Using multiple methodologies in production forecasting

It is observed across the global oil and gas industry that all projects are based on specific production forecasts to provide volumes and ensure continuity in the exploration and production business. The majority of these forecasts fall short or under perform based on promises made to the investment community and various stakeholders. There is evidence that industry expectations are usually higher than historical delivery.

Goal of multiple methodologies

To improve this record, there is a need to provide the best possible range of production forecasts during project appraisal all the way to first production and beyond. The onus lies with the exploration, development and production groups tasked with providing this information to find and use all tools and methods available at their disposal and to make realistic corrections during the forecasting process. It is also important to note that a single model will not perfectly represent reality, but we can rely on multiple paths to represent a realistic range of outcomes.

The importance of providing realistic long-term forecasts cannot be overemphasized because it effectively drives corporate strategies as well as promises made to shareholders. There exists the possibility of combining different production forecasting methods to validate the solution to the problem statement. For example, numerical simulation requires a history match to calibrate inputs. The history-matching process can be tedious and time-consuming, and it requires expertise and good judgment. Other challenges include applying proper mathematical theories and managing the non-uniqueness of solutions contributed by differing geological interpretations. The validity of simulation results can be questioned because of issues such as these. All methods inevitably have their merits when used alone, but the limitations of individual methods can affect the validity of results. Various forecasting methodologies can be combined effectively by leveraging on the strengths of the individual methods to generate a more robust set of production forecasts.

Analogs can be useful in predicting the future performance of new wells based on the performance of existing type wells. However, a big challenge to analog-based forecasting is the inability to focus on reservoir deliverability and quantify the impact of variations in geology, semi-optimum well design, and unforeseen operational challenges on production.

Other methods, such as analytical models, can be used based on the simplicity of the models. A big drawback is in incorporating heterogeneities into an analytical model. There are a few other methods applicable to conventional oil and gas systems such as inflow performance relationship (IPR) curves. IPR curves represent the easiest method that requires a limited set of user input parameters, and they also generate single-point solutions. Material balance methods are easy and quick methods that honor material balance equations.

In the case of unconventional resource plays, which are guided by transient flow regimes, the use of material balance as well as IPR methods may be limited. Looking at other production technologies such as steam-assisted gravity drainage in thermal reservoirs, IPR methods may be used for pump optimization and design tasks but significantly misrepresent the production based on changing reservoir dynamics.

Production forecasts can be improved upon provided there are multiple methodologies that can be used to ultimately narrow down the variability and ultimately converge to some solution. There are various examples in which a combination of methods has been used to narrow down uncertainty in predictions and a few examples follow.

Unconventional shale gas

Both analytical and numerical simulation methods are used to model production from hydraulic fractures in forecasting unconventional resources. Such methods could be combined to mitigate the difficulty of modelling complex fractures using numerical simulation. The combined use of unstructured generated grids and numerical simulation methods supports the generation of production forecasts from fracture networks . A workflow proposed by Mirzaei and Cipolla (Fig 3) identifies the different stages involved in combining methods to generate production forecasts in unconventional shale gas reservoirs.

INSERT Figure 1- Workflow for modeling, simulation and ultimate recovery forecast of a well with hydraulic fractures in an unconventional shale gas reservoir.

There is also a semi-analytical boundary element solution proposed by Zhou et al. 2013 that treats all orthogonal and non-orthogonal fracture networks explicitly.

An example to illustrate seismic to simulation forecasting technique is Barnett Shale gas well. Two models were created in this case as outlined in Figure 2.

INSERT Figure 2- Left side: fracture network model of Stage 1 fractures; right side: the same model using the semi analytical approach by Zhou et al. 2013.  (Pending permission approval)

The semi-analytical approach is also used to assess the validity of fracture network modeling as shown on the graph to the right of Figure 8.15. The expected gas production using four cases Table 1 with varying propped and unpropped fracture conductivity is shown in figure 3. All cases were run using a constant bottomhole pressure of 1250 psi for approximately 33 years. Table 1 shows conductivity values.

INSERT Figure 3 - Effect of fracture conductivity on Stage 1 gas production  (Pending permission approval)

Some reservoir parameters, such as fracture conductivity and formation permeabilities, can be found in SPE 146876. Other parameters assumed by Zhou et al. 2012 are as follows:

INSERT Table 1: Reservoir Properties of Unconventional Shale Gas Example  (Pending permission approval)

INSERT Table 2&3: Reservoir Properties of Unconventional Shale Gas Example  (Pending permission approval)

The cumulative gas production in all of four cases shown in Table 3 can be seen on the right side of Fig 3. The figure compares the results from the UFM model (blue curve) (unstructured grid for numerical reservoir simulation) and the semi-analytical approach (red curve). The production trends when using different conductivities (all 4 cases) and two methods are similar and this increases confidence on the outcome when combining production forecasting methods.

Oil sands example

Another example can be found in SPE 165431, which combines the use of analogs, analytical and numerical modeling techniques. The method described in the paper combines these techniques to see the high and low production bandwidth while working to narrow down the differences in methodologies. The field in question has about 70 producing steam-assisted gravity drainage (SAGD) wells that have production history from 6 to 12 years. The assumption in this case is that all the analog wells are producing from the same point bar system and produce oil with similar reservoir properties.

The selection of analogs was established based on similar production mechanism (gas lift) and maximum operating pressures (MOP). Specific performance-related issues were identified and excluded in the analysis of data. The analogs were then narrowed down to a handful that specifically matched the type of operating conditions the test wells would be put into.

The analytical model that was used to forecast future production is based on work of Miura et al. 2010. The methodology takes into account geological parameters, such as porosity and saturation, and other time-dependent variables, such as injection and production pressures and lift mechanism. The analytical method assumes gravity drainage enhances the flow of hot mixture of bitumen and condensed water. It also assumes conductive heat and steady-state heat transfer occur at the steam/bitumen interface, and accounts for steam chamber rise, spread and coalescence phases. To address uncertainty, a Monte Carlo simulation of the analytical tool was used to generate a statistical range of results using uncertainties found in each of the input variables.

Numerical simulation models were created using reservoir input parameters highlighted in SPE 165555. The simulation model used to forecast the test area was based on a fully history matched model that honored actual field rates and pressure history. The quality-control process followed generally reflects real reservoir conditions and focuses on global changes in the reservoir.

Upon completion of individual forecasts using the different methods described above, the results were then consolidated. Fig 4 shows the interrelationship between methodologies and possibly similarities in trends. It is important to note the central position of a specific decision to the different methods identified in this case.

INSERT Figure 4 - Illustration of integrated ternary model.  (Pending permission approval)

It is critical to obtain a good match between Monte Carlo and analog results in that the reservoir must be well understood to declare a fit. The outer limits of the Monte Carlo simulation (P90 and P10) must cover the analog production curves and the results from the most believable realization of the numerical simulation model (Fig 5). The appropriate probability of production forecast will be selected based on appropriate risk acceptable to the E&P companies.

INSERT Figure 5 - OO Pattern Oil Rate Statistical Range  (Pending permission approval)

Tight gas example

A specific deep gas property was identified and decline analysis was used to estimate its reserves. Several different methods, such as individual and group decline analysis, material balance methods and numerical simulations, were all used in this process. The asset had eight wells producing (Fig 6) at a pool abandonment limit of 250 e3m3/D. Its cumulative gas production as of 2008 was 1.3 Tcf.

INSERT Figure 6- Tight gas field production and EUR using decline analysis. Ni=0.1  (Pending permission approval)

To confirm the full-field production forecasts and EUR, each of the contributing decline curves was aggregated to generate a sum total from individual gas wells. The overall abandonment rate for the individual declines was estimated to be half the abandonment rates of the best producing wells. A comparison of a sum of individual well declines and a group field decline only had a 0.2% difference in estimated reserves.

Numerical simulation was used to ultimately provide a basis for the final pool abandonment pressure and as an estimate of well life expectancy. Fig 7 shows a rate versus time history match (green curve) against actual production history (pink curve). The rate versus cumulative gas plot was generated based on simulation history match results, which estimated about 1.4Tcf reserves.

INSERT Figure 7 - Tight gas field history match vs production history.  (Pending permission approval)

Material balance methods were also used to estimate reserves in the specific tight gas field. An overall pool abandonment pressure of 6,000 kPa was selected from numerical simulations and was used in the material balance calculations.

INSERT Figure 8 - Material balance estimate for tight gas field.  (Pending permission approval)

A final comparison of all methods shows a fairly close range of reserves, based on a combination of methods as can be seen in Table 4.

INSERT Table 4 - Reserves Methodology and Associated Reserves - Estimate for Tight Gas Field (Pending permission approval)

Conclusion

The examples described in this section show the strength of combining different methods and how they ultimately contribute to a more robust set of production forecasts. The important message is that there is usually a range of results obtainable when these forecasts are combined, and the true result will depend on corporate risk taking and its business philosophy.

Noteworthy papers in OnePetro

Cipolla, C. L., Fitzpatrick, T., Williams, M. J., & Ganguly, U. K. 2011. Seismic-to-Simulation for Unconventional Reservoir Development. Society of Petroleum Engineers. http://dx.doi.org/10.2118/146876-MS.

Nandurdikar, N. S., & Wallace, L. 2011. Failure to Produce: An Investigation of Deficiencies in Production Attainment. Society of Petroleum Engineers. http://dx.doi.org/10.2118/145437-MS.

Wang, J., Liu, F. (Changyi), & Morris, P. 2013. A Practical Approach to History-matching Large, Multi-well SAGD Simulation Models: A MacKay River Case Study. Society of Petroleum Engineers. http://dx.doi.org/10.2118/165555-MS.

Wilson, A. 2013. Instilling Realism in Production Forecasting Decreases Chances of Underperformance. Society of Petroleum Engineers. http://dx.doi.org/10.2118/0913-0114-JPT.

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