Almost all the data used in reservoir simulation are subject to uncertainty, which may be quite large especially for an immature reservoir, consequently, reserves cannot be estimated with accuracy for such reservoirs. Current stochastic procedures as the multiple realization approach require large number of simulation runs, such intensive computational effort is sometimes prohibitive. The experimental design method has been used in this study to quantify uncertainty in production forecast by running only a reasonable number of simulation runs.

This study has been carried out as follow, first the uncertain parameters have been identified, a screening design was performed to restrain the analysis to the key parameters, then a more elaborated set of sensitivity runs was designed and run, the design included the key uncertain parameters and the development parameters like well spacing and abandonment pressure, the results of these sensitivity runs have been then used to substitute the reservoir simulation model by a nonlinear function. After verifying the quality of the fit, this nonlinear function was used to carry out Monte Carlo simulation. This procedure was illustrated through Tiguentourine gas condensate field case.

Including the development parameters in the design made possible the estimate of uncertainty in reserves and production profiles for different development scenarios. One major advantage of using experimental design technique is that during the progress of this study as a new well has been drilled, the probability distributions of some uncertain parameters have been reviewed, using the fit of the reservoir simulation model, an update of the production forecasts was possible almost instantaneously and without any extra simulation run.


Reserves estimates are generally highly uncertain at early stage of field development given the sparse geological data.

Considering the large initial capital investments associated with most fields development, decisions should be made in full awareness of geological risk. Many stochastic methods have been used to quantify uncertainty, those methods generally require large number of expensive fluid flow simulation runs, most of them may lead to significant time and manpower commitment. The problem can be exacerbated if different development options have to be considered since reserves should be estimated for each development option separately.

Experimental design was proposed by Damsleth et al1 to get around of this problem by running only a limited a set of well designed sensitivity runs and fit the results of the simulation model by a mathematical function that is then used as a substitute of the real reservoir numerical model to carry out Monte Carlo simulation.

The uncertain development design parameters can be incorporated into the design, so the fit of the reservoir model can be used for the analysis of the different possible development strategies.

This work reports the application of experimental design technique to a real field case where different development scenarios have been under study.


Experimental design is how to conduct and plan experiments in order to extract the maximum amount of information from the collected data in the presence of noise, in the fewest number of experimental runs.

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