A Comparison of Travel-Time and Amplitude Matching for Field-Scale Production Data Integration: Sensitivity, Non-Linearity and Practical Implications
Conference Paper
Overview
Identity
Additional Document Info
Other
View All
Overview
abstract
AbstractTraditional approach to reconciling geologic models to production data involves an "amplitude matching", that is matching the production history directly. These include watercut, tracer concentration and pressure history at the wells. It is well-known that such amplitude matching results in a highly non-linear inverse problem and difficulties in convergence, often leading to an inadequate history match. The nonlinearity can also aggravate the problem of non-uniqueness and instability of the solution. Recently, production data integration via travel-time matching has shown great promise for practical field applications. In this approach the observed data and model predictions are lined up at some reference time such as the breakthrough or first arrival time. Further extensions have included amplitude information via a generalized travel-time inversion. Although the travel-time inversion has been shown to be more robust compared to amplitude matching, no systematic study has been done to examine the relative merits of the methods in terms of the nonlinearities and convergence properties, particularly for field-scale applications.In this paper we quantitatively investigate the nonlinearities in the inverse problems related to travel-time, generalized travel-time and amplitude matching during production data integration. Our results show that the commonly used amplitude inversion can be orders of magnitude more non-linear compared to the travel-time inversion. The travel-time matching is extremely robust and the minimization proceeds rapidly even if the prior geologic model is not close to the solution. The travel-time sensitivities are more uniform between the wells compared to the amplitude sensitivities that tend to be localized near the wells. This prevents over-correction near the wells. Also, for field data characterized by multiple peaks, the travel-time inversion can prevent the solution from converging to secondary peaks, resulting in a better fit to the production response.We have demonstrated our results using a field application involving a multiwell, multitracer interwell tracer injection study in the McCleskey sandstone of the Ranger field, Texas. Starting with a prior geologic model, the traditional amplitude matching could not reproduce the field tracer response which was characterized by multiple peaks. Both travel time and generalized travel time exhibited better convergence properties and could match the tracer response at the wells with realistic changes to the geologic model. Our results appear to confirm the power and robustness of the travel-time matching for field scale production data integration.
