In interpretation of well tests especially model selection and verification, two parallel and equally important sources of diagnosis are widely used by reservoir engineers. They are pressure transient testing itself and the knowledge of the relevant geological and other engineering information on the well/reservoir being tested. The former is based on significant shapes displayed on the test signal, which may correspond to features (behaviour, structures etc) of the tested reservoir. The latter is to predict the reservoir dynamic behaviour on the basis of all the available external geological and engineering data. They are complementary to each other. The interpretation cannot be done correctly using only one of them.
Previous work on the application of artificial intelligence (AI) techniques to this domain only attempted to automate the former diagnostic; nothing has been done on the latter case - an expertise widely used in practice and not yet codified. In addition to further development of the previous work, one of objective of this study is to fill this existing gap. The paper also addresses verification - an equally important task but receiving little attention, whereby the results of interpretation are checked with all the known external data, as well as the well test data, to check for inconsistencies.
As a result, a framework of a knowledge based system (KBS) for well test design and interpretation was proposed and prototyped. It incorporates both pressure transient test data and external geological and engineering data to provide a systematic approach to designing and interpreting the well tests.
The work presented in the paper provides some guidelines in development of new generation well test interpretation software, which will have to interpret the well tests from an systematic point of view and in which the model selection and verification will be essential parts of the whole process.