---

From model-driven optimization to better decisions – a practical shift for subsurface engineering.

Subsurface engineering has long relied on ensemble‑based modelling, data assimilation, and gradient‑based optimization to support decision making under uncertainty. This approach treats decisions as the optimization of a forecasted objective based on current information and a fixed set of controls. This model‑driven perspective has served us well in past scenarios. However, the landscape is shifting.

Many subsurface decisions involve discrete and sequential decisions, that are deeply connected to geological and operational uncertainties. In these cases, classical optimization can be short-sighted. It tends to focus on improving a current forecast without fully considering flexibility, the potential for learning, or the value of information that future observations might provide.

In our talk, we will argue that these aren’t limitations of existing methods but rather came from the way we frame decision making, primarily as a model‑based optimization problem. Our proposal is to move towards a decision‑driven analysis, based on the principles of Decision Quality and Structured Decision Making.

To support this change, we are introducing PRISMA, an open-source application designed to support structured decision making. We envision that PRISMA will complement ensemble modelling, data assimilation and robust optimization with a structured framework, offering transparency, reproducibility and space for continuous learning. We will share some pilot examples on the workflow and analysis to illustrate how this approach works in practise and can elevate decision quality.

Friday Informal Seminar Hour —

Postdoc-run seminar series within the Earth Resources Laboratory (ERL). Features talks by ERL members as well as special guests from academia and industry on earth science and energy transition topics including geothermal energy, carbon sequestration, geologic hydrogen, and critical minerals / in situ mining.

Contact: fish_seminar_organizers@mit.edu