DSpace at VNU: Using response surface design for optimizing operating conditions in recovering heavy oil process, Peace River oil sands

Journal of Petroleum Science and Engineering 117 (2014) 37–45 Contents lists available at ScienceDirect Journal of Petroleum Science and Engineering journal homepage: www.elsevier.com/locate/petrol Using response surface design for optimizing operating conditions in recovering heavy oil process, Peace River oil sands X.H Nguyen a,c, Wisup Bae a,n, Trianto Gunadi a, Yunsun Park b a Sejong University, 98 Gunja-dong, Gwangjin-ku, Seoul, South Korea Myungji University, Seoul, South Korea c Ho Chi Minh City University of Technology, Vietnam b art ic l e i nf o a b s t r a c t Article history: Received 15 March 2013 Accepted February 2014 Available online 16 March 2014 In order to maximize oil recovery with minimal environmental damages and lower production costs for producing heavy oil and bitumen resources in Peace River oil sands, optimal operating conditions were conducted by using design of experiment and response surface methodology This study was aimed to mitigate the risks of incomprehensive economic assessment and engineering in the process operation Simulation responses for 26 design points were estimated based on amount of oil recovery and net present value for each case Response surface methodology was applied to search for promising designs in contour plots and the response surface map The best operating conditions of the Fast-SAGD1 process were an injector–producer spacing of m, injection pressure of 6409 kPa, steam injection rate of 610 m3/d, subcool 1C for the SAGD system; while for the CSS well, injection pressure of 8333 kPa and a steam injection rate of 1007 m3/d In this study, the amount of oil recovery produced through the FastSAGD1 process increased significantly and appeared to be more effective than the conventional SAGD process In addition, it was observed that using lower injection pressures will not yield economical results in either SAGD or FastSAGD processes due to insufficient heat transfer from the steam into the solid bitumen in the reservoir, consequentially causing low NPV and oil recovery The results presented can be widely applied and are practical for the effective recovery of unconventional resources in Alberta's oil sands & 2014 Elsevier B.V All rights reserved Keywords: SAGD fast-SAGD operating condition NPV response surface method Introduction Heavy oil production from unconventional resources has been accelerating dramatically due to rising energy consumption needs Many heavy oil deposits are spread out over the United States, Russia, and various countries in the Middle East However, most of the world's heavy oil deposits are concentrated in two countries, Canada and Venezuela, where estimated heavy oil reserves approximately equal the world's total reserves of conventional crude oil From studies contributing to the development of Canadian oil sands reserves, it was shown that 44% of Canadian oil production was recovered from oil sands with an additional 18% being heavy crude oil, while light oil and condensate had declined to 38% of the total Oil sands represent as much as two-thirds of the world's total liquid hydrocarbon resources, with at least 1.7 trillion barrels in the Canadian oil sands (assuming 10% recovery with current technology) However, the extremely high viscosity of bitumen at normal reservoir temperature is one of the greatest challenges for its recovery process n Corresponding author E-mail address: wsbae@sejong.ac.kr (W Bae) http://dx.doi.org/10.1016/j.petrol.2014.02.012 0920-4105/& 2014 Elsevier B.V All rights reserved The steam-assisted gravity drainage process (SAGD) is an effective method for heavy oil and bitumen production utilizing two parallel horizontal wells, one above the other The top well functions as a steam injector, while the bottom well functions as the oil collector As steam is continuously injected in the upper well, a steam chamber forms in the reservoir and grows upwards to its surroundings, displacing heated oil following a gravity-mechanism drain into the producer (Butler, 2001) Studies in economics highlight the challenges and risks due to the high capital cost of initial investment for surface facilities, operating costs, and uncertainties related to oil and gas prices in the market The risk may be critical in SAGD operations when the operating design is unsuitable In order to predict reservoir performance and profitability, various scenarios using SAGD and Fast-SAGD processes were investigated and simulated The target is to maximize economics in operation, which will significantly affect the operating design of injector–producer spacing, injection pressure, steam injection rate, and subcool temperature Nestor et al (2002) developed a surrogate modeling-based optimization of the SAGD process to maximize oil recovery and net present value Gates and Chakrabarty (2005) used a genetic algorithm to define the optimal operating pressure Their results proposed that higher injection pressure should be applied in SAGD start-up until the growth of steam chamber contacts to the top 38 X.H Nguyen et al / Journal of Petroleum Science and Engineering 117 (2014) 37–45 Abbreviations ANOVA CCD CCFD CSOR CSS DOE analysis of variance central composite design central composite face-centred design cumulative steam oil ratio cyclic steam stimulation design of experiment reservoir, followed by lower injection pressures Yang et al (2009) presented economic optimization and uncertainty assessment strategies of SAGD operations using CMG's designed exploration and controlled evolution (DECE) optimization method The optimum operating conditions were obtained when using a high initial steam rate and high production rate to develop the steam chamber Polikar et al (2000), Gong et al., (2002), and Shin and Polikar (2007) conducted optimal operating conditions of the SAGD process by classical methods based on their numerical simulations and experiments However, there is a lack of confidence level in the optimized conditions because the significance level of operational parameters was not presented Also these studies ignored interactions along with their effects between the considered parameters Based on discrete simulation results and economic indicators, Shin and Polikar (2007) suggested that the operating conditions of the Peace River reservoir should have an injector– producer spacing of 15 m and a steam injection rate of 600 m3/d at a maximum injection pressure of 4510 kPa However, as stated before, these results did not mention the confidence level of the optimization process This might lead to low efficiency issues in field operation Moreover, the economic models in previous studies were not comprehensive enough, with limited consideration on only three factors steam cost, bitumen price and discount rate That approach could be accepted with low prices; however it may not be when extrapolated to future conditions These limitations of the classical method can be avoided by applying central composite design (CCD) and response surface methodology (RSM) that involve statistical design of experiments in which all factors are varied together over a set of experimental runs Hence, the optimal operating conditions should be reevaluated In this study, integrating central composite design and response surface methodology is necessary for indicating the optimal conditions for SAGD and Fast-SAGD processes It was aimed to mitigate the risk of incomprehensive economic assessment on the process operation The study started with the central composite facecentered design (CCFD) to screen variables, and then finding the optimal design based on the response surface method The efficient local optimization was employed by a two-stage approach First, an initial sample of design was obtained using the design of experiment technique (DOE) Simulation responses for 26 design points were estimated by oil recovery and NPV for each case Second, response surface methodology was used to determine the reasonable operating design, represented as contour plots and a response surface map Sensitivity analysis for variables influenced the net present value (NPV) and oil recovery Afterwards, the best choice of operating condition between SAGD and Fast-SAGD processes based on the technical-economic assessment are discussed 1.1 Reservoir model The SAGD and Fast-SAGD models are built on a bitumen reservoir of 101 m  900 m  25 m with no aquifer belonging to Bluesky formation, Peace River region The STARS module of Computer Modeling Group (CMG) software was used to investigate the effects Fast-SAGD fast-steam assisted gravity drainage IPS spacing between injector and producer IP steam injection pressure MSIR maximum steam injection rate NPV net present value RSM response surface methodology SAGD steam assisted gravity drainage Strap steam trap control/subcool temperature of operating parameters The SAGD process employed two horizontal wells, one injector located above the other oil producer Steam is injected continuously into the heated bitumen reservoir causing oil to flow into the producer Reservoir fluids were modeled as oleic, gaseous, and aqueous phases corresponding to three components of hydrocarbon, gas, and water, respectively The oleic phase is constituted of gas and oil, while the aqueous phase is composed of only water The gaseous phase encompasses steam and gas Grid, rocks, and fluid properties of the reservoir model are listed in Table NPV is considered the response variable to measure the degree of production efficiency during the operating process, and therefore this dependent variable is proposed in surface response correlation 1.2 SAGD economic model The economic model was designed based on the previous literatures in the Canadian National Energy Board reports (2006, 2008) The cash flow method in Microsoft Excel spreadsheet is calculated by NPV reflecting property depreciation with a 10% discount rate during 10 years of production phase The production profiles from simulation result included an annual oil rate, a steam injection rate and the amount of water produced The average prices of bitumen and gas are $70/bbl and $4.0/GJ, respectively SAGD pair cost is $5.0 mm, and the cost of a cyclic steam stimulation (CSS) single well is $2.5 mm in the Fast-SAGD process as capital investments Total operating costs covered the electric cost of $1.0 per barrel of produced oil, water handling cost of $3 per barrel, non-gas cost of $6/bbl, emission cost of $1/bbl and field insurance of $0.5/bbl The production profiles are combined with initial capital, operating costs, and the rate of return on capital to calculate the NPV The feasible project economics analysis enforces that at least 36,500 bbl/year must be operated per well pair in order for it to be economic (NPV Z0) The results also presented that the fluctuation of gas prices substantially affects project Table Reservoir properties Parameters Reservoir pressure, kPa Depth to top of reservoir, m Reservoir thickness, m Reservoir width, m Reservoir temperature, 1C Vertical permeability (Kv), D Permeability ratio (Kh/Kv) Porosity Oil saturation Oil viscosity, cp Rock thermal conductivity, kJ/m day 1C Bitumen thermal conductivity, kJ/m day 1C Gas thermal conductivity, kJ/m day 1C Rock heat capacity, kJ/m3 1C Steam quality 4500 630 25 101 18 1.95 0.334 0.28 0.8 220,000 660 11.5 2.6 2400 0.95 X.H Nguyen et al / Journal of Petroleum Science and Engineering 117 (2014) 37–45 operating variables such as spacing between injector and producer (IPS, X1), steam injection pressure (IP, X2), maximum steam injection rate (MSIR, X3), and subcool temperature (Strap, X4) For each of the four variables studied, high (coded value: ỵ1), middle (coded value: 0) and low (coded value: À 1) set points were selected according to the results in Table The four independent variables and their coded levels for using CCFD are presented in Table A number of tests required for the four independent variables are 26 cases to match each NPV response A full second-order polynomial model is accomplished by the multiple regression technique, and the effects of the interactions on the two-factors as well as the main factors included The objective function can be re-written as economics as well as production performance due to the related operating costs Response surface methodology and central composite design Response surface methodology is a statistical method based on the multivariant non-linear model which has been widely used in optimizing the process variables of operating conditions (Box and Wilson, 1951) Furthermore, RSM of designing experiments provide adequate and reliable measurements of the response, developing a mathematical model that has the best fit to the data obtained from the experimental design, and determining the optimal value of the independent variables that generates a maximum or minimum response (Myers et al., 2008) It is also useful in studying the interactions of the various processes affecting parameters In recent years, RSM has played an important role in oil field operations, especially in applications improving the oil recovery factor (Vanegas and Cunha, 2008) The experimental design techniques commonly used process analysis and modeling involve the full factorial, partial factorial and central composite rotatable designs An effective alternative to the factorial design is the central composite design, originally developed by Box and Wilson (1951), and improved upon by Box and Hunter (1957) The central composite design gives almost as much information as a three-level factorial, requires much fewer tests than the full factorial, and has been known to be sufficient in describing the majority of steady-state process responses Nowadays, CCD is the most popular class of designs used for fitting second-order models The total number of tests required for the CCD formula is (2kp ỵ 2k ỵn0), including k as the number of studied variables, 2k points fixed axially at a distance, p the fractionalisation element (full design, p ¼0) from the center to generate the quadratic terms, and replicate tests at the center (n0) Central composite face-centred design is performed to maximize the responses of NPV and the amount of oil recovery from Y ẳ ỵ X1 ỵ X ỵ X ỵ X ỵ 11 X 21 ỵ 22 X 22 ỵ 33 X 23 ỵ 44 X 24 ỵ 12 X X ỵ 13 X X ỵ 14 X1 X4 ỵ 23 X X ỵ 24 X X ỵ 34 X X 1ị The coefficients of the main effects βi – and two-factor interactions (βij) – were estimated from the experimental data by computer simulation programming utilizing the least squares method of @R 12.2.1 software Results and discussion The results showed that amount of oil recovery of case obtained the highest value with 2,318,793 bbls (Table 3), but NPV was lower than other cases due to an incompatible operating condition wherein a large number of steam was injected at high injection pressure While the highest NPV of case 22 was achieved at $54.6 mm under the experimental conditions of IPS 10 m, IP 6000 kPa, MSIR 720 m3/d, and subcool temperature 1C The coefficients of the model are calculated by the multiple regression analysis technique, and the quadratic model was given as follows: Y ẳ 52:85 0:4IPS ỵ 4:9IP þ 6:47MSIR À 1:1Strap À 1:6IPS:IPS À 7:8IP:IP À 7:2MSIR:MSIR 1:8Strap:Strap ỵ 2:6IPS:IP 0:67IPS:MSIR ỵ 0:93IPS:Strap ỵ 0:12IP:MSIR þ 0:94IP:Strap À 0:86MSIR:Strap Symbol Coded levels Injector/producer spacing, m Injection pressure (IP), kPa Steam injection rate, m3/d Subcool temperature, 1C IPS IP MSIR Strap À1 4500 300 0 10 6000 510 ð2Þ The statistical result of the multiple regression function was evaluated by the analysis of variance (ANOVA) in Table The R2 value closer to one represents a good correlation between the observed and predicted values The higher values of R2 (0.98) and adjusted R2 (0.95) also indicated the efficiency of the model suggesting that 98% and 95% variation could be accounted for by the model equation, respectively At the same time, a very low value of the coefficient of the residual standard deviation (RSD ¼2.39) clearly informed a high degree of precision and Table Variables and experimental design levels in the SAGD process Variables 39 ỵ1 16 7500 720 18 Table Central composite face-centered experimental design Case IPS (m) IP (kPa) MSIR (m3) Strap (1C) NPV (mm$) Cum oil (bbl) Case IPS (m) IP (kPa) MSIR (m3) Strap (1C) NPV (mm$) Cumulative oil (bbl) 10 11 12 13 À1 À1 À1 À1 À1 À1 À1 À1 À1 1 À1 À1 1 À1 À1 1 À1 À1 À1 À1 À1 1 1 À1 À1 À1 À1 À1 À1 À1 À1 À1 À1 À1 À1 1 1 26.5 21.3 30.6 35.4 45.9 33.3 45.1 45.6 26 21.3 30 35.4 33.8 1,431,325 1,240,048 1,547,243 1,646,577 2,081,735 1,738,825 2,318,793 2,312,011 1,424,022 1,240,048 1,513,647 1,646,577 1,581,006 14 15 16 17 18 19 20 21 22 23 24 25 26 À1 À1 0 0 0 0 À1 1 0 À1 0 0 0 1 0 0 À1 0 0 1 0 0 0 À1 0 30.4 42.9 48 50.1 52.2 38 51.9 36.6 54.6 52.6 49.4 53.3 53.3 1,608,844 1,991,283 2,119,005 2,189,048 2,190,790 1,867,387 2,206,455 1,690,708 2,253,399 2,263,499 2,075,359 2,205,948 2,205,948 40 X.H Nguyen et al / Journal of Petroleum Science and Engineering 117 (2014) 37–45 reliability of the experimental values and was in relation to the power of prediction, Q2 ¼0.86 Student's t-test of statistics is designed to evaluate quantitative effects of the main factors The regression coefficient values of Eq (2) are listed with standard errors and p-values in Table The p-value is used to check the significance of each coefficient, which in turn may indicate the pattern of interactions between variables It can be seen that the variables of IP and MSIR were significant, with very small p-values (p o0.05) It is noteworthy that a positive sign designates a synergistic effect, while a negative sign represents an antagonistic effect of a factor on the selected response 3.1 Effect of operating variables on NPV A Pareto chart presented standardized effects at p¼0.05 on the NPV responses (Fig 1) All the standardized effects were in absolute values to verify which were positives and negatives This helps to 3.2 Optimization of operating conditions by response surface methodology Table ANOVA for the response surface quadratic model NPV DF SS MS Total Constant Total corrected Regression Residual 25 24 14 10 41909.6 39219.8 2689.75 2632.59 57.1585 1676.38 39219.8 112.073 188.042 5.71585 N ¼ 27 Q2 ¼0.857 R2Adj ¼ 0.949 DF ¼ R2 ¼0.979 RSD ¼2.39 identify important variables that significantly affect the overall outcome of the NPV This graph was divided into two regions: the first region below zero is for negative coefficients of variables (IP.IP, MSIR.MSIR, Strap Strap, MSIR.Strap, IPS.IPS, Strap) that decrease the NPV due to an increase in combining of operating variables The second region above zero is the positive coefficients of individual and combining variables of the operating condition (MSIR, IP, IPS.IP, IP Strap, IPS Strap…) The increase in these variables leads to an increase to the NPV Based on this Pareto chart and the significant of regression coefficients, Table inferred the double interaction of IP and MSIR, the individuals of MSIR and IP were the most important variables that considerably affected the NPV of production performance Based on the nonlinear results shown in the sensitivity chart result, the reasonable operating condition to obtain the maximum NPV should be designed on the IPS of 9–10 m, IP of 6200–6400 kPa, MSIR of 550–600 m3/d, and subcool temperature in the range of 4–7 1C (Fig 2) F P SD 32.8984 10.5864 13.7128 2.39078 Conf lev.¼0.95 Table Estimated regression coefficients for NPV by using coded units NPV Estimate Standard error pvalue Constant IPS IP MSIR Strap IPSnIPS IPnIP MSIRnMSIR StrapnStrap IPSnIP IPSnMSIR IPSnStrap IPnMSIR IPnStrap MSIRnStrap 52.85 À 0.44 4.91 6.47 À 1.06 À 1.62 À 7.82 À 7.17 À 1.77 2.61 À 0.67 0.93 0.12 0.94 À 0.86 1.03 0.56 0.56 0.56 0.56 1.50 1.50 1.50 1.50 0.60 0.60 0.60 0.60 0.60 0.60 1.96E À 13 4.49E À 01 5.52E À 06 4.41E À 07 8.91E À 02 3.04E À 01 3.89E À 04 7.38E À 04 2.64E À 01 1.42E À 03 2.89E À 01 1.50E À 01 8.46E À 01 1.45E À 01 1.82E À 01 Fig The order ranking of factors affecting on the NPV Response surface methodology plays a key role in efficiently identifying the optimum values of the independent variables at which a dependent variable could arrive at the maximum response The 3D response surface and 2D contour plots are represented for prediction between the dependent variable and a set of independent variables The graphical representations of NPV contour plots and response surfaces are given in Fig 3, as injector–producer spacing, injection pressure, steam injection rate, and subcool temperature The value of the predicted maximum on the surface plot is confined in the smallest ellipse in the contour diagram Elliptical contours are generally obtained when there is a perfect interaction between the independent variables The values of independent variables in the smallest contour area and the corresponding maximum are determined to be the optimal operating conditions for the response variable From the response surface plots, the authors recognized that the best choice of operating conditions is located in the smallest region (red) in Fig 3, where the maximum NPV reached over $55.2 mm According to the regression coefficients significance of the quadratic polynomial model and gradient of slope in the 3-D response surface plot, almost all factors significantly affected the SAGD performance Therefore, the best operating conditions were designed using the injector producer spacing of m, steam injection rate of 610 m3/d, injection pressure of 6409 kPa, and subcool temperature 1C within the 95% confidence intervals 3.3 Validation of model equation Sensitivity analysis denoted that the injection pressure and the amount of injected steam were important control parameters in the success of an SAGD operation Scenarios 1, 2, 5, 6, 9, 10, 13, 14, and 19 exhibited that the responses of oil recovery and calculated NPV were quite low because injection pressure operated near the reservoir pressure (Table 3) Low injection pressure operation was not strong enough to push the heat of the steam chamber into the bitumen pay Therefore, the amount of oil recovered may decrease considerably The optimal operating condition was rechecked when the amount of oil recovered was obtained at 2,312,647 bbls in the simulated operation of the SAGD process The oil production rate peaked at 450,000 bbls in the second year, and then was dramatically reduced until the end of the 10th year of operation time (Fig.4c) This production profile will yield a NPV of $55.47 mm, the highest NPV among experimental cases in Table This result X.H Nguyen et al / Journal of Petroleum Science and Engineering 117 (2014) 37–45 41 Fig.2 Effect of operating conditions on the NPV proved that the validity of the RSM model is reasonably adequate to predict the production performance of the SAGD process 3.4 Optimization of operating conditions for Fast-SAGD process The Fast-SAGD model comprises of two full SAGD well pairs and two CSS wells (Polikar et al., 2000) It uses offset wells placed horizontally about 50 m away from the SAGD producer and each offset well These offset wells are operated alternatively as an injector and producer First, the SAGD horizontal pair begins to operate until the steam chamber grows to the top of the reservoir Then, CSS is carried out into the offset well until the steam chamber connects from the horizontal well pairs Afterwards it serves as an additional producer well This change occurs after cycles of CSS The operating conditions of SAGD wells in the Fast-SAGD process are designed using a similar approach to the SAGD system There are three essential cases in screening the operating conditions: 4510, 4600 and 6409 kPa injection pressures The responses of scenarios are listed in Table 42 X.H Nguyen et al / Journal of Petroleum Science and Engineering 117 (2014) 37–45 Fig Optimal operating conditions in the SAGD process (a) Contour plot and (b) response surface plot (For interpretation of the references to color in this figure legend, the reader is referred to the web version of this article.) Offset-well design: the usage of full-factorial design is to choose a reasonable operating condition (Myers et al., 2008) Two important parameters, injection pressure and maximum steam injection rate, are investigated in the range of 7000– 10,000 kPa and 700–1100 m3/d, respectively The CSS wells will start up after 250 days when the growth of the steam chamber touches the top of the reservoir A total of nine cases were generated by full-factorial design with the responses of cumulative oil and NPV, respectively (Table 6) The optimal points of the response surface methodology are determined in the vicinity of IP 8333 kPa and MSIR of 1007 m3/d (Table 7; Fig 4) 3.5 Technical-economic evaluation in the production performances of SAGD and Fast-SAGD processes For the SAGD system, different from the study by Shin and Polikar (2007), operating at low injection pressure as in the SAGD2 Fig Optimal operating conditions in Fast-SAGD process (a) Contour plot and (b) response surface plot case yields the lowest oil recovery The results imply that steam injection pressure should be operated at least at 4600 kPa for the operation to be efficient, and higher than 100 kPa compared to reservoir pressure Using response surface design satisfies in adapting the best operating requirements for achieving the maximum of both oil recovery and economic benefits Simulation results demonstrated that the Fast-SAGD process was significantly more advantageous than the SAGD process Although it incorporates increasing capital costs for additional offset in the Fast-SAGD system, both oil recovery and economic profit increase more dramatically than within the SAGD process, illustrated especially by the highest oil recovery in the Fast-SAGD1 case (Fig 5) In practice, low injection pressure such as in the SAGD2 case is significantly less efficient than the Fast-SAGD2 case, and the responses were not as expected Additionally, the minor difference of 10 kPa between steam injection pressure and reservoir pressure is insufficient to increase the amount of oil recovery in both Fast-SAGD2 and SAGD2 operations X.H Nguyen et al / Journal of Petroleum Science and Engineering 117 (2014) 37–45 43 Table Full-factorial design level of CSS well in the Fast-SAGD process Variables Symbol Coded levels Injection pressure (IP), kPa Steam injection rate, m3/d IP 7000 8500 ỵ1 10,000 MSIR 700 900 1100 CSS well Fast-SAGD Case IP MSIR NPV ($mm) Cum.oil (bbl) À1 À1 À1 0 À1 À1 1 0 À1 60.13 60.64 62.15 61.84 61.71 62.05 60.43 61.97 62.12 2,362,957 2,360,686 2,394,309 2,393,944 2,375,721 2,384,329 2,359,835 2,394,429 2,384,235 MSIR (m3) Cum oil (bbl) NPV ($mm) 1007 2,396,284 62.33 Optimal operating condition IP (kPa) 8333 Table Optimization of operating conditions in the SAGD and Fast-SAGD processes Case SAGD1 (CCF) SAGD2 (low IP) SAGD3 (low IP) IPS, m IP, kPa 6409 10 4510 10 4600 610 101 100 2,312,647 55.47 55.30 610 101 120 2,016,199 43.00 610 101 120 2,208,805 53.47 MSIR, m /d Strap, 1C WPS, m Preheating, day Cumulative oil, bbl NPV (actual), $mm NPV (predicted), $mm Fast-SAGD1 Fast-SAGD2 Fast-SAGD3 6409 (SAGD) 8333 (CSS) 1007 51 (CSS) 4510 (SAGD) 8333 (CSS) 1007 51 (CSS) 8333 (CSS) 1007 51 (CSS) 2,396,283 62.33 2,325,890 53.95 2,383,835 62.89 Fig Cumulative oil Fig Cumulative steam oil ratio Fig exhibits the cumulative steam oil ratio (CSOR) performance change with production time As is known, the CSOR parameter is closely related to steam cost, oil recovery, and NPV A high CSOR means that a large amount of steam volume is injected into the reservoir and this directly affects operating costs This is similar to steam cost, which depends on the change of gas prices of the following season If CSOR increases, oil recovery will accelerate to a certain peak, but at the same time, the economic benefit will be reduced because of an increase in steam cost It was evidenced that the highest CSOR value of Fast-SAGD1 case was 5.2 with high oil recovery (Table 7; Figs and 6) However, the NPV of Fast-SAGD3 case was larger by a fraction than the NPV of Fast-SAGD1 case as a result of lower thermal efficiency In other cases, although low CSOR value is less than or equal to 4.0, both the amount of oil recovery and NPV reduces significantly Examples are shown in the cases of SAGD2, SAGD3, and Fast-SAGD2 44 X.H Nguyen et al / Journal of Petroleum Science and Engineering 117 (2014) 37–45 Accordingly, reasonable control for injected steam mass and injection pressure is an important requirement that determines the success of an operation Production performance with time is illustrated in Fig In the first years, the amount of oil recovery through the Fast-SAGD process is much higher than the conventional SAGD process, estimated to be equal to half of the Fast-SAGD process This is quite a convenience because of payback and profit acquired in a short period of time For this reason, the NPV of the Fast-SAGD cases are greater than the SAGD cases However, the fluctuation of oil and gas prices decides what kind of operation is the best solution In economic aspects of cash flow with time, the minor difference of NPV responses between Fast-SAGD1 and Fast-SAGD3 Fig Oil rate are considered as equivalent Consequently, the Fast-SAGD1 case is the best choice for an operating condition because of its high oil recovery Fig expressed the steam chamber performances in conventional SAGD and Fast-SAGD processes The steam temperature should achieve at least 200 1C, which will heat up the bitumen and decrease oil viscosity enough to be efficiently produced by the producer well For SAGD models, the steam chamber growth contact to the top reservoir occurs after 250 days, 520 days, and 400 days corresponding to the cases of SAGD1, SAGD2, and SAGD3 respectively Afterwards, the CSS well will start to operate in accordance to the Fast-SAGD system The steam chambers of SAGD1 and SAGD3 cases grow in the vertical direction during years, then expand towards the horizontal direction At that time, an increase of heat loss to the overburden and underburden occurs Meanwhile, steam chamber growth of SAGD2 case took a longer time (3 years) to touch the top of the reservoir due to the low injection pressures Therefore the lowest NPV was obtained using the SAGD2 case and this is mentioned in Table The vertical growth of the steam chamber of the CSS well to the top of the reservoir in the Fast-SAGD1 case (520 days) is earlier than both the Fast-SAGD3 (620 days) and Fast-SAGD2 (750 days) cases The steam chambers contact between the SAGD well-pair and CSS well occurred after 600 days for Fast-SAGD1, 750 days for Fast-SAGD3, and 1050 days for Fast-SAGD3 After this time period, heat loss at overburden and underburden will cause higher CSOR and reduce oil recovery The results showed that the operation selection of Fast-SAGD1 will yield the highest oil recovery and harbors an economical efficiency much greater than other cases Nevertheless, this study showed that the application of DOE and RSM were useful and extremely efficient with a high confidence level for finding the Fig Scenario-based time lapse of steam chamber growth in Peace River oil sands X.H Nguyen et al / Journal of Petroleum Science and Engineering 117 (2014) 37–45 optimal operating conditions to maximize profit in a Fast-SAGD process Conclusions This paper showed that the DOE and response surface methodology was employed successfully for indicating operation parameters of SAGD and Fast-SAGD processes Predicted values from the regression function are found to be in good agreement with observed values of R2 at 0.98 and Q2 at 0.86 for the NPV response In order to gain a better understanding the effect of the variables on the SAGD performance and optimal operating conditions, responses are presented on the 2-D and 3-D response surface map For maximizing oil recovery, the best solutions were found on the Fast-SAGD1 scenarios where operating conditions should be designed with injector producer spacing of m, injection pressure of 6409 kPa, steam injection rate of 610 m3/d, subcool temperature of 1C in SAGD operation, CSS is operated at an injection pressure of 8333 kPa, and injection steam of 1007 m3/d with three cycles Results showed that the minor difference of 10 kPa between steam injection pressure and reservoir pressure was insufficient to increase the NPV and oil recovery as Fast-SAGD2 and SAGD2 operations It is recommended that injection pressure should be operated at least at 100 kPa higher than the initial reservoir pressure for the process to be more efficient This study not only supports a decision-making methodology to choose the best technical solution, but also provides an economical way of obtaining the maximum profit in a short period with the fewest number of experiments Acknowledgments This work was supported by the Energy Resources R&D program of the Korea Institute of Energy Technology Evaluation 45 and Planning (KETEP) grant funded by the Korea government Ministry of Knowledge Economy (No 2012T100201728) Moreover, the authors wish to thank Schlumberger K.K for the encouragement of writing this paper References Box, G.E.P., Wilson, K.B., 1951 On the experimental attainment of optimum conditions J R Stat Soc.: Ser A (Gen.) 13, 1–45 Box, G.E.P., Hunter, J.S., 1957 Multi-factor experimental designs for exploring response surfaces Ann Math Stat 28 (1), 195–241 Butler, R.M., 2001 Some recent development in SAGD J Can Pet Technol., Disting Author Ser 40 (1), 18–22 Canada's Oil sands: Opportunities and Challenges to 2015, ver.2006, 2008 Canadian National Energy Board Gates, I.D., Chakrabarty, N., 2005 Optimization of Steam-Assisted Gravity Drainage (SAGD) in Ideal McMurray Reservoir Paper 2005193 Presented at the Canadian International Petroleum Conference, Calgary, June 7–9 Gong J., Polikar M., Chalaturnyk R.J., 2002 Fast SAGD and Geomechnical Mechanism Paper CIPC 2002-163 Presented at the Canadian International Petroleum Conference Calgary, Canada, 11–13 June Myers, R.H., Montgomery, D.C., Anderson-Cook, C., 2008 Response Surface Methodology: Process and Product Optimization Using Designed Experiments, 3rd ed John Wiley and Sons, New York, pp 13–135 Nestor, V.Q., Javier, V.G., Salvador, P., 2002 Surrogate modeling-based optimization of SAGD processes J Pet Sci Eng 35 (1–2), 83–93 Polikar, M., Cyr, T.J., Coates, R.M., 2000 Fast SAGD: Half the Wells and 30% Less Steam Paper SPE 65509, Presented at the International Conference on Horizontal Well Technology Calgary, Canada, 6–8 November Shin, H., Polikar, M., 2007 Review of reservoir parameters to optimize SAGD and Fast-SAGD operating conditions J Can Pet Technol 46 (1), 35–41 Vanegas, J.W., Cunha, L.B., 2008 Prediction of SAGD performance using response surface correlations developed by experimental design techniques J Can Pet Technol 47, 58–64, (Petroleum Society of Canada) Yang, C., Card, C., Nghiem, L., 2009 Economic optimization and uncertainty assessment of commercial SAGD operations J Can Pet Technol 48 (9), 33–40  ... optimal operating conditions should be reevaluated In this study, integrating central composite design and response surface methodology is necessary for indicating the optimal conditions for SAGD... pressure Using response surface design satisfies in adapting the best operating requirements for achieving the maximum of both oil recovery and economic benefits Simulation results demonstrated that the... (DOE) Simulation responses for 26 design points were estimated by oil recovery and NPV for each case Second, response surface methodology was used to determine the reasonable operating design, represented