Multivariate analysis correlates completions, EUR for Canada's Bakken

April 4, 2016
The heterogeneity of oil and gas reservoirs across large resource plays, such as the Bakken, requires operators look at a combination of completion inputs to maximize rates and recoveries. 

Mark Lenko
Samantha Foster

Canadian Discovery Ltd.
Calgary

The heterogeneity of oil and gas reservoirs across large resource plays, such as the Bakken, requires operators look at a combination of completion inputs to maximize rates and recoveries. A variety of completion parameter combinations can deliver similar results, and applying multivariate statistical analyses to these methods can help operators increase estimated ultimate recovery (EUR) and minimize half-cycle cost by identifying which variables obtain better results.

Bakken focus

The Mississippian-Devonian Bakken formation covers roughly 200,000 sq miles across North Dakota, Montana, and Saskatchewan. Bakken operators at Viewfield, Sask. have tried a variety of drilling and completion practices with varying degrees of success in respect to EURs. No single completion technique holds the key, but the right combination of practices can deliver consistent results.

This analysis uses a multivariate statistical procedure to find the optimal completion techniques for the Bakken at Viewfield, maximizing EUR and minimizing finding and development costs. Multivariate statistical analysis can identify an improved completion design, increasing chances of developing a successful well.

Viewfield is the main producing area within the Canadian Bakken. The current set of producing wells includes 2,929 horizontal wells with multi-stage completions, 467 openhole horizontal wells, and one vertical well. This analysis excludes 111 producing wells that have shown a response to waterfloods initiated in 2009.

Data concentration

Canadian Discovery Ltd. developed its well completion and frac database with completion information for more than 37,000 wells in western Canada. The Catalyst Resource Play Evaluation Platform provides well information for all wells in western Canada, including unconventional and conventional wells. The system forecasts EUR estimates for each well by developing type curves for subsections within the larger play. Drilling and completion costs are estimated through an algorithm that employs controllable variables such as length of operation, depth, lateral length, applied technology, fluid volumes, and (sand) tonnage pumped.

This algorithm provides a clear view of completion parameters and the combinations of which result in the highest EURs and lowest half-cycle cost (drilling and completion)/boe. This article limits the data to horizontal wells with multistage completions.

Completion technology

The data set used for this analysis captures nine groups of completions variables including:

• Isolation technology.

• Well depth, lateral length.

• Pressure (breakdown, average, maximum, initial shut-in period).

• Proppant (type, volume, design, tonnage pumped, concentrations).

• Costs (field, authorizations for expenditure, estimated).

• Stages (number, spacing, type).

• Energizer.

• Dates (start, end, problem time).

• Fluid (type, volumes).

Among these controllable variables, completions technology, proppant pumped/stage, and number of stages completed were most significant in maximizing EUR.

Fig. 1 shows a decision tree resulting from the recursive partitioning of EUR based on completion covariates. The highest-impact variables appear at the nodes (top) with resulting EUR groups summarized in box plots (bottom). Each plot shows the median EUR of the group as a solid line in the middle of the box. The variable with the highest impact on EUR is the isolation technology used in the completion.

Four main technologies complete wells in the Bakken at Viewfield (Table 1). The migration from ball-and-seat to coiled-tubing technologies (CT) with time is clear and occurred primarily due to the lower cost of CT completions (Fig. 2b). Operators tried pressure CT and straddle CT systems, but the vast majority of completions have used ball-and-seat or cut-port CT (packer) systems (Fig. 2a).

The two dominant completion technologies, ball-and-seat (1,454 wells) and cut-port CT (713 wells), improve EUR by 33,000 boe or 45%. Higher EURs in combination with lower costs demonstrated the benefit of switching technology.

Proppant application

The amount of proppant pumped/stage is the next most important variable. Figs. 3a, 3b, and 3c, divide the amount of proppant pumped/stage into quartiles. Fig. 3a shows that 10 tonnes proppant/stage achieved the highest median EUR (Table 2). Proppant/stage amounts have trended down over time (Fig. 3b). The map in Fig. 3c shows that the higher tonnage group is concentrated in the play's center, which generally produces higher-EUR wells. The higher median EUR of this group may result from a combination of geology and completions design.

The amount of proppant pumped/stage impacts EUR positively, with a difference of 31,000 boe (44%) between the median EUR and the bottom and top quartiles.

Stages

The number of stages completed is the third major contributor to higher EUR. Figs. 4a, 4b, and 4c, divide the completed stages into quartiles. Analysis shows the highest median EUR comes from completing more than 20 stages, ideally 24 (Fig. 1). Median EURs are clearly different in the lower and upper quartiles (Fig. 4a, Table 3). Fig. 4b shows operators have increased the number of completed stages over time. But there is no geographical concentration with respect to the number of stages completed (Fig. 4c).

The lower two quartiles show no material difference in EUR, but having more than 20 stages appears to have an effect.

Wellbore direction

Fig. 5 shows the effect of wellbore orientation or direction on EUR, with only a slight variation recorded. Median EUR for EW-oriented was 88,000 boe, compared with 74,000 boe for NS-oriented wells (Fig. 5a, Table 4). Operators do not have a discernible preference for direction other than to maintain consistency with adjacent wells (Fig. 5c).

Contractor performance

Seven service companies have provided stimulation services to operators in the Bakken at Viewfield. Median EUR varies widely with respect to each contractor (Fig. 6a). Service providers have entered and withdrawn from the play (Fig. 6b). Results categorized by service company have not been controlled for other variables, such as location in the play or completion design, and should not be considered definitive (Table 5).

Cost analysis

A variety of factors, including depth, lateral length, drilling days, completion days, completion systems, proppant amount, and fluid volumes generated estimated half-cycle costs. The estimated cost for each well was divided by the EUR to calculate the half-cycle cost/boe. Multivariate analysis determined which controllable completion variables were most influential in minimizing cost.

Fig. 7 shows a decision tree resulting from recursive partitioning of cost/boe based on chosen covariates. The highest impact variables appear at the nodes, with the resulting groups summarized in box plots (at bottom). Each plot shows the median EUR of the group as a solid line in the middle of the box. The entire range of data in each group is shown by the dotted vertical lines with outliers shown as single points.

The lowest half-cycle cost was $9.62/boe, brought about by using cut-port CT, pumping more than seven tonnes of proppant per stage, and completing more than 15 stages/well. The next lowest half-cycle cost was $10.77/boe, also done using cut-port CT completion, but pumping between 4-7 tonnes of proppant/stage and completing more than 24 stages/well.

Pressure CT showed a high median EUR of 150,000 boe (Fig. 2a) yet was only used in 2006. One possible explanation for movement away from the technique was a group of wells with median half-cycle cost of $28.19/boe (Fig. 7).

Multivariate statistical analyses used to maximize EUR and minimize half-cycle cost share common variables with the better outcomes. In the Bakken at Viewfield, these include use of cut-port CT completions, pumping more than 7 tonnes of proppant/stage, and completing more than 15 stages/well (with a statistical preference for more than 24 stages).

The authors

Mark Lenko ([email protected]) is the interim managing director and engineering director at Canadian Discovery Ltd. in Calgary, Alberta. He holds a BSc in petroleum engineering from the University of Alberta, and a BA and MA in economics from the University of Calgary. He is a member of the Association of Professional Engineers and Geoscientists of Alberta (APEGA) and the Society of Petroleum Engineers (SPE).

Samantha Foster ([email protected]) is an intermediate engineer at Canadian Discovery Ltd. in Calgary, Alberta. She holds degrees in mechanical engineering (BSc, MSc) from the University of Calgary and is a member of APEGA.