manual sample ports at the meter inlet. The multiphaseflow computer was installed in the operators doghouse on the well-bay deck. Safety Issue (Gamma Densitometer). The gamma densitometer used at the test site was rented from ICI Tracerco and used cesium-137 as the gamma-emitting isotope. The source-energy rating is 24 mCi, which requires special handling and operating procedures and personnel training.
66

Please read the full-length paper for additional detail, illustrations, and references. The paper from which the synopsis has been taken has not been peer reviewed.
MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

NEW PRODUCTION-LOGGING TECHNOLOGY FOR HORIZONTAL WELLS

The main objectives of production logging are to diagnose well-production problems (such as inflow rates and entries of unwanted fluids), supply information for reservoir modeling, and provide data to optimize the productivity of future and existing wells. Determining the inflow profile of oil can help plan a drilling strategy, formulate cleanup methods for current and future wells, determine drainage patterns, and allocate production to sidetracks. Determining the water-entry locations and position of the water cone can provide a better understanding of the reservoir water-transport mechanisms and supply data for potential workovers. Ultimately, use of the results should improve the productivity and long-term recovery from the field. Most of this discussion refers to oil/water systems, with occasional references to gas/liquid systems. Many of the oil-/water-system results are applicable to gas/liquid systems. In a horizontal well, whether it is a barefoot completion or completed with a cemented casing or slotted liner, oil/water flow tends to be segregated by gravitational forces. Along different sections of the wellbore, the heavy- and light-fluid phases segregate according to the following regimes: stratified with a flat interface; stratified with a wavy interface; stratified with a bubbly interface; light phase slugging over the heavy phase; or one phase existing purely as bubbles in the other phase. Except for very heavy oils, stratified flow is normal when the holdup is significant (>20%) for both oil and water and can occur for liquid flow rates ranging from 0 to more than 12,000 B/D. Whatever the density contrast between the heavy and light phase, stratified flow is more likely for flow in sections of wellbore with deviations greater than 90 (i.e., downhill flow). Stratified flow with a bubbly interface can occur with low water holdups and is more likely as deviation decreases below 90. Slugging of the light phase can occur at deviations of less than 90 and is more common in gas/liquid flow. Flow where one fluid phase is mixed as bubbles in the continuous phase tends to occur more often if the heavy and light phases have similar densities or if one fluid phase enters through a jet into the wellbore. Problems encountered in measuring holdup and velocities in multiphase flow in horizontal wells include the following. Sumps and traps change the cross-sectional area open to flow. Segregated flow where different fluids have different velocities can greatly complicate spinner readings. Slight slope deviations from horizontal can cause significant changes in holdup and fluid velocities. Deviations of much less than 90 can cause backflow and circulation. Sensors that relate the density of a column of fluid to the difference in pressure measured at the top and bottom of the column cannot work in horizontal wellbores. A nuclear fluid-density meter is undesirable because it is environmentally hazardous and provides inaccurate measurements (particularly in heavy-oil/water systems of the Northwest shelf of Australia). Slotted liners create complications for accurate calculation of the total flow rate because of the uncertainty introduced by the annulus between the liner and the open hole (e.g., annular flow, changing hole diameter, or variable eccentering of the liner). The spinner behaves insensitively at low flow rates, which typically occur toward the toe of the well where the inflow contributions are often of strong interest. Because of segregated flow, interpretation techniques for vertical wells often are not applicable to horizontal wells. Moreover, existing correlations for horizontal and deviated oil/water flow do not deal with the effects of slight deviations. Therefore, a new tool string consisting of traditional and recently developed sensors has been tested. The possible existence of gas traps and water sumps causes another general difficulty when trying to understand flow behavior in horizontal wells. Given the observed production of these fluids at the surface, it would be dangerous to construe the downhole flow regime of water and gas. For example, when only oil is produced at the surface, it generally would be incorrect to assume that single-phase oil flow exists downhole.
INTEGRATED PRODUCTIONLOGGING TOOL

The tool string shown in Fig. 1 is a stripped-down version of one designed to deal with three-phase flow and is generally adequate for flow that is mainly oil/water. The tool string contains multiple sensors that measure the same quantity. For example, the locations of water inflow into a mainly oil-filled wellbore could be inferred from an interpretation of data from the spinner plus the fluid holdups measured by Schlumbergers Reservoir Saturation Tool (RST) and FlowView Plus tool (FVPT) or from the flow image and bubble counts from the FVPT. All data from all sources must be considered when performing the interpretation.
CONVEYANCE

This article is a synopsis of paper SPE 50178, Application of New-Generation Technology to Horizontal-Well Production LoggingExamples From the North West Shelf of Australia, by A. Carnegie, SPE, Schlumberger, and N. Roberts, SPE, and I. Clyne, Mobil E&P Australia Pty. Ltd., originally presented at the 1998 SPE Asia Pacific Oil & Gas Conference and Exhibition, Perth, Australia, 1214 October.
MARCH 1999

The example wells were logged with the aid of a tractor run above the tool string. It was used to push the string toward the toe of the well. The logging was done while the tool string was pulled out of the wellbore by wireline.
EXAMPLES

Three wells were logged in the Wandoo field, 37 miles offshore Australia on the Northwest shelf. The field contains a thin oil column (72 ft) sandwiched between a small overlying gas cap and a strong aquifer. The oil gravity is 19API, with a reservoir viscosity greater than 15 cp. Reservoir permeability ranges from 500 to 10,000 md.
67

P R O D U C T I O N

O P E R A T I O N S

Combinable productionlogging tool to obtain pressure and temperature.

The RST used to determine water flow; three-phase holdup; formation oil, water, and gas saturations; and gamma ray count.

Flowview tool FVPT

Flowview tool

Full-bore directional spinner used to determine total flow rate.

Combined tools positioned at a 45 offset and used to determine holdup, flow image, bubble map, and bubble velocity.

Fig. 1Tool string used for the Australian Northwest shelf examples (conveyed by tractornot shown).

The field was developed with horizontal wells, 15 oil producers and one gas injector. The average length of openhole section for the oil producers is 3,281 ft. Understanding of water influx into the horizontal wells is crucial to the long-term recovery from the field. Early breakthrough of water was expected because of the thin oil column, unfavorable mobility ratio, and strong bottomwater drive. Results from a previous production-logging campaign indicated that water influx occurred at the heel of the well, the location of maximum drawdown. Recent production logs suggest that this assumption was applicable only for wells intersecting sands of uniform permeability or for wells with higher-permeability sands at the heel. Example 1. Well WB1a intersects higherpermeability sands at the toe of the well. These sands are thought to have incurred large fluid losses during drilling operations, which may have damaged the formation. Test objectives were to determine the flow contribution of the sidetrack, the well inflow profile, and water-entry locations. The tool string consisted of a directional full-bore spinner; an FVPT; and sensors for pressure, temperature, and acceleration. An RST analysis was unnecessary because the downhole flow regime was known to be essentially two phase (oil and water). The holdup in an oil/water flow regime can be determined by the FVPT reliably. The shut-in log data show a strong correlation between parameters: the FVPT bubble counts and water-holdup-analysis results and the low sections of the well trajectory. The spinner indicated that crossflow was insignificant or nonexistent. Data from the FVPT corroborate the spinner with respect to water at low points in the wellbore. But only the spinner can be
68

used to suggest the existence of gas pockets at high sections of the wellbore. Water sumps and gas traps typically are found in shut-in wells with little or no crossflow. During the flowing pass, the full-bore directional spinner indicated that inflow to the wellbore was relatively uniform, probably indicating that the sidetrack was not contributing significantly. The holdup analysis shows that, during flowing conditions, the sumps had been dispersed and smeared except for one at 5,676 ft, where the flow from the toe was probably too low to affect it. The bubblemap log track, which was blank for the shut-in pass, shows an abundance of bubbles under flowing conditions. These bubbles are probably water because they usually occur on the lower side of the liner. Example 2. The next well logged was Well WB4a, which intersects higher-permeability sands at the toe of the well. These sands constitute 9% of the total openhole section of the well. The initial logging passes were shut-in spinner-calibration passes. When the surface flow rates were stable at 3,490 BOPD and 7,265 BWPD, logging was performed. Near the toe of the well, the fluids were stratified vertically (water underneath, hydrocarbon on top) and the bubble count was low (compared with the heel). Moving up the hole, the temperature, bubble count, and water holdup increased dramatically and the flow became more mixed. The bubbles between 6,300 and 6,102 ft were likely water because the water holdup increased dramatically (from 6,300 to 6,234 ft). This influx location coincides with the point where the well intersects the higher-permeability sands at the toe of the well. In the segment of wellbore between 6,102 to 3,281 ft, the water holdup

decreased. The flow became bubbly, and the bubbles probably were oil because most occurred on the high side of the liner. Water and hydrocarbon also became progressively more mixed (i.e., dispersed). From 6,102 to 3,281 ft, all the influx was oil. On the basis of the spinner response, the oil productivity was uniform. Example 3. The third horizontal well surveyed was Well WB9, which was geosteered under an existing producer and intercepted a water cone at 4,921 ft. The tool string consisted of an FVPT and an RST. The FVPT provided the wellbore water-/hydrocarbon-holdup data. These data were necessary to interpret the formation measurements made by the RST. The RST was used to determine the oil, water, and gas saturations in the formation and to provide wellbore holdup data. The results indicated that the residual-oil saturation in the zone from 4,970 to 5,102 ft was approximately 27%, compared with 19% from core analysis. The data from the RST tool, acquired while in dual-burst pulsed-neutron mode, clearly show character that is not an artifact of the wellbore conditions and, therefore, should be a formation response. This response implies that the RST may be used for time-lapse monitoring. The wellbore holdup determination from the RST agrees well with that from the FVPT. These results are being used by Mobil for reservoir modeling and for time-lapse monitoring.

Please read the full-length paper for additional detail, illustrations, and references. The paper from which the synopsis has been taken has not been peer reviewed.
MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

DOWNHOLE SEPARATOR PRODUCES LESS WATER AND MORE OIL

Water management is important in the production of hydrocarbons, especially when water volumes steadily increase as fields age. Novel approaches that can reduce the water volume downhole may supplement the traditional approach to oil/water separation at the surface. Taking produced water out of the well stream downhole increases production-tubing and process-facility capacity for oil and gas. Downhole oil/water separation (DOWS) can reduce the need to upgrade water-treating facilities. Downhole separation offers an alternative to debottleneck constrained water-handling facilities with potentially positive side effects, such as more favorable conditions for separating oil from water, increased productivity as a result of better well hydraulics, reduced discharges of oily water, and maintenance of reservoir pressure. Reducing water at the source also diminishes the need for water treatment and for prevention of corrosion, scale, and hydrates. When wells are already pumped or when produced water is already reinjected, downhole separation will be beneficial, particularly in wells where water shutoff has proved ineffective. New concepts for DOWS have been developed under a joint-industry project run by the Centre for Engineering Research (Canada). The technical feasibility of completing wells with hydrocyclones and downhole pumps to achieve in-well production, separation, and reinjection was demonstrated. The first successful installation outside North America became operational in Germany in 1997.
C ANDIDATE WELLS EQUIPMENT DESIGN

cut (>95%). Wells with a risk of sand production or emulsification must be avoided. The Eldingen field, east of Hannover, Germany, has produced from a shaly sandstone reservoir since the 1950s and meets the screening criteria. Well Eldingen-58 produces light oil from three consolidatedsandstone intervals that are in pressure communication. The reservoir pressure is approximately 72 bar at a 1460-m perforation depth. Production has been lifted by a beam pump at 80 m3/d with 97 to 98% water cut. In preparation of the DOWS installation, a packer was set to isolate the top zone from the two lower zones. The top zone was to be the producing interval and the lower zones the injection interval.
Tubing to surface

Concentrate pump

Motor upper protector

Motor shroud

Motor

Motor lower protector Pump intake

Total-flow pump Bypass tubes Production zone Hydrocyclone Injection pressure sub

Candidate wells have a relatively low production rate (<1000 m3/d) and high water

This article is a synopsis of paper SPE 50617, Downhole Separator Produces Less Water and More Oil, by P .H.J. Verbeek and R.G. Smeenk, Shell Intl. E&P , and D. Jacobs, BEB Erdgas und Erdl, originally presented at the 1998 SPE European Petroleum Conference, The Hague, The Netherlands, 2022 October.
70

Separation packer and locator seal assembly Injection zone

Fig. 1Downhole equipment lineup of well completion and flow paths for Eldingen-58.

The downhole separator was designed in consultation with the equipment supplier on the basis of reservoir and well data. The DOWS for Well Eldingen-58 includes one hydrocyclone and two electrical-submersible pumps (ESPs). Fig. 1 depicts the downhole equipment and flow paths. The high-water-cut oil flows from the production perforations upward to the top of the motor shroud. The bottom of the shroud is coupled to the pump housing by a fluid-tight seal, forcing all fluids over the top of the shroud and downward along the motor into the pump. From the pump intake, all fluids are pumped downward by the total-flow pump (an upside-down ESP with a thrust bearing at the top and discharge at the bottom) into the hydrocyclone where the bulk of the water is separated from the oil. The underflow of the hydrocyclone produces water clean enough for injection into the disposal zone. The overflow, oil with the remainder of the water, flows through bypass tubes into the concentrate pump for production to the surface. These three 20-m-long, 0.9525cm-diameter tubes that bypass the lower pump and motor are sized so that erosion and pressure drop are minimal. A common motor drives both pumps. This motor has protectors at top and bottom, unlike a normal ESP . The motor is powered from the variable-speed drive at the surface through a flat cable, which is strapped to the tubing with metal bands and cross-coupling protectors. With a variable-speed drive and an adjustable surface choke, the system can cover the expected variability in injectivity and productivity. The pump design depends on the flows and pressures required to lift the oil-rich stream compared with those needed to reinject the water. The push-through system used in Well Eldingen-58 is most efficient for dealing with the low reservoir pressure in Eldingen. This concept also avoids any breakout of gas in the hydrocyclone. If the reservoir pressure is sufficiently high, a concentrate pump is not needed. Alternatively, the well stream may be separated before being pumped; this so-called
MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

pull-through concept can be applied provided the bubblepoint pressure is high enough to prevent gas breakout in the hydrocyclone. In some crude oils, the latter concept would avoid emulsions and poor injection-water quality as observed in earlier DOWS trials with heavy oil. Solids carried by the separated water are the biggest concern for sustained injectivity. However, the consolidated reservoir in Eldingen has little sand production. For efficient oil/water separation, the split between overflow and underflow of the hydrocyclone can be controlled by chokes. The oil-in-water content in the underflow of the hydrocyclone should be 100 to 300 ppm. To ensure this water quality into the disposal zone, a rule of thumb for DOWS is to keep the surface water cut slightly higher than 50%. The separation efficiency depends largely on characteristics of the oil/water mixture, in particular oil droplet size. Knowledge of the droplet-size distribution in oil-in-water emulsions downhole could enable better separation.
FIELD PERFORMANCE

duction to surface has decreased by 64%. In the first year of operation, reinjection of water, separated downhole, did not damage matrix permeability; however, a water-cut increase was observed in the project area.
TECHNOLOGY OUTLOOK

downhole separator aimed at reducing the infrastructure and facilities for offshore fields.
CONCLUSIONS

Recompleting the well has increased net oil production by 300%, while net water pro-

Well re-entry has not been required for corrective action on downhole equipment. The water is injected under matrix-flow conditions, and no sign of permeability damage has been observed. Adjacent wells have experienced an increase in fluid level and water cut. These trends result primarily from the influence of DOWS because these wells produce from the lower zones into which Well Eldingen-58 is injecting. Despite favorable performance, the economics of DOWS is still relatively poor. Assuming U.S. $15/bbl and a production-rate increase of 30 B/D, payback time is approximately 1 year. Important factors include oil price, process-facility capacity, and an increase in tubing oil-flow capacity. Phasing well conversions in accord with increasing water rates also would limit exposure of large up-front investments. The concept, developed originally for debottlenecking production facilities, is being upgraded with a more efficient

Industry still needs to prove that downhole separation is a reliable, cost-effective means to increase oil production from capacity-constrained facilities, potentially lengthening the life of oil fields. The downhole-separation concept, developed originally for debottlenecking onshore facilities, has the potential to reduce the infrastructure and facilities of offshore oil fields. Evidence exists that water, separated downhole, can be injected under matrixflow conditions, which may lead to significant power savings in water-injection systems if sustained. Water-cut development in the Eldingen field indicates that DOWS should be applied in reservoir configurations with flow barriers between producing and injection intervals. Please read the full-length paper for additional detail, illustrations, and references. The paper from which the synopsis has been taken has not been peer reviewed.

P R O D U C T I O N

O P E R A T I O N S

OIL-SOLUBLE SCALE INHIBITORS

Fig. 1 shows a subsea-satellite field on the Norwegian continental shelf, 200 km northwest of Bergen, connected through a 9-in. production line to a concrete gravitybased production platform. Production, which started in 1994, is from four production templates, and water injection, for pressure support, is through two injection templates. Oil production currently is 25 000 m3/d. Recoverable reserves are estimated to be 55106 m3 of oil from Brent group sandstone. Squeeze chemicals can be injected into the wells through a 21/2-in. methanol line from the platform, 16 km from the farthest production template. The individual wells are completed differently, but most are vertical or highly deviated, with 51/2-in. tubing. Placement of chemicals is difficult without the use of coiled tubing and an intervention vessel, which is expensive. The most attractive solution is to bullhead the treatment through the methanol line and into the wells.
THE NEED

line does not exist, the other wells at the template must be shut in after a squeeze treatment to enable the squeezed well to produce through the 9-in. production line. If use of an oil-soluble scale inhibitor (OSSI) is possible, the squeezed well can be backproduced without shutting in the other wells at the same template. Also, fewer relative permeability effects for dry wells means that a squeeze treatment can be performed before water breakthrough without risking deferred production because of a prolonged cleanup period. An extended squeeze life may be possible because the OSSI can be placed deeper into the formation without causing water block, as often observed with water-based products. Better placement, with a higher squeeze rate into the formation, may be possible because of a lack of pressure drop caused by changes in saturation.
OSSI

When a scale-inhibitor squeeze treatment with a traditional water-based product is needed, several operational constraints must be considered. These constraints, caused by the long distance between the platform and the templates and the large volume of fluid that has to be pumped and backproduced, include the following. High friction and low pump rates will be experienced because of the high reservoir pressure and long distance. Poor placement can result because of the low injection rates through the 21/2-in. methanol line into 51/2-in. tubing. Each production template is connected to four producers. Because a separate test This article is a synopsis of paper SPE 50706, Development of an Oil-Soluble Scale Inhibitor for a Subsea-Satellite Field, by Rex Wat, SPE, Harry Montgomerie, and Thomas Hagen, Champion Technologies, and Raymond Breng, Hans Kristian Kotlar, SPE, and Olav Vikane, SPE, Statoil A/S, originally presented at the 1999 SPE International Symposium on Oilfield Chemistry, Houston, 1619 February.
72

OSSIs are better described as oil-miscible scale inhibitors. They were developed initially for gas-lift applications and sometimes applied as combined scale/corrosion inhibitors. Early OSSIs often contained multiple components and mutual solvents to hold the package together. In special applications, undesirable side effects (such as behaving as a surfactant) limited the flexibility of component selection. New OSSIs do not contain a mutual solvent and will dissolve in most hydrocarbons at infinite ratios.
HYDROC ARBON C ARRIER

As part of the test program, the authors examined different carrier fluids (hydrocarbon for the pill) that might be available in the field (e.g., diesel, kerosene, crude,

Production Template

Injection Template Production Template

paraffin, and xylene). While xylene is the least likely candidate, its inclusion provides direct comparison with the others because it often is regarded as the best hydrocarbon solvent and is used in many oil-treating chemicals. The choice of carrier fluid depends on costs, handling, availability, and its effect on the OSSI. The last point is critical because the hydrolysis and partitioning kinetics of the OSSI, when it finally comes into contact with the in-situ water phase, control the success of the squeeze treatment in the field. Also, the selected hydrocarbon carrier must demonstrate full compatibility with the OSSI at different operating temperatures (i.e., platform, seabed, and downhole). In the tests, 10% OSSI solutions were made up in crude, kerosene, diesel, paraffin, and xylene. The solutions were mixed 5:1 with formation water to prepare test samples. The samples were shaken briefly and left in an oven overnight at 80C. The aqueous and the oil layers from each sample were separated and analyzed. None of the solvent carriers appeared to have a major influence on the mass-transfer process of the OSSI. On contact with water, most of the OSSI molecules transferred from the oil phase into the aqueous phase through a combined hydrolysis and partitioning process. The level of transfer was extremely high, with little (<0.1%) or no OSSI detected in the final oil layer. The effect of temperature on the masstransfer process was examined briefly in a separate test. Five duplicate samples (10% OSSI in crude with formation water at a 5:1 ratio) were prepared and subjected to temperatures of 0, 20, 50, 70, and 90C. After mixing, the different layers in each sample were analyzed. In each case, little OSSI was found left in the oil phase. Unlike normal reaction kinetics, the response to temperature appeared to be less sensitive. Both the hydrolysis and partitioning steps are believed to be extremely fast. Indeed, the only problem was with the phase separation at 0C, where the layers had to be separated by centrifuge.
EFFICIENCY

Fig. 1Subsea-satellite-field layout.

Operational experiences from nearby fields with similar formation water suggest a
MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

mild scaling-regime change from CO3 to SO4 scale when seawater breaks through. After transferring into the water phase, the OSSI molecules must demonstrate adequate inhibition efficiency. Performance tests were carried out by use of a conventional tube-blocking-test apparatus to determine the minimum inhibition concentration (MIC). Precolumn Test. A second tube-blockingtest procedure uses a precolumn sandpack inserted into one of the water flowlines placed upstream of the test coil. The column simulated a squeeze treatment in the sandpack before the test. The arrangement is suitable for screening and ranking different scale inhibitors for squeeze treatment before the more expensive and time-consuming core tests. If the effluent samples can be preserved and stabilized, such an arrangement can provide simultaneous, inexpensive comparisons of inhibition efficacy and retention characteristics of the scale-inhibitor chemicals. This apparatus possibly can be expanded so that a reservoir-conditioned-core container is coupled with a standard tubeblocking-test apparatus. By monitoring the pressure drop, the level of scale-inhibitor residuals, and the scaling ions in the effluents, a more representative MIC level and squeeze life might be determined for field applications. Once the OSSI molecules hydrolyzed and partitioned in the connate water, they migrated toward the sand surface, where the adsorption process took place. Their return is not affected by the passage of hydrocarbon but follows a desorption profile similar to that of a water-based scale inhibitor. Poorer performance with the water-based scale inhibitor seems contradictory at first. However, after reviewing other data generated during this study, it became clearer and is explained by the unique mechanism of enhanced partitioning. When injecting a water-based scale inhibitor, the best outcome is complete replacement of the connate water by the injected squeeze pill. The maximum concentration gradient that can drive the adsorption process is equal to the pill concentration. However, for the OSSI, the maximum concentration that can be attained by the connate water is governed by the mass-transfer process between the oil and water phases. During the course of this study, it was observed that the equilibrium concentration in the aqueous phase can be significantly higher than that in the original OSSI
74

solution. This may explain the fact that, while the same activity of scale inhibitor was present in both cases, the desorption profile of the OSSI seemed to be better because of the much higher concentration gradient that drove the adsorption process in the first place. Hence, the time required to scale up the test coil was longer. Reservoir-Conditioned-Core Tests. The results from the precolumn tube-blocking tests cannot predict accurately what will happen in the field. As part of the productdevelopment program, a reservoir-conditioned-core test is essential to determine whether the OSSI can offer a reasonable squeeze life when deployed in the field. Also, any potential damage to the formation that might be caused by the OSSI chemical must be assessed. To examine the retention characteristics of the OSSI, the effluents from both the adsorption and desorption stages were analyzed. None of the samples showed any detectable level of scale inhibitor. The results from a separate analysis also confirmed that less than 1% of the injected OSSI remained in these samples. It appears that all the OSSI injected had been adsorbed and that very few of these adsorbed molecules had been released during the kerosene flush. A likely explanation for such a unique phenomenon is that the hydrolysis and partitioning of the OSSI were highly efficient within the porous media. With only 2.7 pore volumes (PV) of OSSI injected, all injected scale inhibitor had been retained. The desorption profile with formation brine is more like that of traditional scale inhibition. After peaking at approximately 44 000 ppm, the scale inhibitor returned to approximately 100 ppm after 50 PV and finally down to 1 ppm after 600 PV of brine injection. For the injectivity and formation-damage study, the differential pressure across the core was monitored during the injection of the spearhead, main pill, and the initial backflow of kerosene and formation brine. A small pressure rise was registered when the spearhead and desorption brine were injected. In both cases, the injection fluid was immiscible with the in-situ fluid. A small, gradual rise in the differential pressure occurred toward the end of the OSSI injection, which coincided with the breakthrough of a minute quantity of water. The authors believe this likely was caused by the end effect and redistribution of the water phase. If an interaction between the OSSI chemical and the core material occurred, which might have

caused formation damage, a sharp increase in the injection pressure or a cessation of flow after the locked-in period would be expected. Effect of Water Saturation. For a mainly water-wet formation, the connate-water saturation in the reservoir can vary between 5 and 25%, even if the field is not yet producing water. In formations where immobile-water pockets exist, the localized saturation can be even higher. One of the final tests carried out was to combine the OSSI pill (10% solution in kerosene) with formation water in various ratios. Nine samples, with OSSI/water ratios ranging from 1:9 to 9:1, were prepared. Apart from one sample with a 1:9 OSSI/water ratio showing minor turbidity, all other samples remained fully compatible. Mass transfer of OSSI molecules was observed in all cases. For the high-water-cut sample (90%), the observed minor incompatibility can be overcome in the field by use of a properly sized hydrocarbon spearhead.
CONCLUSIONS

OSSI molecules hydrolyze and partition readily on contact with an aqueous phase. Once partitioned, the hydrolyzed OSSI molecules exhibit inhibition efficiency and retention characteristics similar to those of generic water-based products. The masstransfer process of the OSSI molecules from the oil phase to the water phase appears to be irreversible. Depending on the phase saturation, enhanced partitioning can be achieved. A larger hydrocarbon spearhead should be considered for wells with a 10 to 25% water cut. The combination of an OSSI squeeze pill and a suitable hydrocarbon overflush will minimize many flowback problems associated with relative permeability and water block. Also, the cleanup period will be much shorter, leading to quick restoration of oil production to its presqueeze rate. This restoration of production rate is particularly beneficial to dry or nearly dry wells and to reservoirs with poor lift energy. For offshore environments, an OSSI squeeze minimizes water handling and subsequent discharge to the sea, thereby reducing the commonly observed oil-inwater problem after a conventional waterbased squeeze treatment.

Please read the full-length paper for additional detail, illustrations, and references. The paper from which the synopsis has been taken has not been peer reviewed.
MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

CHEMICAL MITIGATION OF SULFIDE IN WATER-INJECTION SYSTEMS

Uncontrolled growth of sulfate-reducing bacteria (SRB) in oilfield systems can create safety, environmental, and operational problems (such as microbiologically influenced corrosion, solids production, and biogenic H2S generation). Anthraquinone, a nontoxic biodegradable substance, uncouples the electron-transfer process in SRB required for the bacterias respiration with sulfate. When this metabolic pathway is blocked, SRB are incapable of reducing sulfate to H2S; therefore, the reaction of H2S with soluble iron also is blocked. Anthraquinone is essentially insoluble in water; however, the chemically reduced form (anthrahydroquinone) is soluble in a caustic solution. Anthraquinone, as treatments are referred to in this paper, was injected as a 10-wt% solution of the soluble anthrahydroquinone disodium salt in caustic. Injection of this solution into a flowing water stream forms submicron-sized particles of the inhibitor. The small particles coat the interior surfaces of pipelines in injection systems and subsequently become incorporated into the biofilm. Once the particles are incorporated into the biofilm, they partition into the cell membrane of the bacterial cells and inhibit sulfate reduction. These relatively insoluble, nonreactive particles are believed to provide a timereleased treatment within the biofilm and a continual source of sulfide inhibitor. Periodic treatments are required to replace anthraquinone that has been biodegraded or dissolved in flowing untreated water. Anthraquinone is not biocidal. Bacteria other than SRB also may be harmful in oilfield water systems and should be controlled with conventional biocides. Consequently, anthraquinone treatments This article is a synopsis of paper SPE 50741, A New Chemical Approach To Mitigate Sulfide Production in Oilfield Water-Injection Systems, by M.D. Johnson, M.L. Harless, and A.L. Dickinson, Baker Petrolite, and E.D. Burger, SPE, EB Technologies, originally presented at the 1999 SPE International Symposium on Oilfield Chemistry, Houston, 1619 February.
MARCH 1999

are designed to be used as a supplement in these applications to extend the life of traditional biocide treatments.
FIELD DESCRIPTION

The facility is composed of two separate systems, A and B. Both systems receive water from the same source. System A is a single injection plant pumping approximately 29,000 B/D of produced water to 28 injection wells. System B has three injection plants that pump a total of approximately 28,000 B/D of filtered produced water to 51 injection wells. Water lines that were out of service during the field trial were inspected visually for solids deposition and found to be fouled heavily with accumulated solids. The presumption was that this condition was representative of the lines treated during the trial.
L ABORATORY STUDIES

Measurable sulfide production from the untreated field-initiated bottle tests began 4 days after the wellhead-water samples were taken, while samples treated with the anthrahydroquinone solution were inhibited for at least 12 more days. Laboratory-initiated bottle tests used a synthetic medium with the cultured SRB. This test was designed to evaluate the effect of Fe2+ on the inhibition because Fe2+ forms a weak equimolar complex with anthrahydroquinone. The Fe2+ content in the field produced water was approximately 100 kmol, while the anthrahydroquinone concentration injected during the field trial was approximately 500 kmol. The laboratory study was run with approximately equimolar Fe2+ and anthrahydroquinone (500 kmol and 440 kmol, respectively) and with excess Fe2+ (500 kmol) at the same 440-kmol anthrahydroquinone level. The results indicate that the iron had negligible impact on the inhibition effect of the anthrahydroquinone, although high iron levels did affect the ultimate sulfide level obtained. Results from two sets of dynamic biofilm-inhibition studies show significant inhibition of sulfide production for about 3 days. After sulfide production increased,

subsequent repeat treatments (500 ppm of the anthrahydroquinone solution for 2 or 4 hours) directly into the biofilm column restored inhibition for at least 1 day. During the second test, two treatments were applied before inhibition ceased, but these treatments did not appear to extend the inhibition period beyond that observed for the first test. The initial treatment of the influent SRB flow allows the anthrahydroquinone solution to contact the SRB intimately for approximately 1 minute before entering the biofilm column. This delay allows molecules of anthrahydroquinone to partition into the SRB cell membrane and inhibit sulfate respiration after an initial lag period. The inhibition duration for this laboratory system with a synthetic medium apparently is limited to approximately 3 days. Subsequent treatments of the developing biofilm are not as effective as the initial treatment because of the hydrodynamics of the laboratory system. The drop in pH of the treatment solution as it is injected into the medium causes the formation of rather large particles of anthrahydroquinone that cannot penetrate the biofilm well because of the low shear stress at the wall of the biofilm column. In the field situation, where pipeline Reynolds numbers and shear stresses are high, however, the anthrahydroquinone particles that form as the pH drops are colloidal and are transported easily to the biofilm on the pipe wall by shear dispersion. Additionally, small particles are necessary to obtain high levels of sulfide inhibition.
FIELD TRIAL

A California facility was chosen for an anthraquinone-treatment program because of the very active SRB population and resulting production of iron sulfide solids. The active SRB population in the produced waters of this facility required daily treatments with acrolein. The field trial was conducted during the summer of 1997 to determine whether cotreatment with anthraquinone could extend the interval between acrolein treatments. H2S concentrations for Systems A and B were monitored daily. In System A, the pro75

P R O D U C T I O N

O P E R A T I O N S

duced water flowing to Injection Well A-2 soured the most rapidly during each treatment cycle because of the relatively long residence time in the 31/2-in., 1,900-ft pipeline from the header to the well. Most other injectors had smaller-diameter flowlines or shorter lengths from the headers. Well A-3 was the only other System A well to experience a significant increase in H2S in the injection water during each cycle. In System B, no injection water soured during the first treatment cycle, although the diatomaceous-earth-filter outlet water had high H2S on 16 July because of exceptional fouling. The problem was resolved by backwashing the filter. Only the water transported to the most remote System B well, B1, soured significantly during the second and third cycles. In System B, H2S concentrations increased most rapidly during the third cycle, possibly because the ambient temperature increased almost 18F (to 104F during the daytime) throughout that cycle. Because most pipelines are not buried, flowing-water temperature also increased, probably contributing to higher SRB activity. This high activity may account for the shorter period before H2S began to increase. During generation of baseline (control) data, produced water collected from System A Injection Well A-3 had a significant increase in H2S after 1 day. Produced water collected from System A Injection Wells A1 and A-2 had increased concentrations of H2S after 2 days. Produced water collected from all System B injection wells had increased concentrations of H2S 1 to 2 days after the start of the control period. These results confirmed that daily acrolein treatments were required to maintain stable H2S levels in both injection systems.
TOTAL SUSPENDED SOLIDS (TSS)

the control period are probable causes of these increased H2S and TSS levels.
SRB MEA SUREMENTS

Results from SRB serial dilutions for System A indicate that the population remained relatively stable throughout the trial and control periods for the sample sites monitored. The SRB levels in the System B influent water varied more than those in System A, although, overall, they were slightly lower than those entering System A. This variability most likely was caused by growth of SRB in the filter cake of the diatomaceous-earth filter coupled with backwashing frequency. Except for two wellhead water samples, the SRB levels were between 101 and 103 cells/mL throughout the treatment and control periods.
ANTHRAQUINONE RESIDUALS

Water samples were collected at various locations in each system during the treatment periods to determine system use of the chemical and to confirm that the chemical traveled through the system. These data indicate that the anthraquinone concentration decreased rapidly immediately downstream of the injection location, then slowly decreased as the pipe branched to remote wells. Deposition of the anthraquinone in the biofilm was confirmed by the decrease in concentration within the pipeline segments. Monitored parameters following the treatments indicate that sufficient anthraquinone generally reached all parts of the system during each injection period.
CONCLUSIONS

No correlation between TSS and the acrolein/anthraquinone treatments was noted during the field trial or control period. Random fluctuations in TSS were noted in produced waters collected from each of the injection wells monitored. Downstream TSS levels correlated closely with the TSS levels entering the systems. TSS levels were elevated slightly only during the third treatment cycle in System B, corresponding to the increased H2S levels observed during that cycle. Also, both H2S and TSS levels were higher than those of the initial two treatment cycles during the System B control period. Again, increased SRB activity cause by elevated water temperatures during the third treatment cycle and absence of acrolein/anthraquinone treatments during
76

Laboratory studies confirmed field results that biogenic sulfide production within this California oil fields water-injection system can be inhibited with anthraquinone treatments. Extended-duration inhibition was obtained in the laboratory. The presence of iron does not appear to affect sulfide inhibition. Simple laboratory studies were difficult to perform with this type of inhibitor because of the need for more realistic hydrodynamic conditions to keep the insoluble inhibitor particles small and bioavailable, as they are in a field pipeline system. During the field trial, H2S concentrations remained stable for up to 9 days in both Systems A and B following each acrolein/anthraquinone treatment cycle. After these stable periods, sharp increases in H2S concentrations indicated that the available anthraquinone concentrations within the biofilm had dropped below inhibitory levels. H2S level appears to be the most responsive parameter for monitor-

ing treatment efficacy. Significant increases in wellhead-water H2S levels could be detected more easily and reliably than TSS or SRB levels. As with H2S concentrations, steady increases in TSS were expected during the cycle period but were not observed during the field trial. Instead, TSS concentrations fluctuated throughout the trial and control periods. Therefore, no correlation could be made between TSS concentration and each acrolein/anthraquinone treatment or control period. The observed TSS concentration fluctuations likely were caused by changes in influent-water quality rather than the effects of downstream SRB activity. The SRB population in the wellheadwater samples generally remained stable throughout both water-injection systems. As with TSS levels during the first and second treatment cycles, variability most likely was the result of changes in the influentwater quality. The increased H2S concentrations toward the end of each acrolein/anthraquinone treatment, coupled with a relatively stable SRB population throughout the field trial, indicate that anthraquinone was acting as a sulfate-reduction inhibitor rather than as a biocide. If the treatment program was performing as a biocide, a decline in the SRB population would have been observed immediately after each treatment. This decrease would have been followed by an increase in SRB population over time. The anthraquinone treatment was acting to control any further growth and reproduction of the SRB population, resulting in a stable population over the period of each acrolein/anthraquinone treatment. Additional field work is required to determine the treatment life of anthraquinone in other systems. Anthraquinone must penetrate biofilms to contact sessile SRB, and, therefore, treatment periods and concentrations may be influenced significantly by the biofilm thickness. It is possible that less-heavily-fouled systems would require fewer anthraquinone treatments or lower concentrations to achieve adequate inhibition of SRB sulfate reduction for extended periods.

Please read the full-length paper for additional detail, illustrations, and references. The paper from which the synopsis has been taken has not been peer reviewed.
MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

NONDAMAGING POLYMER PLUGS FOR TEMPORARY WELL ISOLATION

Oil and gas production in the Appalachian basin is characterized by mature, water-sensitive wells. Many wells have been producing for more than 25 years. Most of the storage wells have been in use 40 years or more. These wells often are plagued with entry problems caused by restrictive fittings, valves, or tubing that hinder the repair and replacement of corroded, faulty, or undersized wellheads and casing top joints. The conventional approach uses coiled tubing (CT) with inflatable packers that are set through the restriction into the good casing downhole, allowing uphole repairs. These tools are expensive and pose stability and safety problems in old casing. An alternative is to use CT to place crosslinked-polymer plugs to protect the formation from the kill fluid used to isolate the formation pressure during repair operations.
BACKGROUND

1.2 wt% when mixed in fresh water. The system is buffered with an organic acid to a pH of 3.5. To ensure complete hydration, the polymer is preslurried in isopropyl alcohol. A water-soluble zirconium salt is added at 0.15 vol% as the crosslinker. The system has a delayed, or retarded, crosslink that occurs as a function of time and temperature. This delay allows proper placement of the plug downhole before crosslinking. The plug can be placed through casing, tubing, or CT, and a wide variety of breaker systems can be used to remove the plug.
L ABORATORY DATA

Many additives and fluid systems have been introduced to control fluid loss or to provide a nonmechanical means to isolate intervals. These systems usually are highviscosity fluids and can contain solid particulates. Tests have shown that these systems can be difficult to remove and can damage the intervals they are designed to protect. Crosslinked-polymer plugs provide a clean method of protecting a producing zone from damaging workover fluids. These crosslinked gels contain higher concentrations of polymer than other crosslinked fluids, such as fracturing fluids.
CROSSLINKED -POLYMER PLUG

The crosslinked-polymer plug used in this application consists of a carboxymethylhydroxyethylcellulose at a concentration of This article is a synopsis of paper SPE 51054, Novel Application of Nondamaging Polymer Plugs With Coiled Tubing Improves Efficiency of Temporary Well-Isolation Projects, by Brian B. Beall, SPE, and Thomas E. Suhy, SPE, BJ Services Co., originally presented at the 1998 SPE Eastern Regional Meeting, Pittsburgh, Pennsylvania, 911 November.
78

The crosslinked-polymer plug has been successfully hydrated and crosslinked in many fluids, including fresh water, 2% potassium chloride (KCl), 3% ammonium chloride, and 16.0-lbm/gal fluid spiked with zinc bromide. Crosslinking of the system is both a function of temperature and pH. The optimum pH of the system is between 3.5 and 4.0. The system is normally placed as a linear fluid, with crosslinking occurring once the fluid is in place. Typically, the higher the temperature and lower the pH, the faster the crosslink. These variables need to be taken into account when designing job procedures. The crosslinked-polymer plug is stable at downhole temperatures up to 200F for several days. For temperatures higher than 175F , gel stabilizers and extra polymer can be added for stability. Tests indicate that the crosslinked-polymer plug can help prevent damaging workover fluids from entering a productive zone even in high-permeability formations. When zonal isolation is no longer required, the crosslinked-polymer plug must be removed without causing damage to the formation. Breaker systems evaluated included oxidizers, enzymes, and acids. Return permeability of almost 100% was obtained in all cases.
C A SE HISTORIES

Wellhead Changeout and Pipe Repair. The first case history with the crosslinkedpolymer plug was for wellhead changeouts

on 12 Pratt storage-pool wells in Greene County, Pennsylvania. The wells were drilled in the late 1920s and recompleted as storage wells during 194550. The storagepool sand is a moderately clean conglomerate. Well depths range from 2,700 to 3,000 ft. Pool pressure during the project ranged from 450 to 500 psi. The wells were completed with 51/2-in. casing with the cement top 1,000 ft off-bottom. Approximately half the wells were openhole completions, and the rest were openhole with 31/2-in. perforated liners. All the wells have a restricted entry above the T caused by a welded swedge and a 4-in. restricted inside-diameter valve. Changeout Procedure. Wellhead changeouts were performed with a CT unit. Once on bottom, the CT was pulled up 25 to 30 ft and 250 gal of the crosslinked-polymer plug was mixed. The pH of 4.0 provided adequate pumping time. To water wet the pipe, 1 to 2 bbl of 2% KCl was pumped; this was followed by the crosslinked-polymer plug. The plug was displaced out of the tubing with 10 bbl of 2% KCl, and the CT was pulled to the estimated top of the plug at 2,600 ft. After approximately 1 hour, 2% KCl water was pumped at 1.0 bbl/min until the pressure increased to 200 psi higher than the pool pressure to allow the crosslinked-polymer plug to set fully. The pressure was bled off into the flowback tank, and pumping resumed at 1.0 bbl/min. This procedure was continued until the well was killed at 750 psi overbalanced pressure. The CT was pulled to surface, and the CT unit was moved off the well while the operator changed the wellhead and made repairs to the top joints of pipe without experiencing any problems. Removing the Crosslinked-Polymer Plug. Nitrogen was used to unload the well while tripping in with CT to the top of the polymer plug. Then, 15% HCl was pumped in as an external breaker. As the gel started breaking, the acid was jetted to the bottom. Nitrogen was used to circulate the hole clean to total depth. The CT was tripped out of the hole while pumping nitrogen and the wellhead pressure monitored as it returned to pool pressure.
MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

Shale Control. A crosslinked-polymer plug was used to control caving shales. While drilling the Berea formation at approximately 2,400 ft, a 4-MMcf/D open-flow gas zone was encountered. Rather than cement the 41/2-in. casing, the decision was made to set the casing on a formation packer at 1,350 ft, just above a sloughing shale. Below the shale, the well was completed as a 61/4-in. open hole. After producing the well for several months, cavings from the Sunbury shale bridged off over the producing zone and shut off gas production. Several attempts were made with a conventional service rig to remove the bridge and reset the 41/2-in. casing approximately 120 ft lower to stop the caving. After each cleanout attempt, the shale caved back in before the tool string could be removed and the casing released and lowered. CT Cleanout. CT was tripped into the well with a 2.875-in. drill motor on 11/4-in. tubing. The bridge was drilled out at 2,500 ft while circulating with gelled water. Once the 10- to 20-ft bridge was drilled through, the tubing and tools were run in the hole until a solid bridge was encountered at 2,900 ft. While continuing to circulate, 500 gal of crosslinked-polymer plug was

pumped through the tubing and drill motor. The tubing was then pulled into the casing to allow the polymer plug to set. After 1 hour, the CT was tripped out of the hole and the 41/2-in. casing packer released. Four joints of pipe were added, and the casing packer was reset at 1,470 ft. The CT was tripped back into the hole with a jetting nozzle, and 500 gal of 15% HCl was used to break the gel and remove any additional bridges. The well was cleaned out with nitrogen, the tubing tripped out of the well, and the well returned to production. Fluid-Loss Control for Cementing Water Zones. While drilling a well near Brandywine, West Virginia, two water-producing zones were encountered in high-fluid-loss shales before total depth could be reached. A crosslinked-polymer plug was pumped ahead of the cement plug as a fluid-loss material to avoid losing the cement volume. A volume of 250 gal/zone was pumped through the drillpipe, each zone plugged off successfully, and the well completed as planned. Cement Retainer To Plug and Abandon a Salt-Solution Mine. This well was drilled

in the 1950s and used to produce salt-brine solution for the retrieval of minerals and chemicals. Repairs to the casing were performed frequently because of the corrosive environment. The 6,500-ft well was completed with 7-in. casing to the top of the salt cavern. The plant was abandoned in the late 1980s, and the well and plant reverted to the original owner. Plugging and abandonment was required. The 7-in. casing was highly scaled, with the inside diameter reduced to 2 in. Usually, inflatable packers are set in the casing just above the cavern; in this case, however, no CT unit was available to set the plug. A 500-gal crosslinkedpolymer plug was pumped ahead of the cement and balanced in the casing with 3,200 psi cavern pressure. The polymer plug held the cement in place during the set time without any problems.

Please read the full-length paper for additional detail, illustrations, and references. The paper from which the synopsis has been taken has not been peer reviewed.

MARCH 1999

79

P R O D U C T I O N

O P E R A T I O N S

WIRELINE-TRACTOR PRODUCTION LOGGING IN HORIZONTAL WELLS

The growth in horizontal-well drilling in recent years has renewed interest in electrically powered traction devices that connect in tandem with conventional downhole wireline equipment and assist progression into the well. Production-log quality should benefit from improved depth control and reduced alteration of the flow regime compared with results achieved with coiled-tubing intervention. A direct-current (DC) -powered tractor compatible with standard wireline equipment has been developed that is capable of horizontal-well intervention through a wide range of completion strings. In parallel with the tractor development, the dynamics of downhole-wireline-tractor well intervention have been modeled to quantify the feasibility and range of application of downhole wireline tractors in existing and future well designs.
TRACTOR

The well environment imposes two major constraints on tractor design. The amount of continuous electric power that can be delivered safely through a monoconductor cable from the surface imposes a limit on the mechanical output, which in turn imposes limits on the depth and speed of tractor intervention. Also, tractor dimensions must be small enough to allow free passage through restrictions in the completion string and at the same time must be able to open far enough to maintain traction against the wall of the largest casing size. These simultaneous requirements for high efficiency and small size introduce opposing constraints into the design process. Fig. 1 shows the tractor layout. The tractor is connected between the logging head and the passenger equipment. Traction is This article is a synopsis of paper SPE 51612, Wireline-Tractor ProductionLogging Experience in Australian Horizontal Wells, by Eddie Local, SPE, ORAD Ltd., and Thomas L. Searight, SPE, Sondex, originally presented at the 1998 SPE International Conference on Horizontal Well Technology, Calgary, 14 November.
80

achieved by forcing two drive wheels against the casing wall. The drive wheels are mounted on arms positioned on opposite sides of the tractor and are actuated by a deployment module. When opened, the arms extend to the casing wall and apply sufficient force to create traction without slippage. The magnitude of the force is adjusted automatically in response to the applied load so that traction is sustained throughout the entire load range. When closed, the arms and drive wheels retract fully within the tractor body. For safety reasons, whenever electrical power is removed from the cable, the drive wheels retract automatically. A secondary shear-pin release mechanism allows overpull at the cable head to increase the retraction force if required. The profile of the arm is designed to push well debris (e.g., sand) sideways, helping to avoid a buildup ahead of the tractor that would impede progress along the well. To minimize friction on the tractor itself and to maximize the net payload, the tractor is fitted with in-line rollertype centralizers. A key element of the tractor design was the development of a DC permanent-rareearth-magnet motor that has a power-transfer efficiency greater than 90%. The motor can maintain continuous full-load output at temperatures up to 150C. Torque is transmitted from the motor to the drive wheels through an arrangement of gears built into the arms. The motor can drive in both uphole and downhole directions. Speed varies continuously and changes
CENTRALIZER

approximately 1.5 m/min for every 100-V change at the cable head. Tractor load is derived from the cable current, which increases by approximately 0.35 A for every 1000-N change of applied load. The tractor can sustain loads up to 3000 N without exceeding the electrical ratings of standard 0.56-cm monoconductor cables. The tractor is microprocessor controlled. Data transfer and communication between the downhole electronics and surface control equipment is achieved through a monoconductor cable while power is delivered to the drive motor simultaneously. The drive-motor and arm-deployment functions are initiated by keyboard commands from the surface computer. The tractor provides various downhole measurements, including head voltage and tension, tractor speed, casing-collar location, and temperature and pressure readings, all of which are displayed at the surface as the tractor progresses along the well. This information allows the tractors performance and interaction with the downhole environment to be monitored in real time, which facilitates operational control and safety. In addition, measurement of the drive-wheel rotation provides a means to compute the distance traveled by the tractor. Solid-state switches under microprocessor control allow the cable to be connected selectively to the tractor or to the passenger equipment.
FIELD EXPERIENCE

The first successful application of the tractor was to convey production-logging

CENTRALIZER

Fig. 1Tractor layout showing major components.

MARCH 1999

P R O D U C T I O N

O P E R A T I O N S

TRACTOR LOAD AND CABLE SURFACE TENSION, n

tool weight to the cable tension increases as well deviation decreases and, consequently, for this particular well profile, the cable tension increases while pulling out of the well (contrary to normal experience in a vertical well).
MODEL APPLIC ATIONS
TRACTOR TOTAL LOAD AND HEAD TENSION HOLDUP DEPTH

MEASURED DEPTH, m

equipment in Mobils Wandoo field on Australias northwest shelf. Before the job, a computer model was developed to assess the feasibility of the operation by calculating tractor load and cable tensions expected during the survey. Data recorded during the job were used later to fine tune the model. Tractor Performance. Production logs were run in three oil-producing horizontal wells, and the objectives were accomplished. Fifteen tractor runs were made, and more than 8300 m was tractored; the longest interval was 1190 m. The maximum cable current was approximately 0.6 A, less than 50% of the specified cable rating. The maximum surface voltage was 600 V DC, and average speed was 5 m/min. The essentially monobore wells have similar deviation profiles, vertical depths of approximately 630 m, and horizontal sections 1000 to 1200 m long. Maximum well deviation was 92. The passenger equipment was 14.6 to 23.6 m long and weighed 95 to 159 kg, including deployment bars. Tractoring started from holdup depths, and logging tools were conveyed to programmed survey depths. Various sections of log were run, including stationary readings, and three additional tractor passes were required.
THEORETIC AL MODEL

intervention objectives and to confirm that the cable can retrieve the tools from the well without exceeding acceptable safety limits. The underlying theory is similar to that of investigating borehole friction on casing strings. The model makes use of the well-survey data. The friction coefficients are not known precisely and, in the absence of field experience, a range of values must be assumed. A diverse range of values for borehole-friction coefficients is published in the literature. Tractor data collected from the Wandoo field wells were used to quantify friction coefficients by finding best-fit values between field data and the theoretical model. The tractor data allow independent estimation of cable- and tool-friction coefficients. Because the head-tension measurement is independent of the tool friction, it is used to estimate the casing-friction coefficient; and the difference between the tractor load and the head tension is used to estimate the tool-friction coefficient. Field Data vs. Simulation Model. Fig. 2 shows the simulation model applied to Well B9. Two surface-cable-tension plots are shown, one running into the well and the other pulling out. The tractor-load and head-tension data are plotted from holdup to total depth. The field and simulated data were in good agreement. Good agreement also was found between the calculated and actual holdup depths. Cable tensions recorded at pickup checkpoints during the tractor run and when pulling out of the hole also fit closely with the model. The contribution of the

The simulation software is integrated with the tractor-control software, and actual cable tensions and tractor loads may be compared with simulation-model predictions during the tractor operation. This feature contributes to general operational safety by allowing anomalous events to be identified more rapidly. The model provides a useful planning tool. By comparing the tractor-pulling capacity with computed tractor loads at the toe of the well or at the deepest survey depth, the feasibility of downhole-wirelinetractor conveyance may be assessed and compared with other methods of deployment. The model also is used to confirm that the wireline may be retrieved from the well without exceeding acceptable safety limits. Although larger cables have a higher pull rating, this advantage is offset partly because more pull is required to retrieve the cable itself from the well and, in particular, because heavier cables give rise to more cable drag in the horizontal pipe, which in turn increases the load on the tractor. For example, the drag from cable weight for a 5/16-in. cable is approximately double that for a 7/32-in. cable. The model may be used to compare different well trajectories and to optimize the well design to facilitate downhole-wirelinetractor intervention. Ideally, holdup depth should be as deep as possible, preferably deeper than any restrictions or abrupt changes in tubing diameter that are likely to impede tractor progress. Further field experience is required to fine tune the selection of parameters for the well model, and appropriate safety margins should be adopted, particularly in extended-reach wells where errors in the friction coefficients are compounded over long distances. Factors, such as well flow and the presence of debris and sand, that increase tractor loading also should be taken into account.

A three-dimensional computer model has been developed that simulates both tractor loading and cable tensions that arise during a downhole-wireline-tractor operation. The model provides a means to assess quantitatively the capability of the tractor to achieve
MARCH 1999

TVD, m

Please read the full-length paper for additional detail, illustrations, and references. The paper from which the synopsis has been taken has not been peer reviewed.
81