Abstract
Rate transient analysis (RTA) is routinely used to analyze the production history of unconventional wells. It is used to determine the time at the end of linear flow (time to depletion), to estimate drainage volume, and to predict the estimated ultimate recovery (EUR) of wells. RTA routinely employs a rate-normalization technique to characterize how the well rate changes with a changing bottomhole flow pressure. However, rate normalization is an approximation to a more general technique known as pressure deconvolution. This work evaluates and compares the performance of rate normalization and pressure deconvolution for both synthetic and tight-oil examples.
We generate synthetic examples with known errors in the time-varying bottomhole flowing pressure (BHP). First, we apply: (a) rate-normalization, and (b) pressure deconvolution to obtain an equivalent rate at constant BHP conditions: the unit-pressure-drop rate. This study uses a regularized exponential basis function inverse scheme to deconvolve the pressure variations. Second, we fit and hindcast production using the single-phase slightly compressible rate-time model to the: (a) normalized rate, and (b) deconvolved unit-pressure-drop rate. Third, we cast back pressure variations into the fitted model by (a) multiplying by the pressure drop for each time, and (b) applying time superposition. Finally, we compare the results in terms of the reservoir properties and EUR predictions. We conclude by illustrating the application of this procedure to tight-oil wells.
For the synthetic cases, the hindcast of the single-phase rate-time model using pressure deconvolution can accurately estimate the time of end of linear flow, the hydrocarbon pore volume, and the EUR. In contrast, the rate normalization single-phase model fit cannot accurately estimate these three quantities. In addition, this technique is not able to check, and eventual correct errors present in the BHP measurements. Regarding the tight-oil wells, pressure deconvolution can identify flow regimes present in the well and provides excellent fits to the oil rate histories. Conversely, rate normalization is not always able to correctly identify the flow regime(s) and produces poor history-matches to the oil production. We show that pressure deconvolution produces more realistic estimates of model parameters, time of the end of linear flow, and EUR.
This paper compares the application of rate normalization and pressure deconvolution to history-match and forecast unconventional oil production using a rate-time model. Based on the results of this work caution is required when applying rate normalization for main two reasons. First, rate normalization is a first-order approximation to pressure deconvolution. Second, rate normalization does not provide a method to correct for possible errors present in the BHP measurements. For these reasons, rate normalization might lead to inaccurate results. Therefore, pressure deconvolution should be the preferred approach.