Attain Accurate Developed, Proved and Possible Deterministic Reserves with SEDA
July 25, 2016
Boyd Russell, P. Eng.
Type Curves, Decline Analysis, Oil and Gas Forecasting, Reservoir Engineering
The preferred method of reserves evaluation for producing wells is deterministic. However, when deterministic methods are used, evaluators cannot reliably assign reserves that satisfy the prescribed certainty for each reserves category. In addition, they have no method to quantify the impact that aggregation will have on reserves certainty. My colleague Randy Freeborn and I completed a research project, technical paper and presentation to demonstrate how evaluators can overcome these deficiencies and improve the accuracy of their reserves reporting through a new method of statistically enhanced decline analysis (SEDA).
The basic premise of SEDA is three-fold:
- A best estimate forecast, with adjustment for bias, is 2P reserves.
- A best estimate production forecast is dynamic and will change with time. A best-fit forecast with a year of history will be different from one with three years or one after you have seen boundary-dominated flow.
- For a specific time-period, there is a constant ratio between the remaining reserves categories.
In developing SEDA, we followed this game plan:
- We found analog wells with 13 to 14 years of production history and forecasted each of these wells at various times in their life cycle.
- For each well and time, we determined the ratio of the remaining reserves calculated from the truncated forecast to the remaining reserves calculated from a forecast using all 13 to 14 years of data (our assumed known answer). The formula for remaining reserve ratio (RRR) follows:
- Using Monte Carlo simulation, we calculated remaining reserves ratios and their probability distributions for groups of wells to represent the effect that aggregation will have on the results.
Analog Wells and Forecasting
For use as our analog, we selected 850 older horizontal Barnett shale wells: we used 550 wells to develop the correlations to implement SEDA, and the remaining 300 to verify the results.
We felt the forecasts were incredibly accurate when compared with our assumed known answer. The range of EUR for truncated data was from -4% to +5%. The six-month and the one-year forecasts used Analog Forecasting™ (SPE 167215) and were within 2-5% of the assumed known answer.
Distribution of Remaining Reserve Ratio
We show the distributions of remaining reserve ratio, plotted, from six months to eight years. As expected, the slope of the distribution gets steeper with more history, reflecting greater certainty in the results. Of most interest are the black lines. They represent the results for six-months and one-year of data. Thanks to Analog Forecasting, it’s probably the best data.
Application of Remaining Reserve Ratio
To use SEDA for proved remaining reserves, we multiply the remaining reserves from our best estimate forecast. In this example, 1.05 is the P50 remaining reserve ratio.
Because we base the remaining reserve ratio on a forecast using all of the production history, SEDA has the effect of transferring that forecast knowledge to wells with less production history.
For proved reserves, the multiplier to our best estimate forecast is the ratio of RRR at P90 to that at P50 (0.38 / 1.05). This multiplier is lower than that usually seen in reserves reports. We will see that aggregation will increase this multiple. In a similar manner, the multiplier for possible reserves is 5.51 / 1.05.
The terms P10, P50 and P90 describe the level of certainty or the percentage of the time forecast value will exceed the amount specified. So if SEDA determines a P90 remaining reserve, then SEDA will underestimate the reserves 90% of the time.
We took this same ratio and aggregated it for 10 wells. What does this mean? From the 550 wells, we randomly selected ten wells and determined the remaining reserve ratio for those 10 wells in aggregate. We do this same thing for another 10 random wells, and another 10, repeating the process 500,000 times.
The distribution of remaining reserve ratio for 10 wells is the blue line on the plot. The slope is much steeper than that of the distribution without aggregation, reflecting improved certainty in the proved remaining reserves when the number of wells is increased.
Ten wells is simply an example. When applying SEDA, you would aggregate to the number of wells in a project, field or other entity allowed for aggregation by the regulation. Thus, we create an aggregated distribution for various well counts, and extract the information for commonly used probabilities and display it on a trumpet plot.
There will be a trumpet plot for each time period analysed. In this example, the time is 5 years.
Application of SEDA to a Well
To use SEDA, you first determine the number of wells that will be aggregated. The number of wells will determine the x-axis value to use on the trumpet plot. The producing life for each well will determine which trumpet plot to use (interpolate between two as required).
Forecast each well to obtain the best fit. Obtain multipliers for P10, P50 and P90 from the appropriate trumpet plot and calculate the remaining reserve for each reserve category.
Use an appropriate method to adjust the rate-time profile of the best fit to match the remaining reserves for each reserve category. As seen on the graphs, this can be done by altering the time to boundary dominated flow for unconventional resources or altering the Arps’ decline parameters for conventional wells.
Does SEDA Work?
We created the SEDA trumpet plots using 550 wells, reserving the remaining 300 wells to confirm that the method works.
We chose a specific number of wells for aggregation and ran 10,000 Monte Carlo trials for that number of wells. For each Monte Carlo trial, we would make the following random choices:
- The trial wells as a subset of those reserved for testing.
- The years (or months if 6 months) of production history available for forecasting. A different value for each well.
For each Monte Carlo trial, we compared the SEDA calculated aggregated remaining reserve to the almost known answer (13 to 14 years of data) and counted the number of times the SEDA remaining reserves exceed the known for each reserve category.
We repeated this process for various values for the number of wells in the aggregation and show the results graphically. The SEDA calculated results precisely match the desired expectation of certainty for all reserve classes. This independently demonstrates SEDA as reliable technology for the Barnett Shale.
Using SEDA will:
- Resolve deficiencies of deterministic forecasting.
- Improve certainty and accuracy.
- Increase 1P and 2P reserves.
- Provide the statistical framework for aggregation of deterministic reserves for regulatory submissions.
Boyd Russell is the original designer of Value Navigator and AFE Navigator. In his role, he provides valuable leadership support and vision to the organization, keeping current through research and by attending conferences, workshops, and forums. Boyd Russell is a practicing petroleum engineer with over 35 years’ experience, affirmed as an expert by the Alberta Court.