Around 80% of historical PVT studies are not of sufficient accuracy for use in a full field simulation model. This figure is  based upon the reviews of many, many PVT reports and supported by other experts. Before embarking on the task of incorporating the fluid compositions into a reservoir model, it is essential that a review of the existing studies be performed so that non representative data can be weighted accordingly. Reasons for the inaccuracies include:

Non representative PVT samples.

For mature reservoirs, it is essential that a well conditioning program be implemented to return the reservoir fluid to initial conditions (if possible) before taking samples. If there has been no significant production, then a review of the existing data can technically and financially optimise future programs as evidenced in many of the projects I have completed.

Inaccurate recombination of separator samples.

Whereas the collection of separator samples is one of the least expensive methods of sampling, the accuracy of the subsequent reservoir fluid data can only be as accurate as the GOR/CGR used for the recombination. For black oils, there are often inaccuracies in GOR measurements. In just one example, single phase bottom hole samples were collected at the same time as separator samples. The analyses of the single phase samples confirmed that they were representative. The measured bubble point of the recombined separator samples was around 1000 psi too low purely because of an inaccurate field measured GOR. For gas condensates, the inaccuracies in CGR measurements are often very significant.

Inaccurate PVT Properties.

Unfortunately, there are no industry standard methods for measuring PVT data. Consequently, each service provider has developed their own techniques and often use different equipment platforms. Analyses of the pressurised samples is complex and demands fine attention to detail. 

Inaccurate Fluid Compositions.

These days, the single phase fluid composition is arguably the most important fluid property. This is normally performed by flashing the pressurised fluid into a gas and liquid phase; the two phases are then analysed and the resultant compositions are mathematically combined using the flashed GOR (CGR). There are industry standard methods for measuring the gas composition. However, there are no such standards for performing the flash. In fact, the preferred methods are different for black oils and gas condensates. Again, there are different methods for obtaining the compositions of the liquids in the industry. The end result is that a lot of reported compositional data is inaccurate and not suitable for use with a PVT simulator.

During a recent project, my client requested that we perform a comparative study between two separate service providers on duplicate samples. The comparison of the resultant PVT data on this lean gas condensate was very acceptable. However, the two compositions were very different. Our normal supplier used preferred procedures for analysing the sample and had very precise QC and calibration procedures for the equipment. The other supplier did not even use the industry standard method for analysing the flashed gas phase. The client decided not to send any more samples to the second laboratory.