Method for quantifying uncertainties related to continuous and discrete parameters descriptive of a medium by construction of experiment designs and statistical analysis

• US 20030220775A1
• Filed: 04/02/2003
• Published: 11/27/2003
• Est. Priority Date: 04/02/2002
• Status: Active Grant

• Current Assignee: Institut Francais Du Petrole (IFP Energies Nouvelles)
• Original Assignee: Institut Francais Du Petrole (IFP Energies Nouvelles)
• Inventors: Zabalza-Mezghani, Isabelle, Manceau, Emmanuel, Jourdan, Astrid
• Granted Patent: US 7,590,516 B2
• Time in Patent Office: Days
• US Class Current: 703/2
• CPC Class Codes: E21B 2200/22 Fuzzy logic, artificial intelligence, neural networks or the like
• CPC Class Codes: E21B 44/00 Automatic control systems specially adapted for drilling operations, i.e. self-operating systems which function to carry out or modify a drilling operation without intervention of a human operator, e.g. computer-controlled drilling systems; Systems specially adapted for monitoring a plurality of drilling variables or conditions
• CPC Class Codes: G01V 1/30 Analysis G01V1/50 takes precedence
• CPC Class Codes: G06F 17/18 for evaluating statistical data , e.g. average values, frequency distributions, probability functions, regression analysis forecasting specially adapted for a specific administrative, business or logistic context G06Q10/04

Abstract

Method for quantifying uncertainties related to continuous and discrete (qualitative) parameters descriptive of a medium such as an underground zone and/or for managing the selection of a scenario from a series of possible scenarios relative to this medium, by construction of experiment designs and results analysis suited to the experiment designs constructed.

The method essentially comprises constructing the factorial part of the experiment designs by folding a regular fraction for the quantitative factors and assignment of at least one modality of a qualitative factor to each one of the blocks formed by folding, determining the axial part of the qualitative factors according to a D-optimality criterion (preferably on a limited number of scenarios) and analysing the results by combining a sensitivity analysis and a risk analysis involving marginal models (models adjusted to each scenario) and a global model (model adjusted to all of the scenarios). The method allows for example, in an otherwise uncertain context, to compare different production scenarios (reservoir crossed by open or closed faults, enhanced recovery by water injection or WAG type alternating injection, etc.) in order to better understand the role of non-controllable discrete parameters (status of a fault, etc.) and/or to select the scenario which optimizes production in the case of controllable discrete parameters (well addition, completion levels, water injection or WAG, etc.).

Application: reservoir exploration or engineering for example.

Claims

1. ) A method for quantifing uncertainties related to continuous and discrete parameters descriptive of an underground zone and/or for managing the selection of a scenario from a series of scenarios, by construction of experiment designs comprising a factorial part, a central part and an axial part, which take account of quantitative and qualitative factors, and a results analysis suited to the experiment designs constructed, characterized in that:

the factorial part of the experiment designs is constructed by folding a factorial design fraction for the quantitative factors and assignment of at least one modality of a qualitative factor to each block formed by folding, and

the results are analysed by combining a sensitivity analysis and a risk analysis involving marginal models and a global model.

2. ) A method as claimed in claim 1, characterized in that the axial part of the qualitative factors is determined according to a D-optimality criterion.

3. ) A method as claimed in claim 1 or 2, characterized in that the sensitivity analysis is carried out by means of marginal models to detect the terms that influence each scenario and by means of the global model to detect the terms that globally influence all of the scenarios.

4. ) A method as claimed in any one of claims 1 to 3, characterized in that the risk is analysed by localized prediction of a response in form of a prediction interval for a fixed set of values of said parameters, by means of the global model.

5. ) A method as claimed in any one of claims 1 to 3, characterized in that the risk is analysed by predicting responses from a large number of sets of values of said parameters, randomly selected in the variation range thereof.

6. ) A method as claimed in claim 5, characterized in that the global model is used when the or each discrete parameter is not controllable.

7. ) A method as claimed in claim 5, characterized in that the marginal models are used for risk analysis when each discrete parameter is controllable and when the global model has not detected any very influential quantitative-quantitative interaction that cannot be detected by the marginal models.

8. ) A method as claimed in claim 5, characterized in that the global model is used for risk analysis when each discrete parameter is controllable and when the global model has detected a very influential quantitative-quantitative interaction that cannot be detected by the marginal models.

9. ) A method as claimed in claim 5, characterized in that the global model and the marginal models are used to determine the respective influence of the aliased quantitative-quantitative and quantitative-qualitative interactions.

10. ) Application of the method as claimed in any one of the previous claims to reservoir exploration or engineering scenarios.