Search

Summary The leaching of a salt cavern will trigger a series of rock-fluid interactions, including salt rock dissolution, cavity expansion, and brine transport caused by convection, turbulence, and diffusion effects. These interactions have influences on one another. The primary objectives of this study include developing a 3D multiphysical coupled model for horizontal salt cavern leaching and quantifying these interactions. The species transport equation and standard κ-ε equation were combined to describe the brine transport dynamics within the cavity. Based on the velocity and concentration distribution characteristics predicted, the interface movement equation implemented with mesh deformation techniques was applied to describe the cavity expansion. Next, the Volgograd cavern monitored data were collected for model validation. The predicted results are consistent with the field data. The average relative errors are 11.0% for brine displacing concentration and 4.5% for cavity volume. The results suggest that the cavity can be divided into three regions, including the main flow region, circulation region, and reflux region. The results also suggest that the brine concentration distribution is relatively uniform. With the dissolution threshold angle and anisotropic dissolution rates considered, the resultant cavity cross section is crown top and cone bottom. The results also show that the cavity can be divided into dissolution and erosion sections according to its position relative to the injection point.

Abstract The horizontal salt cavern is the ideal place for energy storage built in bedded salt formations. The construction of the multi-step horizontal salt cavern is a complicated process of fluid dynamics and chemical dynamics, including salt boundary dissolution, cavern development, brine flow, heat transfer, and species transport. Considering the influence of the heat transfer, the multiple governing equations are coupled to simulate the construction process of the 3D horizontal salt cavern. The influence of heat transfer on fluid flow, brine concentration, and cavity expansion is analyzed. According to the results, heat transfer accelerates the transport and diffusion of the brine, which can increase the dissolution rate of salt rock. The direction of the forced thermal convection is perpendicular to the direction of fluid flow and the brine concentration gradient in the cavity. The results also show that the formed cavity previously will continue to expand in the next leaching stage, but the height of expansion decreases gradually. This work helps predict the multi-step horizontal salt cavern development and concentration distribution under reasonable accuracy, which may guide to the scheme design and engineering practice of the multi-step horizontal salt cavern construction for underground gas storage.

Salt cavern solution mining is a complicated process of fluid dynamics and chemical dynamics, including salt boundary dissolution, cavern expansion, brine flow, and species transport. The reaction processes occur simultaneously and interact with each other. In this study, a multiphysical coupled model is established to evaluate the real-time three-dimensional salt cavern shape expansion, the velocity field, and the brine concentration distribution. Then, the predicted results are compared with the field data of a Jintan Gas Storage Well in China. The average relative deviations with the turbulent flow are 5.7% for outlet brine concentration and 4.0% for cavern volume. The results show that salt cavern can be divided into four regions, including the shock region, plume region, reflow region, and suction region. The results also indicate that the turbulent flow will stimulate the formation of the vortex, thus affecting the distribution of brine concentration. And, the brine concentration distribution primarily influences cavern corrosion. The results suggest that adjusting the inject velocity and the tube position can change the cavern construction rate and the cavern shape. Overall, these results have guiding significance for the design and engineering practice of salt cavern construction for energy storage.

Understanding stress and/or temperature sensitivity on permeability helps in predicting permeability distribution in space and its change with production. While permeability evolution with pressure...

Summary Gas adsorption, desorption, and displacement occur throughout the coalbed methane (CBM) and enhanced coalbed methane (ECBM) recovery process, causing the coal pores to deform and affecting injectivity and productivity. The primary objectives of this study include determining the pressure at which the monolayer adsorption transitions to multilayer adsorption and linking the swelling strain to both classes of adsorption. In this study, the simplified local density (SLD) theory was first modified and applied to describe the characteristics of both types of adsorption and to determine the transition pressure. A strain model coupled with the SLD theory was then developed to describe adsorption-induced deformation. Next, the measured methane (CH4) and carbon dioxide (CO2) adsorption isotherms, and strain data on the same coals, were collected for model validation. Results suggest that gas adsorbed on coal surfaces at the very beginning (monolayer adsorption) and the adsorption on other gas molecules continues once the surface has been filled (multilayer adsorption). Results also suggest that swelling strain is proportional to both types of adsorption, with the multilayer case being larger than the monolayer case. This difference may be due to the additional repulsion between the adsorbate and multilayer liquid film.

Coal particles caused by well drilling and completion, hydraulic fracturing, and formation pressure decreasing would migrate into the horizontal well carried by fracturing flow-back fluid, formation water and CBM, and then deposit on the low side of pipe leading to the decrease of cross section area of pipe for flow and even blocking of the pipe. The purpose of current research was to obtain the flow pattern while there was stationary solid particle bed on the low side of the pipe and an explicit equation to predict the height of solid particle bed was proposed. In this study, with the coal particle, water, and air as the solid, liquid, and gas phases respectively, gas/liquid/solid three-phase flow experiments in horizontal well were carried out and 4 crucial factors including different water flow rate, gas flow rate, coal particle size, and coal particle mass concentration were considered. Fluid's real apparent velocity was defined and used to analyze the distribution of flow pattern coordinate points. Results show that the flow pattern coordinate points determined by real gas apparent velocity and real water apparent velocity distribute near the flow pattern transition lines of Taitel and Dukler flow pattern map. Besides, coal particle size, water flow rate, and gas flow rate have obvious impact on the height of coal particle bed and the coal particle mass concentration has a negligible effect while the height of interface between gas layer and water layer is only dramatically affected by the gas flow rate. At last, an explicit equation for predicting the cross section area of coal particle bed based on the experimental data was proposed which is crucial to predict the flow pattern and calculate the pressure drop in horizontal well. The predicting cross section area of coal particle bed is in good agreement with the measured cross section area of coal particle bed and the average relative differences are 9.38% and 13.45% for stratified smooth flow and stratified wavy flow respectively. The equation is applicable for a range of coal particle size (8–30 mesh) and water apparent velocity (0.025–0.123 m/s).

Oil-water two-phase flow through the complex pipeline, consisting of varying pipes and fittings in series or parallel, is commonly encountered in the petroleum industry. However, the majority of the current study is mainly limited to single constant-radius pipe. In this paper, a unified model of oil-water two-phase flow in the complex pipeline is developed based on the combination of pipe serial-parallel theory, flow pattern transformation criterion, two-fluid model, and homogenous model. A case is present to verify the unified model and compare with CFD results. The results show that the proposed unified model can achieve excellent performance in predicting both the flow distributions and pressure drops of oil-water two-phase flow in the complex pipeline. Compared with CFD results for water volumetric fractions ranging from 0% to 100%, the highest absolute percentage error of the proposed model is 14.4% and the average is 9.8%.

In coal-bed methane (CBM) horizontal wells, coal particles generally deposit on the low side of the wellbore to form a particle bed, which increases well pressure loss and even blocks the well. The coal particle bed undergoes changes in height and geometry with the fluctuation of gas and water production. Besides, with a low water flow rate, the water layer in CBM well is extremely thin. Both the dynamically changing particle bed and thin water layer make the conventional gas/water interfacial friction factor (fi) prediction method for stratified flow in horizontal pipe not suitable for that of CBM well. Based on the large-size multiphase complex flow experimental equipment, a gas/liquid/solid three- phase flow experiment in horizontal pipe was carried out. In the experiment, gas flow rate, water flow rate and coal particle size were the three main factors considered. During the experiment, the gas/water flow pattern, migration of coal particles and evolution of blocked section could be investigated through the transparent pipe. Furthermore, due to the significant effect of coal particle bed on the height of gas-water interface, a non-negligible gravitational pressure drop calculated using water level differences is taken into consideration when obtaining the friction pressure drop. In the experiment, three gas/water flow patterns including slug flow, stratified smooth flow and stratified wavy flow were observed. The coal particles mainly transport by means of rolling and saltating in the thin water layer under stratified flow and migrate by way of dispersing in water under slug flow. Besides, for particles no larger than 20-30 mesh, the blocked section tended to be overall driven forward like fluid by local high differential pressure and then led to the discontinuity of coal particle bed. However, for blocked section with particles not less than 10-20 mesh, instead of being driven forward, the coal particles on the top surface of it would be carried away layer by layer until the raised part was eroded off. Furthermore, as with the situation in gas/liquid two-phase stratified flow in horizontal pipe, fi in the experiment increases with liquid Reynolds number (ReL). However, it is much larger than that of gas/liquid stratified flow under the same ReL,. The analysis of experimental data presents a closely negative correlation between fi and the equivalent diameter of effective flow channel. As for the coal particle size, it has some indirect effects on fi through the equivalent diameter of effective flow channel. This work is helpful for coal particle management and productivity prediction during CBM development, which may provide guidance to particle-bailing operation and also serve as a basis for theoretical mechanistic models to predict the gas/liquid interfacial friction factor of gas/liquid/solid three-phase stratified flow.

Gas adsorption and desorption and displacement has a significant effect on coal deformation and permeability evolution during the primary recovery of coalbed methane (CBM) and enhanced coalbed methane recovery (ECBM). The objectives are to (1) quantify the coal deformation and permeability change caused by methane (CH4) displacement with carbon dioxide (CO2) and (2) model the transportation of CH4 and CO2 in deformed coalbed. In this study, the gas adsorption and desorption and displacement, coal deformation, and permeability evolution during CBM and ECBM recovery were described by an internally consistent adsorption-strain-permeability model, of which the simplified local density (SLD) adsorption theory, a theoretical strain model, and a matchstick-based permeability model were rigorous coupled. The coupled model was then verified with all of the CH4 and CO2 measured gas adsorption and desorption and coal strain data published in the past 60 years. Next, sensitivity analysis was further conducted on the ...

Due to geological complexity and the heel–toe effect, the production profiles of long horizontal wells are usually imbalanced, and as a result, premature water breakthrough is usually encountered. Once water breakthrough occurs, this phenomenon will reduce oil production. To maximize oil recovery, inflow-control devices (ICDs) are widely used to create a uniform inflow profile. To date, known ICDs cannot meet all the ideal requirements throughout the well’s life. In this study, based on the combination of a successive restriction mechanism and water swelling rubber, a novel autonomous inflow control device is proposed. Then, the rules of oil–water two-phase flow through the novel design are studied by a numerical simulation based on optimized structural parameters, and the fluid property sensitivities are analyzed. Upon integration of the novel design into the test apparatus, flow tests are conducted. The influences of water content, inflow rate, and injection rate on the pressure drop are further analyze...

Natural gas produced from coals, or coalbed methane (CBM), is a significant component of the energy portfolio for many countries. One challenge associated with CBM production is associated water. Specifically, coalbeds in situ contain significant amounts of water, and ideally this water is removed by pumping prior to the primary recovery of CBM to lower pressure and stimulate methane desorption. Such a prior water production can be challenging because desorption depends on the occurrence state of methane and water in situ, e.g., how much of each fluid is adsorbed or otherwise. Accordingly, primary objectives of this analysis include quantifying both the occurrence state of methane and water of different coals for a range of coalbed properties and conditions, and specifically quantifying the impact of coal moisture on methane desorption. Ultimate and proximate analysis and methane adsorption tests were first conducted on several coal samples from different basins. Simplified local density (SLD) theory was ...

Coal is known as a dual-porosity media composed of cleat and matrix pore. Methane can be stored in the cleats or adsorbed on the inner surface of matrix pore. While fluid mobility is mainly controlled by the developed cleat network, methane desorption has a significant effect on cleat deformation. In the process of coalbed methane recovery, both reservoir compaction and matrix shrinkage will occur and have opposite effects on permeability evolutions. A variety of analytical permeability models have been developed to describe the transient characteristics of permeability in coals. In this study, three common permeability models are first revisited and evaluated against the experimental data under uniaxial strain condition. Shi–Durucan (S&D) model demonstrates the best performance among these models. However, constant cleat volume compressibility was used to assume for S&D model, and the generalization of S&D model is significantly limited. For ease of generalization, the relation between cleat volume compressibility and effective horizontal stress is re-derived and introduced to the derivation of permeability model. Since coal reservoirs usually demonstrate strong anisotropy and heterogeneity, the influences of elastic and adsorption properties are further tested to reveal the overall trend of permeability. The results show that S&D model and its modification with the main variable of effective horizontal stress have the best performances in matching the experimental data under uniaxial strain. The relationship between cleat volume compressibility and effective horizontal stress can be better reflected by the inverse proportional function. In addition, the strengths of reservoir compaction effect relative to matrix shrinkage effect in different models only vary with Poisson’s ratio, while their magnitudes are also affected by Young’s modulus. For a typical coal reservoir, the C&B and P&M models will observe a stronger permeability decline at the initial, while the improved P&M model will receive an earlier and more rapid rebound than the S&D and W&Z models.

Recovery of coalbed methane (CBM) can trigger a series of coal-gas interactions, including methane desorption, coal deformation, and associated permeability change. These processes may impact each other. A primary objective of this analysis is to simultaneously quantify these interactions and their impacts during CBM recovery. To achieve this and other objectives, a rigorously coupled adsorption–strain–permeability model was developed. Gas adsorption, coal deformation, and cleat permeability characteristics were described using simplified local density (SLD) adsorption theory, a theory-based strain model, and matchstick-based permeability models, respectively. The strain model was verified against measured methane-adsorption-induced coal strain data published during the past 60 years, and the coupled model was tested using well test data measured in the San Juan Basin of New Mexico in the United States. Results suggest that the strain model is very consistent with measured coal deformation data for fluid ...

Numerous oil wells, especially in their middle-late periods, are becoming less economic due to the high lifting costs and reduced recoveries. The downhole oil–water separation (DOWS) system is aimed to reduce the production cost, mitigate the environment impact, and enhance the oil recovery. However, current separators are of either poor separation effects or poor separation efficiencies. In this paper, a novel oil–water separator design is proposed based on the combination of two different flow resistance mechanisms and pipe serial-parallel theory, with the restrictive path restricting the heavier water, while the frictional path impeding the more viscous oil. Based on the combination of the flow pattern transformation criterion, homogenous model, two-fluid model, and pipe serial-parallel theory, a unified model of oil–water two-phase flow is developed to predict both the flow rate and water content distributions in different paths, which is then compared with the computational fluid dynamics (CFD) results. Unlike the CFD results, each path has a specific flow rate and water content, and as a consequence, specific flow regime and flow pattern. Both the CFD and model results show that the flow rate distributions in different paths of the separator will be adjusted automatically according to the fluid's property, while the model can also predict the water content distributions at the same time. And the average relative deviation between the CFD and model results for flow rate distribution is 14.24%, while that for water content distribution is 42.03%. Specifically, oil, being more viscous, mainly takes the restrictive path; while water, being heavier, tends to take the frictional path instead. To sum up, this autonomous function directs oil and water to different paths, hence oil and water is well separated.

When cementing horizontal wells, it is essential to keep the displacement interface steady, in order to achieve good cementing quality, especially in eccentric annuli. Considering the displacement behavior between shear-thinning fluids in a horizontal eccentric annulus, the Lattice-Boltzmann method (LBM) was used to investigate the displacement interface shape. The classical Hele-Shaw approach was used to simplify the model from 3D flow into a 2D model. In the simulations, the effects of eccentricity and density ratio on the displacement interface shape and on the interface length versus time were analyzed in detail, resulting in an understanding of the fingering mechanism. Comparison of LBM with the FLUENT solutions showed that the results with LBM are essentially in agreement. In general, a greater density ratio and a lower eccentricity will result in a shorter displacement interface i.e. better displacement in a horizontal eccentric annulus. For a given density ratio, there is an optimum eccentricity to achieve a steady interface. The study provides guidance to the design of wellbore cement jobs and demonstrates that LBM is a promising tool for modeling of the fluid displacement process.

Summary In this article, a more-general flow-pattern classification of oil/water two-phase flow in the horizontal wellbore is proposed first according to the theoretical analysis and previous research achievements, on the basis of which a simplification is then performed through reasonable incorporation, and the ultimate flow patterns considered for modeling are reduced to two categories containing only six standard patterns. By use of the classical two-fluid and homogeneous modeling methodologies stemming from oil/water two-phase flow in conventional pipes, combined with the simplified classification, a mechanistic model is developed to predict the flow characteristics including the flow patterns and pressure losses for oil/water two-phase variable-mass flow in the horizontal wellbore. Model implementation is performed on the basis of the universal principle that a system will stabilize to the equilibrium state of minimum energy. Overall performance of the mechanistic model is then validated against the new data sets measured upon a large-scale experimental apparatus at the China University of Petroleum (CUP), which is designed and constructed to simulate the gas/oil/water multiphase flow in horizontal wellbores with wall mass transfer. Results show that the model developed in this paper can not only properly predict the flow patterns of oil/water two-phase flow in the horizontal wellbore, but also has high prediction accuracy for the pressure drops. Compared with the new experimental data for oil/water two-phase variable-mass flow that covers a series of input water-volumetric fractions ranging from 10 to 90%, the highest absolute average percentage error of the new unified model is 12% and the whole error is 9.2%, which demonstrates an acceptable performance. Investigations conducted in this study further enrich and develop the theory of hydrodynamic calculation for oil/water flow in the horizontal wellbore with wall influx.

Coalbed methane (CBM) is an important part of energy in the world for its larger reserves, lower production cost. More and more countries and oil companies spent more and more money for CBM production in the past decades. Hydraulic fracturing is an important development method for the CMB production and the fracture parameters optimization is the key step for the hydraulic fracturing design. In this study, a fracture parameters optimization method is put forward which is based on the unified fracturing design (UFD) method. The maximum dimensionless productivity index and optimal dimensionless fracture flow conductivity of fractured wells for a certain proppant number are calculated, and then the optimal half-length and the width of the fracture, the dimensionless production index of fractured horizontal well are calculated correspondingly. All of the above is concluded under the condition that the non-Darcy effect in fracture is considered. Based on the result, the variation law of Reynolds number, the effective permeability, the optimal fracture half-length, the optimal fracture width and the optimum dimensionless production index to the free-gas saturation in CBM reservoir are obtained. We can conclude that the non-Darcy turbulent flow effect can decrease the proppant effective permeability in the fractures by mean of the generation of additional pressure drop, which will decrease the fracture diverting capacity and the production rate.

The prediction accuracy of the five horizontal wellbore single-phase drop models (Siwon, Asheim, Su, Yuan, Ouyang) is evaluated and compared based on the high-quality experimental data measured by a self-developed experimental system with large size horizontal wellbore (ID: 139.7 mm). The results show that the pressure drop prediction accuracy of the five models is, from high to low, Ouyang, Siwon, Asheim, Yuan, and Su. Generally, the Su and Yuan models have high predictions and the Siwon and Asheim model have low predictions. The Ouyang model has good prediction performance, its average relative error is only 5.5%, and the example also proves its good prediction effect, but it can't represent complex mechanisms such as the resistance effect caused by wall perforation. The models for predicting the pressure drop of single-phase flow in a horizontal wellbore need to be further improved to completely represent complex mechanisms of the variable mass flow in the horizontal wellbore, such as the resistance effect caused by wall perforation and lubrication flow drag reduction effect caused by wall influx. Key words: pressure drop model, horizontal wellbore, perforated completion, large-scale wellbore, prediction accuracy evaluation

In long horizontal wells, premature water or gas breakthrough is usually encountered due to the imbalanced production profile. This imbalanced phenomenon could be caused by the heel–toe effect, reservoir anisotropy, reservoir heterogeneity or natural fractures. Once coning occurs, water/gas fast track will be generated, leading to the reduction in oil production. Inflow control devices (ICDs) are usually installed in the completion sections to maintain a uniform inflow by generating an additional pressure loss. However, none of current ICDs are perfect enough to meet all the ideal requirements throughout the well׳s life. In this paper, a novel autonomous inflow control device (AICD) design is proposed based on the combination of two fluid dynamic components, with the splitter directing the flow, and the restrictor restricting the flow. Based on the combination of the flow pattern transformation criterion, homogenous model, two-fluid model, and pipe serial–parallel theory, a unified model of oil–water two-phase flow is developed to predict both the flow distributions and pressure drops through the splitter, which is then compared with the computational fluid dynamics (CFD) results. Also the rules of oil–water two-phase flow through the disk-shaped restrictor are studied by numerical simulation. The results show that the unified model compares well with the CFD results. The average error percentage between the model and CFD results for flow distribution is 10.02%, while that for pressure drop is 11.25%. Both the model and CFD results show that the flow distributions in different paths of the splitter will be adjusted automatically according to the fluid׳s specific property, thus different fluids will enter the restrictor differently, and result in varying flow resistances. Specifically, oil, being more viscous, tends to take the restrictive path, enter the restrictor radially, and result in minimal flow restriction; while water, being less viscous, tends to take the frictional path, enter the restrictor tangentially, begin spinning rapidly near the exit, and result in obvious flow restriction. This autonomous function enables the well to continue producing oil for a longer time while limiting the water production; hence the total oil production is maximized. The investigation conducted in this study also further enriches the theory of hydrodynamic calculation for oil–water two-phase flow in complex pipelines.

Multi-layered CBM (Coalbed Methane) reservoirs contain dozens (20–40) of thin coal seams (0.5 m–10 m) and sandstone seams and a commingled production strategy is preferred. Gas crossflow between coal seams and sandstone seams contributes much to the commingled gas production. Our study is focused on gas crossflow between coal and sandstone with fused interface. The experimental apparatus designed for this study consisted of a sandstone sample holder and a coal sample holder, where nitrogen was injected at a constant pressure difference between the sandstone and coal. The gas crossflow rate between coal and sandstone was monitored. Factors including confining stress, effective horizontal stress, axial stress and the permeability of sandstone influencing the gas crossflow were analyzed based on the experimental results. Based on the interface characteristics, a crossflow model for fused interface between coal and sandstone was proposed and gas crossflow resistant coefficients were defined. The experimental results show that gas crossflow rates decreased dramatically with the increment of confining stresses loading on the coal sample and increased linearly with the increment of average pore pressure on the condition of constant effective horizontal stresses. Compared with stresses loading on sandstone samples, the axial stresses and confining stresses loading on coal make larger influence on the gas crossflow rates. With the increment of the sandstone permeability, gas crossflow rates become less sensitive to the change of sandstone permeability. The comparison between model results and experimental results demonstrate that the proposed mathematical models can effectively predict gas crossflow between coal and sandstone with fused interface.

The Autonomous Inflow Control Device (AICD) adjusts the inflow of the horizontal wellbore by adding additional resistance to the fluid, thereby prolonging the water seepage time and enhancing oil recovery. The performance of AICD mainly depends on structural parameters. The computational fluid dynamics software is used to analyze the influence of structural parameters on the performance of AICD. The results show that the diameter of the variable diameter section, the diameter of the restrictor and the diameter of the outlet are the main factors affecting the performance of the inflow control device. The fluid parameter sensitivity simulation results show that the oil phase viscosity and water content have a great influence on the performance of AICD, while the influence of oil phase density on the performance of the device is negligible; the pressure drop under the pure water condition of this type of AICD is more than twice that of pure oil condition, so it has better water control ability.

Numerous oil wells, especially in their middle-late periods, are becoming less economic due to the high lifting costs and reduced recoveries. The downhole oil-water separation (DOWS) system is aimed to lower the production cost, reduce the environment impact, and enhance the oil recovery. However, current separators are of either poor separation effects or poor separation efficiencies. In this paper, a novel oil-water separator design is proposed based on the combination of two different flow resistance mechanisms and pipe serial-parallel theory, with the restrictive path restricting the heavier water, while the frictional path impeding the more viscous oil. Based on the combination of the flow pattern transformation criterion, homogenous model, two-fluid model, and pipe serial-parallel theory, a unified model of oil-water two-phase flow is developed to predict both the flow rate and water content distributions in different paths, which is then compared with the computational fluid dynamics (CFD) results. Unlike the CFD results, each path has a specific flow rate and water content, and as a consequence, specific flow regime and flow pattern. Both the CFD and model results show that the flow distributions in different paths of the separator will be adjusted automatically according to the fluid's property, while the model can also predict the water content distributions at the same time. And the average relative error for flow distribution is 17.71%, while that for water content distribution is 32.66%. Specifically, oil, being more viscous, mainly takes the restrictive path; while water, being heavier, tends to take the frictional path instead. To sum up, this autonomous function directs oil and water to different paths, hence oil and water is well separated.

In long horizontal wells, production rate is typically higher at the heel than that at the toe. The resulting imbalanced production profile may cause early water or gas breakthrough into the wellbore. Once coning occurs, well production may be severely decreased due to limited flow contribution from the toe. To eliminate this imbalance, inflow control devices (ICDs) are placed in each screen joint to balance the production inflow profile across the entire lateral length and to compensate for permeability variations. Currently, there are three different passive ICD designs in the industry: nozzle-based, helical channel, and tube-type. They use restriction mechanism (nozzle-based), friction mechanism (helical channel) or cooperating the two (tube-type) to achieve a uniform inflow profile. However, the reality is that none of these ICDs alone meets the ideal requirements of an ICD designed for the life of the well: high resistance to both plugging and erosion, high viscosity insensitivity, and high density sensitivity. Therefore, the selection and optimization of ICDs for a specific reservoir requires for further study. In this paper, three computational fluid dynamic based, numerical models of these ICDs with same flow resistance rating (FRR) were developed to characterize the flow performance. The results show that the throttle pressure drop depends on fluid properties, and geometries of each ICD. The weight of each factor that affects pressure drop was determined by maximizing deviations combining with analytic hierarchy process, and subsequently optimal ICD under specific reservoir condition was selected with the help of fuzzy comprehensive evaluation, thereby an ICD selection diagram was built. For a specific reservoir, we will have the selected ICD with best pressure drop composition by optimizing its structural parameter, which has best corrosion resistance and least viscosity sensitivity.