Wednesday 12 December 2018, 13.00 – 16.00, The Technical University of Denmark, Building 303A, room 49

**Principal supervisor**:
Associate Professor John Bagterp Jørgensen, DTU Compute.

**Co-Supervisors**:
Associate Professor Niels Kjølstad Poulsen, DTU Compute
and Freelance computational scientist Andrea Capolei.

**Examiners**:
Associate Professor Mirza Karamehmedovic, DTU Compute.

Associate Professor John Hedengren, Brigham Young University, United States of America.

Associate Professor Jiri Mikyška, Czech Technical University in Prague, Czech Republic.

Moderator: Assistant Professor Dimitri Boiroux, DTU Compute

**Summary:**

Oil remains the world’s leading fuel, and it is expected to account for a significant part of the
world’s energy consumption for several decades. However, current recovery techniques do
not recover all of the oil that is present in the oil reservoirs.
Furthermore, the production of oil
from a given reservoir may be uneconomical depending on the oil price
which is very volatile.

In this PhD project, we consider nonlinear model predictive control (NMPC) for improving the
economics of oil recovery processes. The objective of NMPC for oil reservoirs management is
to compute a field-
wide
closed-loop feedback control strategy which optimizes a long-term
financial measure of the recovery process, e.g. the total amount of recovered oil or the net
present value over the life-time of the reservoir.
Whenever new measurements become available, the NMPC algorithm uses 1) state
estimation to estimate the state of the reservoir (as well as parameters in the model) and 2)
dynamic optimization to compute a new updated field-wide production strategy. Reservoir flow
models are used in both the state estimation and the dynamic optimization. Accurate reservoir
flow models are therefore key to the effectiveness of the NMPC algorithm.

In this work, we present thermodynamically rigorous models of thermal (varying temperature)
and isothermal (constant temperature) compositional reservoir flow processes. Models of such
processes are based on two main principles: 1) conservation of mass and energy, and 2)
phase equilibrium. The conservation of energy is related to the first law of thermodynamics,
and phase equilibrium is related to the second law of thermodynamics. The phase equilibrium
problems that are relevant to the thermal and the isothermal models are the UV flash and the
VT flash, respectively. We formulate these phase equilibrium problems as equality cons
trained
optimization problems and the phase equilibrium conditions as the first order optimality
conditions. Furthermore, we demonstrate that the thermal and the isothermal models are in a
semi-explicit differential-algebraic form, and we formulate algorithms for state estimation,
dynamic optimization, and NMPC of systems in this form.
These algorithms are relevant to other phase equilibrium processes as well because it is
natural to model such processes using differential-algebraic equations in the same semi-explicit form.

A copy of the PhD thesis is available for reading at the department

All are welcome

Wed 12 Dec 18

13:00 - 16:00

13:00 - 16:00

DTU Compute