Deconvolution
Deconvolution in pressure transient analysis is a mathematical operation that uses downhole flow-rate measurements (or surface flow-rate measurements when downhole rates are not available) to transform bottomhole pressure measurements distorted by variable production rates into the equivalent constant-rate pressure-transient response that can be analyzed using standard analytical pressure transient theory — the operation effectively removes the variable-rate distortion from the pressure data and produces a clean transient that can be interpreted using Horner plots, type curves, or numerical reservoir simulation as if the well had been produced at constant rate; deconvolution has the major advantage over the alternative analytical approach (convolution, where a model is fit to the variable-rate pressure data through forward simulation) that it does not assume any particular reservoir model for the pressure-transient response — the deconvolved transient can reveal whatever flow regimes (radial flow, hemispherical flow, linear flow, boundary effects, dual porosity) are actually present in the reservoir without bias toward an assumed model; the simplest forms of deconvolution (direct division of pressure transform by flow rate transform in the frequency domain) often give noisy results due to amplification of measurement noise, while more complex approaches (regularized deconvolution, Levitan deconvolution, Schroeter deconvolution) produce cleaner results at the cost of computational complexity and the requirement to balance denoising against signal preservation; modern pressure transient analysis software including KAPPA Saphir, Schlumberger Eclipse, Computer Modelling Group GEM, and others includes built-in deconvolution algorithms that allow practical application to typical well-test data.
Key Takeaways
- Variable-rate pressure response problem in conventional pressure transient analysis arises whenever the well being tested has been produced at varying rates rather than at the constant rate assumed by classical analytical methods — typical sources of rate variation include changing operating conditions during the test (deliberate rate changes for multi-rate testing), uncontrolled rate variations from production system upsets, and the unavoidable variations between flow and shut-in periods of any practical well test sequence; the pressure response at any time during the test depends on the entire history of rate variations through the convolution integral P(t) = integral from 0 to t of [Q(tau) * dPunit(t-tau)/d(t-tau)] dtau, where Punit is the unit-rate response and Q(tau) is the production rate history; the convolution integral cannot be solved analytically for arbitrary rate histories, so traditional analysis required either the assumption of constant rate (introducing systematic errors when rates actually varied) or the application of superposition methods that approximated the rate variation as discrete rate changes (introducing approximation errors).
- Deconvolution algorithm approaches include direct frequency-domain methods, time-domain regularized methods, and explicit Volterra integral equation methods — frequency-domain deconvolution uses the Fourier transform property that convolution in time domain corresponds to multiplication in frequency domain, allowing division of the pressure transform by the flow rate transform to recover the unit-rate response transform, which can then be inverse-transformed to give the deconvolved time-domain response; this approach is mathematically simple but suffers from noise amplification when the flow rate transform has values near zero at high frequencies; regularized methods (Tikhonov regularization, von Schroeter regularization) add a smoothing constraint that reduces noise amplification while accepting some loss of signal accuracy; the Levitan method (developed by Mike Levitan in 2005) uses a constrained least-squares formulation that incorporates physical constraints (positive deconvolved response, monotonic increase) to produce stable, physically reasonable deconvolved transients.
- Pressure-rate convolution and superposition theory underlies the deconvolution mathematics and provides the framework for analyzing variable-rate well tests — the key principle is that for a linear system (flow in porous media is linear in the pressure-flow rate relationship under most conditions), the response to any rate history can be computed as the convolution of that rate history with the unit-rate response of the system; the deconvolution operation is the mathematical inverse of this convolution, recovering the unit-rate response from observations of the actual pressure history and the actual rate history; the linear system assumption breaks down for systems with strong nonlinear effects (high-pressure-drop conditions, gas-condensate behavior, multiphase flow with relative permeability changes), and deconvolution analyses for these systems require appropriate corrections; for the common case of single-phase or modestly compressible single-phase flow, the linear assumption is generally valid and deconvolution provides accurate results.
- Deconvolution applications in pressure transient analysis include extending the effective duration of analysis beyond the longest single flow or shut-in period, allowing identification of late-time flow regimes (boundary effects, formation depletion) that would not be visible in any single segment of the pressure history; resolving multiple flow regimes that overlap in time within different rate periods; and providing the unit-rate response for type-curve matching even when the actual test was conducted with multi-rate sequences; modern PTA practice routinely applies deconvolution to extend the effective analysis duration of typical well tests, improving the reservoir characterization information obtained from limited test durations.
- Practical limitations of deconvolution include the requirement for accurate, time-synchronized flow rate measurements (typically requiring downhole flow rate measurement or careful surface rate accounting), the noise sensitivity of the deconvolution result that requires regularization parameters to be properly tuned, and the linear system assumption that may not be satisfied in some applications; modern deconvolution software typically uses regularization parameters that are automatically tuned based on the noise characteristics of the input data, but the analyst should verify that the deconvolved response is physically reasonable (monotonically increasing, no unphysical oscillations, consistent with expected flow regimes) before relying on the deconvolved transient for reservoir characterization conclusions; for cases where the deconvolved response shows pathological behavior, alternative analysis approaches (segment-by-segment analysis, numerical simulation matching) may be more appropriate.
Fast Facts
Deconvolution methods for pressure transient analysis evolved through several generations of algorithm development, with key contributions from researchers including Lawrence Aanonsen, Mike Levitan, Rolf von Schroeter, and others over the period from the 1980s to the 2010s. The Levitan deconvolution algorithm, introduced in 2005, became a widely used standard due to its combination of computational efficiency, stable handling of noisy data, and incorporation of physical constraints. Modern commercial pressure transient analysis software (KAPPA Saphir, Schlumberger Eclipse, Computer Modelling Group GEM, Halliburton EasyTSE) includes deconvolution capability as a standard feature, making sophisticated deconvolution-based analysis available to practitioners without requiring specialized algorithm development. Deconvolution is now a routine element of pressure transient analysis in major petroleum company workflows, particularly for offshore and HPHT applications where the test cost justifies the additional analytical complexity.
What Is Deconvolution?
Conventional pressure transient analysis assumes that a well test consists of clean constant-rate flow periods followed by clean shut-in periods, allowing classical Horner plot or type-curve methods to be applied directly. Real well tests rarely satisfy this idealization — production rates vary during flow periods due to operational fluctuations, the bridge between the test and any prior production may not be cleanly defined, and operational events (separator changes, choke adjustments, upsets) introduce additional rate variations. Deconvolution is the mathematical operation that transforms this messy reality into the clean constant-rate response needed for conventional analysis.
The mathematics is based on convolution theory, which states that the pressure response of a linear system (porous media flow under typical conditions) to any arbitrary rate history can be computed by convolving that rate history with the system's response to a unit rate change. Deconvolution is the inverse of this convolution: given the actual pressure history and the actual rate history, deconvolution recovers what the pressure response would have been if the well had been produced at constant rate from time zero. The deconvolved response can then be analyzed by standard methods, providing the formation properties (permeability, skin, boundaries) that the test was designed to determine.
Deconvolution Application in Modern Pressure Transient Analysis
Modern PTA practice routinely applies deconvolution to extract the maximum information from well-test data. The workflow involves loading the pressure-time and rate-time data into PTA software, selecting the analysis time window, applying the deconvolution algorithm with appropriate regularization parameters, and inspecting the resulting deconvolved transient on diagnostic log-log plots. The deconvolved transient should show physically reasonable behavior including a clear pressure derivative pattern that reveals the flow regimes present in the reservoir (early-time wellbore storage, intermediate-time radial flow, late-time boundary effects). Analysis of the deconvolved transient using type curve matching or analytical methods provides the formation properties without the systematic bias that variable-rate effects would introduce. The deconvolution is particularly valuable when the well test has not reached the late-time radial flow regime within the available test duration — deconvolution can extend the effective analysis duration substantially, revealing flow regimes that would not be visible in any single segment of the actual test history.
Deconvolution Use Across International Pressure Transient Analysis
Canada (AER / WCSB): Canadian pressure transient analysis practice for major fields routinely includes deconvolution as part of high-quality test interpretation, particularly for unconventional well tests where multi-rate sequences are common; major operators integrate deconvolution into their reservoir simulation workflows with the deconvolved response providing constraints on the reservoir model.
United States (API / EIA): US unconventional shale well testing extensively uses deconvolution to interpret the complex variable-rate pressure responses characteristic of post-fracture flowback and early-time production; the technical sophistication of US PTA practice supports widespread use of deconvolution as a standard analytical tool.
Norway (Sodir / NORSOK): NCS exploration and field development tests routinely use deconvolution as part of comprehensive PTA analysis; Norwegian operators have invested in advanced deconvolution capability for HPHT applications where the test cost justifies the additional analytical effort.
Middle East (Saudi Aramco): Aramco's PTA workflow includes deconvolution as a standard analytical method for major field tests; the company's research at EXPEC ARC has contributed to advances in deconvolution methodology adapted to Aramco-specific reservoir conditions.