My research for the PhD period is about the analysis and application of TPWL (trajectory piecewise linearization) method in reservoir simulation. TPWL is one type of ROM (Reduced-Order Modeling) methods. It enables one to run many simulation scenarios with small perturbations VERY FAST.

One important issue is to characterize how far this perturbation can go to ensure the accuracy even applicability of the TPWL method. My current research is to delve into this issue and analyze the theory behind the engineering practice, hence find out better way to utilize the TPWL method.

This topic is interesting and important, also because it can shed light on the reservoir simulator architecture and fast simulation methods through many aspects. It involves previous work done on efficient nonlinear solver (e.g. Rami Younis, Xiaochen Wang, Boxiao Li), proper orthogonal decomposition, and linear solver. It can also be used in the context of reservoir history matching, production optimization (Marco Cardoso, Jingcong He), etc.


My research during the Master's period is an attempt to use Proper Orthogonal Decomposition (POD) based Reduced-Order Modeling (ROM) method as a linear-solver preconditioner for pressure systems in the reservoir simulator.

The aim of ROM method is to achieve faster performance in reservoir simulation. This is a favorable method particularly for optimization or history matching, where one needs to run a large number of similar models with similar control schemes.