Local Grid Refinement

The Well Is Where the Action Is

Pressure in a reservoir does not change evenly. Far from any well it varies slowly across hundreds of metres, but right at a wellbore it plunges over a metre or two as fluid rushes into the perforations. A grid coarse enough to be affordable across the whole field is far too coarse to capture that steep near-well drawdown, and a model that smears the pressure around the well will misjudge its flow rate and its water or gas breakthrough. The near-well region is exactly where the grid most needs fine cells.

Refine Locally, Not Everywhere

Local grid refinement (LGR) answers this without bankrupting the simulation. Instead of shrinking every cell in the model, it subdivides only a small block of cells around the well into a fine sub-grid, leaving the rest of the field coarse. Move the well in the widget and toggle the refinement: a handful of coarse cells each split into a fine patch, and the readout compares the resulting cell count with the cost of refining the entire model to the same near-well resolution. The difference is routinely one or two orders of magnitude.

Spend Cells Where They Change the Answer

LGR is the budgeting idea from the previous section applied surgically. The fine cells go where the pressure gradient is steepest and the flow physics is most demanding, and nowhere else. It is the standard way to model near-well behaviour, such as water or gas coning toward a producer and the detail of a deviated or horizontal completion, inside a field-scale grid that must still run overnight. Used well it buys local accuracy almost for free; scattered carelessly across the model, it gives back the cost it was meant to save.