Reservoir Simulation
1. Since IMPES (implicit pressure calculation carried out @ each time step followed by an explicit saturation calculation for black-oil reservoir simulation) scheme remains to be the fastest possible approach on a per-time step basis (by avoiding the ‘slower’ and ‘stable’ ‘coupled implicit approach’, where, non-linear functions in the inter-block flow terms are evaluated at the end of each time step); - would it remain feasible to curtail/avert the 'stability issues' that restrict the time step size using ML/AI for various production scenario?
OR
Sequential implicit scheme (IMPES with implicit saturation) remains to be the best - for large models, with stability consideration - in the absence of any need for ML/AI?
OR
ML/AI would be of critical use in applying coupled semi-implicit and fully-implicit approaches?
OR
ML/AI could play a crucial role in equalizing the number of unknowns per grid-block to remain to be equal to the number of phases rather than as a function of number of components (+1)?
2. For compositional simulation, whether the basic “non-Newton-Raphson, IMPES-type procedure” (where, the pressure equation remains not only based on physical principles but also includes the compositional effects) – extended to a 'sequential implicit compositional formulation' – could efficiently make use of ML/AI towards addressing
(a) estimation of inter-block total fluid velocities with the estimated pressure @ new time level;
(b) treatment of relative permeabilities and capillary pressure implicitly by using Buckley-Leverett phase velocity and inter-block total velocities - towards estimating saturation @ new time level; and thereby subsequently evaluating the inter-block phase velocities; &
(c) estimation of inter-block component transport and the amount of each component in each block @ the end of the time-step using the phase velocities?
While doing the above process, whether ML/AI knows clearly that the pressure equation remains essentially to be a ‘volume balance’, while also knowing that the estimation of saturation equations would remain to be based on the volume of a single-phase rather than on the volume of the total mixture (although, it remains possible to apply implicit methods used in black-oil simulation directly to compositional problems)?
- Badges
- Science topic
- Similar topics
- Linguistics
- Composition Studies
- Publication