Balancing short and long term production strategies

Balancing short and long term production strategies

This is a page showing ongoing activity by the IO Center. It involves highlighted research activities which are currently being carried out, description of goals, activities and expected results

5
Average: 5 (1 vote)

1 users have rated this content. We would love to have your vote as well. Log in and rate!



Reservoir engineers make decisions with time horizons of months to decades, while production engineers are thinking in days and weeks. Integrating long- and short-time horizons is an important but complex task. In the IO-center, we have had good progress in developing decision support tools using model based optimization methods for both the short and long horizons. However, due to large uncertainty both in geological and economic models, a seemingly optimal long term plan will hardly have a short term impact if it severely contradicts with short term goals. Balancing short and long term production strategies, on may very well be able to find strategies predicting adequate NPVs both on a short- and long-term horizon.
Content

Reservoir engineers make decisions with time horizons of months to decades, while production engineers are thinking in days and weeks. Integrating long- and short-time horizons is an important but complex task. In the IO-center, we have had good progress in developing decision support tools using model based optimization methods for both the short and long horizons. However, due to large uncertainty both in geological and economic models, a seemingly optimal long term plan will hardly have a short term impact if it severely contradicts with short term goals. At the same time, always thinking in short term, may very well lead to long-term sub-optimal management strategies, like leaving behind unswept pockets of oil, early water breakthrough or high GOR.  

Going to the extreme, one could try to find a good balance between short term production (days to weeks) and the life-cycle NPV through using optimization on top a coupled reservoir and network simulator, however due to the vast difference in time-scale we do not see this as a viable or fruitful path. In stead, we focus on embedding simplified reservoir models in short term planning to account for medium term effects, but also on balancing medium and long term goals (on time-scales it makes sense to use full reservoir simulations). For an illustration of the latter, we consider a simple example:
 

Example

Consider simplistic (part of a) homogeneous oil-water reservoir with one horizontal producer and one horizontal injector as shown in the cross-section in Figure 1. The task is to optimize the net-present-value (NPV) over the next six years by varying the water injection rate and keeping the producer BHP fixed. For the sake of example, we calculate the net cash flow over a time period dt as

 

 

R = dt (100 qo - 10 qpw - 10 qiw)

 

where qo is the oil production rate, qpw is the water production rate and 10 qiw is the water injection rate. We use an NPV which is the accumulated net cash flows with a yearly discount rate of 10%. We divide the six years into 12 half-year control steps. With this setup, one can easily find the optimal injection strategy by using an optimization algorithm on top of the reservoir simulator, but here we want to illustrate how the injection strategies will vary depending on the length of the optimization horizons which we vary from 1 (½ year)  to 12 (6 years) control steps. This means that for every control step, we optimize the NPV for the next fixed number of control-steps in a receding horizon fashion. The resulting optimal strategies for some of the cases are plotted in figure 2. When maximizing NPV only half a year ahead, the optimal strategy is to not inject water at all. Accordingly, the cost of injecting water is not compensated for by increased oil production during the subsequent 6 months. For this example, this is not only due to compressibility, but also because of water coning effects. When the horizon is increased to two years, the long term NPV is very close to that obtained by optimizing over the whole 6 years.

If we in addition assume that the cost associated with installing the injector for this part of the reservoir is 50M $, we would require a horizon of at least 3 years in order for the installation to be profitable (see Figure 3).  In addition, we have plotted a reactive strategy, which is simply to install the injector when the production rate drops below a prescribed threshold. Even though the cost of installing the injector 3 years into the future is considerably discounted, this strategy does not pay off in the 6-year NPV.

This example illustrates that the length of the planning horizon can be important for the long term performance, but it does not give a definite answer what is the best option. It merely suggests that if one wants to maximize profit over the next six years and there is sufficient confidence in the models, the injector should be installed right away. On the other hand, if there is lack of confidence, and/or one has a much shorter profit horizon, one should probably wait.

Research and case study

In real life, the situation is of course much more complex than in the above example. As a consequence, it is much more difficult to detect which effects (if any) cause mismatch between short- and long term goals. But, the complexity is also good news since it introduces many degrees of freedom; research point to that the optimal long term strategy is not likely to be a distinct peak on the response surface, but rather lies on ridges or plateaus. This means that one very well may be able to find strategies predicting adequate NPVs both on a short- and long-term horizon. One example of this is shown in Figure 4, where the long term NPV was used as a non-linear constraint in optimization of the short term NPV (E. Suwartadi, S. Krogstad, and B. Foss, 2011). To visualize the trade-off between short- and long term strategies, a methodology for computing a pareto-like plot was developed (A. Hasan, B. Foss, S. Krogstad, V. Gunnerud, and A. Teixeira, 2013), see Figure 5. The method was tested on a simplified reservoir model of the South wing of the Voador field operated by Petrobras. A gradient-based optimization on top of an adjoint simulator in MRST was used for the numerical experiments. Recently, the adjoint capabilities in MRST was extended to 3-phase black oil models, and the case study with the Voador field, is now continuing with the un-simplified model. MRST has also recently been linked to the ResOpt framework for embedding simplified reservoir models in short term planning to account for longer term effects.

 

 

Other key information

284 results
Below, you will find related content (content tagged with same topic(s) as this ongoing activity)
Content type: Report

The Integrated Operations Maintenance and Modification Planner (IO-MAP) The first usability evaluation – study and first findings

The Integrated Operations Maintenance and Modification Planner (IO-MAP) is a software test bed, which is being designed in an iterative process.

5
Content type: Ongoing activity

A benchmark for integration of reservoir engineering and geophysics

This activity facilitates the collaboration between reservoir engineers and geophysicists

0
Content type: Publication

A Derivative-Free Approach for the Estimation of Porosity and Permeability Using Time-Lapse Seismic and Production Data

This is a scientific publication written in collaboration with the IO Center.

0
Content type: Publication

A Lagrangian-Barrier Function for Adjoint State Constraints Optimization of Oil Reservoirs Water Flooding

This is a scientific publication written in collaboration with the IO Center.

0
Content type: Report

A Method for Function Analysis and Allocation in IO-driven MTO Systems: Challenges and Prospects

Describes use of the IO MTO method in functional analysis and allocation

0
Content type: Report

A new telemedicine practice on-the-go

Development and demonstrations of a telemedicine prototype that enables the key features of the “Virtual Examination Room - On the go”.

0
Content type: Publication

A Robust Scheme for Spatio-Temporal Inverse Modeling of Oil Reservoirs

This is a scientific publication written in collaboration with the IO Center.

0
Content type: Publication

A Sparse Basis POD for Model Reduction of Multiphase Compressible Flow

This is a scientific publication written in collaboration with the IO Center

0
Content type: Publication

A structured approach to improved condition monitoring

A paper describing a systematic approach to select and implement appropriate condition monitoring for systems, structures and components.

0
Content type: Presentation

Acceptance criteria for dynamic risk analysis

Presentations at IO Center Work Shop Sept 2013

0

Pages