Models Cross-Validation as a Best Practice for Optimization of Well Stimulation

Bochkarev Alexander,Chernyshev Dmitriy,Gaysina Narkes,Lisitsyn Andrey

DOI: https://doi.org/10.2118/202876-ms

2020-11-09

Abstract:Abstract Development of low-permeable oil reservoirs with complex bedding often requires application of production stimulation methods such as hydraulic fracturing. Success of well stimulation greatly depends on quality of geomechanical and hydraulic fracture models. Several different workflows can be used to build a certain model when the input data available are necessary and sufficient. However, in the absence of some of the key data, it is necessary to use rational assumptions instead, but doing so automatically regards model output as probabilistic. This paper concentrates on steps necessary to constrain initial assumptions made within models and demonstrates the result obtained. We present an integrated workflow used for cross-validation of geomechanical and hydraulic fracturing models based on production history match as a constrain. Relatively thin 35 deg API oil-bearing ∼ 0.3 mD hydrostatically pressurized zone, represented by interbedding of heterogeneous sedimentary layers of gross height of ∼ 15 m out of which about 5.5 m is pay, with argillitic barriers at top and bottom (see Fig. 2) has active water-bearing formation 45-50 meters above. Production history available for stimulated horizontal wells has shown a significantly increased water-cut compare to one expected from the zone of interest. Earlier frac company stress models were made by tying pressure gradients - "regional experience" - to gamma ray and SP logs and attributing E, KIc to lithology. These fracture models were consistently showing hydraulic fracture containment by barriers of the target zone and, thus, fracturing into water zone was not viewed as a highly probable cause of an increased water cut. Our target was to resolve the water cut problem by means of geomechanical and retrospective hydraulic fracturing modeling. These two models were initially built on incomplete data due to the absence of: pore pressure measurements mechanical properties (static Young's modulus) from core analysis. The models were then cross validated by constraining on production history while utilizing Young's modulus as variable. PEBI-grid based reservoir simulator is fully coupled with finite difference planar full-3D hydraulic fracturing simulator and allows direct transfer of fracture geometry (conductivity) for multiphase flow simulation.