Abstract
Karachaganak field, one of the largest accumulations of gas and condensate in the world, is in production since 1985. The field is an isolated carbonate platform with a hydrocarbon column of about 1500 m.
The analysis and integration of the vast amount of geological and production data allowed the building of a reservoir model characterised by a high quality History Match.
According to the usual approach, the following uncertainty analysis consists in the building of several models representing possible alternative scenarios endorsed by a similar quality HM. However, this approach was dropped because not feasible for a so large reservoir with a long production history, where the achievement of a high quality HM is time consuming.
The workflow has been, therefore, turned upside down using the HM reference model as benchmark, modifying its characteristics only in the areas not controlled by well data, without perturbing significantly the achieved HM. Estimating possible petrophysical changes (Phi, Kh, Kv, Swi), using the "distance" from productive wells and the magnitude of the HM perturbations as control parameters, "end members" models, capable to give a significant spread in the production forecast, were identified.
Finally combining these "end members" as "corner points", an Experimental Design technique allowed to build the PROXY model for relevant response variable. The validated PROXY models have been then used to generate, with Monte Carlo approach, random sampled profiles which give a reliable picture of the uncertainties.
