Optimizing exploration decisions under geologic uncertainty in basin and petroleum system modeling
Abstract/Contents
- Abstract
- Basin and Petroleum System Modeling (BPSM) is coupled-physics approach that tracks over the course of basin history, the evolution of basin geometry, compaction, pressure, fluid flow, temperature, and chemical transformation of organic matter to quantitatively predict petroleum generation, migration and accumulation. The measured data from petroleum wells, conversely, can help us improve our knowledge of the basin's geologic history. For basin modeling, the initial model building requires parameters that are derived from our geologic knowledge of various aspects of the basin history through time (typically tens to hundreds of millions of years), like stratigraphy, geochemistry, timing of tectonic events, and boundary conditions like heat flow. But because of the spatially and temporally changing depositional environments in a basin, it is very challenging to accurately know the large number of input parameters required to represent the basin history. In addition, current workflows of constraining the inputs to measured data or evidence often do not account for the various non-unique possibilities that can create the outcome that is the present. To address this challenge, we demonstrate the use of data from drilled wells and basin models in Bayesian networks to create a workflow that provides a quantitative way to: 1) Vary model parameters: consider all hypothesis without biasing to one, 2) Reduce uncertainty by calibrating the model to measured evidence without repeated manual adjustments 3) Update the understanding of model parameters when new data becomes available, without re-running computationally heavy coupled-physics simulations, by using Bayesian Networks and 4) Create traceable workflow with an integrated economic analysis to make optimal decisions under the reduced, but still present uncertainty using Influence Diagrams. An example of prolific petroleum producing Jeanne d'Arc basin, offshore Newfoundland, Canada, is used to illustrate how the workflow facilitates constraining the source rock quality, thermal history, and migration pathways. The thesis is comprised of three main chapters. They are written in journal format, each designed to be a standalone chapter: Chapter 1 presents a comprehensive basin study of the Jeanne d'Arc basin. This work examines the past five decades of research of the Mesozoic -- Cenozoic (250 million years ago to present) evolution of the basin. We closely examine the unknowns and uncertainties, and where some studies differ in their findings. We create a 3 -- dimensional numerical basin model spanning an area of about 3200 km2 and use the framework to incorporate large regional fault trends, spatial variation in the quality of organic matter, and to test the conceptual models of elevated basal heat flows associated with the rifting of North America from Africa, Iberia, and Greenland. The model can also help us understand the evolution of neighboring basins: Orphan and Flemish pass, which have a large resource potential. Chapter 2 presents the novel use of Bayesian Network approach to quantify the multi-dimensional uncertainty created from non-linear interactions of basin parameters and insufficient constraints. We show how the Bayes Net structure incorporates expert knowledge about cause-effect relationships like Total Organic Carbon (TOC) and quantity of hydrocarbons produced, as well as the conditional independence of temperature to the TOC. We elucidate with an example why this network representation can summarize the joint probabilities in a compact form. We then illustrate how the relationship between parameters is learnt from data produced by different realizations of the basin model, and how uncertainty in the input parameters is reduced by conditioning to measured evidence. With the 120 basin models created with varying input parameters, we show how this method helps quantitatively reduce uncertainty in both our understanding of geologic history and our predictions of drilled hydrocarbon fluid quantities. Sensitivity analysis shows that hydrocarbon accumulation is more sensitive to fault sealing properties than the basal heat flow in the range of present uncertainty. Our analysis finds that the time-varying permeability of faults largely impacts the leakage and filling of deep and shallow reservoirs, and hence their accumulation volumes. Chapter 3 illustrates a structured decision-making process that is informed by a quantitative evaluation of risks and returns from exploration decisions using influence diagrams. Once we learn the probabilities of different predictions of accumulation volumes from the methods developed in Chapter 2, a question arises: how do we use these probabilities to quantitatively inform decisions and actions? We compare influence diagrams to the more conventional decision trees and then use data from different times in the exploration history of the Jeanne d'Arc to demonstrate the use of influence diagrams to calculate the value of information and predict optimal survey, drilling, and production decisions. Finally, we argue that the graphical formulation is an excellent communication tool that can incorporate quantitative uncertainties, expert knowledge, and decision maker preferences for different types of decision scenarios. Our illustration with real data paves the path for incorporating this workflow in large organizational settings.
Description
| Type of resource | text |
|---|---|
| Form | electronic resource; remote; computer; online resource |
| Extent | 1 online resource. |
| Place | California |
| Place | [Stanford, California] |
| Publisher | [Stanford University] |
| Copyright date | 2021; ©2021 |
| Publication date | 2020; 2020 |
| Issuance | monographic |
| Language | English |
Creators/Contributors
| Author | Chheda, Tanvi Dhiren |
|---|---|
| Degree supervisor | Graham, S. A. (Stephan Alan), 1950- |
| Thesis advisor | Graham, S. A. (Stephan Alan), 1950- |
| Thesis advisor | Mukerji, Tapan, 1965- |
| Thesis advisor | Scheirer, Allegra Hosford |
| Thesis advisor | Shachter, Ross D |
| Degree committee member | Mukerji, Tapan, 1965- |
| Degree committee member | Scheirer, Allegra Hosford |
| Degree committee member | Shachter, Ross D |
| Associated with | Stanford University, Department of Geology |
Subjects
| Genre | Theses |
|---|---|
| Genre | Text |
Bibliographic information
| Statement of responsibility | Tanvi D. Chheda. |
|---|---|
| Note | Submitted to the Department of Geology. |
| Thesis | Thesis Ph.D. Stanford University 2021. |
| Location | https://purl.stanford.edu/kg930wk8262 |
Access conditions
- Copyright
- © 2021 by Tanvi Dhiren Chheda
- License
- This work is licensed under a Creative Commons Attribution 3.0 Unported license (CC BY).
Also listed in
Loading usage metrics...