Response Matched

Key Takeaways

• Seismic-to-well tie through response matching is the critical quality control step that validates whether a seismic interpretation is geologically consistent with the well data, or whether a systematic phase, polarity, or time-depth conversion error is corrupting the interpretation — to create a synthetic seismogram for seismic-to-well tie, the log analyst uses the P-wave sonic log (which provides interval velocities) and the density log (which provides bulk density) from a well to compute acoustic impedance (the product of velocity and density) at each depth; the reflection coefficients between adjacent impedance contrasts (formation boundaries) are convolved with the seismic wavelet (extracted from the seismic data near the well) to produce a synthetic seismogram that should look like the seismic trace at the well location if the seismic data, the logs, and the depth-time conversion are all correct; when the synthetic does not match the seismic trace (wrong event polarity, time misalignment, amplitude mismatch), the mismatch reveals a specific interpretation problem that must be diagnosed and corrected before any seismic attribute extraction or reservoir property mapping can be trusted; a poor seismic-to-well tie means the "reservoir" being mapped on seismic is not actually the reservoir being drilled, with obvious and expensive consequences for well placement.
• Response matching in well test interpretation requires both a mathematical model that can compute the pressure response for a given reservoir description and an optimization procedure that adjusts the model parameters until the computed and measured responses agree within an acceptable tolerance — modern well test interpretation software (Kappa Ecrin, IHS Harmony, Saphir) provides a library of analytical models (homogeneous radial, dual porosity, dual permeability, channel, closed boundary, constant pressure boundary) and numerical models (for complex geometries) and allows the interpreter to match the measured pressure and pressure derivative versus time data by adjusting model parameters; the match is evaluated visually (does the simulated log-log derivative curve match the measured derivative's shape and level during radial flow, transitional flow, and boundary flow periods?) and quantitatively (what is the sum of squared residuals between measured and simulated pressure at each time point?); a non-unique match (multiple different models and parameter sets that all match the data equally well) is the fundamental challenge of response matching in well testing, requiring the interpreter to impose geological constraints (known boundary distances from seismic, known formation type from log analysis) to select among competing models that mathematically fit the data but are geologically different.
• Reservoir simulation history matching uses response matching across years or decades of production and pressure history from dozens or hundreds of wells to constrain a reservoir model that is then used to predict future performance — the history matching process adjusts uncertain reservoir model parameters (permeability distribution, fault transmissibility, aquifer strength, relative permeability) until the simulator reproduces the observed field production (oil rate, water cut, gas-oil ratio at each well over time) and pressure history (measured bottomhole pressures and well test results); a response-matched history match means the simulation's predictions agreed with actual field behavior during the historical period, which provides some confidence (though not certainty) that its predictions of future behavior under development scenarios are also correct; history matching is time-consuming (hours to months of computational time for large models, iteratively adjusted and re-run), non-unique (many different permeability distributions can reproduce the same production history), and can be achieved by compensating errors rather than correctly representing geology; a history match that was achieved by incorrect physics (wrong rock properties that happen to produce the right total production by coincidence) can give dramatically wrong predictions under different development scenarios; this is why history matching is an engineering discipline requiring geological judgment, not just an optimization problem to be solved by software.
• Response matching of formation evaluation logs (log modeling) allows petrophysicists to verify their interpretation of formation properties by computing what the logging tools should have measured if the rock had those properties, and comparing the prediction to the actual log data — if a petrophysicist interprets a sandstone interval as having 25% porosity, 40% water saturation, and 15% clay content based on log analysis, they can use a rock physics model to compute the expected density, neutron porosity, resistivity, and gamma ray values for a rock with those properties and compare the predictions to the actual logs; if the model predictions match the logs, the interpretation is internally consistent; if the predictions do not match (the computed density is higher than measured, or the computed resistivity is lower than measured), the discrepancy identifies a specific error in the interpretation model — the porosity or saturation estimates are wrong, the mineralogy model is incorrect, or the tool response model is using wrong environmental corrections; this log response modeling approach is particularly valuable in complex lithologies (mixed carbonates, tuffaceous sands, conglomeratic reservoirs) where standard interpretation methods are unreliable and the response matching approach provides the only rigorous check on interpretation consistency.
• Non-uniqueness in response matching is the fundamental limitation that prevents any single matched model from being definitive, and managing this non-uniqueness requires either additional independent data to distinguish competing models or explicit uncertainty quantification — in seismic-to-well tie, a synthetic can often be made to match the seismic by adjusting the wavelet phase and frequency within reasonable bounds, but different phase corrections produce different geological interpretations of which reflections correspond to which bed boundaries; in well test interpretation, a dual-porosity model and a finite-conductivity fracture model may produce nearly identical log-log pressure derivative responses, making them indistinguishable from the transient data alone; the correct approach to non-uniqueness is not to pick the "best match" and treat it as the truth, but to quantify the uncertainty in the matched parameters (using Monte Carlo analysis, stochastic history matching, or explicit exploration of the alternative matching models) and propagate that uncertainty into the engineering decisions that depend on the matched parameters; a permeability estimate with 50% uncertainty has very different implications for development planning than one with 10% uncertainty, and communicating that difference honestly is as important as performing the response matching correctly.