History matching complex reservoirs with hundreds of active wells over decades of production history may reveal to be tedious and complex given the amount of data to be honoured, the level of uncertainty as well as software and hardware requirements. This paper aims at sharing ADCO's experience in achieving a satisfactory match quality on a giant carbonate oil reservoir in Middle East. The approach encompasses manual as well as assisted history matching techniques in a two way communication leading to consistent reservoir images that honour all observed data and mitigate any potential pitfalls which may lead to geologically unrealistic solutions.

Main outputs enabled a better understanding of key influencing reservoir uncertainties that can improve the match quality as well as identifying a few alternative reservoir images that can replicate the observed data both at well and full field level, in addition to an impact assessment on production forecasting uncertainty. Furthermore key remaining uncertainties were pin-pointed so that they can be acted upon.

The reservoir model history matching was undertaken on a single deterministic geological realisation via manual history matching to start with, and then assisted history matching was introduced as a way to rank uncertainty parameters based on their impact on match quality and to provide some directionality for history matching improvement. Moreover, the approach enabled identifying alternative equally history matched reservoir models.

Given the high number of mis-match parameters to satisfy both at reservoir and well levels and the number of reservoir uncertainties to be tested, a structured (staged) assisted history matching approach was implemented. A physically-sound and proper set of parameters with realistic ranges have been considered at each stage of the history-match process in a logical order.

Ranges of selected parameters were designed in a way that satisfactory solutions at each stage can be carried over to the next stage to continue the history-match. This process started with the Latin Hypercube technique as a browsing phase to assess uncertainty impact on response parameters and then discard those with no impact on match quality. The next step evolved proxy modelling - a mathematically / statistically defined functions based on existing realisations that replicated the simulation model outputs for selected input parameters. This helped reducing time required to get the solution to converge. Afterwards, sequential adjustments have been implemented from global then flow units, followed by local changes in model properties using Evolutionary Strategy algorithm. During each stage of this process the "heavy-hitters" with the highest impact on the history-match process were identified and thouroughly investigated.

You can access this article if you purchase or spend a download.