Skip to main content

Production forecasting with uncertainty quantification

Objective



Objectives

It is becoming increasingly important to quantify accurately the economic risk associated with hydrocarbon development
projects. In particular, the marginal fields in the North Sea require optimal use of all available information to assess the associated economic risks. These data need to be combined in a formally unbiased way that honours the information content inherent in the data. Statistical theory provides the formal framework in which the 'Uncertainty' can be 'Managed'. PUNQ is aimed at further developing a methodology to quantify and reduce the uncertainty in production forecasts obtained from computer simulations of hydrocarbon reservoir models.

Technical Approach

Reservoir simulation models are generally conditioned directly to static information obtained from well logs, and indirectly to the reservoir's observed historical dynamic performance
such as well pressures, water-gas ratios, etc. PUNQ follows a Bayesian framework in which the unknown reservoir parameters, expected to have a first order effect on predicted
performance, have to be estimated using a priori information on these parameters, and using an inversion process routinely referred to as History Matching (HM). Geologists and
reservoir engineers integrate their respective views at an
early stage by jointly selecting the HM-parameters, and by
defining a priori probability density functions (PDF) of these parameters. A likelihood function, or HM-response surface, is then constructed, describing the probability density that a
given parameter combination can reproduce the observed
production behaviour.

The main technical objectives of the project are as follows: - development of techniques for mapping HM-response surfaces in the multi-parameter space;
- development of techniques for projecting HM-response
surfaces to forecasting uncertainty;
- evaluation of the value of underlying parameters for HM & quantification of forecasting uncertainty;
- efficient optimization of noisy objective functions;
- the development of efficient Fast Simulators.

Expected Achievements and Exploitation

The main challenges addressed in this project comprise
quantitative geological definition, flow simulation and
statistical formalism including optimization and search
techniques. The main deliverables are:
- an algorithm to compute the PDF of future production;
- method validation through a geologically (near)-realistic, heterogeneous synthetic case.

An objective of the PUNQ project is to pave the way for future exploitation of the method. Statistical theory and
optimization theory are becoming just adequate to allow the
development of formalisms capable of modelling the
uncertainties associated with hydrocarbon field development.

Funding Scheme

CSC - Cost-sharing contracts

Coordinator

Nederlandse Organisatie voor Toegepast Natuurwetenschappelijk Onderzoek
Address
97,Schoenmakerstraat
2600 JA Delft
Netherlands

Participants (9)

Amoco Exploration Company
United Kingdom
Address
Amoco House Westgate
W5 1XL London
ELF EXPLORATION UK PLC
United Kingdom
Address
Buckingham Gate 30
SW1E 6NN London
GEOLOGICAL SURVEY OF DENMARK AND GREENLAND
Denmark
Address
8,Oester Voldgade 10
2400 Koepenhagen
Imperial College of Science Technology and Medicine
United Kingdom
Address
Prince Consort Road
SW7 2BP London
Institut Français du Pétrole
France
Address
1-4 Avenue Du Bois Préau
92506 Rueil-malmaison
NORWEGIAN UNIVERSITY OF SCIENCE AND TECHNOLOGY
Norway
Address
Alfred Getz Vei 1
7491 Trondheim
Norwegian Computing Center
Norway
Address

0314 Oslo
Technische Universiteit Delft
Netherlands
Address
120,Mijnbouwstraat 120
2600 GA Delft
University of Liverpool
United Kingdom
Address

L69 3BX Liverpool