Pseudocomponent Selection for Compositional Simulation
- T.M.J. Newley (BP Research Centre) | R.C. Merrill Jr. (BP Research Centre)
- Document ID
- Society of Petroleum Engineers
- SPE Reservoir Engineering
- Publication Date
- November 1991
- Document Type
- Journal Paper
- 490 - 496
- 1991. Society of Petroleum Engineers
- 5.4.2 Gas Injection Methods, 5.8.8 Gas-condensate reservoirs, 5.2 Reservoir Fluid Dynamics, 5.4.9 Miscible Methods, 5.4.3 Gas Cycling, 4.6 Natural Gas, 5.2.2 Fluid Modeling, Equations of State, 5.5 Reservoir Simulation, 4.1.4 Gas Processing, 5.2.1 Phase Behavior and PVT Measurements
- 1 in the last 30 days
- 433 since 2007
- Show more detail
- View rights & permissions
|SPE Member Price:||USD 12.00|
|SPE Non-Member Price:||USD 35.00|
This paper describes a scheme for combining the components in amulticomponent equation of state (EOS) into fewer components for compositionalstudies. The method is based on minimizing the difference between the K valuesof the original components and the K value of the pseudocomponent to which theyare assigned. New rules to determine pseudocomponent to which they areassigned. New rules to determine pseudocomponent properties are presented. Therules, based on K-value pseudocomponent properties are presented. The rules,based on K-value weighting, ensure that information about both coexistingphases is used in the derivation of pseudocritical temperatures and pressures.Results for the lumped models developed by this method are compared with theoriginal multicomponent models. The method is shown to work for a variety ofgas-condensate, miscibility, and gas-injection problems.
EOS models can yield accurate phase-behavior predictions for oil and gasmixtures provided that the heavy fraction is taken into account properly. Thepresence of trace heavy components can have a significant effect on phasebehavior. The literature indicates that extending the analysis to approximatelyC40, with the techniques outlined in Refs. 1 and 2, for example, will ensurethat sufficient components have been included to allow modeling of the phasebehaviors of most systems with reasonable accuracy. The use of extendedanalyses leads to fluid models with a large number of hydrocarbon components.In compositional reservoir simulation, however, the number of components mustbe minimized to reduce the simulation cost. Extensive effort has been expendedin the development of schemes to group components in the extendedsingle-carbon-number distribution to form pseudocomponents. The problem oflumping a detailed analysis of fluid composition into pseudocomponents has beenattacked in several ways. The simplest methods pseudocomponents has beenattacked in several ways. The simplest methods assign pseudocomponents on thebasis of mole or mass fractions. More complex pseudoization schemes also havebeen proposed; for example, Whitson proposed a method in which themolecular-weight range spanned by the components is divided into intervals andcomponents with molecular weights in each interval are grouped. Li et al.proposed a method in which intervals are again defined, on the basis of the logof the component K values, and components in each interval are grouped. Many ofthese schemes give acceptable groupings when applied to thesingle-carbon-number distribution generated from the extended analysis. In manyapplications, further pseudoization of an existing multicomponent model thatalready contains grouped components mabe necessary. For example, a reasonablydetailed multicomponent model (about 20 components) may be preferred tosimulate surface facilities or to study displacement processes in more detailin a limited section of the reservoir, while for full-field simulation, we maybe constrained to use a model with few pseudocomponents (e.g., six to sevencomponents). Transfer between the multicomponent and pseudocomponent models isfacilitated if the pseudocomponent model is generated directly from themulticomponent model. pseudocomponent model is generated directly from themulticomponent model. Any lumping rule applied to an existing multicomponentEOS must give a satisfactory pseudoization of the already-grouped components.Unsatisfactory results can be obtained when some of the existing schemes forchoosing pseudocomponents that were developed for single-carbon-numberdistributions are applied. Assignment of an already-grouped component to aninappropriate pseudocomponent or deficiencies in the rules for determiningpseu-docomponent properties can lead to significant changes in phasepseu-docomponent properties can lead to significant changes in phase behavior.In the pseudoization process, the agreement with data is often lost, andregaining an acceptable match may not always be possible. In this paper, wepresent a scheme for selecting pseudocomponents that can be appliedsuccessfully to a multicomponent EOS model containing grouped components. Aweighted averaging rule to determine pseudocomponent properties is alsoproposed for use in conjunction with the scheme. properties is also proposedfor use in conjunction with the scheme. The pseudocomponent models that resultfrom the scheme are in excellent agreement with the original multicomponentmodels. Refitting the lumped models is often unnecessary.
The pseudocomponent selection method proposed uses the K values of thecomponents predicted by the detailed EOS. The scheme aims at minimizing thedifference between the K values of the original components and the K value oftheir pseudocomponent. This ensures that the predicted phase behavior is notaltered significantly and that the important phase-behavior features of anydisplacement process change very little. The schemes presented here weredevised process change very little. The schemes presented here were devisedprimarily for gas-condensate systems, but they also can be applied primarilyfor gas-condensate systems, but they also can be applied to crude oil systems,as shown in the examples below.
|File Size||1 MB||Number of Pages||7|