Identification of an appropriate data assimilation approach in seismic history matching and its effect on prediction uncertainty

Abstract

Reservoir management may be improved if the present state of the field is known and if changes can be predicted. The former requires information about current fluid sweep and pressure change, while the latter requires accurate reservoir description and a predictive tool such as a simulation model. With this information, important decisions can then be made, including facility maintenance and well optimisation. We apply an automated history matching method which updates a parameter such as permeability, barrier transmissibilities and NTG (Net:Gross) by matching 4D seismic predictions from the simulations to observed data. Firstly, we look at the choice of starting model in the history matching process by testing our parameterisation and updating scheme to see whether it can convert a realisation into a better representation resembling reality. We set up some synthetic test cases to validate the history matching and parameterisation scheme. We find that, if we use a pilot point separation that is equivalent to the range of the variogram used in a generation of permeability distributions, we can obtain a good representation of the model. Secondly, we investigate the impact of successively updating barriers by adding new data to our observed dataset and comparing this to a single history match where all data is used. We demonstrate the method by applying it to the UKCS Schiehallion reservoir. We update an upscaled version of the operator’s model for increased speed. We consider a number of parameters to be uncertain, including barrier transmissibilities. Our results show a good match to the observed seismic and dynamic well data with significant improvement to the base case. The best result occurs when early data is used in short simulations first as we learn about optimum parameter values. Later data may be added for fine tuning or to explore new parameters. We investigate the value of seismic data in reducing forecasting uncertainty. The aim here is to look at the reduced uncertainty that we obtain in Schiehallion when we add 4D seismic to the history matching procedure. We look at the change to parameters and then take some of the best models and predict the behaviour of an in-fill well. We quantify the accuracy of history match predictions and the impact of time-lapse seismic data.