Optimization sequence of infill well-drilling using Latin hypercube plus radial basis function network

Document Type : Research Paper


1 Petroleum Research and Development Center, Iraqi oil ministry, Baghdad, Iraq

2 Petroleum Engineering Department, college of engineering, Baghdad University, Baghdad, Iraq



Infill drilling is the first choice to increase the recovery factor, but the mission of selecting the best well location is considered a major challenge with the huge area of the reservoir and the time consumption to conducting the simulation runs that may reach hundreds to thousands. This paper adopted the design of an experiment plus a proxy optimization technique to solve this problem. Where the Latin Hypercube represents the DoE while the radial basis function network represents the artificial intelligence proxy model. The proxy model mimics the reservoir model to reduce the computation time and speed up the well-placement optimization process. The Latin Hypercube approach is used to generate data to train the proxy model to construct a reliable artificial intelligence model to predict the best wells locations. The results are very optimistic and encouraging to rely on using the art-of-state to construct a proxy model to conduct the infill wells drilling optimization. Where the increase in cumulative oil production for the optimized case is more than the un-optimized case by 6.45% and the decrease in the field cumulative water production for the optimized case is less than the un-optimized case by 16.11% from 2020 to 2040.


Articles in Press, Corrected Proof
Available Online from 03 March 2023
  • Receive Date: 07 November 2022
  • Revise Date: 14 February 2023
  • Accept Date: 02 March 2023
  • First Publish Date: 03 March 2023