Hydraulic Fracturing Design Optimization - Bakken Case Study

Hydraulic Fracturing Design Optimization - Bakken Case Study

Big Data vs Type Curves Bakken Case Study

The case study is prepared to illustrate the unified approach to optimize the design of multiple transverse fractures in a horizontal well using the optiFracMS software.

The goal of the study is to optimize a fracturing design for a Bakken formation.

Two fracture design cases were considered:

1. Operating company fracture design based on a Big Data application
2. Optimized fracturing design based on the Type Curves

For the both cases a production forecast was calculated and cumulative oil production was compared.

Optimized fracture design case predicts 61% more cumulative oil in the first year of well production.

Inputs

The following inputs are used in both fracture design cases.

Fracturing Design Parameters

Fracture Production Forecast Parameters

Parameters Values Proppant Properties    Mass, 1000 lbm 10000 Proppant type 30/70 TX Brown Proppant permeability, md 57661 Mean diametr, in 0.01 SG 2.65 Bulk volume, lbm/ft3 95.5 Porosity, % 40.4 Gel damage, % 90 Reservoir and rock data   Permeability, md 0.04 Thickness net, ft 45 Fracture height, ft 150 Drainage width, ft 1320 Drainage length, ft 10000 Well-bore radius, ft 0.25

Parameters Values Reservoir Properties    Initial pressure, psia 3500 Initial temperature, F 140 Porosity, % 6 Oil saturation, % 100 Rock compressibility, 1/psi 6e-06 Total compressibility, 1/psi 1e-05 Fluid Properties   Permeability, md 0.04 SG oil, API 30 Bubble point, psia 1000 Oil formation volume factor, bbl/stb 1.11 Solution gas ratio, scf/bbl 162.8 Oil compressibility, 1/psi 4e-06 Oil viscosity, cP 3.66 Oil density, lb/ft3 50.5

Hydraulic Fracturing Design Cases

Parameters	Big Data	Type Curves
Proppant number	14.8	14.8
Dimensionless productivity index, JD	18.1	46.7
Number of fractures, n	200	28
Dimensionless fracture conductivity, CfD	60	8
Fracture half length, xf, ft	64	455
Fracture width, wf, in	0.032	0.032
Fracture penetration, Ix	0.1	0.7

Operating company design - Big Data

The Bakken Big Data study on the area compiled by the company petroleum engineers suggests an optimum completion design being 30 stages with 10 clusters per stage. Based on working experience, engineers estimate to get about 200 fractures as a result of the well stimulation.

Design agrees with the established within the company design practice of increasing the reservoir contact by adding more fractures to increase the oil rate.

Optimized fracturing design - Type Curves and optiFracMS

optiFracMS online petroleum engineering software at www.pengtools.com was run to optimize the fracture design.

The resulted design calculates the optimum number of fractures to be 28.

Note that optiFracMS was also used to calculate the design parameters (JD, CfD, xf, wf, Ix) of the Big Data design based on number of fractures of 200.

The model of the case study is available for review online at www.pengtools.com.

Hydraulic Fracturing Production Forecast

Both cases were simulated using the analytical model by Brown et al. using the commercial software IHS Harmony.

1 year production forecasts were compared as a result:

Parameters	Big Data	optiFracMS
Cumulative oil production in 1 year, Mstb	70	113

Bakken oil rate fracturing forecast

Bakken cumulative oil rate fracturing forecast

Bakken Case Study Discussion and Conclusions

The main conclusion

Optimized fracture design case predicts 61% more cumulative oil in the first year of well production.

Discussion

• The number of fractures was optimized from 200 to 28, which resulted in increase of the fractures half lengths from 64 to 455 ft, which increased the reservoir stimulation area and recovery.
• The Big data case produces high conductivity fractures (CfD =60 vs 8) with short fracture half lengths (xf = 64 vs 455 ft). This results in high initial rate which is rapidly declines due to lack of the reservoir contact area. The fracture underperforms.
• Make sure that technical optimum performance data is represented in your Big Data range. Otherwise care should be taken when applying the Big Data. If the data range is represented with the poor performance data only, the result will be the poor performance average.