Black-oil: how to handle fluxes crossing PVT region boundaries
andlaus opened this issue · comments
The black-oil model's current treatment of mass fluxes that cross PVT region boundaries is incorrect:
for two PVT regions A and B, imagine that a fluid flux goes from A to B. if the selected quantities to conserve are the "surface volume" of the components, the conserved quantity does not correspond to a quantity that is conserved in the physical world anymore: if the density of e.g. gas at surface conditions is 1 kg/m^3 in region A but 10 kg/m^3 for region B, conserving the "surface volume" of the gas component in the looked uppon case adds 9 kg of gas per cubic meter that flows from region A to region B. obviously this is an extreme example, but the same argument applies for any A and B that exhibit different surface densities.
the interesting question is what can and should be done about this: on one hand, volume fluxes that cross region boundaries could be multiplied by rho_s,B/rho_s,A (this would make sure that mass is conserved), on the other hand, the black-oil model uses pseudo components that consist of a mixture of a multitude of chemical species anyway. This implies that conserving "pseudo component mass" is still incorrect because PVT regions essentially capture different compositions of these pseudo components: e.g., as far as the model is concerned, methane could be converted into butane by simply crossing a PVT region boundary.
I think that you are correct in your view that this is wrong with the standard black-oil model. There is a fairly common (meaning implemented i multiple different simulators) extension to black-oil called API tracking (not sure what API is short for), which solves exactly the problem you outline by keeping track of (in a simplified way compared to compositional simulation) the composition of the oil in the cells, perhaps we should consider implementing that if there is sufficient interest?
yes, we should probably consider implementing API tracking. it sounds quite elaborate, though...
another possibility would be to abandon black-oil with multiple PVT regions and use a fully compositional model for such cases. Unfortunately that's also very elaborate to implement.
@joakim-hove, @alfbr: do you have any idea if reservoir engineers are aware of above problem?
do you have any idea if reservoir engineers are aware of above problem?
I'll add my 2 cents to this as well... I am sure they are, and some are probably abusing the feature to get effects in the simulator they otherwise could not!
I suppose we decided that our approach here is "we don't care". closing.