At the appraisal phase of an offshore field development, there are limited amount of hard data for reservoir description. For that reason significant uncertainty exists not only in the variability of the observed petrophysics and facies data (heterogeneity uncertainty) but also in the representativeness of the observed data (model parameter uncertainty).
The aim of this study is to quantify the effect of these uncertainties on the dynamic behaviour of the reservoir and provide input for the overall uncertainty analysis. These uncertainties are important at this field development stage, contributing to decisions being made with respect to gas contracts, facility design and commercial evaluation.
It is difficult to incorporate the model parameter uncertainty by simply generating a certain number of realisations out of one global set of model parameters - i.e. a standard stochastic modelling approach. Thus, different parameter distributions were utilised as input to different heterogeneity models provided by the disciplines involved (sedimentology, petrology, petrophysics and geology). In order to ensure equi-probability of the realisations generated, a higher hierarchical model with two individual levels of uncertainty was used.
The problem arising with a hierarchical model is the large number of possible model parameter combinations. Hence, the Latin Hypercube technique was used for experimental design when planning for realisations. 99 realisations of the reservoir plumbing have been generated including facies and petrophysics modelling. However, because of the complexity of the fluid system and the need for compositional reservoir simulation very few of the realisations could be evaluated with the full field simulation model. The realisations were therefore ranked on element models using front-tracking and simplified black oil models.
The results of this analysis were applied in the overall uncertainty analysis. In addition, the results contributed significantly to the understanding of the reservoir and its potential dynamic behaviour.
Reservoir simulation is an indispensable tool in modern reservoir management. However, large uncertainties are known to exist in the production profiles thus calculated. One reason is that heterogeneities are generally poorly represented, due to lack of knowledge of the spatial distribution of the architectural elements in the reservoir. Specially in offshore reservoirs, the large distances between wells result in geological models which rarely represent the heterogeneities influencing the fluid flow. In recent years, a large effort has been dedicated to representing more correctly the geological features of the reservoir and the associated uncertainty. However, to date, little has been done in the modelling to correctly handle uncertainty of the input parameters which is critical to the outcomes of the geological architecture.