The compositional, full-field model featured eight hydrocarbon
components over several hundred thousand active grid cells—including more
than 100 hydraulically fractured wells. The model was calibrated against
historical performance, including effects such as condensate banking and
velocity stripping, and was used to predict various production scenarios. These
scenarios were then optimized over all uncertain parameters.
The project was powered by a fully configured, 128-core high-performance
computing (HPC) cluster, deployed in an optimum hosting environment. The HPC
cluster used Intel Xeon X5672 processors with up to 96 GB of RAM per node.
Interconnected InfiniBand was employed as required for parallel ECLIPSE
calculations. The cluster was maintained at maximum performance by Schlumberger
staff, who also supported remote access. All transferred data was encrypted and
all data stored on the server was secured. Daily automatic backups ensured data
The Petrel uncertainty optimization process established the impact of
all uncertain parameters—geological or dynamic—using an objective
function (e.g., recovery of gas reserves). It then converted the results into a
proxy model that was optimized using Monte Carlo methodology.
The resulting uncertainty analysis workflow enabled a combination of
geologic and engineering variables to be evaluated, which was previously
impossible. For the first time, the impact of geological parameters—such
as absolute and relative permeability, and condensate gas ratios—were
investigated and ranked. These variables were complemented by evaluating the
impact of engineering parameters, such as varying fracture length, for a more
coherent analysis of the factors that affect cumulative gas production.
Wintershall was able to perform a comprehensive analysis of all uncertain
parameters to improve production forecasting and field development
E-mail firstname.lastname@example.org or contact your local Schlumberger representative to learn more.