3. Black Oil or Compositional Model? (in Design) Black Oil: A black oil model assumes that the fluids consist of a liquid phase and a gas phase only. The amount of gas that dissolves in the oil is dependent on pressure and temperature. Black oil models should only be used in steady state simulations in which the API gravity is less than 45 and the GOR is less than 14000 standard m3/m3 of stock tank oil (2 500 scf/stb). The following minimum data shall be used to set up a black oil model: 1. Oil SG. 2. Gas SG. 3. GOR 4. Water cut This data should be obtained from a laboratory multistage flash analysis (normally 3 stages). The sum of the gases evolved from each of the laboratory flash stages is known as the producing GOR. The density of the oil produced from the final flash stage is defined as the stock tank oil density. The gas SG is based upon the weighted average gravity from each stage. This analysis should be performed at conditions that closely match field operating conditions i.e.: 1. The 1st and 2nd stage pressure and temperature conditions in the test should closely match the 1st and 2nd stage separators in the field. 2. The 3rd and final stage of separation is performed at stock tank conditions, 1,01 bar, 15°C (14,7 psia, 60°F). 3. Even if the field operation only has a single stage of separation, the laboratory test shall also include a second stage at stock tank conditions. Using the GOR, oil SG, gas SG data, and standard black oil correlations the physical property information required for a multiphase flow analysis can be evaluated. Physical property correlations should be tuned to match laboratory data at the bubble point condition and the 1st stage separator. Solution GOR and volume formation factor information is available from the multiphase flash data at these 2 conditions. Live oil viscosity data that is generally available at the bubble point condition should be used to tune the oil viscosity correlation. Compositional Model: A compositional model shall be used for any fluids that lie close to the critical point, such as highly volatile oils, as well as for gas-condensate systems. The industry standard, transient multiphase simulator, OLGA, currently only works with compositional descriptions of the fluids. Compositional modelling should be used for systems in which the GOR is > 2500 or the API gravity is > 45. Detailed fluid analysis, as provided from the process design, shall be used. A compositional model requires a detailed analysis of the fluids. However, attempting to model the fluids in terms of the components e.g. C10, C11, does not produce an adequate characterisation of the fluid, rather the heavy end components beyond approximately C7 require a full characterisation in terms of their normal boiling point (NBP), molecular weight (MW), and SG. Reservoir engineers often have available heavy end component characterisation in terms of critical pressures and temperatures; however these can be manipulated using standard routines to provide the characterisation in terms of NBP, MW, and SG. The characterised fluid data can be used either to generate tables of properties, within which design software packages have to interpolate, or to enable some software programs to calculate the required properties at any temperature and pressure, directly. Fluid viscosity values shall be obtained from laboratory analysis if fluid samples are available for testing. If fluid samples are not available, PVT package generated viscosity values may be used. Historically PVT package determined viscosity values have proven to be significantly underestimated. Current BP preferences for PVT packages are:
  • PVTsim from CalSep
  • MultiFlash from InfoChem

Whichever package is selected, the equation of state used to characterise the fluid shall be matched to as much laboratory data as possible before any design work is undertaken. Validation exercises have shown that BWRS is the most accurate method for predicting a range of fluid properties. However the BWRS method is not supported by many PVT packages. If the PVT package does not support BWRS methodology, either the Peng-Robinson or SRK methods should be used. With either of these methods, both liquid and gas property predictions are significantly improved if Peneloux shift parameters are applied.

[link Point to another website Only the registered members can access]