Copyright © the author(s). This work is licensed under a Creative Commons Attribution 4.0 International License. DOI:10.14800/IOGR.426 Received July 2, 2018; revised September 23, 2018; accepted October 13, 2018. *Corresponding author: liuss.gwdc@cnpc.com.cn 1 Comparison Of Fracturing Treatment Design With An In-house Code, MFrac And FracPro Shanshan Liu*, Great Wall Drilling Company, Beijing, China Abstract Hydraulic fracturing software is widely used in our industry nowadays for fracturing treatment design. Some of them are fracturing simulators that can actually mimic fracture growth while some kind is solely for treatment design. Although being different, they can provide reasonable ultimate fracture geometry and design procedure. This project aims to compare design results from three different fracturing software. Four case scenarios are studied, which involve sensitivity of consistency index of fracturing fluid, out-of- zone fluid loss multiplier, minimum horizontal stress and reservoir permeability. Each parameter is studied with individual software without interference. Results from this project can help users to understand how to cope with design changes when there is dramatic input change. Introduction Three software are used for this study. They are M23(a self-developed program), MFrac and FracPro. MFrac and FracPro are fracturing simulator that can provide fracture properties under every time step so that users can see how the fracture grows. Meanwhile, they also provide treatment design. Not being a simulator for M23, it will not provide how fracture grows during the treatment, but it will give the ultimate fracture geometry and treatment design. No matter which software is used for treatment design, users must cope with different input changes. Once there are input changes, the ultimate fracture geometry and pumping sequence are likely to change. So, users have to understand the physics behind the software so that these changes can be properly treated. To achieve this, this project selects four representative parameters and investigate how they influence fracture design. This paper is written in such a way that there is no interference between each software. All four parameters are studied by each individual software, as shown in Figure 1. Input Data The layer data are shown in Table 1. It is a sandstone reservoir with shale layer laminated. Data for proppant and fracturing fluid are shown in Table 2 and 3. Table 4 shows reservoir property. Study through M23 In this section, we studied the effect of consistency index, out-of-zone fluid loss multiplier, minimum horizontal stress of top layer and reservoir permeability by using M23. We first run a base case and get the results. Then we run separate case by changing one of the parameters above, and compare results with the results of the base case. For this section, the results are summarized in Appendix A. mailto:liuss.gwdc@cnpc.com.cn 2 Table 1—Reservoir layer data. Top ft Thickness ft Stess psi KIC psi in1/2 Perf E 106psi v k md Lith 1 9000 500 7585 1000 FALSE 4.28 0.3 0.001 Shale 2 9500 100 7831 1000 FALSE 4.28 0.3 0.001 Shale 3 9600 15 7110 1200 TRUE 2.80 0.26 0.5 Sand 4 9615 50 7905 1000 FALSE 4.28 0.3 0.001 Shale 5 9665 10 7156 1200 TRUE 2.80 0.26 0.5 Sand 6 9675 30 7946 1000 TRUE 4.28 0.3 0.001 Shale 7 9705 10 7158 1200 TRUE 2.80 0.26 0.5 Sand 8 9715 15 7972 1000 TRUE 4.28 0.3 0.001 Shale 9 9730 10 7204 1200 TRUE 2.80 0.26 0.5 Sand 10 9740 100 8028 1000 FALSE 4.28 0.3 0.001 Shale 11 9840 500 8274 1000 FALSE 4.28 0.3 0.001 Shale Table 2—Proppant data. Proppant mass 400000 lbm Proppant permeability 50000 md Proppant relative gravity 2.65 Stressed proppant porosity 0.3 Unstressed proppant porosity 0.38 Table 3—Fracturing fluid data. Power law fracturing fluid Consistency index, K 0.1 lbf·ft-2·sn Flow behavior index, n 0.6 Table 4—Reservoir property. Reservoir area, Ad 40 acre Proppant mass, M2w 400000 lbm Proppant permeability, kf 50000 md Reservoir permeability, k 0.5 md Net pay, hn 45 ft Figure 1—Project Framework. Case 1: Consistency index of fracturing fluid decrease 10 times. Fracturing fluid is power law fluid. Based on PKN model, the fracture width of power law fluid is calculated with Eq. 1. ��,0 = 9.15 1 2�+2 × 3.98 � 2�+2 1+2.14� � � 2�+2� 1 2�+2 �� �ℎ� 1−��� �' 1 2�+2 ,…………………….......………………….(1) We can see fracture width is function of fluid rheology properties, fracture height, rock plane modulus and pumping rate. If fluid consistency index decreases by 10 times, the fracture width will become approximately 2 times smaller, as shown in Eq. 2. �1 �2 = �1 �2 1 2�+2 = 10 1 2∗0.6+2 = 2.05,..………………………………………………………….…..…….(2) In M23, the ultimate fracture length is pre-designed fracture length, which comes from UFD optimization. Fracture height is coupled with fracture width through net pressure. With the changed parameter, M23 cannot give a treatment design. It is not feasible, so we need to focus on M10. In M10, the fracture length and height are kept the same as base case. Decreasing fracture width will decrease fracture volume. If proppant mass is kept the same, the slurry concentration in some stage will be larger than movable slurry concentration. The calculation is through Eq. 3. ��ℎ� = ��−1���� ����ℎ��� ，��ℎ� > ���������,…………………………………..………………………..….…….(3) 4 In order to solve this problem, we need to decrease slurry concentration. We can either increase fracture volume or decrease proppant mass. After calculation, the following two methods are suggested.  Design a longer fracture with half length 670 ft  Reduce mass of proppant to 245,000 lbm. However, both methods produce a lower JD. Case 2: Increase fluid loss multiplier outside pay zone from 0.25 to 0.45. In order to get the optimal JD, UFD method is used to get the optimum fracture length and conductivity. The fluid loss multiplier outside the pay zone does not affect the UFD optimization. So, the designed fracture half-length will not change. It is still 463 ft. Due to the layer data, net pressure and fluid density remain the same, the fracture height will also not change. Based on Eq. 1, the fracture width will not change either. However, the fluid loss multiplier affects the slurry mass balance equation. �� ℎ��� � − 2��� � − (��� + 2��) = 0,…...…………………………………………………….…………(4) ������� = ���� ,...………….…..…………….……………………………………………………….….(5) ������� = ������� − �����,..…….…………….…………………………………………..……..………(6) � = ����� ������� ,.…………………………………….………………………………………………....…….(7) ���� = ���� ,…………………………………………………………………………………………….(8) The increase of CL and Sp results in a larger pumping time. In this case, parameters that are related with pumping will change based on Eqs. 4 to 8. Generally, as the slurry volume, liquid volume, pumping time and pad time increase, the slurry efficiency decreases. The results can be seen in Appendix A. By comparing the results between base case and this changed case, we can see M23 can help users to decrease the negative effect of changing fluid loss multiplier and give an adjusted design. Case 3: Increase minimum horizontal stress in Layer 1 by 1000 psi. Minimum horizontal stress is involved when calculating net pressure, and further this will affect the calculation of stress intensity factor, as shown in Eq. 9. The equilibrium fracture height will be affected. ��+ = 1 �� −� � ��(�) �+� �−� ��� ,….……..…………………………………………………………...…….(9) If we look at the fracture height of the base case, the upper tip is at 9,510 ft. The first layer is from 9,000 ft to 9,500 ft. That means the fracture upper tip does not penetrate into Layer 1, as shown in Figure 2. Therefore, before doing the design, we can guess that changes of minimum horizontal stress in Layer 1 will not affect the fracture upper tip position. And the whole fracture geometry will remain the same. Correspondingly, the treatment schedule will also be the same. This speculation is proved by running M23 with the changed parameters. The results comparison is shown in Appendix A. Case 4: Decrease reservoir permeability by 5 times. Reservoir properties determine the fracture optimization design. Fracture needs to be designed to produce maximum productivity index for given amount of proppant. Therefore, change of reservoir properties will need a new fracture optimization. Figure 3 describes the effect of reservoir permeability. A new fracture optimization and treatment design is obtained by running M23 with changed reservoir permeability. The results are shown in Appendix A. We can see the new JD is larger than the base case. The comparison between this case and base case shows that M23 can give adjusted new design if reservoir permeability changes. Figure 2—Fracture profile of base case. Change of reservoir permeability Change of proppant number Change of JD, Cfd-opt and xf-opt Change of whole pumping schedule Figure 3—Effect of reservoir permeability. Study through MFrac In this section, we studied the effect of consistency index, out-of-zone fluid loss multiplier, minimum horizontal stress of top layer and reservoir permeability by using MFrac. Likewise, we first run a base case and get the results. Then we run separate case by changing one of the parameters above, and compare results with the results of the base case. For this section, the results are summarized in Appendix B. Case 1: Consistency index of fracturing fluid decrease 10 times. In MFrac, the final proppant concentration is set to stop the simulation process. If the proppant concentration across the fracture is larger than this value, the simulated pumping continues until this concentration is reached. In this study, we set the final proppant concentration to be 10 ppga. If we change consistency index of fracturing fluid, the fracture width during pumping becomes smaller. In order to accommodate the proppant, the fracture length has to become longer. In other words, MFrac will always produce a treatment design, with fracture geometry being very different. In this case, the fracture half-length is around 300 ft longer than base case. Meanwhile, the fracture height in MFrac is obtained from an aspect ratio with fracture length. If the length increase due to pumping, the fracture height will also keep increasing correspondingly. The calculated results are shown in Appendix B. We can see MFrac can provide user an adjusted design if the consistency index changes. Case 2: Increase fluid loss multiplier outside pay zone from 0.25 to 0.45. Fluid loss data are modified layer by layer in MFrac. Modification is only applied to Shale zone. Provided the same amount of proppant, same fracturing fluid, and same final proppant concentration, the simulation process (fracture propagation process) will keep going until the final proppant concentration reaches the pre-set value. In 6 other words, the fracture volume is the same as base case. The fracture width in MFrac is calculated through Radial Model (Eq.10) � ∝ 1−�2 � ���� 2−� 2 2 3�+6 ,..………………………………………………..………..…………..…….(10) We can see the fracture width will not change due to fluid loss change. In the meanwhile, the fracture height and fracture length is coupled. So for the same fracture volume, if fracture width does not change, neither the fracture height nor length will change. The mass balance equation in MFrac is also described with Eqs. 4 through 8. So larger fluid loss coefficient leads to a larger pumping time, volume pumped and pad time. The slurry efficiency will become smaller. We can see, MFrac can adjust another design if fliud loss parameter changes. The results are shown in Appendix B. Case 3: Increase minimum horizontal stress in Layer 1 by 1000 psi. In Base case, the fracture upper tip is at 9349 ft, which is in Layer 1. So if the minimum horizontal stress in Layer 1 increases, the fracture height will decrease according to Eq. 6. The simulation results are shown in Appendix B. We can see the fracture height decreases from 503 ft to 380 ft. In order to accommodate the proppant, the product of fracture length and width should increase. We can see MFrac can adjust treatment design if minimum horizontal stress changes. Case 4: Decrease reservoir permeability by 5 times. MFrac is fracture treatment design software. It cannot give fracture optimization design. If we change the reservoir permeability but keep the proppant mass, fracturing fluid the same and final proppant concentration the same, the ultimate fracture geometry will not change. In the meanwhile, the fluid loss parameters are not correlated with permeability in MFrac, so the simulated pumping parameters will also be the same. This can be seen in Appendix B. However, as a designed, we always want to have a maximum JD. So, the user should be knowledgeable to have a rough estimation towards the fracture geometry and dimensionless conductivity. MFrac cannot substitute users at this point. Study through FracPro In this section, we studied the effect of consistency index, out-of-zone fluid loss multiplier, minimum horizontal stress of top layer and reservoir permeability by using FracPro. Like previous two cases, we first run a base case and get the results. Then we run separate case by changing one of the parameters above, and compare results with the results of the base case. For this section, the results are summarized in Appendix C. Case 1: Consistency index of fracturing fluid decrease 10 times. In FracPro, the target Cfd is set to stop the simulation process. If the Cfd is not reached after limited number of iterations, the software will stop. Compared with base case, the Cfd is still 4.5 in this changed case. FracPro cannot give a treatment design because the calculated fracture width is small. In order to have a treatment, users need to decrease the target Cfd. In this case, the Cfd is changed to 0.2, and a treatment design is calculated, as shown in Appendix C. We can see the proppant mass decrease around 20 times as compared with base case. The fracture geometry remain close. This means the fracture permeability becomes smaller. In general, FracPro cannot adjust its treatment design if fracturing fluid consistency index changes. User need to make adjustments based on own knowledge. Case 2: Increase fluid loss multiplier outside payzone from 0.25 to 0.45. Fluid loss data are correlated with formation permeability in FracPro. So, the change of fluid loss multiplier will have formation permeability changed. In this case, I changed shale layer permeability to have the fluid loss coefficient out of payzone changed. The calculated results are shown in Appendix C. We can see the results are close to the base case, expect the slurry volume, liquid volume, injection time and pad time. The slurry efficiency is a little lower. These parameters are correlated with mass balance, as we have discussed in previous sections. As can be seen, FracPro can adjust its treatment design if fluid loss parameters change. Case 3: Increase minimum horizontal stress in Layer 1 by 1000 psi. In base case, the fracture upper tip is at 9,514 ft, which is in Layer 2. Therefore, if we change the minimum horizontal stress of Layer 1, the results will not be changed. The results are shown in Appendix C. We can see the results are the same as base case. We cannot conclude FracPro can adjust its treatment if minimum horizontal stress changes by just running this case. However, M23 give correct result for this case. Case 4: Decrease reservoir permeability by 5 times. A target CFD is set for this simulation. If reservoir permeability decreases, the product of fracture width and fracture permeability should also decrease. In this case, fracture width remains the same because we based on Eq.1, so the fracture permeability should decrease, which cause the decrease of proppant concentration inside of fracture. Also, the mass proppant concentration in injected slurry will also decrease. In the meanwhile, the fluid loss parameters will also decrease for the pay zone. Thus, the parameters related with mass balance will change. In the results, the slurry volume, liquid volume, injection time and pad time decreases and slurry efficiency increases. As we can see, FracPro can help to adjust the treatment design if reservoir permeability changes. Table 5—Summary. M23 MFrac FracPro Consistency index N Y N Fluid loss parameter Y Y Y Minimum horizontal stress in top layer Y Y Y Reservoir permeability Y N Y Conclusions We can draw the following conclusions: 1) In general, different software have different ability to cope with input changes. It can be summarized in Table 5 (Y is the software can adjust and N is not). 2) When we analyze sensitivity of parameters, first we need to know how the fracture geometry changes. 3) Based on fracture geometry, we then analyze how the pumping schedule changes. Conflicts of Interest The author(s) declare that they have no conflicting interests. References User Manual, MFrac, Meyer Associates User Manual, FracPro, CarboCeramics 8 Appendix A Base Case K1 10 times smaller flmult from 0.25 to 0.45 1000Psi increase k decrease 5 times Mass of proppant injected, lbm 400000 400000 400000 400000 400000 Mass proppant concentration in injected slurry (ppga) 10.2 NA 10.2 10.2 5.42 Slurry volume injected (gal) 159000 NA 226000 159000 264000 Liquid volume injected (gal) 141000 NA 208000 141000 246000 Injection time (min) 108 NA 154 108 180 Pad time (min) 44.64 NA 86.8 45.1 77.8 Frac slurry efficiency 0.36 NA 0.253 0.36 0.348 Net frac pressure (psi) 312 NA 312 312 319 Half length (ft) 463 NA 463 463 618 Upper frac height (ft) 160 NA 160 159 169 Lower frac height (ft) 137 NA 137 137 152 Total frac height (ft) 296 NA 296 296 321 Max frac width (in.) 0.534 NA 0.534 0.534 0.591 Average frac width (in.) 0.336 NA 0.336 0.336 0.371 Average surface concentration (lbm/ft^2) 1.46 NA 1.46 1.46 1.01 Upper tip location (TVD) (ft) 9510 NA 9510 9510 9500 Lower tip location (TVD) (ft) 9810 NA 9810 9810 9820 Treating pressure at reference depth (psi) 7860 NA 7860 7860 7870 Base pressure to calculate net pressure (psi) 7550 NA 7550 7550 7550 Dimensionless Productivity Index 0.943 NA 0.943 0.943 1.47 Dimensionless fracture conductivity 3.02 NA 3.02 3.02 12.51 Appendix B Base Case K1 10 times smaller flmult from 0.25 to 0.45 1000Psi increase k decrease 5 times Mass of proppant injected, lbm 400000 400000 400000 400000 400000 Mass proppant concentration in injected slurry (ppga) 10 10 10 10 10 Slurry volume injected (gal) 188080 326410 293070 178940 188080 Liquid volume injected (gal) 169980 308310 274970 160840 169980 Injection time (min) 127.95 222.04 199.37 121.73 127.95 Pad time (min) 75.07 188.12 143.66 68.25 75.07 Frac slurry efficiency 0.32 0.188 0.206 0.332 0.32 Net frac pressure (psi) 297 186.85 295.26 402.13 297 Half length (ft) 450 742.62 451.51 517.26 450 Upper frac height (ft) 320.58 361.35 323.29 182.79 320.58 Lower frac height (ft) 182.43 146.61 182.31 197.58 182.43 Total frac height (ft) 503.01 507.96 505.6 380.37 503.01 Max frac width (in.) 0.436 0.265 0.433 0.568 0.436 Average frac width (in.) 0.268 0.162 0.267 0.3 0.268 Average surface concentration (lbm/ft^2) 1.11 0.656 1.101 1.26 1.11 Upper tip location (TVD) (ft) 9349.4 9308.7 9346.7 9487.2 9349.4 Lower tip location (TVD) (ft) 9852.4 9816.6 9852.3 9867.6 9852.4 Dimensionless fracture conductivity 2.48 0.91 2.46 2.42 2.48 10 Appendix C Base Case K1 10 times smaller flmult from 0.25 to 0.45 1000Psi increase k decrease 5 times Mass of proppant injected, lbm 469400 22000 476000 467900 112400 Mass proppant concentration in injected slurry (ppga) 14 10 14 14 2 Slurry volume injected (gal) 150528 119070 186984 149352 127596 Liquid volume injected (gal) 129301 118079 165459 128197 122514 Injection time (min) 102.2 81 126.9 101.5 86.8 Pad time (min) 50.3 36.6 75.5 49.6 26.5 Frac slurry efficiency 0.41 0.37 0.32 0.41 0.52 Net frac pressure (psi) 1131 1058 1093 1132 1147 Half length (ft) 498 482 510 497 495 Upper frac height (ft) 157 142 158 156 159 Lower frac height (ft) 140 126 141 140 145 Total frac height (ft) 297 268 299 296 304 Max frac width (in.) 0.68 0.56 0.65 0.68 0.71 Average frac width (in.) 0.43 0.35 0.4 0.43 0.44 Average surface concentration (lbm/ft^2) 2.21 0.11 2.24 2.22 0.48 Upper tip location (TVD) (ft) 9514 9528 9512 9514 9510 Lower tip location (TVD) (ft) 9810 9796 9811 9810 9815 Treating pressure at reference depth (psi) 8287 8214 8249 8288 8303 Base pressure to calculate net pressure (psi) 7156 7156 7156 7156 7156 Dimensionless fracture conductivity 4.26 0.2 4.23 4.27 4.58 Shanshan Liu is a petroleum engineer in the Drilling Fluid Division of Great Wall Drilling Company, a subsidiary of China National Petroleum Corporation. She got both MBA and bachelor’s degree from Shenyang University, China. Abstract Introduction Input Data Study through M23 Study through MFrac Study through FracPro Conclusions Conflicts of Interest References Appendix A Appendix B Appendix C