The Analysis and Prediction of Electric Submersible Pump Failures in the Milne Point Field, Alaska
- S.J. Sawaryn (BP-Amoco) | K.N. Grames (Centrilift Baker-Hughes) | O.P. Whelehan (BP-Amoco)
- Document ID
- Society of Petroleum Engineers
- SPE Production & Facilities
- Publication Date
- February 2002
- Document Type
- Journal Paper
- 53 - 61
- 2002. Society of Petroleum Engineers
- 3.1.2 Electric Submersible Pumps, 2.2.2 Perforating, 2.5.2 Fracturing Materials (Fluids, Proppant), 3.1 Artificial Lift Systems, 1.6 Drilling Operations, 3.2.5 Produced Sand / Solids Management and Control, 3 Production and Well Operations, 5.2 Reservoir Fluid Dynamics, 4.1.5 Processing Equipment, 5.2.2 Fluid Modeling, Equations of State, 2.4.5 Gravel pack design & evaluation, 5.5.8 History Matching, 2.4.3 Sand/Solids Control, 4.1.2 Separation and Treating, 4.3.4 Scale
- 3 in the last 30 days
- 768 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 5.00|
|SPE Non-Member Price:||USD 35.00|
Electric submersible pumps (ESPs) are the predominant lift method used in the Milne Point field, Alaska. Each year, approximately 50 ESP failures occur, adversely affecting lifting costs, rig utility, and production. Statistical methods based on the Weibull distribution and the bathtub reliability model have been used to analyze ESP reliability data collected over a 13-year period. This analysis differs from earlier works by treating a portion of the infant failures (the early part of the bathtub model) as a series of Bernoulli trials. Analysis shows that the three main factors affecting ESP reliability in the Milne Point field are reservoir, sand control, and whether the ESP is the first installed in a well or is a replacement. Further, the ESP failure rate is shown to be dynamic, influenced by the delivery of new wells and the time lag between failure and replacement. The results have been incorporated into a computer-based failure simulator that uses Monte Carlo techniques to simulate the failure rates. Numerical values for the statistical parameters are included with the details of the simulator.
The simulator is currently used within the Milne Point field asset as a planning tool to establish changes in rig utility, production loss, and ESP replacement strategies in a cash-constrained environment. It is also used to assess improvements in ESP performance and to secure commercial terms for ESP lease and purchase options. In the future, it is recommended that the simulator be used, with suitable analog data, to predict the economic impacts of ESP failures on development programs before sanction.
There are more than 112 ESPs operating in the Milne Point field, producing fluids from four distinct reservoirs. As the field has matured and reservoir fluid composition has changed, completion practices have been improved to help increase the productivity of the wells and the longevity of the ESPs. Over time, the number of production wells has increased, and additional ESPs have been brought on line. Upchurch1 has demonstrated that the variations inherent in such dynamic conditions are reflected in the ESP run times.
Analysis reveals that the run lives of the Milne Point field ESPs vary considerably. The current mean time to failure is 330 days, with the age of the longest surviving pump being in excess of 10 years. In 1997, 46 ESP failures were recorded, and a total of 67 failures occurred in 1998. A rig workover, costing in the range of U.S. $300,000, is required to replace a failed ESP. The burden on the asset's operational cost is large, exceeding U.S. $20 million per annum. The ESP failures also result in lost production, which is estimated at 1,700,000 bbl per year.
Earlier qualitative methods used in the asset to predict the number of ESP failures for a given year were not successful. The number of failures predicted for 1997 and 1998 were underestimated by 50 and 40%, respectively. This underestimation resulted in a shortfall in the asset's production, overruns in the operating budget, and disruption to the rig schedules. For these reasons, a more formal analysis1-4 was attempted.
Well Categorization and ESP Data
Well conditions and completion types have been recorded along with the number and types of ESPs installed throughout the history of each of the Milne Point field production wells. The data collected on the 320 ESPs installed up to 31 July 1998 were used to categorize each well into a class type and to calculate run times.
The reservoirs/formations from which the wells produce include Prince Creek, Schrader Bluff, Kuparuk, and Sag River. Each formation has distinct reservoir characteristics, which contribute significantly to the completion design and operation of the individual ESPs. The temperature of the main reservoirs, Schrader Bluff and Kuparuk, are 75° and 170°F, respectively.
The type of sand control is identified on the basis of completion. With full sand control, all production zones are fully screened and fracture or gravel packed. Partial sand control means that either resin-coated proppant has been used to control proppant back production and sand production, or at least one, but not all, zones have been screened. With no sand control, the well has been cased and perforated, and nothing has been put into place to prevent sand or proppant from being produced. Kuparuk wells have either full or no sand control; Schrader Bluff wells have all three. All water-source wells have full sand control, and all Sag River wells have no sand control.
The type distinguishes between a new ESP completion vs. a second, or subsequent, ESP installed in a well. It is known that the run life of the first ESP is typically shorter than the run life of the second, or subsequent, ESP.
This enables ESP performance to be tracked under different reservoir and completion conditions.
In the Kuparuk and Sag River reservoirs, approximately 75% of the producers have been stimulated with propped fractures. In Schrader Bluff, virtually all conventional producers have been fractured, with a minority that have been simply gravel packed or left in a natural state.
With an increased water cut, scale has become a significant contributing factor of ESP failures. An aggressive scaleinhibition program has been established to combat this effect.
Reason for ESP Failure.
Each ESP failure has been coded to identify the component that failed and the assessed reason for the failure.
Currently, the N sands in the Schrader Bluff exhibit the highest sanding tendencies and create the most operational problems.
More than 95% of the ESPs installed in the Milne Point field have been provided by the same vendor. No further analysis of this classification has been made.
|File Size||1 MB||Number of Pages||9|