Computational Strategies for Multiperiod Mixed-Integer Nonlinear Programming Model for Optimal Carbon Capture and Storage

High Resolution Image Download MS PowerPoint Slide The paper describes an advanced optimization model developed and applied to multiperiod reservoir simulations for carbon capture and storage (CCS). The primary goal of this work is to scale up the model to longer time horizons while minimizing the computational time. The model determines the number of injection wells, buffer tanks and amount of CO 2 injected into the reservoirs to minimize overall investment and operational costs, while meeting the daily target injection rate of cryogenic CO 2 . A major improvement in the solution strategy is the use of sampling time indices approach, where there is a transition from a daily time step to a weekly/after-10-days analysis. This shift from a full-scale Mixed-Integer Nonlinear Programming (MINLP) Model to a Sampled MINLP Model leads to a very significant improvement in computational efficiency for longer time horizons. Moreover, the study combines a Greedy Heuristics Algorithm with the Sampling Time Indices approach, to scale the multiperiod problem from a time horizon of 1 to 30 years, with order of magnitude reductions in computational time. The model results have shown that the application of these computational methods can be extended to complex CCS configurations and multiperiod optimization problems.