Naturally fractured reservoirs (NFR) are highly complex from their characterization to thier exploitation; their behavior depends on two systems: the fracture and the matrix. This complex nature hinders the development and adjustment of numerical models; most parameters present high uncertainties. Simulation engineers spend a lot of time obtaining a representative and reliable model. This work was developed with the purpose to establish an assisted history matching methodology using an evolution strategy algorithm (ESA) to accelerate the construction of this reliable model. Evolution algorithms were first showcased in the 1960's and in recent years have expanded their use as a mechanism for acceleration of history matching or production optimization.
Additionally we used ESA to relocate already programmed wells and thereby new potential areas for exploitation were identified, where wells weren't considered in the initial development strategy, managing to increase the ultimate recovery factor (URF). Evolution Strategy Algorithm was used successfully in a field of the south of Mexico with a numerical model available, which has a 4 year production history through 14 active wells. The field is a NFR with high production potential and strong water breakthrough in the wells. Starting from a previous simulation model, we establish 3 stages for the proposed work flow:
Analysis of geological variables, the discrete fracture network (DFN) and identifying and weighting the impact on the dynamics of both fluids systems.
Application of ESA using the variables identified on stage 1, analyzing several simulation models and using an objective function to quantify the miss match from our ideal model.
Optimization of the proposed wells, evaluation of areas with exploitation potential and proposing of new wells in these areas using ESA.
Using ESA we manage to optimize simulation times and to achieve a reliable and representative model in 75% less time than with our previous effort. This history matched model includes a DFN and reproduces the water breakthrough behavior on 90% of the wells and in 100% of wells with higher oil and water production. The result model was used to relocate the next two proposed wells into a new area applying ESA; where the cumulative oil production was maximized at the end of the simulation period, increasing URF in 2.5% respect to the original development plan and the interference with neighborhood wells were minimized.
Finally as consequence of proposed well relocation, we found a new area in the field with potential to allocate an additional well. Setting the ESA to maximize cumulative oil production yield to the location optimization on this new well, thus increased URF another 1.2%. Assisted history matching using ESA reduces substantially time analysis compared with traditional methods and, it was possible to have a reliable model with DFN that reproduces water breakthrough and captures the heterogeneity of the fractured system. The methodology proposed helps to accelerate decision-making with more technical support and has the flexibility to reevaluate the exploitation strategies and to maximize projects profitability.