When it comes to evaluating hydrocarbon reservoirs and predicting production performance, one thing is certain: Uncertainty.
This fact has long plagued geoscientists and engineers as they strive to quantify myriad uncertainties in the complex subsurface environs, such as porosity, permeability, structural surfaces and more.
“Evaluating uncertainty is becoming more and more important,” said Jan Inge Tollefsrud, technical manager responsible for uncertainty solutions at Roxar. “This is because new discoveries are getting smaller even though the number of discoveries is still high.”
To begin at the beginning, it’s imperative to understand the underlying factor that impacts your ability to quantify what the uncertainty is.
It’s not the reservoir; it’s you.
“Uncertainty is a function of your state of knowledge and not an intrinsic quantity of the physical world,” said Mike Christie, professor of reservoir simulation at Heriot-Watt University in Edinburgh. “It’s a function of your lack of knowledge of what’s going on.”
There is, in fact, a lot going on -- across the board.
“In the case of fluid flow alone, you’re looking at determining how much oil is there, the connectivity between wells,” Christie said, “and the conductance, which is related to porosity and permeability. Relative permeability and capillary pressure can be big uncertainties.
“Even measured data are not necessarily that certain,” he noted.
The problem is universal.
“Uncertainty exists across the workflow within all the disciplines,” said David Hardy, product manager at Roxar. “Despite the rapid uptake of 3-D model technology, uncertainty isn’t commonly considered part of the 3-D modeling process, even though everyone knows they should be doing something about it.
“Lack of time or available tools to do the job properly are blamed for the inattention.”
Bottoms-Up
A whole new attitude toward the problem is emerging today, as methodologies and tools are being developed to allow the complete uncertainty chain to be evaluated within the reservoir model.
“For the subsurface, this includes tools that work on the 3-D static and dynamic models geoscientists and engineers are building,” Hardy said. “The integrated approach also allows the uncertainty to be handled across all the disciplines together, which is essential.”
Industry interest in uncertainty modeling is evident at Heriot-Watt, where a number of major oil companies support an industry research group comprising two post-doctoral and four doctoral students looking at the mathematics and computational aspects of uncertainty quantification and also the geological aspects.
Regarding the geology, the idea is to be able to capture the uncertainties in the geology in the modeling process in a way that’s as recognizable to a geologist as to a mathematician.
Early approaches to handling uncertainty in 3-D tended to focus on the reservoir simulation, in large part because the relatively small simulation models afford an easier place to start. The newer trend is to work back toward the geological models, Hardy noted, which is where the problem needs to be addressed; after all, this is the source of the data input.
In fact, there currently are two principle approaches to modeling uncertainty: top-down and bottoms-up.
Basically, the top-down approach typically focuses on a simple reservoir model, which is not burdened with details of the geology; it begins coarse, and detail is added if and when needed.
Conversely, the bottoms-up approach to quantifying uncertainty utilizes a detailed model, which captures all the geological uncertainties.
Roxar’s uncertainty management solution can best be defined as classic bottoms-up, according to Hardy. The solution will be released early in the fall of this year as a module in the company’s integrated reservoir management product.
This tool spans an expanded and integrated reservoir characterization workflow, and the uncertainty model will allow users to evaluate uncertainty across the entire workflow, which includes:
- Structural framework modeling.
- Fault seal analysis.
- Geological property modeling (porosity, permeability, water saturation, etc.).
- Reservoir simulation.
Ease of use is a goal of the tool.
“It’s covering the whole workflow from static model to dynamic model, including flow simulation.
“What you don’t know is what you’re trying to capture,” Tollefsrud noted. “You have to make estimates and then induce some screening of elements in building the models and then find which are the most critical.
“We try to find out the most important elements and then we focus on those and try to reduce that uncertainty,” he said. “For example, if you find the velocity model in the overburden is the most critical element in defining the in-place volumes, you try to find ways to improve the velocity models.”
‘Transformational Technology’
An alternative approach to incorporate reservoir uncertainty in model construction and performance prediction is being used at BP.
It makes use of top-down reservoir modeling (TDRM) technology -- a proprietary technology developed at BP.
“BP has developed a pragmatic approach to thinking about uncertainty over the whole workflow, which includes the tools and the philosophy,” said Glyn Williams, technical manager in the development organization responsible for uncertainty solutions.
The philosophy is to initiate investigations with the simplest model and simulator suited to the business solution.
“BP uses the simplest appropriate model for the decision, and that’s the key thing,” Williams said. “It’s not about building a large complex model with a mountain of information.
“Our approach covers the aspects from thinking about the uncertainties and which of these are key,” he noted. “We design the appropriate model and calibrate the model in the sense of history matching to come up with alternative models that match the data. We use the alternative models to actually make decisions based on those uncertainties -- and it works.
“It’s a transformational technology, which has gone from an embryo -- the research idea -- to worldwide operations at BP,” Williams said. “It’s transformed how BP thinks about uncertainty,” he noted.
The company has successfully applied TDRM technology in hundreds of cases, according to Williams. These include myriad types of reservoirs in all stages of development.
Benefits achieved through TDRM application include reduced risk via better understanding of uncertainty and faster work cycle time. The estimated NPV (net present value) for projects has been elevated by as much as 20 percent, Williams noted.
Blurring the Boundaries
As uncertainty management tools and technology continue to evolve, the boundaries between top-down and bottoms-up applications appear destined to blur.
“Fundamentally, we’re all trying to get at the same problem,” Hardy said. “I think we’ll be meeting in the middle and may pass each other in different directions.
“At the moment Roxar is focusing on geological uncertainty and the parameterization of that and allowing you to build multiple models,” he said, “which we can then check against production history.”
The top-down uncertainty management technique tends to move between the model and the reservoir simulator, allowing some comparisons of simulated results versus historical production data. The information is fed back into the reservoir model in order to build it better and come closer to a solution.
Technology that focuses on improving results and accelerating reservoir simulation can be combined with the bottoms-up uncertainty management technique to make the most of both worlds.
For instance, automated history matching -- which is a specialty focus at U.K.-based Energy Scitech -- accelerates reservoir simulation and can be used to try to assess uncertainty at the simulation end, i.e., the coarse engineering end of the process. Hardy noted the two companies are working toward combining elements of Roxar’s uncertainty approach with Energy Scitech’s EnAble technology in a manner that will somewhat resemble the top-down uncertainty solution, at least in terms of the fundamentals of the approach.
“In effect, the net outcome is that we can create a shared earth model under uncertainty,” said Neil Dunlop, director of engineering business at Energy Scitech. “Components in the two technologies allow you to do a shared earth modeling workflow.
“For a long time, the industry has been seeking a way to generate reservoir simulation models, which are themselves consistent with all known geological information,” Dunlop said. “This workflow lets you do that.
“You can sensitize the geomodels,” he said, “and if you have production data from a producing field, you can use that data to generate understanding of the distribution of production behavior and calibrate it based on actual observations.”
In the case of a field under production, the users can take field information from that history period and project it into the future using the same geomodel they’ve matched it with. Combining the two constitutes a top-down uncertainty solution.
Mitigating the Risk
Dunlop believes that in a pure exploration situation, there’s only really a bottoms-up approach possible; no calibration or information is available.
“What we’re doing with the Roxar system is we’re adding new functionality not available before,” Dunlop said. “That’s to calibrate our understanding of the field in such a way that the reservoir simulation model and the geological model are coherent with each other -- people have long complained their models never are.”
Uncertainty evaluation, capture and integration is not a one-shot event.
Indeed, it must be conducted over the life of the reservoir, according to Emmanuel Gringarten, reservoir characterization engineer at Earth Decision Sciences.
Food for thought: Uncertainty quantification has no value in its own right, according to Hardy.
“It is improved decision-making that everyone is aiming for,” he noted. “Uncertainty management is all about making better, more informed decisions to help make plans to mitigate the risk.
“The better able you are to quantify the risk, the more you can improve the financial performance of the company.”