Multiwell deconvolution converts pressure and rate histories from interfering wells into
constant-rate pressure derivative responses of individual wells as if they were producing
alone, and provides interference effects from any one well to any other. The deconvolved
responses have the same duration as the corresponding pressure histories and allow identifying reservoir features not visible during individual build ups.
The multiwell deconvolution algorithm used in this study was developed over several
years as part of Imperial College Joint Industry Project (JIP) on “New Developments in
Well Test Analysis”. A latest development was the introduction of constraints to decrease
the non-uniqueness of the solutions and ensure their physical validity.
The objective of the present study is (1) to further test the constrained multiwell
deconvolution algorithm; and (2) to develop a practical workflow for deconvolution of well
test data in interfering wells, covering the most common conditions encountered in actual
fields, in particular highly depleted gas fields, and incomplete or erroneous flow rates.
Deconvolution can only be applied to linear systems, i.e. to oil reservoirs directly,
and to gas reservoirs with low to moderate pressure depletion after pressures have been
replaced by pseudo-pressures to correct for non-linearity. Gas reservoirs with high pressure
depletion, however, are highly non-linear systems for which linearization requires the use
of pseudo-times instead of actual times in addition to the use of pseudo-pressures. In this
study, pseudo-times are calculated numerically with a consistent, stable and fast numerical
technique, called generalized integral quadrature (GIQ), which can handle sensitivity to numerical computations.
A methodology supported by theory is introduced to correct for erroneous or missing
rates. The use of multiwell deconvolution for oil reservoirs and gas reservoirs with low and
high pressure depletion, and for erroneous or missing rates is illustrated with theoretical
and field examples.
Finally, this thesis investigates a number of practical issues in the use of multiwell
deconvolution, such as the influence of numerical magnitude on deconvolution; and the need to convert all pressures to a common datum prior to deconvolution.