The big loop work flow concept adds tools that lead to more accurate reservoir simulations and better reservoir management. (All graphics courtesy of Roxar)

While it is the new oil and gas developments in the early production phase that tend to get most of the attention, today more than 70% of the world’s production comes from fields that are more than 30 years old.

Maturing reservoirs come with a unique set of reservoir management challenges, from increased water cuts and gas-oil ratios through to aging technologies and environmental, health and safety implications.

Another major reservoir management challenge is that maturing reservoirs often require much more detailed forecasts of production behavior, even though production volumes and revenues are lower.

With the importance of understanding and forecasting the production behavior of the reservoir increasing as the oil and gas becomes harder to extract, production costs rise. This, together with the increased volumes of data from maturing reservoirs being harder to assimilate and analyze manually, reinforces the need for the operator to look to new means of predicting production behavior.

History matching

The most effective solution for predicting production behavior is based on using a reservoir simulator tuned to production history by the process called history matching. Yet is this solution rising to the challenges posed by maturing reservoirs?

History matching, the act of adjusting a reservoir model until it closely reproduces its past behavior, is probably the number one topic of interest in the reservoir simulation community today. It is critical for monitoring displacement processes, constructing good reservoir models, predicting future performance and estimating reservoir uncertainty.

History matching also plays a key role in developing an integrated approach to reservoir management because it allows the static geological model to be rationalized with production data.

However, too often manual history matching tends to be characterized by inaccuracies that cause losses in time, money and productivity as well as the time it takes to explore all scenarios.

Furthermore, aligned to this is the dependency of history matching on the quality of the model and accompanying production data as well as the difficulty in selecting important conditioning parameters. With such a large number of unknown parameters to consider, traditional manual history matching remains very much a work in progress.

Reservoir simulation

Too often in the past reservoir simulation has been dominated by small, highly specialized teams of engineers running complex tools that take months to understand and years to master.

Simulation software packages have also been characterized by inefficient, manual workflows, incompatibility with other data sets and third-party simulation engines, and, like history matching, long amounts of time required to explore all scenarios.
At a time of skills shortages and the need to get new employees productive as quickly as possible, such time-consuming processes are simply not an option today.

Developments

There are technological developments in both history matching and reservoir simulation, however, that are helping address today’s reservoir management challenges. The rise of computer-assisted history matching is an example of this.

Computer-assisted history matching allows the engineer to focus on developing an understanding of reservoir mechanisms and their relative impact on production behavior. Through such history matching tools, match modifiers are updated intelligently, there is automatic parameter sensitivity analysis and runs are even submitted when engineers are away from their desks.

Roxar’s history matching and uncertainty estimation software product, EnABLE, is an example of the embracing of computing and stochastic optimization technology in history matching. The computer assistance provided by the software makes it possible to consider more information when developing a history match or sensitizing an appraisal analysis.

With manual history matching it is just too difficult and so time consuming as to make impossible the evaluation of all the aspects of the reservoir description that could have an effect on behavior.

However, today reservoir engineers with access to such computer-assisted tools can evaluate large numbers of modifiers in a full physics reservoir simulator with fewer runs. Because the software interacts with the engineer, engineering judgment can still be applied to effectively interpret and reconcile information and alleviate any concerns of automation.

In summary, the software takes information from the geological descriptions of the reservoir and the engineer and then uses powerful statistical techniques to determine multiple matches of the reservoir to production history and to model the uncertainty.

These results are then used with the simulator to predict how a field will perform and give measures of the uncertainty about these predictions. In this way, it brings geological modeling and simulation closer together and provides valuable information on the economics of the reservoir.

There are also positive changes in the development of reservoir simulation tools.

The rise of desktop parallel processing via the new 64-bit multicore chips, and increasingly affordable clusters, means that multimillion cell reservoir simulation models are now increasingly common.

Algorithmic developments such as single-grid dual-porosity modeling and software developments like the integration of simulation programs with other tools such as surface network modeling and assisted history matching are all leading to higher resolution and more accurate, finer-scale simulation of maturing oil and gas reservoirs.

Probably the single biggest development in reservoir simulation tools, however, is the growing accessibility and adoption of reservoir simulation programs on the less “expert” engineer’s desktop.

Reservoir simulation today and its increased emphasis on cost; ease of use; ongoing support; and an accessible, streamlined workflow; is starting to involve the entire subsurface community rather than a few highly specialist reservoir engineers.

Take the software company’s reservoir simulation product Tempest, for example. With a graphical interface and four integrated software models covering the different stages of reservoir simulation including data preparation, fluid flow simulation and economic evaluation, the majority of reservoir engineers can engage in the simulation process in confidence without the need for prolonged training.

Closing the loop

Accurate predictions of field performances and better targeted capital expenditure are not, however, simply down to the development of history matching and simulation software products as described above.

They are down to the workflow and an integration of history matching and simulation with geologically accurate models of the reservoir.

Traditionally, the modeling workflow from geologic model to the history-matched simulation model has often been a disjointed one with modifications made to the simulation model to achieve a history match not relating back to the original geologic model, and the geologic model is seldom updated to reflect these modifications.?

This traditional “little loop” workflow is often time-consuming and does not always lead to a geologically sound simulation model, reflecting uncertainties within the reservoir.

Our goal is to develop a “big loop” workflow of reservoir management that carries uncertainties and details
in the geologic model through to simulation. Using assisted history-matching technology in conjunction with geologic modeling and simulation software, key geologic modeling uncertainty parameters can be modified in such a way as to ensure that a history match is achieved by the models and the uncertainties in forecasts identified.?

Perhaps the most important need driving the big-loop development is to ensure that we provide all the information that good uncertainty estimation tools require; a good geologic model may provide multiple matches to production data, but are there other geologic models that also provide at least equally good history matches?

If so, they must be involved in the history matching in order to be able to estimate uncertainty. With efficient assisted history matching linked to geomodeling tools, we have the capability to test such possibilities within acceptable time and resource limits.

Changes to the geologic model that are necessary to obtain the history match will therefore be consistent with the underlying geological interpretation and a better understanding of uncertainty will be provided. In addition, although the process is computer-assisted, the variation of the parameters is defined by the asset team, which can decide which parameters are allowed to change, and what the limits of the variation are.

One of the areas that made consistency with geological interpretation difficult in the past was when modifications or updates to the structural framework of the geological model were required to achieve an improved history match. However, with a structural modeling toolkit, this has become a much easier, faster and more reliable process and should largely remove this barrier in adopting and working with big loop-style workflows.

As stated above, the big loop workflow allows the investigation of more models and more uncertainties, which will provide an improved quantification of uncertainty in reservoir performance predictions and improved reservoir management decisions.?

By integrating history matching, reservoir simulation tools and modeling applications we can provide the first commercial solution where numerous geological scenarios can be examined and history matched to create simulation models that are fully consistent with their underlying geological interpretation.

And at a time when the industry is looking for fast, easy-to-use, powerful reservoir management solutions to
optimize production from increasingly marginal assets and to make better decisions over the allocation of capital and resources, the timing could not be better.

Reservoir simulation and history matching are finally rising to the challenge!