This paper demonstrates how a combination of old theory (decline methods, statistics) and new technology (computer graphics) can enhance decline-curve forecasts for multi-well groupings (aggregated production). Production and well-count forecasts are presented for four very different-sized groups of aggregated production.
The first example is for a small Mannville pool and the second is for the Pembina Cardium, Alberta's largest oil pool. The third example is an aggregate of all reef pools of Keg River age. The final example is all wells reporting conventional oil production in the province. In each case the graphical results show the historical and the forecast trends for the statistical distribution of well rates, the median well rate, the aggregated rate, and the number of producing wells.
The examples illustrate the utility of enhanced decline-curve methods for making forecasts of Alberta's aging oil production.
Our society thrives on information and forecasts on such diverse areas as commerce, sport, medicine, and the weather. The challenge in all areas is how to present huge amounts of data in a comprehensible form suitable for forecasting and decision making. The oil industry has used decline curves for this purpose for over 60 years. Decline curves have served the industry well; however, they can be significantly enhanced by a combination of old theory (decline methods, statistics) and new technology (computer graphics). The objective here is to demonstrate the enhancement and in the process provide some insight to the question asked in this paper's title.
The first thing one should determine about any decline curve is if it is for a single well or a group of wells. Decline curves for single wells have the least potential for misunderstanding because data smoothing only involves adjustments for producing hours (eg the raw data is normally for a calendar month but calendar-day and/or operated-day rates are often plotted). If ratio curves are shown they should be based on actual oil, gas, and water production. Decline curves and forecasts for single wells are widely used and well understood because
the theoretical basis is strong,
they display all the raw data,
the decline rate is self-evident and changes are usually caused by changes in the total fluid production rate,
the economic limit is directly applied to estimate reserves,
the relationship of well rate to cumulative production is fundamental.
On the other hand, decline curves and forecasts for multiple wells (ie. aggregated production) are widely used but often misunderstood because
the theoretical basis is weak,
they display only a fraction of the raw data,
individual wells have different decline rates than the aggregate,
the economic limit is loosely applied to estimate reserves,
the relationship of well rates and well count to cumulative production is fuzzy.
A previous paper1 demonstrated that the historic deficiencies for the multi-well case can be largely overcome by using a lognormal distribution to quantify the changes in well rates over time.
