There are several methodologies that use 4D seismic (4DS) information as deterministic dynamic observed data to improve the accuracy of reservoir simulation models. However, 4DS data also have uncertainties due to errors caused by resolution, acquisition, processing and interpretation. Deterministic 4DS information does not consider these uncertainties properly. Thus, the objective of this work is to evaluate the application of a methodology that incorporates 4DS and simulation data probabilistically, considering uncertainties from both data sets.

Through the combination of uncertain reservoir parameters, multiple simulation models are generated/simulated, providing multiple pressure change (ΔpSIM) maps. In this study, the available 4DS data are multiple pressure change (Δp4DS) maps yielded from a probabilistic seismic inversion. We first transfer the Δp4DS maps to the simulation scale and then, for each reservoir position, we build two probabilistic density functions: (1) PDFSIM using ΔpSIM maps and (2) PDF4DS from Δp4DS maps. We then compare PDFSIM and PDF4DS, establishing which one is the most precise in each reservoir position, and use this information to select the most precise ΔpSIM and ΔpSEIS maps.

We used a synthetic reservoir with moderate complexity, eight predominant uncertainties (such as permeability and porosity), 11 producers and eight water injectors. Two sets of ΔpSIM were studied: (a) 500 ΔpSIM maps generated from 500 simulation models at the beginning of a probabilistic history-matching procedure using wells and (b) 97 ΔpSIM maps from history-matched models using well production data. Two sets of 4DS data were considered: (c) 125 Δp4DS maps yielded from a probabilistic inversion considering reservoir uncertainties and noiseless synthetic seismic data (d) 500 Δp4DS maps from a probabilistic inversion considering reservoir uncertainties and noisy synthetic seismic, representing a more realistic 4DS data. Using as input data (a) and (c), we identified positions where Δp4DS maps were more precise. We then selected the ΔpSIM maps using Δp4DS values at these positions. As result, selected ΔpSIM maps presented more precise estimates of pressure than the initial ΔpSIM maps. Applying (a) and (d), more precise estimates of ΔpSIM were also obtained. Using (b) and (c+d), we identified, through our methodology, reservoir positions where simulation data are more precise than 4DS data. The selected Δp4DS maps using ΔpSIM values presented more accurate and precise estimates then the initial Δp4DS maps.

The main contribution of the work was to apply a methodology incorporating simulation and 4DS data, considering uncertainties from both. Applying the studied methodology, we can use seismic information to update simulation models, as well as simulation data to reinterpret 4DS data (employment of engineering data to constrain 4DS).

This content is only available via PDF.
You can access this article if you purchase or spend a download.