Improved Integrated Asset Simulation Surface-Reservoir Using PID Control in Adaptive Time Stepping Conference Paper uri icon


  • Copyright 2018, Society of Petroleum Engineers. Modern, capital intensive upstream projects require integrated dynamic modeling to best project production and economics. Surface-subsurface coupling, or simply coupling, is utilized to model integrated production systems, i.e. model the reservoir, wellbore and surface facilities as one system. Three general methods of coupling exist: Explicit, partially implicit and fully implicit. The explicit method requires the least computational effort, however solutions are inaccurate and often unstable. Fully implicit coupling yields the most accurate and stable solutions, however, fully implicit systems are computationally expensive and are often impractical in operational settings due to high development costs. The partially implicit method offers the best compromise between accuracy and stability. Yet, partially implicit methods still encounter instability. The general objective of this work was to create an easily implementable partially implicit scheme that was accurate and stable. The authors hypothesized that adaptive time stepping could provide a stable solution as well as reduce computational requirements. In this work, a PID control loop is utilized to perform adaptive time stepping in coupled simulation. The objective of the PID control loop was to control coupling error via the manipulation of the time step size. A PID controller was adapted from prior literature which aimed to provide a stable and accurate solution method to numerically integrate ordinary differential equations. This controller was utilized in a commercial software package, and it overrode the commercial software's own time step selection algorithm. The commercial coupling software was utilized to act as a framework and baseline to compare with the novel PID time stepping algorithm. The only requirement of any proposed controller-software system is that the time step must be programmable at every iteration in the underlying coupling software. Simple algorithms may be used to transfer the updated time step from the controller to the coupled simulation, and ensure automatic communication between both systems. The coupled system included a 100mD, 10X10X12 grid block reservoir. Each grid block was 250X250X10 ft. The controller gains were varied to optimize the controller for the given simulation. Results showed that the novel coupling algorithm significantly reduced computational effort, while maintaining solution accuracy and stability. In one scenario, the PID controlled simulation was 297% faster than the commercial simulation, while incurring less than 0.5% error in cumulative production. Results showed that controller gain values were critical to producing fast and stable coupled solutions. An adaptive gain schedule yielded the fastest and most accurate solution, providing motivation to develop an algorithm to select controller gains at each time step. This work represents the first time that PID control has been used to perform adaptive time stepping in coupled simulation. This work shows that a novel time stepping scheme can be successfully incorporated to existing coupling software.

name of conference

  • SPE Europec featured at 80th EAGE Conference and Exhibition

published proceedings

  • SPE Europec featured at 80th EAGE Conference and Exhibition

author list (cited authors)

  • Redick, B., & Gildin, E.

publication date

  • January 1, 2018 11:11 AM