Well Production Forecasting by Extrapolation of the Deconvolution of Well Test Pressure Transients

Abstract

The deconvolution of well test pressure transient data provides a constant rate pressure response that applies for the duration of the test. By extrapolating this response in time, the future performance of a well can be predicted. The extrapolation can include constraints defined by estimates of the volume of reservoir connected to the well but otherwise, relies only on information that is normally available at the time of testing. Uncertainty in either the deconvolution or the connected volume will result in a variety of extrapolations that defines the range of expected well performance. The advantage of this procedure is that a well's future production profile can be estimated very quickly and easily. Such a scoping exercise helps to understand a well's potential and to define the expected range of outcomes from more detailed modelling. Furthermore, it can be an efficient method for screening prospects when time and resources are limited. By monitoring the actual well performance against its prediction, underperforming wells can be identified for work-over. If permanent downhole pressure gauge data are available, the deconvolution can be continuously updated thus improving the production forecast. Where a production well is planned that is of a different type to the tested well (e.g. a horizontal or hydraulically fractured well instead of a normal vertical well), a technique is described to modify the tested well's deconvolved response to account for the change in well type whilst preserving the pressure transient response due to the reservoir and its boundaries. Two field examples of gas well tests are presented which demonstrate the method and the predicted production profiles are compared with observed data. Introduction A well's production can be estimated by building flow models and simulating future production for a given set of well and facility constraints. Simulation relies on what can often be uncertain geophysical, geological and petrophysical information. Generally, it is more accurate if it has been calibrated (history matched) to existing production data. Alternatively, the previous production history can be examined directly, trends identified and extrapolated to forecast production assuming that the operating conditions remain unchanged. This type of decline curve analysis uses no model but does require significant flow history in order to establish the trends for extrapolation. Production data analysis using simple analytical models is also possible (e.g. Fetkovitch, Blasingame) but also requires significant production data. In the initial stages of reservoir development, there is no production and the reservoir simulation model may be quite uncertain or unavailable because of insufficient information. This paper presents a method to forecast well production based on information obtained almost entirely during well testing without the need to build a dynamic model. The method involves four steps:AcquisitionDeconvolutionExtrapolationPrediction Each step is described in the following sections and the underlying assumptions are highlighted. Two field examples are presented and conclusions drawn.

Characteristic Value Method of Well Test Analysis for Horizontal Gas Well

“…Equipped with Laplace transformation and boundary element method, Zerzar and Bettam [12] addressed an analytical model for horizontal wells with finite conductivity vertical fractures. By extrapolating the transient pressure data, a simplified approach to predict well production was presented by Whittle et al [13].…”