Reservoir simulation models in production forecasting

The purpose of reservoir simulation is to predict field performance and ultimate recovery for various field development scenarios to evaluate the effects on recovery of different operational conditions and compare economics of different recovery methods.

The Simulation method is a spatial, 3D approach. It is more robust than Analogous, Empirical and Decline or Type Curve methods, and it provides the dimensionality that Material Balance methods cannot. It is the only forecasting method that incorporates areal and vertical distributions of non-uniform rock properties, coupled rock-fluid properties (relative permeabilities) and resulting rates, pressures and saturations. It allows assessments of multiple development alternatives, which is the level of rigor that is required for most projects.

The following topics will be focused in this chapter:

• Type of reservoir simulation models
• Main principles for building reservoir simulation models.

Types of reservoir simulation models

Reservoir simulation models are often referred to by types of models:

• Black-oil
• Compositional
• Thermal
• Single-porosity or dual-porosity (for fractured reservoirs)

The application of compositional models is discussed briefly in reservoir rock and fluid properties.

Reservoir simulation model building workflow

Reservoir simulation model is one of the major “building blocks” in an integrated reservoir modelling process, as illustrated in Fig 1.

INSERT Figure 1 - Integrated Reservoir Modelling “building-blocks” (Pending permission approval)

Reservoir simulation grid construction

The main objective in building a reservoir simulation grid is to construct a reservoir simulation model that resolves changes in reservoir flow properties, pressure, and fluid saturation with enough precision to provide accurate performance predictions.

The most important geological features are faults, variation in reservoir parameters, and the layering style, or stratigraphy.

Well completions must also be resolved in the reservoir simulation grid. These features are preserved by carefully defining the areal grid and the layering by use of processes described below.

Input data from geological modelling.

Input data from geological modelling includes:

• Structural modelling (geological grid, faults and fault properties, reservoir zonation, vertical communication, fluid contacts)
• Property modelling (porosity, permeability, net-to-gross ratio, initial water saturation, end-point fluid saturations).

Areal grid and layering

To represent spatial variation in the reservoir properties, gridblocks must be smaller than the features being represented. For example, representing sinuous channels requires that the gridblock dimensions be at most one-third of channel width. In the absence of specific geologic objects, a rule of thumb is that the reservoir simulation cell dimension should be approximately twice that of the geological model cell in X- and Y-directions in areas where hydrocarbons and wells are presented.

Variation of saturation and pressure around the wells depend strongly on reservoir description, fluid properties and flow rates. Therefore, the best way to decide on grid resolution near the wells is to conduct a gridding study in a finer scale element model representing the area around a well.

Another issue is the proper representation of fluid contacts. A thin oil rim cannot be resolved in the layer thickness or areal grid-block size if it is larger than the dimensions of the oil rim.

Layering style in the reservoir simulation model usually is set to be the same as that in the geological model. The objective in layer selection is to determine the number of layers to preserve the heterogeneity seen in the geological model. Choice of layering is also affected by the displacement processes. It is important that the layers in the model be thin enough to resolve the location of well completions and vertical movement of fluid to those completions.

Vertical communication

Vertical communication between reservoir simulation grid layers is important for describing correct flow displacement in reservoirs, both vertically and areally. This input should be based on the structural and petrophysical analysis and should provide vertical transmissibility multipliers to be used in the reservoir simulation model.

Faults and fault properties

Most reservoirs are faulted and faults often must be represented in reservoir simulation models. Sometimes the model can be simplified by treating faults as vertical. This decision depends on the detailed fault geometry, and how it intersects with the fluid contacts, the wells, and the property variation on either side of each fault.

If the sand or shale thicknesses are the same as the fault throw, it will be important to represent the juxtaposition of reservoir properties across the faults, and slant faults may be necessary. Similarly, if the fluid contacts impinge on the faults or if wells are near faults, it will be important to represent fault geometry more accurately.

Aquifer properties

As much of the rock containing the aquifer as practical should be included in the reservoir simulation grid with the corresponding aquifer properties assigned: porosity, permeability and net-to-gross. This is needed to establish an appropriate dynamic picture in the reservoirs.

However, if there is no data available to describe an aquifer structure “sufficiently”, it can be modeled by the aquifer modelling facilities available in commercial software (“analytical aquifers”).

Upscaling static properties

Reservoir properties often need to be up-scaled if the geological static model is built on a finer scale in terms of cell size than is intended for the simulation model. This property upscaling should be done after the 3D-reservoir simulation grid has been constructed.

Porosity and Net-to-Gross

The up-scaled porosity and NTG should preserve the pore volume of the underlying geological model. In this case the bulk-volume weighted averages give the correct porosity.

Saturations

The up-scaled saturations should preserve the volume of each phase in the volume associated with each grid-block. In this case pore-volume-weighted averaging will give the correct result.

Permeability

As a flow property, permeability should be defined as a diagonal tensor and up-scaled by using flow based up-scaling techniques. Tensor upscaling can be time consuming for large / complicated models, and scalar upscaling is often employed. It can be verified if arithmetic-harmonic (horizontal and vertical permeability) upscaling yields a reservoir simulation model with reasonable behavior and similar to that with tensor upscaling, which in some cases tends to underestimate vertical communication and is therefore trickier to use.

Reservoir rock and fluid properties

Rock compressibility

Rock compressibility is one of the most important input parameters for initialization of any reservoir simulation study. This parameter should be based on core analysis and rock mechanics studies.

It is recommended not to change this parameter during the history matching process, if possible. If modified, it should be checked with the team members representing the rock mechanics discipline. This needs to be done to ensure that final rock compressibility applied in the model has a physical meaning.

Fluid PVT data

The two most common types of reservoir fluid models are black-oil models and compositional models. The black-oil models are based on the assumptions that the saturated phase properties of two hydrocarbon phases (oil and gas) depend on pressure only.

Compositional models also assume two hydrocarbon phases, but they allow the definition of many hydrocarbon components. The time cost of running a compositional simulator increases dramatically with an increase in the number of components modeled, but the additional components make it possible to more accurately model complex fluid phase behavior. If compositional model results are to be used in a process engineering model, it is often necessary to compromise on the number of components to be used for each application.

The typical fluid PVT properties used in reservoir simulation study, are: formation volume factor, viscosity, solution gas-oil-ratio and water compressibility.

Relative permeability and capillary pressure

Relative permeability curves represent flow mechanisms, such as drainage or imbibition processes, or fluid wettability. Relative permeability data should be obtained by experiments that best model the type of displacement processes in the reservoirs. For example, water/oil imbibition curves are representative of waterflooding, while water/oil drainage curves describe the movement of oil into a water zone. The modelling team should recognize that the relative permeability curves used in a flow model may be very dependent on the experiment that was used to measure the curves. Applying these curves to another type of displacement mechanism can introduce significant error.

Capillary pressure

Capillary pressure is usually included in reservoir simulation studies. The relationship between capillary pressure and elevation is used to establish the initial transition zone in the reservoir. The oil/water transition zone, for example, is the zone between water-only flow and oil-only flow. It represents the part of the reservoir where 100% water saturation grades into oil saturation with irreducible water saturation. Similar zones may exist at the interface between other pairs of immiscible phases.

Capillary pressure data is primarily used for determining initial fluid contacts and transition zones. If petrophysical evaluation and property modelling concludes that there is a relative thick transition zone, then capillary pressure should be used; however, if the transition zone is considered to be negligible, then no major benefit can be expected from the use of capillary pressure data.

Dynamic properties and well data

Dynamic properties and well data are primarily required for the following:

• Reservoir simulation model quality control
• History matching process
• Predictions

Historical production/ injection and pressure data

The typical historical data used in reservoir simulation models includes the following: 

• Oil, water and gas rates; water and/or gas injection rates
• Measured shut-in pressures, water-cut, RFT/PLT data, saturation data.

The historical well tubing head pressure, well bottomhole pressure, well productivity index (PI), and well flow performance tables are required to calibrate reservoir simulation models when reservoir simulation models are used for predictions (in other words, when the wells are changed from volume control to pressure control).

Well data

The well data (trajectory and perforation intervals) are required to assign well perforation intervals to the reservoir simulation grid and simulate well performance.

Production, injection and pressure constraints for facilities and wells

These data are required, particularly when reservoir simulation models are used for predictions.

Reservoir simulation model initialization and validation

Reservoir simulation model initialization

To initialize a reservoir simulation model, the initial oil, gas and water pressure distribution and initial saturations must be defined in the reservoir model. Pressure data are usually referenced to some datum depth. It is convenient to specify a pressure and saturation at the datum depth and then to calculate phase pressures based on fluid densities and depths.

The initialization of the reservoir simulation models is the process where the reservoir simulation model is reviewed to make sure that all input data and volumetrics are internally consistent with those in the geomodel. The reservoir simulation model should normally be in dynamic equilibrium at the start of production, but there might be some exceptions to that rule. Non-equilibrium at initial conditions may imply some data error or the need to introduce pressure barriers (thresholds) between equilibrium regions.

Reservoir simulation model validation

At this step, the main objective is to verify that the reservoir simulation model accurately represents the structure and properties in the geologic model. The following validation steps are recommended:

Visualize reservoir simulation grid, each grid layer and each cross-section, to ensure that simulation grid is constructed correctly and all gridblocks are suitable for reservoir simulations.

Compare reservoir simulation grid with the geological grid and make sure that reservoir simulation grid layers and fault geometries are consistent with the structural depth maps used.

Visualize and compare reservoir simulation model properties (porosity, permeability, net-to-gross ration and fluid saturation) with those in the geological model.

Compare reservoir simulation model gross-rock-volume, pore volume, and hydrocarbon in-place volumes with the geological model volumes

Verify that the wells are consistently represented in the reservoir simulation grid.

History matching

The main objective of the history match is to achieve a reasonable agreement between the simulated and observed historical field/well behavior to establish a satisfactory quality reservoir management tool. This is done under the premises that the geological model, the reservoir parameters, and other static and dynamic data used have a “defendable” quality.

Manual vs. assisted history matching

Two approaches can be applied for performing history matching study: manual history matching and assisted history matching using specialized software. Traditionally, history matching is performed by a trial-and-error approach. In this case, a lot of manual tasks are involved, such as changing the reservoir simulation model, running reservoir simulations, plotting curves and comparing to observed data. The main advantage of assisted history matching is to automate those manual tasks, such as reservoir simulation model modifications, running reservoir simulations, comparison of observed and reservoir simulation data, etc. however care should be taken in setting parameter range limits, etc. in automated history match to ensure any solutions are physically valid.

History matching input data

The following historical (measured) input data for individual wells or reservoirs are typically used in history matching process:

• RFT pressures (measured pressure points vs. depth)
• Shut-in pressure (measured pressure vs time)
• Historical production / injection rates vs. time
• Allocated or measured well GOR and WCT vs. time
• Fluid saturation profiles from well logs

History matching steps

The following steps are recommended for performing history matching:

• Match average reservoir pressure and field rates to have a good understanding about material balance in the reservoir.
• Match individual well RFT pressure to have control on compartmentalization and flow barriers.
• Match individual well gas/oil ratio, water-cut and shut-in pressure to have a good control on flow dynamics in reservoir and well performance.

History match quality

There are several ways to decide if a match is satisfactory. In all cases, a clear understanding of the study objectives should be the reference for making the decisions. For example, if a coarse study is being performed, the quality of the match between observed and simulated parameters does not need to be as accurate as it would be for a more detailed study.

Quality of Modifications Made. If the model has a good match but the changes made were not realistic, then the model results should be viewed with skepticism. Remember that the ultimate objective of reservoir simulation is not achieving a history match; it is being able to reasonably predict the future performance of the reservoir. The history match is only an intermediate step in the modelling process.

Reservoir simulation model prediction capability

The reservoir simulation model-building process and history matching are intended to provide a working model of the reservoir and establish a level of confidence in the validity of a flow model. Therefore, the final history matched model is usually re-configured to predict the behavior of the reservoir into the future.

When a reservoir simulation model is changed from history matching to prediction mode, the phase rate profiles should be smooth, provided new wells are not added or existing wells shut-in, and the fundamental constraints on the wells are not changed. There should not be a shift up or down in rates at this point. Such a shift is usually indicative of non-calibrated wells.

It is recommended that the last year of history is run in prediction mode and the actual production compared with the simulated prediction. While this should not be expected to give a perfect match, it will help to highlight major discrepancies in the model.

When a reservoir simulation model is used for predictions, the limitations and uncertainties involved in the reservoir simulation models should be recognized. If the geological model, for example, is not reasonable and observed data quality is poor, not much quality can be expected from reservoir simulation model, no matter the quality of the history match.

Noteworthy papers in OnePetro

Mattax, C. C., & Dalton, R. L. 1990. Reservoir Simulation (includes associated papers 21606 and 21620 ). Society of Petroleum Engineers. http://dx.doi.org/10.2118/20399-PA

Noteworthy books

Aziz, K., & Settari, A. 2002. Petroleum reservoir simulation. Calgary: K. Aziz & A. Settari. http://www.worldcat.org/oclc/49352809

Carlson, M. R. 2006. Practical reservoir simulation : using, assessing, and developing results. PennWell Publishing Company. http://www.worldcat.org/oclc/939069181

Crichlow, H. B. 1978. Modern reservoir engineering: A simulation approach. Englewood Cliffs - N.J: Prentice-Hall. http://www.worldcat.org/oclc/636084916

Fanchi, J. R. 2001. Principles of applied reservoir simulation. Houston, Tex: Gulf Pub. http://www.worldcat.org/oclc/471519143

Noteworthy books

Society of Petroleum Engineers (U.S.). 2011. Production forecasting. Richardson, Tex: Society of Petroleum Engineers. WorldCat or SPE Bookstore

See also

Production forecasting glossary

Aggregation of forecasts

Challenging the current barriers to forecast improvement

Commercial and economic assumptions in production forecasting

Controllable verses non controllable forecast factors

Discounting and risking in production forecasting

Documentation and reporting in production forecasting

Empirical methods in production forecasting

Establishing input for production forecasting

Integrated asset modelling in production forecasting

Long term verses short term production forecast

Look backs and forecast verification

Material balance models in production forecasting

Probabilistic verses deterministic in production forecasting

Production forecasting activity scheduling

Production forecasting analog methods

Production forecasting building blocks

Production forecasting decline curve analysis

Production forecasting expectations

Production forecasting flowchart

Production forecasting frequently asked questions and examples

Production forecasting in the financial markets

Production forecasting principles and definition

Production forecasting purpose

Production forecasting system constraints

Quality assurance in forecast

Reservoir simulation models in production forecasting

Types of decline analysis in production forecasting

Uncertainty analysis in creating production forecast

Uncertainty range in production forecasting

Using multiple methodologies in production forecasting