HYBRID MODELS BASED ON PHYSICS AND AUTOMATIC LEARNING FOR TANK SIMULATIONS

摘要

The invention relates to a system and methods for simulating fluid flow during downhole operations. Measurements of an operating variable at one or more locations inside a formation are obtained from a downhole tool placed in a wellbore inside the formation during a step during a downhole operation carried out along the wellbore. The measures obtained are applied as inputs to a hybrid training model. The hybrid model consists of physics and machine learning models coupled together within a simulation grid. The fluid flow inside the formation is simulated, based on the inputs applied to the hybrid model. A response from the operating variable is estimated for a next step in the downhole operation along the wellbore, based on the simulation. The flow control parameters for the next step are determined based on the estimated response. The next stage of the operation is carried out according to the determined flow regulation parameters.