Blowout Risk Analysis of Gas-Lift Completions
- D.D. Grassick (Enterprise Oil plc) | P.S. Kallos (Enterprise Oil plc) | Simon Dean (Baker-Jardine and Assocs. Ltd.) | S.D.J. King (Baker-Jardine and Assocs. Ltd.)
- Document ID
- Society of Petroleum Engineers
- SPE Production Engineering
- Publication Date
- May 1992
- Document Type
- Journal Paper
- 172 - 180
- 1992. Society of Petroleum Engineers
- 3.1.6 Gas Lift, 2 Well Completion, 4.1.5 Processing Equipment, 3 Production and Well Operations, 4.3.4 Scale, 1.14 Casing and Cementing, 5.4.2 Gas Injection Methods, 5.6.3 Deterministic Methods, 5.6.4 Drillstem/Well Testing, 7.2.1 Risk, Uncertainty and Risk Assessment, 5.1.2 Faults and Fracture Characterisation, 4.1.2 Separation and Treating
- 3 in the last 30 days
- 355 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 10.00|
|SPE Non-Member Price:||USD 30.00|
A dynamic risk analysis was performed on four different gas-lift completionsusing a computer simulator. Results are presented to allow comparison of thedesigns in terms of blowout risk and minor leakage during production and toprovide insight into the effect of critical components production and toprovide insight into the effect of critical components and operations on theoverall well safety.
Any oil well capable of flowing to the surface presents a risk of blowout,which could damage facilities or cause pollution, injury, or loss of life. Therisks and consequences of such an incident may be amplified in some gas-liftedwells by the large inventory of high-pressure gas in the well. This paperpresents a risk analysis of four different completion designs to investigatethe effect of various design features on the risk of blowout. In addition, therisk of minor (finite) leaks and the effects of different testing andinspection intervals on both blowout risk and production availability areexamined. The study was conducted for platform wells during the production modeonly. Blowout risks during workover and wireline modes are not analyzed. Theanalysis was performed with the MAROS simulator, which allows the well systemto be modeled in finer detail and with fewer assumptions than required for suchconventional methods as fault-tree and event-tree techniques. While everyeffort was made to model the four completion designs in accurate detail, theoverall risk of blowout is highly dependent on the component reliability dataused. Therefore, the results for blowout risks should not be treated inabsolute terms but as a means of comparing the relative risk of blowout betweenthe four completion designs.
Each well completion design is described in the model by a definition ofevery component's reliability distribution and a system logic model thatcharacterizes the interaction of the components. Blowout paths are defined byspecifying every combination of simultaneous paths are defined by specifyingevery combination of simultaneous component failures that would result in ablowout. The simulator proceeds through a life cycle in a series of timesteps,examining proceeds through a life cycle in a series of timesteps, examining thestate of each component at each timestep. Studying the results from a largenumber of life-cycle scenarios and altering key parameters in the design makesit possible to converge on the optimum parameters in the design makes itpossible to converge on the optimum design solution. The well-completion modelsare constructed in two fundamental parts. First, all relevant failure modes foreach well component are parts. First, all relevant failure modes for each wellcomponent are defined, each with an appropriate failure and repairdistribution. Maintenance, testing, and inspection are included in the model toreflect the potential duration that components are in a failed condition butthe failures remain undetected. The second part of the model is used to detectsituations where combinations of components are in the failed statesimultaneously. These situations represent blowouts and are defined as theblowout paths derived from the physical design of the well completion. physicaldesign of the well completion. The simulation approach offers many advantagesover conventional deterministic techniques, such as fault trees and reliabilityblock diagrams. 1. Performance simulators account for continuous changes in thestate of the system over its expected life; equipment functionality, differentfailure modes and consequences, operating and maintenance philosophies, and theavailability of services and personnel are all taken into account. personnelare all taken into account. 2. Although the majority of deterministic methodsprovide a single expected value, simulation can provide a distribution ofresults from which more insight into system behavior can be extracted. 3.Design changes can be assessed very quickly once the basic model isconstructed, making design optimization rapid.
The base-case models are used as reference points throughout the sensitivityanalyses to assess the changes in risk of blowout and leakage as a result ofselective changes in the models. From previous studies on single- anddual-string completions (Figs.1 and 2), it was recognized that the presence ofhigh-pressure lift gas beneath the tubing hanger in the single-string designcould be detrimental to the overall safety of the system. The dual tree,however, was regarded as adding undesirable complexity to the design.Consequently, two other designs have been included for comparison. The first ofthese is a "hybrid" design (Fig. 3), which consists of a dual tubingstring combined with a single tree. The injection gas is supplied to the shortstring directly through the tubing spool. The second design is a concentriccompletion (Fig. 4), which also uses a single tree. Lift gas is injectedthrough the tubing spool and down the annulus created by the additional casingstring and the production tubing. The base component reliability data areidentical for the four comletion types. This ensures that the results willstill be comparable even if there is uncertainty in the absolute values of thedata. The total simulation duration was 500,000 years for each simulation run,equating to 25,000 life cycles of 20 years, which is the assumed well life forthe purposes of this study. The only scheduled event in the models is a majorworkover at 4-year intervals. In reality, workovers are unlikely to bescheduled this way, but they provide a means of fault detection and anopportunity for repairing failed components. This interval represents arealistic average. Conditional events were used for modeling the repair logicof the tubing-retrievable subsurface safety valve (SSSV), which allows repairby insertion of a wireline-set safety valve, and blowouts, which require anumber of simultaneous failures for their occurrence. The single and hybridcompletions were modeled with the assumption that no wireline access to theannular safety valve was available. Relevant data used in the base-case modelsthat were subsequently modified for the sensitivity analyses presented in thispaper are given below. 1. The SSSV was tested at 3-month intervals. 2. SSSVreliability was assumed to have a Weibull distribution with an 18-yearcharacteristic life with a shape factor of 2.5. 3. Production-tubingreliability was assumed to have a Weibull distribution with a 20-yearcharacteristic life and 1.8 shape factor below the upper packer and a 1.9 shapefactor above it. Gas-Injection-tubing characteristic life was 40 years with ashape factor of 2.0. 4. All wellhead and tree components used exponentialfailure disributions, modified by factorizing for the sensitivity analyses.Component reliability data were taken from Refs. 2 and 3, as well as fromproprietary sources. The Weibull slope factors used to describe the failuredistributions were based on available failure data.
|File Size||588 KB||Number of Pages||9|