A centripetal turbine optimization design method based on advanced optimization algorithm

Abstract

The present application relates to the technical field of centripetal turbine optimization design, and particularly discloses a centripetal turbine optimization design method based on an advanced optimization algorithm, which comprises the following steps: S1: a one-dimensional aerodynamic thermodynamic mean diameter design model is constructed, and the model is integrated with loss models including flow passage loss, blade tip clearance loss, tail edge loss and outlet kinetic energy loss; S2: the one-dimensional mean diameter design model of step S1 is coupled with a particle swarm optimization algorithm with the maximum total static efficiency of the centripetal turbine as the target, and global optimization is performed on dynamic parameters, thermodynamic parameters and geometric parameters affecting the performance of the turbine to obtain a one-dimensional preliminary design parameter set maximizing the total static efficiency; and a hierarchical collaborative optimization framework is constructed to significantly improve the design efficiency and precision of the centripetal turbine, realize systematic approximation of global optimal design under the premise of guaranteeing physical authenticity, and provide a solution for high-performance turbine product development.

Description

TECHNICAL FIELD

[0001] The application relates to the technical field of optimization design of centripetal turbines, and particularly relates to a centripetal turbine optimization design method based on an advanced optimization algorithm. BACKGROUND

[0002] With the deepening of global energy transformation, the development and utilization of renewable energy and industrial waste heat are increasingly valued. Organic Rankine cycle technology has become one of the key technologies in the fields of geothermal energy, solar thermal energy and industrial waste heat recovery, because it can efficiently convert medium and low temperature heat energy into electric energy;

[0003] According to the search of the public number CN114896699B, a multi-disciplinary optimization design method of a centripetal turbine impeller in an aero-engine is disclosed. In this technology, a variance analysis method based on a response surface is used to analyze the design space, and the design variables that significantly affect the required performance are selected for multi-disciplinary optimization design, which reduces the design space dimension and reduces the time cost to a certain extent. A global optimization algorithm is used to solve the multi-disciplinary optimization design problem of the centripetal turbine, which can significantly improve the design level of the centripetal turbine, fully explore the design potential, and meet the vibration performance requirements of the centripetal turbine, and improve the economy and reliability of the centripetal turbine stage.

[0004] The existing centripetal turbine design process usually regards one-dimensional preliminary design and subsequent three-dimensional modeling and optimization as two relatively independent stages. This disconnection leads to inherent limitations in the design process. First, one-dimensional design often relies on simplified empirical loss models and the selection of isolated parameter ranges, which has limited accuracy and is difficult to accurately reflect complex separation, secondary flow and other key physical phenomena in three-dimensional flow. When three-dimensional modeling is directly based on one-dimensional results, the initial scheme may deviate from the global optimal region. Secondly, if parameter optimization is directly performed in three-dimensional space to improve performance, there is a huge challenge in computational cost, because each design iteration needs to perform high-precision three-dimensional flow field simulation again. This computational burden makes it almost impossible to achieve systematic and comprehensive global exploration of blade geometric parameters (such as installation angle) in engineering practice, and the design process is forced to rely on local adjustment or empirical trial and error, making it difficult to ensure that the final scheme reaches the desired performance level. SUMMARY

[0005] In view of the deficiencies of the prior art, the application provides a centripetal turbine optimization design method based on an advanced optimization algorithm, which significantly improves the design efficiency and accuracy of the centripetal turbine by constructing a hierarchical collaborative optimization framework, and realizes the systematic approximation of global optimal design under the premise of ensuring physical authenticity, thereby providing a solution for high-performance turbine product development.

[0006] To achieve the above object, the application is implemented by the following technical scheme: a centripetal turbine optimization design method based on an advanced optimization algorithm, comprising the following steps:

[0007] S1: a one-dimensional mid-passage design model is constructed, and the model integrates loss models including flow passage loss, tip clearance loss, trailing edge loss and outlet kinetic energy loss;

[0008] S2: the one-dimensional mid-passage design model of step S1 is coupled with a particle swarm optimization algorithm to maximize the total static efficiency, and the global optimization is performed on the dynamic parameters, thermodynamic parameters and geometric parameters affecting the turbine performance to obtain a one-dimensional preliminary design parameter set maximizing the total static efficiency;

[0009] S3: based on the one-dimensional preliminary design parameter set maximizing the total static efficiency obtained in step S2, a three-dimensional turbine initial geometric model is generated;

[0010] S4: based on the three-dimensional turbine initial geometric model generated in step S3, the blade installation angle is used as the optimization variable, and the Latin hypercube sampling method based on space filling is used to sample in the design space to generate multiple sets of three-dimensional blade geometric variant schemes;

[0011] S5: three-dimensional computational fluid dynamics numerical simulation is performed on each set of three-dimensional blade geometric variant schemes generated in step S4 to obtain the total static efficiency and net output power;

[0012] S6: based on the multiple sets of simulation data obtained in step S5, a Kriging response surface proxy model is constructed with the blade installation angle as the input and the total static efficiency and net output power as the output;

[0013] S7: the total static efficiency and net output power predicted by the Kriging response surface proxy model are used as the optimization target, and the particle swarm optimization algorithm is applied again to optimize the blade installation angle to determine the final three-dimensional blade optimal installation angle;

[0014] S8: according to the three-dimensional blade optimal installation angle determined in step S7, the three-dimensional turbine model is updated to complete the final optimization design of the centripetal turbine.

[0015] Preferably, in step S1, the flow passage loss in the loss model is calculated by the following formula:

[0016]

[0017] wherein, is an empirical coefficient, and are the hydraulic length and hydraulic diameter of the rotor, and respectively rotor inlet radius and outlet meridional mean radius, is rotor outlet mean relative flow angle, is rotor outlet blade height, and respectively rotor inlet and outlet relative velocity.

[0018] Preferably, in step S2, the key design parameter range that maximizes the total static efficiency is determined by global optimization, including: isentropic speed ratio is 0.68-0.71, rotor inlet absolute flow angle is 66°-78°, degree of reaction is 0.45-0.65, rotor meridional velocity ratio is 0.65-1.0.

[0019] Preferably, in step S2, the particle swarm optimization algorithm is an improved algorithm with inertia weight, and the particle update formula is:

[0020]

[0021]

[0022] wherein, is the inertia weight, and is the acceleration constant, and is a random number in the interval [0, 1], is the individual historical optimal position of the particle, is the global historical optimal position of the particle swarm.

[0023] Preferably, in step S4, the blade installation angle includes a blade leading edge angle and a blade trailing edge angle; wherein the blade leading edge angle refers to the rotor blade inlet geometric angle, and the blade trailing edge angle refers to the rotor blade outlet geometric angle; the sampling range of the blade leading edge angle is 60°-110°, the sampling range of the blade trailing edge angle at the hub is 50°-90°, and the sampling range of the blade trailing edge angle at the tip is 35°-75°.

[0024] Preferably, in step S5, the three-dimensional computational fluid dynamics numerical simulation adopts a k-ω SST turbulence model, the working fluid is organic working medium R245fa, the real gas properties are provided by the REFPROP database, and the independence of the calculation grid is verified.

[0025] Preferably, in step S6, the correlation coefficient between the prediction results of the Kriging response surface surrogate model and the three-dimensional computational fluid dynamics numerical simulation results is not less than 92.3%.

[0026] Preferably, in step S7, the optimization result shows that when the working medium attack angle determined by the blade leading edge angle is in the range of -10° to 0°, the flow loss is the smallest; when the blade leading edge angle is greater than 74°, the increase of the attack angle will cause the blade suction surface flow separation to be intensified, and the flow loss to be increased.

[0027] The application provides a radial turbine optimization design method based on an advanced optimization algorithm.

[0028] 1. The application constructs a hierarchical collaborative design system of "one-dimensional global optimization-three-dimensional fine optimization". Through one-dimensional rapid analysis combined with intelligent algorithms by integrating physical loss models, the macro high-efficiency design direction is efficiently locked, the search space and calculation burden of subsequent three-dimensional optimization are significantly reduced, the problems of large dimensional span and high calculation cost in traditional methods are overcome, and the first balance between design efficiency and accuracy is realized, thereby providing a feasible technical path for the automatic optimization of complex turbine systems.

[0029] 2. The application converts the expensive and time-consuming three-dimensional CFD simulation into an efficient optimization engine through high-precision proxy model technology; the Kriging response surface model constructed based on a small amount of high-fidelity simulation data has a prediction accuracy related to the CFD result of 92.3% or more, so that it is possible to complete thousands of times of performance evaluation of blade geometric parameters within an acceptable calculation cost, and the fine global optimization of the three-dimensional blade installation angle is realized, thereby significantly improving the performance potential of the final design.

[0030] 3. The application refines design criteria and mechanism understanding with clear physical significance from the optimization process; the best matching range of key aerodynamic parameters and the internal relationship between the blade leading edge angle and flow loss are revealed through system optimization, thereby providing scientific guidance that can be directly applied to radial turbine design. These findings not only improve the one-time success rate of design, but also deepen the understanding of internal flow mechanism, thereby laying a theoretical foundation for subsequent performance improvement and technical innovation. BRIEF DESCRIPTION OF DRAWINGS

[0031] Figure 1 The application provides a design method flowchart. DETAILED DESCRIPTION

[0032] The technical solutions in the embodiments of the application will be described clearly and completely below with reference to the drawings in the embodiments of the application. Obviously, the described embodiments are only part of the embodiments of the application, rather than all the embodiments of the application. Based on the embodiments in the application, all other embodiments obtained by those skilled in the art without creative labor fall within the protection scope of the application.

[0033] Please refer to Figure 1The application provides a technical scheme: a centripetal turbine optimization design method based on an advanced optimization algorithm, comprising the following steps:

[0034] S1: a one-dimensional aerodynamic thermodynamic mean diameter design model is constructed, and the model is integrated with loss models including flow passage loss, blade tip gap loss, trailing edge loss and outlet kinetic energy loss;

[0035] S2: the one-dimensional mean diameter design model of step S1 is coupled with a particle swarm optimization algorithm to maximize the total static efficiency, global optimization is performed on dynamic parameters, thermodynamic parameters and geometric parameters affecting the performance of the turbine, and a one-dimensional preliminary design parameter set maximizing the total static efficiency is obtained;

[0036] S3: based on the one-dimensional preliminary design parameter set maximizing the total static efficiency obtained in step S2, a three-dimensional turbine initial geometric model is generated;

[0037] S4: based on the three-dimensional turbine initial geometric model generated in step S3, the blade installation angle is used as an optimization variable, a Latin hypercube sampling method based on space filling is used to sample in the design space, and a plurality of three-dimensional blade geometric variant schemes are generated;

[0038] S5: three-dimensional computational fluid dynamics numerical simulation is performed on each three-dimensional blade geometric variant scheme generated in step S4, and the total static efficiency and net output power are obtained;

[0039] S6: based on the plurality of simulation data obtained in step S5, a Kriging response surface proxy model with the blade installation angle as the input and the total static efficiency and net output power as the output is constructed;

[0040] S7: the total static efficiency and net output power predicted by the Kriging response surface proxy model are used as optimization objectives, the particle swarm optimization algorithm is applied again to optimize the blade installation angle, and the final three-dimensional blade optimal installation angle is determined;

[0041] S8: according to the three-dimensional blade optimal installation angle determined in step S7, the three-dimensional turbine model is updated, and the final optimization design of the centripetal turbine is completed.

[0042] In this embodiment, the application builds a hierarchical, multi-scale and optimized closed-loop collaborative design framework, systematically integrating the complete design chain from macroscopic aerodynamic layout to microscopic three-dimensional geometry; through the integration of one-dimensional intermediate model for rapid analysis and global optimization based on intelligent algorithm, the high-efficiency macroscopic design direction is locked under the premise of physical authenticity, the design space is greatly reduced and the optimization starting point is significantly improved; on this basis, the three-dimensional refinement design stage is entered, the high-precision performance proxy model is built through space sampling and high-fidelity simulation, and the secondary intelligent optimization is implemented on the model, so as to complete the fine tuning of three-dimensional blade geometry with affordable computing cost; the whole process forms a complete iteration of "macroscopic screening-microscopic optimization-closed-loop verification", effectively solving the problems of disconnection between one-dimensional and three-dimensional design, and difficult balance between calculation efficiency and design precision in traditional turbine design; through the rapid screening of one-dimensional model, the blindness and computational burden of subsequent three-dimensional optimization are significantly reduced, and through the proxy model technology, the expensive three-dimensional simulation is converted into an efficient optimization engine, so as to realize the optimization balance of design efficiency and final performance as a whole, making it possible to systematically approach the global optimal design within the range of engineering feasibility, and significantly improving the aerodynamic design level and product development efficiency of radial turbine.

[0043] Specifically, in step S1, the flow passage loss in the loss model is calculated by the following formula:

[0044]

[0045] wherein, is an empirical coefficient, and are the hydraulic length and hydraulic diameter of the rotor, respectively, and are the inlet radius and outlet meridian plane average radius of the rotor, respectively, is the average relative airflow angle at the outlet of the rotor, is the blade height at the outlet of the rotor, and are the relative velocities at the inlet and outlet of the rotor, respectively.

[0046] In this embodiment, the comprehensive influence of the friction of the inner wall of the flow passage is quantified by introducing the ratio of the hydraulic length of the rotor to the hydraulic diameter; the square term of the ratio of the radius of the rotor outlet to the radius of the rotor inlet is used to represent the mismatch of the centrifugal force and the pressure gradient caused by the change of the curvature of the flow passage, so as to capture the main cause of the secondary flow loss; at the same time, the cosine value of the average relative airflow angle of the rotor outlet and the ratio of the outlet blade height to the hydraulic length are combined, which can effectively reflect the influence of the contraction or expansion form of the flow passage on the flow separation trend and the development of vortex system; finally, the weighted average of the square term of the relative velocity at the inlet and outlet of the rotor is directly taken into account, which directly takes into account the unevenness of the main flow velocity distribution and the conversion efficiency of kinetic energy in the flow passage; the formula is not a purely empirical correlation formula, but a semi-empirical model with clear physical meaning, each term of which corresponds to a specific flow loss mechanism; compared with the simple loss coefficient estimation method, the model can more accurately predict and quantify the energy loss in the flow passage under different design schemes, and is especially suitable for evaluating the complex flow effects caused by the change of geometric parameters (such as the curvature of the flow passage, the blade height, etc.), thereby providing a solid physical basis for subsequent one-dimensional global optimization based on the loss model and ensuring the correctness of the one-dimensional optimization direction.

[0047] Specifically, in step S2, the key design parameter range that maximizes the total static efficiency determined by global optimization includes: the isentropic speed ratio is 0.68-0.71, the absolute airflow angle of the rotor inlet is 66°-78°, the degree of reaction is 0.45-0.65, and the meridian velocity ratio of the rotor is 0.65-1.0.

[0048] In this embodiment, the optimized value range of the determined isentropic speed ratio, rotor inlet absolute flow angle, degree of reaction and rotor meridian velocity ratio is a high-efficiency cooperative matching relationship among the key aerodynamic parameters revealed by the global optimization; the isentropic speed ratio defines an optimal matching window of turbine rotating speed and available energy of working medium, ensuring that the impeller peripheral speed is in effective conversion of enthalpy drop while avoiding excessive friction and secondary flow loss; the selection of the rotor inlet absolute flow angle directly determines the angle of attack and angle of incidence of airflow entering the rotor, and its optimal range ensures that the airflow enters the rotor passage smoothly, minimizing the inlet impact and separation loss; the degree of reaction defines the distribution ratio of expansion work between the static blade and the moving blade, and the optimized range makes the pressure gradient and velocity distribution in the flow passage more reasonable, ensuring sufficient energy conversion capacity and controlling the subsonic flow loss and the rationality of the blade root reaction degree; and the rotor meridian velocity ratio reflects the expansion or convergence trend of the flow passage in the meridian plane, and a value close to 1 means that the flow passage tends to be an equal cross-section, which is conducive to maintaining the stability of the boundary layer and suppressing flow separation; these parameters are not a simple collection of isolated optimal values, but a parameter matching system with optimal cooperation; in the method of the application, the particle swarm optimization algorithm is used to perform global cooperative optimization on the above parameters, which is equivalent to automatically finding a design point that minimizes the comprehensive loss of complex flow (including secondary flow, shock wave, separation, etc.) in the turbine; this parameter range can not only be used as the core design criterion of the application to directly guide the initial design of the high-efficiency centripetal turbine, avoiding the large number of blind attempts of the traditional trial-and-error method, but more importantly, it proves that through systematic global optimization, the design limitations based on isolated parameter experience can be overcome, and the global performance optimal region under the cooperative action of multiple parameters can be found and locked.

[0049] Specifically, in step S2, the particle swarm optimization algorithm is an improved algorithm with inertia weight, and the particle update formula is:

[0050]

[0051]

[0052] wherein, is an inertia weight, and are acceleration constants, and are random numbers in the interval [0, 1], is the individual historical optimal position of the particle, is the global historical optimal position of the particle swarm.

[0053] In this embodiment, through the synergistic effect of the inertia weight term, the individual cognitive term and the social cognitive term introduced and dynamically adjusted, the particle swarm (i.e. multiple parallel search design schemes) is guided to efficiently and intelligently explore and converge in a complex high-dimensional multi-peak design space; the inertia weight gives the particle a tendency to maintain its previous motion direction and speed, and setting a larger value in the early optimization stage helps to enhance the global exploration ability, widely scan the design space and avoid premature convergence to local optimum; while reducing its value in the later optimization stage, it can strengthen the local fine search and promote the particle swarm to accurately locate the optimal solution near the discovered superior area; the individual cognitive acceleration term drives the particle to learn from its own historical optimal position, retaining excellent individual experience; the social cognitive acceleration term drives all particles to gather around the historical optimal position discovered by the entire population, achieving effective sharing of population information and synergistic evolution; the multiplication of the two acceleration constants and two random numbers in the interval [0, 1] determines the random weight of the influence of individual experience and social experience on the flight direction of the particle in each iteration, and this randomness enhances the potential of the algorithm to jump out of the local optimum; compared with the basic particle swarm algorithm, it exhibits superior global search ability and convergence stability when solving complex optimization problems such as centripetal turbine design, which has high dimensionality, strong nonlinearity, multivariate coupling and possibly multiple local extrema. Its adaptive adjustment of the exploration-development balance mechanism enables the algorithm to effectively overcome the defects of traditional gradient-based optimization methods, such as being prone to fall into local optimum and being sensitive to initial value, in both one-dimensional global parameter optimization (step S2) and three-dimensional blade angle fine optimization (step S7) stages; thereby significantly improving the probability of the method described in the present invention to reliably approach or find the global optimum or high-performance approximate solution in the entire optimization design process, ensuring the superior performance of the final design result.

[0054] Specifically, in step S4, the blade installation angle includes a blade leading edge angle and a blade trailing edge angle; wherein the blade leading edge angle refers to the rotor blade inlet geometric angle, and the blade trailing edge angle refers to the rotor blade outlet geometric angle; the sampling range of the blade leading edge angle is 60°-110°, and the sampling range of the blade trailing edge angle at the hub is 50°-90° and at the tip is 35°-75°.

[0055] In this embodiment, the blade leading edge angle directly determines the initial geometric angle of attack of the working medium entering the rotor passage, and its value has a decisive influence on the impingement loss of the cascade inlet, the starting of flow separation and the passage shock wave structure; the blade trailing edge angle controls the outflow direction and absolute speed of the gas leaving the rotor, significantly affecting the residual velocity loss and the uniformity of the downstream flow field; due to the centrifugal effect, the secondary flow intensity and the passage pressure gradient change sharply along the blade height direction, resulting in significant differences in flow field structure from the hub to the tip, so different optimization ranges are set for the trailing edge angle at the hub and the tip to allow the blade profile to be adaptively twisted in the spanwise direction, thereby more targetedly optimizing the local flow at different radius positions; the leading edge angle sampling range is set relatively wide to fully explore its control potential on the inlet flow state; and the trailing edge angle is differentiated at the hub (a relatively large range) and the tip (a relatively small range) to effectively reconcile the contradiction between suppressing strong secondary flow at the hub area and optimizing flow deviation at the tip area; so that the subsequent Latin hypercube sampling and response surface construction process can explore in a scientifically defined and physically meaningful high-dimensional design subspace, avoiding wasting computing resources in an invalid or inefficient geometric space, ensuring that the series of three-dimensional blade geometric variant schemes generated have the potential for high performance, laying a solid sample foundation for efficiently constructing an accurate proxy model and ultimately locking the globally optimal three-dimensional blade geometry, and is a key link for successful transition from one-dimensional macroscopic parameter optimization to three-dimensional refined modeling optimization.

[0056] Specifically, in step S5, the three-dimensional computational fluid dynamics numerical simulation adopts a k-omega SST turbulence model, the working fluid is organic working medium R245fa, the real gas properties are provided by a REFPROP database, and the independence of the calculation grid is verified.

[0057] In this embodiment, by using the k-omega SST turbulence model, taking the organic working fluid R245fa provided by the REFPROP database as the working fluid, and performing grid independence verification, three-dimensional computational fluid dynamics numerical simulation is carried out. The principle is to build a high-fidelity physical simulation environment to accurately capture the real flow details inside the radial turbine, which are highly complex and may exist transonic, strong three-dimensional, strong shear and strong curvature effects. The k-omega SST turbulence model intelligently combines the high accuracy of the standard k-omega model in the near-wall region and the robustness of the k-epsilon model in the mainstream region, and introduces shear stress transport limitation, making it particularly suitable for simulating the internal flow field of turbomachinery with strong adverse pressure gradient, flow separation and transition. It can more accurately predict the boundary layer development, separation starting position and vorticity dissipation, so as to reliably calculate the flow loss caused by complex vortex system structure such as secondary flow, tip leakage vortex and passage vortex; Using the real gas model and calling the REFPROP database to obtain the properties of the organic working fluid R245fa is to accurately take into account the significant nonlinear changes of specific heat capacity, sound speed, density and other thermophysical properties with temperature and pressure during the turbine expansion process, especially under near-critical or trans-critical conditions. This change will deeply affect the shock position, expansion process, and finally the isentropic efficiency and output power. Ignoring its real gas effect will introduce significant errors; The grid independence verification is to ensure that the calculation accuracy of the numerical solution is not subject to the subjective influence of the discrete grid size. By systematically encrypting the grid until the change of the key performance parameters (such as efficiency and pressure ratio) is lower than the acceptable threshold, it is confirmed that the simulation results obtained are reliable approximations of the physical flow, rather than artifacts of numerical discretization; Provide accurate, reliable and physically consistent data sources for the final performance evaluation and blade angle optimization of the entire optimization process.

[0058] Specifically, in step S6, the correlation coefficient between the prediction results of the Kriging response surface surrogate model and the three-dimensional computational fluid dynamics numerical simulation results is not less than 92.3%.

[0059] In this embodiment, by statistically interpolating the complex nonlinear relationship between the design parameters and the performance indicators, an approximate model with high calculation efficiency and accurate prediction is established. Its core advantage is to drive a large number of low-cost optimization iterations with a small amount of high-cost simulation data: this model can replace expensive CFD simulation, allowing subsequent optimization to perform thousands of performance evaluations in a short time, enabling fine global optimization of blade installation angle and providing reliable protection for obtaining high-performance design schemes.

[0060] Specifically, in step S7, the optimization results show that when the working fluid attack angle determined by the blade leading edge angle is in the range of-10° to 0°, the flow loss is the smallest; when the blade leading edge angle is greater than 74°, the increase of the attack angle will lead to the aggravation of the flow separation on the blade suction surface and the increase of the flow loss.

[0061] In this embodiment, the geometric design of the blade leading edge angle directly determines the actual angle of attack of the incoming flow into the rotor passage; when this angle of attack is in a small negative value range (-10° to 0°), it means that the airflow enters the flow passage in a way of slightly "ramming" the blade pressure surface; this flow state is most conducive to forming a smooth flow velocity distribution in the blade leading edge area, which can effectively avoid premature flow separation on the suction surface, and ensure that the pressure surface does not have excessive flow stagnation, so that the boundary layer in the entire passage remains attached and stable, and the flow loss is minimized; when the blade leading edge angle is designed too large (more than about 74°), even if the incoming flow direction does not change, the geometric angle of attack will increase significantly to a positive angle of attack; at this time, the airflow will directly impact the blade suction surface leading edge, forming a strong adverse pressure gradient area, which is prone to induce flow separation on the suction surface side, and even generate strong separation vortex; these separation structures not only block the effective flow area of the flow passage, but also cause huge energy dissipation by mixing with the main flow, resulting in a sharp increase in passage loss, secondary flow loss and mixing loss; by converting the optimized empirical conclusion into a physical criterion that can directly guide the design, the "forbidden zone" (avoiding too large) and "high efficiency zone" (pursuing a small negative angle of attack) of the leading edge angle design are clearly pointed out, so that the designer can actively guide the blade leading edge geometric design to the low loss area that is conducive to suppressing flow separation and maintaining laminar attachment; not only improves the scientificity and first success rate of blade design, but more importantly, this flow mechanism understanding refined from global optimization deepens the understanding of the source of internal losses in the radial turbine rotor, providing theoretical support for subsequent performance improvement and new design.

[0062] It should be noted that, in this document, the terms such as first and second are used merely to distinguish one entity or action from another, and do not necessarily require or imply that these entities or actions are in any way mutually exclusive or in any way arranged in any sequence. Moreover, the terms "comprise", "include" or any other variant thereof are intended to cover non-exclusive inclusion, so that a process, method, article or device including a series of elements includes not only those elements, but also other elements not explicitly listed or inherent to such a process, method, article or device.

[0063] Although embodiments of the present application have been shown and described, it will be understood by those skilled in the art that various changes, modifications, substitutions and alterations can be made therein without departing from the principles and spirit of the application, the scope of which is defined by the appended claims and their equivalents.

Claims

1. A method for optimization design of a radial turbine based on an advanced optimization algorithm, characterized in that, The method comprises the following steps: S1: constructing a one-dimensional aerothermodynamic mean diameter design model, the model integrating loss models including flow passage loss, blade tip clearance loss, trailing edge loss and outlet kinetic energy loss; S2: coupling the one-dimensional mean diameter design model of step S1 with a particle swarm optimization algorithm to globally optimize dynamic parameters, thermodynamic parameters and geometric parameters affecting turbine performance to obtain a one-dimensional preliminary design parameter set maximizing total static efficiency; S3: generating a three-dimensional turbine initial geometric model based on the one-dimensional preliminary design parameter set maximizing total static efficiency obtained in step S2; S4: sampling in a design space by using a space-filling-based Latin hypercube sampling method to generate multiple sets of three-dimensional blade geometric variant schemes with blade installation angle as an optimization variable based on the three-dimensional turbine initial geometric model generated in step S3; S5: performing three-dimensional computational fluid dynamics numerical simulation on each set of three-dimensional blade geometric variant schemes generated in step S4 to obtain total static efficiency and net output power thereof; S6: constructing a Kriging response surface proxy model with blade installation angle as input and total static efficiency and net output power as output based on the multiple sets of simulation data obtained in step S5; S7: taking total static efficiency and net output power predicted by the Kriging response surface proxy model as optimization objectives, and again applying the particle swarm optimization algorithm to optimize the blade installation angle to determine a final three-dimensional blade optimal installation angle; S8: updating the three-dimensional turbine model according to the three-dimensional blade optimal installation angle determined in step S7 to complete the final optimization design of the radial-inflow turbine.

2. The centripetal turbine optimization design method based on advanced optimization algorithm according to claim 1, characterized in that: In step S1, flow passage loss in the loss model This is calculated by the following equation: ； wherein, is an empirical coefficient, and are the hydraulic length and hydraulic diameter of the rotor, respectively, and are the rotor inlet radius and the outlet meridional mean radius, respectively, is the rotor outlet mean relative flow angle, is the rotor outlet blade height, and are the relative velocities at the rotor inlet and outlet, respectively.

3. The method of claim 1, wherein: The key design parameter ranges for maximum overall static efficiency, determined through global optimization in step S2, include: isentropic speed ratio 0.68-0.71, rotor inlet absolute flow angle 66°-78°, degree of reaction 0.45-0.65, rotor meridional velocity ratio 0.65-1.

0.

4. The centripetal turbine optimization design method based on advanced optimization algorithm according to claim 1, characterized in that: In step S2, the particle swarm optimization algorithm is an improved algorithm with inertia weight, and the particle updating formula thereof is: ； ； where, is the inertia weight, and is the acceleration constant, and is a random number in the interval [0, 1], is the individual best position of a particle, is the global best position of the swarm.

5. The method of claim 1, wherein: In step S4, the blade installation angle includes a blade leading edge angle and a blade trailing edge angle; wherein the blade leading edge angle refers to a rotor blade inlet geometric angle, and the blade trailing edge angle refers to a rotor blade outlet geometric angle; the sampling range of the blade leading edge angle is 60°-110°, and the sampling range of the blade trailing edge angle at the hub is 50°-90° and at the blade tip is 35°-75°.

6. The method of claim 1, wherein: In step S5, the three-dimensional computational fluid dynamics numerical simulation adopts a k-ω SST turbulence model, the working fluid is organic working medium R245fa, the real gas properties are provided by a REFPROP database, and the calculation grid is verified for independence.

7. The method of claim 1, wherein: In step S6, the correlation coefficient of the prediction results of the Kriging response surface proxy model and the three-dimensional computational fluid dynamics numerical simulation results is not less than 92.3%.

8. The method of claim 1, wherein: In step S7, the optimization results show that when the working medium attack angle determined by the blade leading edge angle is in the range of -10° to 0°, the flow loss is the smallest; when the blade leading edge angle is greater than 74°, the increase of the attack angle will cause the blade suction surface flow separation to be intensified, and the flow loss to be increased.

Images