Modern Advances in Reservoir Simulation; A Future Where Simulators Write Themselves and Never Fail

Friday, August 31, 2012 from 03:30 PM to 04:30 PM

Please join us at 3:30 p.m. Friday, August 31, 2012 in Keplinger Hall, room #M2, for the McDougall School of Petroleum Engineering's Graduate Seminar featuring Prof. Rami Younis, Assistant Professor.

Overview

To date, there is a clear trend in the way that we develop, maintain, and periodically, retire commercial-grade reservoir simulators. Every decade or so, a next-generation simulator is developed to encompass under one hood all of the problems of current interest. In front of the backdrops of an aggressively changing hardware architecture scene and exponentially growing complexity and nonlinearity, the business-as-usual model is failing us. Current undertakings to develop the next greatest predictive simulation tools are teaching us that we fundamentally need new ways of developing simulation software and treating nonlinearity.

The first part of this talk concerns the software implementation of the formulation and linearization of nonlinear simultaneous equations; the archetypical inflexible component of all large scale simulator software. The proposed solution is an Automated Simulator Generator System (ASGS). Dr. Younis will describe the ASGS and layout the development plan for the first-ever prototype. He will also present a concrete step towards the first ASGS; an algorithmic framework and library of data-types called the Automatically Differentiable Expression Templates Library (ADETL). The ADETL provides generic representations of variables and discretized expressions on a simulation grid, and the data-types implement algorithms employed behind the scenes to automatically compute the sparse analytical Jacobian matrix. A key technical challenge that is addressed by the ADETL is in enabling this level of abstraction and programming ease while making it easy to develop code that runs fast. Faster than any of several existing automatic differentiation packages, faster than any purely Object Oriented implementation, and at least as fast as hand-differentiated code that is delivered by a professional development team.

The second part of this talk introduces an evolving portfolio of nonlinear solution methods and analysis results that exploit an understanding of the nature of the physics of travelling waves to deal with the current and upcoming challenges in nonlinear stiffness. While the Fully Implicit Method offers unconditional stability of the discrete approximations, its stability comes at the expense of transferring the inherent physical stiffness onto the coupled nonlinear residual equations that are solved at each time-step. Current nonlinear solution methods cannot guarantee convergence, nor provide estimates of the relation between convergence rate and time-step size, nor quantitatively anticipate their computational efficiency. We establish an alternate class of nonlinear solution methods that converge all the time with a well-characterized, ideal computational efficiency.

Contact:
Prof. Rami Younis
rami-younis@utulsa.edu
(918) 631-2426