Exploring Proppant Concentration in the Bakken with DI Transform

Share

While the price of WTI crude oil was in the process of dipping below $44 in March, most operators were tasked with decreasing costs as much as possible. Even with the current recovery hovering near $60, it’s probably not the best time to suggest spending an extra $1 million dollars for more proppant in each well. However an analysis of proppant concentration in the Bakken suggests that this may be a great time to have this discussion.

When planning for how heavy to engineer wells there are a few general scenarios to consider. Should we over-engineer the best acreage or will our best acreage produce enough with moderate engineering? Can we over-engineer the poorer rocks to get them to produce like higher tiered acreage or will this be a waste of money? This study takes a statistical approach to quantifying answers to these questions.

For this analysis we need to start with a geological framework. This will allow us to analyze the effects of proppant within similar geology. The DI Play Assessment for the Bakken provides this geological framework which includes stratigraphic correlations along with structure, thickness, and property maps. This framework is delivered in the DI Transform platform. DI Transform includes multivariate analytics which gives users the ability to analyze effects of engineering parameters within specific geology. Finally, incorporating refined well data from DI Analytics sheds time by allowing the user to start off with clean, aliased data necessary for statistical analysis. DI Analytics is also the source for two other key components in this study, proppant amounts and our graded acreage map.

First let’s take a look at proppant/ft in the North Dakota portion of the Bakken play. Here we see a 0.379 correlation between proppant per foot and initial 12 month oil production.
Exploring Proppant Concentration in the Bakken with DI Transform
Figure 1: This crossplot compares proppant/ft (X-Axis) to initial 12-month oil production (Y-Axis).

Next, let’s see how proppant concentration has evolved over the course of Bakken development by analyzing proppant/ft versus spud date.

Exploring Proppant Concentration in the Bakken with DI Transform
Figure 2: The crossplot on the left shows how much proppant/ft (Y-Axis) has been used over the development of the play by comparing this to spud date (X-Axis). The plot on the right shows this same data filtered on 3 key Bakken operators.

We can see that proppant use has increased significantly since early 2012. The cross-plot on the right shows that one key operator (with purple data points), led this push. We also see two other key operators in red and blue starting to significantly increase proppant concentration.
For this study, let’s break out the samples by their graded acreage. We’ve set up three tiers. These include grades 1-3, grades 4-6, and grades 7-9. A map of this tiered acreage is included below.

Exploring Proppant Concentration in the Bakken with DI Transform
Figure 3: Grades 1-3 displayed in red, Grades 4-6 in green, Grades 7-9 in blue. Rigs are also shown.

Now that we’ve broken up the play into three tiers based on the geology, we will analyze proppant in each tier. When looking at the bi-variate cross-plots for each of these three tiers, we can see varying degrees of positive correlations. Tier two acreage has the best correlation with a 0.497 relationship. Tier one is second with 0.290. Tier three has just a slight 0.0596 relationship.

Exploring Proppant Concentration in the Bakken with DI Transform
Figure 4: These graphs show proppant/ft (X-Axis) versus initial 12 month oil (Y-Axis). The top left graph is tier 1, the top right is tier 2, the bottom left is tier 3, and the bottom right is all tiers.

The increase in oil production for the top two tiers of Bakken acreage is very encouraging, but not so encouraging for the bottom tier acreage. If we follow the correlation line, we gain approximately 50k barrels by increasing from 400 lbs/ft of proppant to 900 lbs/ft of proppant in tier 1 acreage. In tier two acreage we gain approximately 70k barrels for the same increase. For a 10,000 ft lateral, common in the Bakken, an increase of 500 lbs of standard proppant would add approximately $1 million to the well. This is based on an assumption of 20 cents/lb for standard proppant.

With WTI prices around $59 and assuming a return of approximately $40 per barrel after royalties and other fees, we would increase our return by approximately $1 million per well in the first year by increasing the proppant from 400 lbs/ft to 900 lbs/ft in tier 1 acreage. In tier 2, this net increase is approximately $1.8 million per well. Increasing the proppant in tier 3 acreage would result in a loss of around $850k per well.

For an even more granular look, we analyzed proppant/ft within a multi-variate statistical model. This will normalize the effects of horizontal length along with key geological attributes so that we can specifically focus on proppant/ft within the model. Using this analysis, the upside of additional proppant/ft is even greater. Tier 1 now reveals an increase of approximately 70k barrels, or an increased ROI of approximately $1.8 million within the first year. For tier 2, these values increase to 100k bbls and $3 million. We still show a large loss when trying to pump more proppant into the lowest grade acreage. These models are seen in the next few images.
Exploring Proppant Concentration in the Bakken with DI Transform
Figure 5: Multi-variate model for tier 1. The graph on the left shows the overall model. The graph on the right shows the effect of proppant/ft on the model if we fix all other values to their averages within this sample size.

Exploring Proppant Concentration in the Bakken with DI Transform
Figure 6: Multi-variate model for tier 2. The graph on the left shows the overall model. The graph on the right shows the effect of proppant/ft on the model if we fix all other values to their averages within this sample size.
Exploring Proppant Concentration in the Bakken with DI Transform
Figure 7: Multi-variate model for tier 3. The graph on the left shows the overall model. The graph on the right shows the effect of proppant/ft on the model if we fix all other values to their averages within this sample size.

Here are a few takeaways from this study. Stepping up on proppant concentration in quality acreage appears to provide outstanding returns, but over-engineering the bottom tier acreage looks like a waste of sand. Another interesting observation is that from a statistical perspective we have yet to see a turnover where increased proppant concentration no longer matters in the quality acreage. There still looks to be better ROI with more and more proppant in the top two tiers of acreage. Eventually these plots will plateau at the top indicating diminishing returns, but this hasn’t happened yet.

For a more detailed look this analysis could quickly be run in each grade of acreage or around your specific holdings. It could also be run on other completions metrics that you feel are key contributors. It’s more important than ever to not think it terms of strictly cutting costs, but rather in terms of optimizing your completions for ROI.

So maybe it’s time to march into your boss’s office and request that you double down on proppant in the Bakken. Tell him or her that it will only cost an additional $1 million per well. If you can present your analysis before you’re placed in a strait jacket then you might just be a hero.

Your Turn

What do you think? Leave a comment below.

The following two tabs change content below.

Clint Barefoot

Clint Barefoot is a Solutions Architect at Drillinginfo. He is responsible for working with clients to help maximize ROI using Drillinginfo products. Clint has a Master's degree in Geology from Oklahoma State University and worked at Chesapeake Energy prior to joining Drillinginfo.