1. Trang chủ
  2. » Thể loại khác

DSpace at VNU: Prediction of the viscosity of water-in-oil emulsions

7 134 0

Đang tải... (xem toàn văn)

THÔNG TIN TÀI LIỆU

Thông tin cơ bản

Định dạng
Số trang 7
Dung lượng 652,9 KB

Nội dung

Petroleum Science and Technology ISSN: 1091-6466 (Print) 1532-2459 (Online) Journal homepage: http://www.tandfonline.com/loi/lpet20 Prediction of the viscosity of water-in-oil emulsions Saeid Nasery, Seyedahmad Hoseinpour, Le Thi Kim Phung & Alireza Bahadori To cite this article: Saeid Nasery, Seyedahmad Hoseinpour, Le Thi Kim Phung & Alireza Bahadori (2016) Prediction of the viscosity of water-in-oil emulsions, Petroleum Science and Technology, 34:24, 1972-1977 To link to this article: http://dx.doi.org/10.1080/10916466.2016.1233248 Published online: 09 Dec 2016 Submit your article to this journal Article views: View related articles View Crossmark data Full Terms & Conditions of access and use can be found at http://www.tandfonline.com/action/journalInformation?journalCode=lpet20 Download by: [Gazi Universitesi] Date: 11 December 2016, At: 04:01 PETROLEUM SCIENCE AND TECHNOLOGY , VOL , NO , – http://dx.doi.org/./.. Prediction of the viscosity of water-in-oil emulsions Saeid Naserya , Seyedahmad Hoseinpourb , Le Thi Kim Phungc , and Alireza Bahadorid,e a Young Researchers and Elite Club, Ahvaz Branch, Islamic Azad University, Ahwaz, Iran; b Department of Petroleum Engineering, Petroleum University of Technology, Ahwaz, Iran; c Department of Chemical process and Equipment, Faculty of Chemical Engineering, Hochiminh City University of Technology, Hochiminh City, Vietnam; d Southern Cross University, School of Environment, Science and Engineering, Lismore, Australia; e Australian Oil and Gas Service Pty Ltd, Lismore, Australia ABSTRACT KEYWORDS Water-in-oil (W/O) emulsions occur in different parts of petroleum recovery and since their properties are different from those of crude oil and water, it is essential to find information about their physical properties The water causes some problems and heightens the cost of oil production Separation of water from oil, water treatment, and disposal of water are the steps that make oil production costly Among various physical properties of W/O emulsions, viscosity is the most important Because of droplet crowding, emulsions indicate nonNewtonian behavior, which is why their viscosity is higher than the viscosity of oil and water As a result, it is essential to have knowledge about the viscosity of W/O emulsions in order to extract and process petroleum properly In this study, an intelligent model, namely a COA-LSSVM, is presented to accurately prognosticate the viscosity of W/O emulsions The results indicated that the values estimated by the developed model is in great consistency with the laboratory data by R2 of 0.9972 and MSE of 1.1762 ∗ 10−5 A comparison with another model, which was recently introduced, revealed that the developed model in this study is superior Cuckoo optimization algorithm; least square support vector machine; viscosity of water-oil emulsion; water-oil emulsion Introduction Water-in-oil (W/O) emulsions occur in different parts of petroleum recovery and since their properties are different from those of crude oil and water, it is essential to find information about their physical properties The amount of water in W/O emulsion varies, and it can be less than 1% or more than 80% The water causes some problems and heightens the cost of oil production Separation of water from oil, water treatment, and disposal of water are the steps that make the oil production costly In addition, crude oil should be in compliance with specific product rules for sale; as a result, separation of water from oil is an essential process (Becher, 1983; Schramm, 1992) It is difficult to treat emulsions and also some operational difficulties in gas/oil separating plants and wet-crude handling facilities may result from emulsions Because of W/O emulsions, pressure decreases highly in flow lines, which in return leads to a rise in demulsifier use, and also sometimes trips or upsets are created in wet-crude handling facilities During winter, due to lower surface temperatures, emulsions are more problematic (Schramm, 1992) The tendency for emulsifying is different in crude oils Some of crude oils not emulsify or their emulsions are loose; consequently, their separation is not difficult On the other hand, others form stable CONTACT Alireza Bahadori alireza.bahadori@hotmail.com Southern Cross University, School of Environment, Science and Engineering, PO Box , Lismore, NSW, Australia Color versions of one or more of the figures in the article can be found online at www.tandfonline.com/lpet ©  Taylor & Francis Group, LLC PETROLEUM SCIENCE AND TECHNOLOGY 1973 emulsions and their separations are difficult In an untreated emulsion, if there is a sufficient difference between the oil and the water, an amount of water will be separated from the oil through natural coalescence and settling In order to separate the water from oil completely, some methods of treatment should be utilized, otherwise a small amount of water will remain in the oil (Schramm, 1992; Manning and Thompson, 1994) Among various physical properties of W/O emulsions, viscosity is the most important Because of droplet crowding, emulsions indicate non-Newtonian behavior, which is why their viscosity is higher than the viscosity of either the oil or the water In in situ thermal recovery, heavy oils are extracted from reservoirs in the form of W/O emulsion Because emulsion is more viscous than crude oil and water, having knowledge about emulsion viscosity is crucial in order to estimate oil recovery (Chung and Butler, 1988) Several researchers have conducted some experiments for W/O emulsions, and several models have been developed based on these experiments Doing experimental researches is expensive and time consuming; as a result, it is essential to propose some models to predict W/O emulsions (Farah et al., 2005; Li et al., 2016) Einstein (1911) developed a model for prediction of emulsion viscosity that can be used for dilute suspension system This model indicates that there is a positive linear relationship between the relative viscosity of emulsion and volume fraction of dispersed phase Taylor (Taylor 1932) extended the work of Einstein and considered the effect of both dispersed and continuous phase in order to improve the accuracy of emulsion viscosity prediction This model is applicable for emulsion systems in which the concentration of dispersed droplets is low Brinkman developed a model to prognosticate the viscosity of emulsions with spherical surface droplets Krieger and Dougherty (1959) proposed a correlation that is applicable for emulsions in which the dispersed phase has high concentration Pal (1998) developed a theoretical model to predict the viscosity of both Newtonian and non-Newtonian fluids This model uses the flocculation of dispersed droplets and hydration effect Richardson (1950) mentioned that there is exponential relationship between emulsion viscosity and volume fraction of dispersed phase Ronningsen (1995) suggested a model as function of the water cut and the temperature to predict the viscosity of W/O emulsions Krynke and Sek (2004) introduced three models with specific range of applicability for forecasting the viscosity of emulsions In these models, volume fraction and droplet size of the inner phase and viscosity and density of the outer phase are considered as the parameters that affect the viscosity of emulsions Farah et al (2005) extended the ASTM equation (ASTM International, 2001), method D-341, which describes viscosity as a function of temperature, in order to incorporate the changes of dispersed phase volume fraction Their correlation showed satisfactory accuracy for prediction of the viscosity of W/O emulsions Li et al (2016) modified the model proposed by Ronningsen (1995) to prognosticate the W/O emulsions viscosity They showed that their modified model indicates better performance in comparison with previous models Although the previously mentioned models are valuable, more investigations should be carried out in order to develop simple model with high accuracy Recently, some artificial intelligence models are adopted with various researchers in order to predict different parameters with great precision in petroleum industry Naseri et al (2015) applied radial basis function artificial neural network in order to predict asphaltene precipitation Their model showed great consistency between the output of the model and experimental data Naseri et al (2016) predicted choke flow coefficients with high accuracy using radial basis function artificial neural network Least squares support vector machine (LSSVM) method was utilized by Hemmati-Sarapardeh et al (2014) for prediction of reservoir oil viscosity This model also prognosticated the experimental data accurately Many other researches can be found in the literature that use artificial intelligence models in order to predict experimental data with high accuracy (Tatar et al., 2015a; Tatar et al., 2015b; Tatar et al., 2016a; Tatar et al., 2016b) Due to the great ability of expert systems for prediction, the Cuckoo optimization algorithm LSSVM (COA-LSSVM) is used in order to prognosticate the viscosity of W/O emulsions After development of the model using the dataset provided by Li et al (2016), its performance will be assessed by different parameters and graphs 1974 S NASERY ET AL Details of intelligent model 2.1 SVM strategy The SVM is a consistent and efficient approach developed from the machine-learning community (Suykens and Vandewalle, 1999; Eslamimanesh et al., 2012) An SVM is capable of analyzing data and recognizing patterns and can be utilized to analyze regression When an LSSVM method with RBF kernel function is used, two tuning parameters should be obtained by minimizing the deviation of the LSSVM model from laboratory data It is valuable to mention that the main merit of LSSVM over the SVM is the modification of the inequality constraints of Eq (3) to the equality constraint of Eq (1) 2.2 COA COA is an evolutionary optimization algorithm proposed by Rajabioun (2011) The idea of this optimization method is the behavior of a cuckoo bird Similar to other evolutionary algorithms, an initial population of cuckoos is used in order to initiate the model Cuckoos try to find the best places in which more eggs survive When remaining eggs grow and mature cuckoos are born, some societies are made by them Each society lives in a specific region and the best habitat will be chosen by the cuckoos of other societies Egg-laying radii is determined by the number of eggs of cuckoos and the distance of cuckoos to the best habitat After that time, eggs are laid by cuckoo in several random nests in the egg-laying radius Until the best place with maximum profit value is achieved and the majority of cuckoo population immigrates to that place, the process keeps on (Rajabioun, 2011) Result and discussion 3.1 Data acquisition It is necessary to use valid data that includes a wide range of variables in order to build an accurate and broad model In this study, the viscosity of W/O emulsion is obtained from the open literature (Li et al., 2016) The proposed model, a COA-LSSVM, aims to prognosticate the viscosity of W/O emulsion precisely At this model, temperature, dispersed phase volumetric fraction, shear rate, and oil properties including API gravity, total acid number (TAN), resins content, asphaltenes content, carbon residue, ash content, and sulfur content are the inputs of the developed model and the viscosity of W/O emulsion is the output In this data set, there are three types of petroleum and their characteristics are mentioned in Table 3.2 Model development Matlab 2014 (The MathWorks, Natick, MA) was utilized to construct the suggested COA-LSSVM model This model has two important tuning parameters, namely regularization factor (γ ) and kernel sample variance (σ ) The optimum values of the tuning parameters should be specified for minimization of the Table  Properties of the crude oil samples (Li et al., ) API gravity @ °F TAN Resin content Asphaltenes content Carbon residue Ash content Sulfur content Crude oil A Crude oil B Crude oil C . . . . . . . . . . . . . . . . . . . . . PETROLEUM SCIENCE AND TECHNOLOGY 1975 Table  Comparison with model presented by Li et al () COALSSVM Li et al () Train Data Test Data All data Crude oil A Crude oil B Crude oil C R MSE . . . . . . .∗ − .∗ − .∗ − — — — mean square error (MSE) between the target values and the model’s output To obtain these optimum values, trial and error can be used by tuning each of them and assessing the MSE; however, it is better to apply an optimization algorithm In this model, COA is used as an evolutionary optimization method to find the optimal values of the tuning parameters At first, train dataset and test dataset were selected in a random manner from the dataset To construct the model, the training dataset was adopted Also, test dataset was used to evaluate the performance of the model in prediction of unseen data The optimum values for σ and γ are determined to be 0.01532 and 954.797, respectively 3.3 Accuracy of the proposed model and validation Statistical and graphical methods were used to assess the accuracy of the developed model The developed COA-LSSVM model was compared with the research presented by Li et al (2016) Table shows the parameters for the train, test, and total dataset In addition, in order to indicate the precision of the proposed model graphically, two figures are presented The first one, Figure 1, represents the crossplot of the predicted data versus laboratory data In this figure, the vertical axis is the output of the model and the horizontal axis is the laboratory data To achieve an excellent accuracy, most of the Figure  Crossplot of experimental data versus predicted data by the proposed COA-LSSVM 1976 S NASERY ET AL Figure  Relative error deviation between real and predicted values for COA-LSSVM data points should be located near the 45° line Based on Figure 1, it is obvious that the predicted and experimental data are in great consistency Another useful plot to evaluate the performance of the proposed model is the relative error deviation (Figure 2) On the basis of this plot, it can be found that the majority of the data points have error deviation of less than 0.1 Finally, the introduced model is compared with the model presented by Li et al (2016) For conducting the comparison, the values of R2 were used It should be noted that Li et al (2016) carried out the experiments on three kinds of petroleum and presented a separate model for each kind of petroleum However, in this study, a general model is presented using the data of three types of petroleum and it can be used to predict the viscosity of different W/O emulsions at various temperatures and pressures In addition, based on the Table 2, it is evident that the proposed COA-LSSVM model is superior to the model presented by Li et al (2016) Conclusion In this study, COA-LSSVM model was successfully developed to forecast the viscosity of W/O emulsions The dataset presented by Li et al (2016) was used to develop and validate the model The dataset included three types of petroleum; as a result, it is a general model that can be used for petroleum with different properties A total of 587 data points were used for the development of the model and 147 data points were used for the validation of the model Petroleum properties including API gravity, total acid number (TAN), resins content, asphaltenes content, carbon residue, ash content, and sulfur content, along with temperature, dispersed phase volumetric fraction, and shear rate are the inputs of the model The results of the model show that the introduced model estimates the viscosity of W/O emulsions In addition, the comparison between the proposed model and the models presented by Li et al (2016) indicates that COA-LSSVM model has better performance in predicting the viscosity of W/O emulsions References ASTM International (2001) Annual book of ASTM standards, Vol 03.01: metals-mechanical testing; elevated and lowtemperature tests metallography 399-90 (Reapproved 1997) West Conshohocken, PA: ASTM Becher, P (1983) Encyclopedia of emulsion technology, vol 1: basic theory New York, NY: Marcel Dekker Brinkman, H C (1952) The viscosity of concentrated suspensions and solutions J Chem Physics 20:571–571 Chung, K., and Butler, R (1988) Geometrical effect of steam injection on the formation of emulsions nn the steam-assisted gravity drainage process J Can Pet Technol 27:36–42 PETROLEUM SCIENCE AND TECHNOLOGY 1977 Einstein, A (1911) Elementary consideration of the thermal conductivity of dielectric solids Ann Phys 34:591 Eslamimanesh, A., Gharagheizi, F., Illbeigi, M., Mohammadi, A H., Fazlali, A., and Richon, D (2012) Phase equilibrium modeling of clathrate hydrates of methane, carbon dioxide, nitrogen, and hydrogen+ water soluble organic promoters using support vector machine algorithm Fluid Phase Equilib 316:34–45 Farah, M A., Oliveira, R C., Caldas, J N., and Rajagopal, K (2005) Viscosity of water-in-oil emulsions: Variation with temperature and water volume fraction J Pet Sci Eng 48:169–184 Hemmati-Sarapardeh, A., Shokrollahi, A., Tatar, A., Gharagheizi, F., Mohammadi, A H., and Naseri, A (2014) Reservoir oil viscosity determination using a rigorous approach Fuel 116:39–48 Krieger, I., and Dougherty, T (1959) Concentration dependence of the viscosity of suspensions Trans Soc Rheol 3:137– 152 Krynke, K K., and Sek, J P (2004) Predicting viscosity of emulsions in the broad range of inner phase concentrations Colloids Surf A: Physicochem Eng Aspects 245:81–92 Li, C., Huang, Q., Ma, S., and Ji, C (2016) An experimental study on the viscosity of water-in-oil emulsions J Dispers Sci Technol 37:305–316 Manning, F., and Thompson, R (1994) Oilfield processing (Vol 2) Tulsa, OK: PennWell Naseri, S., Tatar, A., Bahadori, M., Rozyn, J., Kashiwao, T., and Bahadori, A (2015) Application of radial basis function neural network for prediction of titration-based asphaltene precipitation Pet Sci Technol 33:1875–1882 Naseri, S., Tatar, A., and Shokrollahi, A (2016) Development of an accurate method to prognosticate choke flow coefficients for natural gas flow through nozzle and orifice type chokes Flow Measur Instrument 48:1–7 Pal, R (1998) A novel method to correlate emulsion viscosity data Colloids Surf A: Physicochem Eng Aspects 137:275–286 Pelckmans, K., Suykens, J A., Van Gestel, T., De Brabanter, J., Lukas, L., Hamers, B., De Moor, B., and Vandewalle, J (2002) LS-SVMlab: a matlab/c toolbox for least squares support vector machines Tutorial KULeuven-ESAT Leuven, Belgium Rajabioun, R (2011) Cuckoo optimization algorithm Appl Soft Comput 11:5508–5518 Richardson, E (1950) The formation and flow of emulsion J Colloid Sci 5:404–413 Ronningsen, H P (1995) Correlations for predicting viscosity of W/O-emulsions based on North Sea crude oils SPE 28968, SPE International Symposium on Oilfield Chemistry, San Antonio, Texas, February 14–17 Schramm, L L (1992) Fundamentals and applications in the petroleum industry Adv Chem 231:3–24 Suykens, J A., and Vandewalle, J (1999) Least squares support vector machine classifiers Neural Process Lett 9:293–300 Tatar, A., Naseri, S., Bahadori, M., Hezave, A Z., Kashiwao, T., Bahadori, A., and Darvish, H (2016a) Prediction of carbon dioxide solubility in ionic liquids using MLP and radial basis function (RBF) neural networks J Taiwan Instit Chem Eng 60:151–164 Tatar, A., Naseri, S., Bahadori, M., Rozyn, J., Lee, M., Kashiwao, T., and Bahadori, A (2015a) Evaluation of different artificial intelligent models to predict reservoir formation water density Pet Sci Technol 33:1749–1756 Tatar, A., Naseri, S., Sirach, N., Lee, M., and Bahadori, A (2015b) Prediction of reservoir brine properties using radial basis function (RBF) neural network Petroleum 1:349–357 Tatar, A., Nasery, S., Bahadori, M., Bahadori, A., Bahadori, M., Barati-Harooni, A., and Najafi-Marghmaleki, A (2016b) Prediction of water removal rate in a natural gas dehydration system using radial basis function neural network Pet Sci Technol 34:951–960 Taylor, G I (1932) The viscosity of a fluid containing small drops of another fluid Proc R Soc Lond Ser A 138:41–48 ... minimizing the deviation of the LSSVM model from laboratory data It is valuable to mention that the main merit of LSSVM over the SVM is the modification of the inequality constraints of Eq (3) to the. .. with temperature, dispersed phase volumetric fraction, and shear rate are the inputs of the model The results of the model show that the introduced model estimates the viscosity of W/O emulsions. .. viscosity of emulsions In these models, volume fraction and droplet size of the inner phase and viscosity and density of the outer phase are considered as the parameters that affect the viscosity of emulsions

Ngày đăng: 16/12/2017, 17:41

TÀI LIỆU CÙNG NGƯỜI DÙNG

TÀI LIỆU LIÊN QUAN