This paper discusses specific issues encountered when pressure tests are analyzed in reservoirs with complex geological properties. These issues relate to questions concerning the methodology of scaleup, the degree of aggregation, and the reliability of conventional methods of analysis. The paper shows that if we desire to use pressure-transient analysis to determine more complex geological features such as connectivity and widths of channels, we need a model that incorporates reservoir heterogeneity. This complexity can lead to significantly more computational effort in the analysis of the pressure transient.
The paper demonstrates that scaleup criteria, based on steady-state procedures, are inadequate to capture transient pressure responses. Furthermore, the number of layers needed to match the transient response may be significantly greater than the number of layers needed for a reservoir-simulation study. The use of models without a sufficient number of layers may lead to interpretations that are in significant error.
The paper compares various vertical aggregation methods to coarsen the fine-grid model. The pressure-derivative curve is used as a measure of evaluating the adequacy of the scaleup procedure. Neither the use of permeability at a wellbore nor the average layer permeability as criteria for the aggregation was adequate to reduce the number of layers significantly.
The objectives of this paper are to demonstrate the impact of the detailed and small-scale heterogeneities of a formation on the flow characteristics that are obtained from a pressure test and how those heterogeneities affect the analysis of the pressure test. The literature recognizes that special scaleup procedures are required in the vicinity of wells located in heterogeneous fields. Our work demonstrates that these procedures apply only to rather small changes in pressure over time and are usually inadequate to meet objectives for history-matching well tests. Using a fine-scale geological model derived by geological and geophysical techniques, this work systematically examines the interpretations obtained by various aggregation and scaleup techniques. We will demonstrate that unless care is taken, the consequences of too much aggregation may lead to significant errors on decisions concerning the value of a reservoir. Current scaleup techniques presume that spatial (location of boundaries, location of faults, etc.) variables are maintained. In analyzing a well test, however, one of our principal objectives is to determine the relationship between the well response and geometrical variables. We show that a limited amount of aggregation will preserve the spatial and petrophysical relationships we wish to determine. At this time, there appears to be no method available to determine the degree of scaleup a priori. Because the objective of well testing is to estimate reservoir properties, the scaleup process needs to be made a part of the history-matching procedure.
By assuming a truth case, we show that too much vertical aggregation may lead to significant errors. Comparisons with traditional analyses based on analytical techniques are made. Whenever an analytical model is used in the analysis, unless otherwise stated, we use a single-layer-reservoir solution.