به کارگیری الگوریتم‌های بهینه‌سازی و پارامتری سازی در تلفیق داده‌های لرزه‌ای و نگاره‌ای چاه‌ها در فرایند ساخت و به روزرسانی مدل‌های رخساره‌ای

ﺷﻨﺎﺧﺖ ﺧﺼﻮﺻﯿﺎت ﻓﯿﺰﯾﮑﯽ ﯾﮏ ﻣﺨﺰن ﻫﯿﺪروﮐﺮﺑﻨﯽ اﻣﺮی ﻏﯿﺮﻗﺎﺑﻞ اﺟﺘﻨﺎب در ﻣﺪﯾﺮﯾﺖ ﻣﺨﺰن ﻃﯽ ﻣﺮاﺣﻞ ﻣﺨﺘﻠﻒ ﻋﻤﺮ آن ﻣﯽﺑﺎﺷﺪ. ﺣﺼﻮل ﺷﻨﺎﺧﺖ ﮐﺎﻓﯽ از ﻣﺨﺰن در ﮔﺮو ﺗﻠﻔﯿﻖ ﻣﻨﺎﺑﻊ ﻣﺨﺘﻠﻒ اﻃﻼﻋﺎﺗﯽ در ﻓﺮآﯾﻨﺪ ﻣﺪلﺳﺎزی ﮐﺎﻣﭙﯿﻮﺗﺮی ﻣﺨﺰن ﻣﯽ ﺑﺎﺷﺪ. در اﯾﻦ راﺳﺘﺎ اﯾﻦ ﭘﮋوﻫﺶ ﻣﺴﺌﻠﻪ ﺗﻠﻔﯿﻖ دادهﻫﺎی ﻧﮕﺎرﻫﺎی ﭼﺎهﻫﺎ و ﻟﺮزهای دوﺑﻌﺪی/ﺳﻪﺑﻌﺪی در ﻓﺮآﯾﻨﺪ ﻣﺪلﺳﺎزی رﺧﺴﺎرهای ﻣﺨﺰن را ﻣﻮرد ﺑﺮرﺳﯽ ﻗﺮار داده اﺳﺖ. ﺑﻪ اﯾﻦ ﻣﻨﻈﻮر دو روش از دﺳﺘﻪ روشﻫﺎی ﻣﻮﺳﻮم ﺑﻪ ﭼﺮﺧﻪ اﻧﻄﺒﺎق ﺑﺎ داده ﻫﺎی ﻟﺮزهای ﻣﻌﺮﻓﯽ ﺷﺪه اﺳﺖ. در روش اول، از اﻟﮕﻮرﯾﺘﻢ ﺑﻬﯿﻨﻪﺳﺎزی ازدﺣﺎم ذرات ﺑﻪ ﻣﻨﻈﻮر ﭘﯿﺪا ﮐﺮدن ﻣﻘﺪار ﺑﻬﯿﻨﻪ ﭘﺎراﻣﺘﺮ ﺗﻐﯿﯿﺮ روش آﺷﻔﺘﮕﯽ اﺣﺘﻤﺎل اﺳﺘﻔﺎده ﺷﺪه اﺳﺖ. ﺑﻪﮐﺎرﮔﯿﺮی روش آﺷﻔﺘﮕﯽ اﺣﺘﻤﺎل ﺑﻪ ﻣﻨﻈﻮر ﺗﺒﺪﯾﻞ ﻣﺴﺌﻠﻪ ﺑﻬﯿﻨﻪﺳﺎزی ﺑﺎ n ﭘﺎراﻣﺘﺮ ﺑﻪ ﯾﮏ ﻣﺴﺌﻠﻪ ﺑﻬﯿﻨﻪﺳﺎزی ﺑﺎ ﯾﮏﭘﺎراﻣﺘﺮ ﻣﯽﺑﺎﺷﺪ. در روش دوم، در ﻏﯿﺎب روشﻫﺎی ﭘﺎراﻣﺘﺮیﺳﺎزی، ﻣﺴﺌﻠﻪ ﺑﻪروزرﺳﺎﻧﯽ ﻣﺪلﻫﺎی رﺧﺴﺎرهای، ﯾﮏ ﻣﺴﺌﻠﻪ ﺑﻬﯿﻨﻪﺳﺎزی ﺑﺎ n )ﺗﻌﺪاد ﺳﻠﻮلﻫﺎی ﻣﺪل ﻣﺨﺰن( ﭘﺎراﻣﺘﺮ ﻣﺠﻬﻮل ﺧﻮاﻫﺪ ﺑﻮد. واﺿﺢ اﺳﺖ ﺑﺎ اﻓﺰاﯾﺶ ﺗﻌﺪاد ﭘﺎراﻣﺘﺮﻫﺎی ﻣﺠﻬﻮل ﺑﻬﯿﻨﻪﺳﺎزی، دﻗﺖ اﻟﮕﻮرﯾﺘﻢﻫﺎی ﺑﻬﯿﻨﻪﺳﺎزی در ﯾﺎﻓﺘﻦ ﺟﻮاب ﺑﻬﯿﻨﻪ ﮐﺎﻫﺶ ﻣﯽﯾﺎﺑﺪ. ﯾﮑﯽ از روشﻫﺎی ﻏﻠﺒﻪ ﺑﺮ اﯾﻦ ﻣﺸﮑﻞ، ﻃﺮاﺣﯽ اﻟﮕﻮرﯾﺘﻢﻫﺎﯾﯽ ﺑﺎ ﺗﻮاﻧﺎﯾﯽ ﺑﺎﻻﺗﺮ ﻣﯽﺑﺎﺷﺪ. در روش دوم ﺳﻌﯽ ﺷﺪه اﺳﺖ ﺑﺎ ﺗﻠﻔﯿﻖ ﻋﻤﻠﮕﺮ ﺗﻘﺎﻃﻊ در اﻟﮕﻮرﯾﺘﻢ ﮐﻠﻮﻧﯽ زﻧﺒﻮر ﻣﺼﻨﻮﻋﯽ، ﺗﻮازن ﻣﻨﺎﺳﺒﯽ ﻣﯿﺎن ﺗﻮاﻧﺎﯾﯽﻫﺎی اﮐﺘﺸﺎف و اﺳﺘﺨﺮاج آن ﺑﺮﻗﺮار ﺷﻮد. ﺑﺮای ارزﯾﺎﺑﯽ دﻗﺖ ﻋﻤﻠﮑﺮد روشﻫﺎی ﭘﯿﺸﻨﻬﺎدی، از ﯾﮏ ﻣﺪل ﻣﺼﻨﻮﻋﯽ ﺳﻪﺑﻌﺪی ﻣﺨﺰن )ﻣﺪل ﻣﺮﺟﻊ( اﺳﺘﻔﺎده ﺷﺪ. ﻣﺪلﻫﺎی رﺧﺴﺎرهای ﺑﻪروزرﺳﺎﻧﯽ ﺷﺪه ﺑﻮﺳﯿﻠﻪ روشﻫﺎی آﺷﻔﺘﮕﯽ اﺣﺘﻤﺎل ازدﺣﺎم ذرات و ﮐﻠﻮﻧﯽ زﻧﺒﻮر ﻣﺼﻨﻮﻋﯽ ژﻧﺘﯿﮏ ﺑﻪﺗﺮﺗﯿﺐ دارای ﯾﮏ ﺗﻔﺎوت 6/65 و 0/99 درﺻﺪی ﺑﺎ ﻣﺪل رﺧﺴﺎرهای ﻣﺮﺟﻊ اﺳﺖ. ﺑﺮای ﻧﺸﺎندادن ﺗﻮاﻧﺎﯾﯽ اﻟﮕﻮرﯾﺘﻢﻫﺎی ﭘﯿﺸﻨﻬﺎدی در ﺳﺎﺧﺖ و ﺑﻪ روزرﺳﺎﻧﯽ ﻣﺪلﻫﺎی رﺧﺴﺎرهای، دو روش ﺳﻨﺘﯽ زﻣﯿﻦآﻣﺎری ﻧﯿﺰ ﺑﺮ روی ﻣﺴﺌﻠﻪ ﻣﻮردﻧﻈﺮ ﭘﯿﺎدهﺳﺎزی ﺷﺪه اﺳﺖ. ﻧﺘﺎﯾﺞ ﺣﺎﺻﻞ ﻧﺸﺎن داد ﮐﻪ ﺑﻪﮐﺎرﮔﯿﺮی روشﻫﺎی آﺷﻔﺘﮕﯽ اﺣﺘﻤﺎل ازدﺣﺎم ذرات و ﮐﻠﻮﻧﯽ زﻧﺒﻮر ﻣﺼﻨﻮﻋﯽ ژﻧﺘﯿﮏ ﺑﻪﺗﺮﺗﯿﺐ ﺑﺎ ﯾﮏ اﻓﺰاﯾﺶ دﻗﺖ 18/8 و 24/46 درﺻﺪی در ﺗﻔﺎوت ﺑﺎ ﻣﺪل رﺧﺴﺎرهای ﻣﺮﺟﻊ، ﻧﺴﺒﺖ ﺑﻪ روشﻫﺎی زﻣﯿﻦآﻣﺎری ﻫﻤﺮاه ﺑﻮد. در ﭘﺎﯾﺎن ﻋﻤﻠﮑﺮد روش ﮐﻠﻮﻧﯽ زﻧﺒﻮر ﻣﺼﻨﻮﻋﯽ ژﻧﺘﯿﮏ ﺑﺮ روی دو ﻣﺪل ﻣﺨﺰن ﻣﺼﻨﻮﻋﯽ ﺑﺰرگﺗﺮ و ﭘﯿﭽﯿﺪهﺗﺮ ارزﯾﺎﺑﯽ ﺷﺪ. ﻧﺘﺎﯾﺞ ﮐﻤﯽ و ﮐﯿﻔﯽ ﭘﮋوﻫﺶ ﻧﺸﺎن داد ﻋﻠﯽ رﻏﻢ ﻣﺤﺪودﯾﺖﻫﺎی روشﻫﺎی ﭘﯿﺸﻨﻬﺎدی، روش ﮐﻠﻮﻧﯽ زﻧﺒﻮر ﻣﺼﻨﻮﻋﯽ ژﻧﺘﯿﮏ ﺑﺎ ﻧﺘﺎﯾﺞ ﻗﺎﺑﻞ ﻗﺒﻮلﺗﺮی ﻫﻤﺮاه اﺳﺖ.

application of optimization and parameterization algorithms for the integration of seismic and well logging data in the process of building and updating lithofacies models

summary in this research, integration of well logging and 2d/3d seismic data in the reservoir lithofacies modeling process has been considered. for this purpose, two methods from the so-called seismic matching loop class have been used. in the first method, the particle swarm optimization (pso) algorithm is implemented to find the optimal value of the probability perturbation method (ppm) deformation parameter. the ppm is used to convert an n-parameter optimization problem to a problem with one parameter. in the second method, in the absence of parametrization methods, the problem of updating lithofacies models will be considered as an optimization problem with the n-unknown parameter. obviously as the number of optimization unknown parameters increases, the optimization algorithms ability in finding the optimum solution decreases. one way to overcome this problem is to design optimization algorithms with higher capabilities. in the second method, an attempt has been made to establish a proper balance between the exploration and exploitation capabilities of the optimization algorithm. in this research, the crossover and mutation operators of the genetic algorithm (ga) optimization method have been used to improve the exploration and exploitation capabilities of the pso and artificial bee colony (abc) algorithms. to evaluate the performance of the proposed methods, a 3d synthetic reservoir model (reference model) has been used. the obtained results show that reservoir lithofacies models generated by ppm-pso, pso-ga and abc-ga methods have 6.65%, 10.44%, and 0.99% mismatches compared with the reference lithofacies model, respectively. to highlight the ability of the proposed algorithms in generating and updating the reservoir lithofacies models, two traditional geostatistical methods have also been applied to the specified problem. the results indicate that using the ppm-pso, pso-ga and abc-ga algorithms, respectively, leads to 18.8%, 15.27%, and 24.46% improvement on mismatch values compared to the traditional geostatistical methods. finally, the performance of abc-ga method has been evaluated on two larger and more complex synthetic reservoir models. introduction the realistic and optimal management of the hydrocarbon reservoirs requires maximum understanding of their characteristics, which can be achieved through the integration of various data sources in the reservoir modeling process. seismic data, due to its extensive areal coverage and high lateral resolution compared to well-based data, have always been of interest in static property estimation at locations among wells. in order to use more of the seismic data in the facies modeling process, the seismic matching loop approach, which is based on geostatistical techniques and optimization algorithms, can be used. obviously, by increasing the number of reservoir model grid blocks, the ability of the algorithm to generate the optimal facies model decreases. the main focus of this research is to introduce two approaches to solve this problem. the first approach is the integration of pso algorithm into ppm, which is a parameterization technique. the innovation of this method is the integration of the pso algorithm into the ppm to find the optimal value of its deformation parameter. in the second approach, the unknown parameters of the optimization problem are equal to the number of grid blocks in the reservoir model. the innovation of the second proposed method is the integration of the ga crossover operator in the abc optimization algorithm and the complete elimination of the scout bees phase.