حبیب رستمی

In underground gas storage (UGS) reservoirs, deliverability and velocity of gas flow toward the well is very high and rate-dependent pseudo skin may be a big part of the total skin factor around the wellbore. Therefore, accurate determination of non-Darcy factor, D, can be very important in exact prediction of rate-independent skin (or true skin) factor and deliverability of the well. Multirate tests provide reasonable estimates of reservoir parameters such as true skin factor and non- Darcy factor. However, running a multi rate test is much more expensive and time consuming than single rate tests. Especially in the case of UGS reservoirs, running a multi stage test can be risky, as these reservoirs are usually designed for supply of energy in cold months of the year and any interruption in constant production of gas for running multi rate tests can be critical. Therefore, the use of these tests should be minimized in analysis of UGS reservoirs. The objective of this study is to use back-propagation neural network (BPN) in prediction of non-Darcy factor in some UGS reservoirs by using reservoir properties. Then, based on the proposed correlation and analysis of single rate tests, the reservoir parameters, i.e. non-Darcy factor and true skin factor for each well were calculated. The results indicate that the presented artificial neural network (ANN) is appropriate to estimate skin factor in these reservoirs.