VSD APPLICATION TO INCREASE ROD PUMP EFFICIENCY

IATMI 2006-09
PROSIDING, Simposium Nasional & Kongres IX Ikatan Ahli Teknik Perminyakan Indonesia (IATMI) 2006
Hotel The Ritz Carlton Jakarta, 15-17 November 2006

VSD APPLICATION TO INCREASE ROD PUMP EFFICIENCY

By : Mursalim, Didi Ruchyadi, Nibukat Zaradan, Budi Julianto of PT. Chevron Pacific Indonesia (CPI), and Emir Syahrir of PT. Catur Khita Persada (CKP)


ABSTRACT
This paper will describe low lift producing efficiency of the rod pump that is caused by incomplete (low) pump fill. This phenomenon is the most common operational problem experienced specifically on ramp down wells in Duri Steam Flood which utilizes sucker rod pump as artificial lift method. Incomplete pump fill is the result of having pump capacity exceeds the production rate of the well or having excessive gas/steam producing at the pump intake3.

Excess pumping capacity creates shock loading by fluid pound which result in rod buckling, exces­sive pump and tubing wear, unnecessary friction and stress fluctuations throughout the entire pumping system. Shock loading experienced by surface pumping unit associated by vibration that goes beyond tolerable limit is another problem that will reduce equipment runlife. Pumping unit unbalance and gearbox overloading have worsened the condition.

This paper will further describe about the alternative solution we implemented through the Portable VSD pilot project on 10-wells in DSF area-6 with promising result.

Introduction
One of the most significant operating costs associ­ated with sucker-rod pumping is the expense of pulling and repairing the rods, pump, tubing and the surface pumping unit. Many wells are pulled for repairs so often that they are marginally economic. This problem is made worse by the days of lost production associated with the downtimes.

To find solution to the “over-pumping” problem, we should refer to producing parameter: pump size, speed, frequency and stroke length. Reducing pump size is the latest way to do because of its high cost. Reducing stroke length has small improvement impact to the problem and is less favorable due to limitation in number of crank holes, worn out hole, and many pumping units have been set-up at the maximum stroke. However, good result can be achieved by combining it with speed reduction.

Reducing Pumping Frequency
The custom­ary method of reducing pumping frequency is to pump intermittently. This is done by placing the unit on a timer or by installing a pump-off controller (POC). Both methods have at least four drawbacks:

• Intermittent pumping changes the frequency, but not the pumping speed. When the unit comes back on, the loads associated with motion and fluid pound reoccur.
  
• During the time when the pump­ing unit is off, the reservoir is flowing liquid into the wellbore. This flow builds a liquid level in the casing, which creates backpressure on the producing formation. Maximum flow occurs with minimum backpressure, which is achieved by keeping the col­umn of liquid in the casing as small as possible.
  
• Starting the unit from a dead stop requires the maximum power usage. The start phase also is a significant shock on the pumping unit, rods, and pump.
  
• If the well produces any solids, the downtime allows the solids to set­tle. These settled solids will increase the frequency of stuck plungers.

Therefore, customary methods can­not eliminate all the shocks, unneces­sary pump cycles, and unnecessary power use, nor can they provide the minimum backpressure against the reservoir at all times.

The Solution: "Slow Speed"
The faster a unit pumps, the larger the number of cycles, and the sooner the rods reach a fatigue limit. The effect is much like flexing a wire until it breaks. The number of pump­ing cycles and the stresses caused by motion can be minimized if the unit is pumped at the slowest speed and with the longest stroke possi­ble, thus producing at the flow rate that the reservoir delivers. Long run times are normal with slow and steady pump strokes, making marginal wells economic over a long period of times.

The Duri Steam Flood & Dynamometer Survey
Sucker rod lift systems are the only method used to lift produced fluids in the Duri Steam Flood, the biggest steam flood in the world. This system is used because of its capability handling the combination of high fluid temperatures and significant sand production, which are common to steam flood operation1. Presently, Duri Field produces around 200M BOPD, with average steam injected of 1.017 MM BCWEPD. Total active producers are 4,200wells. Most of production comes from Pertama and Kedua sands, some of areas have developed Rindu as well.

The Giant steam flood records the highest oil production at 350,000 BOPD in 1995 and naturally declines by time to 200,000 BOPD nowadays. Data (Figure 1) shows currently 56.5 % (2372 of 4200) wells are low pump fill. It happens because declining of oil production is not adequately followed by downsizing of producing parameter such as tubing pump size, stroke length and pump speed.

Steam breakthrough, one of steam flood problems, is the condition where a producer well having average wellhead temperature rapidly increase from 200oF up to above 250oF, usually accompanied by decline of oil production. The reservoir pressure has risen to 150 – 200 psi, high pressure that makes it difficult to lift fluid. This condition requires the producer well to be closely monitored and need numbers of efforts to be normal or even “pump-off”.

The “pump-off” means that the pump can lift all fluid supplied by the reservoir. This ideal condition is achieved when the producing parameter meets the production rate and it happens at the lowest operating fluid level near the pump intake at the bottom of the well. The pump-off condition can be determined through the dynamometer survey, which is scheduled once per-well for every two months.

The dynamometer survey also measures the pumping system efficiency, pump fill and loading of its components. This data can be used as a source for optimization. There are two kinds of dynamometer survey, simple dyno and full dyno analysis. Simple dyno survey is to identify mechanical condition of wells and operating fluid level above the pump. Full analysis measures the pumping unit balance, structural and gear reducer loads, rod stress, power consumption and component mechanical efficiency.

From the following chart (based on dynamometer data) we can see that over half of Duri wells are low pump fill:

Figure. 1. Pump Fill Distribution of Duri Wells.

27% of all Duri wells are very low fill (0-30%)
30% wells are low fill (30-60% pump fill)
20% wells are medium fill (60-90% pump fill)
23% wells are high fill (90-100% pump fill)

Speed optimization using vsd
Each area in DSF experiences production phases with rapid incline within first few years then begin slow decline when the steam maturity reached. To accommodate the declining phase, optimization efforts should be aggressively performed, otherwise we will pay for multiple failure costs related to over-pumping condition: repair cost and downtime oil loss.

There are three key points in this optimization effort: dynamometer data for analysis and candidate selection; application of VSD (variable speed drive) for getting optimum pump speed at the production rate; and drive sheave size change.

Well candidate selection starts with grouping of wells by pump fill. We plot very low pump fill for pump downsizing (out-scope of this project); medium fill for a combination of reducing pump speed and stroke length (in-scope); and for high fill we closely monitor until reaching FOP (fluid over the pump) or pump fill ≥100%, the stage we must increase stroke length and/or speed not to let excess fluid level over the pump.

To get an optimum speed at production potential for each well, we can use software simulation but it is not popular here because of less accuracy due to many variant assumptions involved.

Variable Speed Drive (VSD) is the most popular instrument to adjust speed to optimum level. For best result, people install it permanently and individually, one VSD is served for one pumping unit. That’s ideal to accommodate rod load changes to the most suitable speed at anytime. We plan to install it at new wells which are still in ramp-up phase with high production rate.

Nevertheless, for mature wells in ramp-down phase which are marginally economic, we can do differently by utilizing only limited numbers of VSD for all target wells. We install one VSD unit at one well for two days only, just to get the optimum speed. Then followed by changing drive sheave size. This “portable VSD” + dynamometer survey + sheave change costs much less then installing individual/ fixed VSD.

Well performance before and after speed change can be compared through pre & post dynamometer analysis and well test data. For continual improvement, our petroleum engineers and technical assistants monitor and analyze all related data provided by dynamometer survey, well test and operator checklist.

To accommodate dynamic changes of formation fluid level, we schedule to review the pump speed once-a-year for stable wells; and twice-a-year for ramp-down wells. Anyway, we will make adjustment anytime we notice significant changes or when dynamometer data tells us to do so.

Besides, this initiative incorporates a new pumping unit balancing method using a load cell and load indicator. Since long time ago until today, pumping unit unbalance is still a big challenge in Duri which brings impact to well downtime, manhour and material loss (e.g. gearbox, bearings and wristpin broken, belt cut-off, motor burnt-out).
balancing the pumping unit

The customary practice in Duri Field to balance pumping units is with an ammeter. The amp probe is connected to the electric motor leads and measures amperage drawn by the electric motor during the pumping cycle. The counterweights are adjusted until the mean amp reading is about the same on both upstroke and down stroke. However, this method can be time consuming and often inaccurate because we move the weight to indefinite distance then secure the bolts, run the unit and take the amperage. If the balance has not yet achieved, we often redo the process again and again until we get a satisfying ampere reading.

Technically, this method could be effective when the unit is already close to proper balance. However, it may otherwise result in a grossly unbalanced condition. Since the ammeter cannot detect whether the alternating current is leading or lagging the voltage, the motor current will indicate a positive value, regardless of whether the motor is regenerating or motoring. Regenerative currents, therefore, may be mistaken for motoring currents. The result may be a severely unbalanced pumping unit6. More over, moving heavy weights again and again increases safety hazard to technicians. People call it ‘trial-and-error’ process. Another important thing to consider is downhole fluid level increase during the pump down for this balancing work. The longer it down the longer will take time to normalize the load. It leads to wrong amp reading and sometimes takes one hour to get a correct reading until the load stable.

Another helpful method to simplify balancing calculation is using a computer program such as CBALANCE developed by Theta Enterprises. Output of this simulation is the weight distance and direction we should move, how many pounds should be added or reduced; and it contains information of crank and counterweight data of the most commonly used pumping unit brands. If our pumping unit brand and model numbers are not available in the list, we can not use this simulation. Furthermore, to make it works, we must complete all required data of both downhole and surface. That’s why we still have difficulties using this software.

The best alternative solution so far is the use of a load cell and load indicator. The load cell is installed above the carrier bar to record load changes and get CBE (counterbalance effect) data. After pumping unit stops, the counterweights are adjusted to a value in the CBE data with only one time definite adjustment. Much faster balancing process is achieved with using CBE data provided in the dyno full analysis. Compared to amp probe, the load cell method seems to be much easier, faster, better (more accurate) and safer. The effectiveness of this method was proven for the first time at wells 7Q31A, 7Q10A and followed by many wells afterwards.

The pilot project :

1. Collecting data and study: since September 2004.
   
2. Presentation to stakeholders: 27 April 2005.


3. Awarded for a Presentation in the CPI “Operational Excellence Forum 2005” in Rumbai: 27-28 September 2005.


4. Target wells: 10 wells in DSF area-6.


5. Execution time planned: 1 month, Starts 4th Week of October 2005 – Ends 4th Week of November 2005(2 days per wells or 20 days for 10 wells candidate).
   Execution time actual: 7 months, Starts 14 October 2005 – Ends 29 May 2006.


6. Candidate selection criteria (for this pilot):
   Pump fill <> 0.3; tubing pump is mechanically good (i.e. low slip); mature steam (150-250oF); good well test facilities; high historical failures.


7. Obstacles during execution that contribute delays:
   a. NFD (no-fluid detected) during well test are caused by:
   i. Leak by pass valve to CVC line at upstream.
   ii. Failure upstream check valve (clogged, eroded, un-seated or bad seating etc.).
   No upstream check valve near the well.
   iii. Sour line and 3-way valve leak.
   iv. Foamy fluid.
   b. JDE work order and General Work Permit process & approval take time.
   c. Work coordination with Maintenance contractor for balancing PU, drive sheave fabrication, installation and motor replacement.
   d. Contracting issues.
   e. VSD failures:
   i. Power module burnt-out due to bad grounding on TECO VSD.
   ii. “Over-current fault” when using VSD (TECO) but no problem when using VFD (ePac Vector Flux Drive) specifically for less than 3 SPM where current (I) are high and V is low in freq less than 30 Hz.
   iii. Fail to run VSD due to high harmonic load caused by PU unbalance.
   f. New installed drive sheaves at 7P73C and 7P85A loose, it needs to improve fabrication and installation.
   g. Pumping unit problems: longer carrier bar, bad brake, bad selector switch, distribution panel short-circuit. Repetitive unbalance, difficult to balance PU with ammeter only, specifically when balancing PU of high PI well (7P73C) that need ½ hour to get ammeter reading stable.
   h. MFI communication error.
   i. Bad location.
   

Findings during Execution
a. Surprisingly, we get lower water cut at 5R79A where the fluid is foamy. It seems that improved pump fill (higher fluid level than before) creates higher back pressure reducing steam interference.
b. Well test failure indicated NFD is often caused by valves leak.
c. The candidate selection criteria and the whole process need to be reviewed to get optimum result. Target pump fill should be: 30-60%.
d. The slowest speed we can afford is 5.3 SPM (stroke per-minute) due to limitation on electric motor shaft size and inside diameter of sheave.


Result of the 10-well Pilot Project
1. 40% pump speed is reduced averagely from 9.1 to 5.5 SPM which equals to 1.9 million stroke per-unit/year without reducing production.
2. 50% vibration is reduced averagely from 0.68 to 0.32 in/sec to the safe range below 0.5 in/sec.
3. Gearbox loading reduced and balanced.
4. Ampere reading reduced more than 20%, leads to save energy and motor size reduction. All the 10 electric motors are reduced from size 3 to 2.
5. Pump fill increased by 70% and fill ratio increased by 79%.
6. The steam interference reduced.

Conclusion :

1. By improving pump fill with slower speed, we reduce fluid pound/shock loading and friction/wear which will reduce failures and welldown. In other word, we would not get production increase directly through slowing (optimizing) speed, but rather indirectly through reducing welldown. If in the long period production trend shows declining, after speed reduction it would follow the trend without short fall (Figure 5).


2. Balancing pumping unit with load cell contributes to best result for its safer, accurate, easier and faster method.


3. PU gearbox loading and balancing data shows a big challenge: 87% of PU unbalanced, 21% overloaded and 48% oversized (Figure 7).


4. By reducing failures we save much repair costs.


5. Portable VSD will save much budget compared to fixed/ individual VSD (unit price: $14,000).


6. Once-a-year review for mature/stable wells and twice-a-year for ramp down wells is considered adequate to accommodate the fluid level fluctuation.


7. Good teamwork amongst CPI teams and business partners is the key of success of this project.

To accommodate the declining phase of producing formation, optimization efforts should be aggressively performed, otherwise we will pay for multiple failure costs related to over-pumping condition.

This over-pumping condition will keep continue if we still pay less attention to it and only focus to the high production rate wells—the “low hanging fruit”.
10. Addressing abnormal condition will enhance employees' confidence; and increasing facilities' reliability will increase quality of work environment and then increase employees' morale and work spirit.

Acknowledgement :
Author would like to thank the management of PT. Chevron Pacific Indonesia for their permission to publish this paper. Special thanks to Production Management Team (PMT), Well & Facility Maintenance (WFM), Facility Engineering (FE) and our business partner PT. Catur Khita Persada-CKP).

REFERENCES :

1. Nagy, T. A, “Rod Pumping Optimization in Duri Field”, SPE paper 22960 presented at the SPE Asia-Pacific Conference held in Perth, Western Australia, 4-7 November 1991.


2. Gael, B. T., Putro, E. S., Masykur, Akmal, and Lederhos, L. J., “Reservoir Management in the Duri Steamflood”, SPE/DOE paper 27764 presented at the SPE/DOE ninth Symposium of Improved Oil Recovery held in Tulsa, Oklahoma, U.S.A., 17-20 April 1994.


3. McCoy, J. N., Rowlan, O. L., Becker, D. J., Podio, A. L., “How to Maintain High Producing Efficiency in Sucker Rod Lift Operation”, SPE paper 80924 presented at the 2003 SPE Production and Operations Symposium, Oklahoma City, Oklahoma, U.S.A., 22-25 March 2003.


4. Svinos, John G., ”Rod Pumping Optimization”, handbook for in-house training, Duri 2001, Theta Enterprises, 1989-1995.


5. Bommer, Paul M., of Univ. of Texas, Shauner, David, of Bommer Engineering Co., “Benefits of Slow-Speed Pumping” Journal of “Technology Today Series”, October 2006.


6. Peterson, Ron, “Balancing a Pumping Unit with Motor Torque Instead of Motor Current”, Journal of “Oil & Gas Automation Solutions”, 1st edition, October 2002.


7. Peterson, Ron, “Increase Pumping Efficiency by Controlling Sucker Rod Pump Fill”, Journal of “Oil & Gas Automation Solutions”, 5th edition, February 2003.


8. Hunt, Cecil, of Lufkin Industries Inc., “Good Maintenance Can Extend the Life of Your Beam Pumping Unit” hand book for the Southwestern Petroleum Short Course-98.

Figure. 2. Traditional control and VSD Panel of Pumping Unit.

Figure. 3. Production trend still keeps incline after 33% speed reduction.

Explanation to Figure 3:
* Previous Pump Speed: 7.9 SPM (stroke per-minute)
* 21-28 November 2005: Running with VSD Installed.
* VSD recommends 4.75 SPM. We can not go so slow due to electric motor shaft size and sheave inside diameter limitation.
* 18 January 2006: Change Drive Sheave (pulley)
to the new speed 5.3 SPMà the slowest speed we can go