Time-lapse (or 4D) seismic is increasingly being used as a qualitative description of reservoir behaviour for management and decision making purposes. When combined quantitatively with geological and flow modelling, as part of history matching, improved predictions of reservoir production can be obtained. Here, we apply a method of multiple model history matching based on simultaneous comparison of spatial data offered by seismic, as well as individual well production data.
Using a petro-elastic transform and suitable re-scaling, forward modelled simulations are converted into predictions of seismic impedance attributes and compared to observed data by calculation of a misfit. With a similar approach applied to dynamic well data, the total misfit is used to update model probabilities in a Bayesian framework. These are then used to guide the re-sampling of the parameter space, generating multiple models using the quasi-global stochastic Neighbourhood Algorithm (NA) method. This approach improves on gradient-based methods by avoiding entrapment in local minima and can be used to efficiently analyse uncertainty of the matched parameters.
We demonstrate the method by applying it to a UKCS turbidite reservoir, updating the operator's initial model, which had previously been history matched to dynamic well data by conventional methods. We consider a number of parameters to be uncertain. The reservoir's net to gross is initially updated to better match the RMS amplitudes of the observed seismic baseline survey. We match simultaneously for permeability, fault transmissibility multipliers and the petro-elastic transform parameters. Our results show a good match to the observed seismic and dynamic well data with significant improvement to the operator's base case.
Introduction
Reservoir management requires tools such as simulation models to predict asset behaviour. History matching is often employed to alter these models so that they compare favourably to observed well rates and pressures. This well information is obtained at discrete locations, and thus lacks the areal coverage necessary to accurately constrain dynamic reservoir parameters such as permeability and the location and effect of faults, among others. Time-lapse seismic captures the effect of pressure and saturation on seismic impedance attributes giving two-dimensional maps or three-dimensional volumes of the missing information. The process of seismic history matching attempts to overlap the benefits of both types of information to improve estimates of the reservoir model parameters.
We present an automated history matching method that includes time-lapse seismic along with production data, based on an integrated workflow (Figure 1). It improves on the classical approach, where the engineer manually adjusts parameters in the simulation model. Our method also improves on gradient-based methods, such as Steepest Descent, Gauss-Newton and Levenberg-Marquardt algorithms (e.g. Lépine, et al., 1999, Dong and Oliver, 2003, Gosselin et al., 2003, Mezghani, et al, 2004), which are good at finding local likelihood maxima but can fail to find the global maximum. Our method is also faster than stochastic methods such as genetic algorithms and simulated annealing, which often require more simulations and may have slower convergence rates. Finally, multiple models are generated enabling uncertainty analysis in a Bayesian framework. The posterior probability surface is resampled to obtain parameter distributions.
We have applied our method to a UKCS turbidite reservoir, in which time-lapse seismic has shown great promise (Chapin, 2000; Parr et al., 2000; Saxby, 2001). The original geological model was constructed by the field operator using typical approaches where facies objects and static flow properties such as porosity and permeability were distributed stochastically. The model was constrained to well logs and a qualitative match to the baseline seismic survey was obtained. The operator upscaled and then manually history matched the model to the well production data before we applied our method.
