WP 4: From physics- to data-driven models

1. Introduction

The goal of this work package is to develop improved predictive reservoir models that combine and balance physics-based and data-driven models. Our hypothesis is that if properly calibrated with available data, models with less degree of physical complexity can often replace the full physics models partly or completely and thus significant speed up the simulation time in both the history-matching and the optimization loop allowing for more frequent model update, and better representation of uncertainly. 

Fast and reliable workflow for continues optimization

2. Research goals

Key research questions are how to determine the appropriate level of complexity needed for the model to keep its predictable power while still giving significant speed up, and how to couple models with different level of complexity consistently and efficiently.

3. Delivery

The work package will deliver a novel workflow that integrate models of varying complexity into a holistic simulation framework.

WP4 lead

Tor Harald Sandve
Senior Researcher
Energy & Technology Department, NORCE


Task 4.1: Data-driven proxy models for fast and accurate optimisation

Task 4.2: Multi-physics models for coupling near well regions and the reservoir

WP4 staff

Ove Sævareid
Birane Kane
Håkon Hægland

Eirik Keilegavlen
Jakub W. Both (WP deputy)