Numerical design method for centrifugal diffuser

Abstract

The invention relates to a diffuser parameter design technology, and provides a centrifugal diffuser numerical value design method in order to solve the technical problems that a diffuser designed through a traditional centrifugal compressor diffuser design method is limited in working condition adaptability and low in efficiency. Comprising the following steps that S1, initial design is conducted on the diffuser, and the inlet diameter, the outlet diameter, the blade number, the installation angle and the blade profile of the diffuser are determined; s2, determining an optimal mounting angle and the number of blades by using a global optimization method; s3, the geometry of the diffuser is set according to the optimal mounting angle and the number of the blades; s4, parameterizing the blade profile and determining control parameters of the blade profile; s5, determining an optimal blade profile by using a global optimization method through the control parameters in the step S4; s6, further setting an optimal diffuser geometry according to the optimal blade profile; and S7, outputting the optimal diffuser geometry, and ending the design.

Description

Technical Field

[0001] The present invention relates to a method for designing diffuser parameters, and in particular to a numerical design method for a centrifugal diffuser. Background Art

[0002] Centrifugal compressors are fluid machinery widely used in the industrial field. Their performance directly affects energy conversion efficiency and system stability. The diffuser is one of the core components of the centrifugal compressor, which can convert the kinetic energy of the high-speed airflow at the impeller outlet into pressure energy to achieve fluid pressurization.

[0003] Therefore, the design of diffuser blades is of great significance in improving energy conversion efficiency, stabilizing airflow and improving system performance. The traditional diffuser design method is to determine the basic structural parameters of the diffuser based on previous design experience and experimental data, such as the diffuser's inlet diameter, outlet diameter, number of blades and blade angle. Although this method is simple and easy to implement, it is difficult to accurately capture the complex flow mechanism and lacks accurate analysis of the internal flow field. It is only suitable for the design of centrifugal compressor diffusers with relatively simple working conditions and low performance requirements. It has the defects of limited adaptability to working conditions, low efficiency, and difficulty in responding to new technologies and new demands. Summary of the Invention

[0004] The purpose of the present invention is to solve the technical problems of limited adaptability to working conditions and low efficiency of diffusers designed by traditional centrifugal compressor diffuser design methods, and to provide a numerical design method for centrifugal diffusers.

[0005] To achieve the above objectives, the present invention provides the following technical solutions:

[0006] A numerical design method for a centrifugal diffuser is characterized in that it comprises the following steps:

[0007] S1: Performing an initial design of the centrifugal diffuser according to existing design methods, determining the inlet diameter, outlet diameter, number of blades, installation angle, and blade profile of the centrifugal diffuser, and obtaining the initial design parameters of the centrifugal diffuser;

[0008] S2: Based on the initial design parameters of the centrifugal diffuser, the optimal installation angle and number of blades are determined using a global optimization method;

[0009] S3: Preliminarily setting the geometry of the centrifugal diffuser according to the optimal installation angle and number of blades, and determining whether its aerodynamic performance meets the first preset requirement. If so, executing step S4; if not, returning to step S2 and re-determining the optimal installation angle and number of blades;

[0010] S4: parameterize the blade profile and determine the control parameters of the blade profile;

[0011] S5: Optimizing the blade profile parameters using a global optimization method based on the control parameters in step S4 to determine the optimal blade profile;

[0012] S6: updating the centrifugal diffuser geometry in step S3 according to the optimal blade profile to obtain a new centrifugal diffuser geometry, and determining whether the aerodynamic performance of the new centrifugal diffuser geometry meets the second preset requirement. If so, executing step S7; if not, returning to step S5 and re-determining the optimal blade profile.

[0013] S7: Output the new centrifugal diffuser geometry and complete the centrifugal diffuser numerical design.

[0014] Furthermore, step S2 is specifically as follows:

[0015] S2.1, using the installation angle and the number of blades as design variables and, based on the initial design parameters of the centrifugal diffuser determined in step S1, using a Latin hypercube test to determine an experimental sample with the installation angle and the number of blades as design variables, and calculating six performance parameters of the experimental sample, specifically including near-surge point efficiency, near-surge point total pressure ratio, design point efficiency, design point total pressure ratio, near-blocking point efficiency, and near-blocking point total pressure ratio;

[0016] S2.2 Based on the experimental sample data, the response surface model of the corresponding design variables and performance parameters is established as follows:

[0017] y(x)=f(x) T β+z(x)

[0018] Where x is the input vector, which consists of the installation angle and the number of blades; f(x) is the regression basis function vector; β is the regression coefficient vector; f(x) T β is the overall trend of the response variable; z(x) is a zero-mean Gaussian random process, and y(x) is the output vector, which is the six performance parameters mentioned above.

[0019] S2.3 Based on the response surface model of design variables and performance parameters, a multi-objective genetic algorithm is used to find the optimal solution to obtain the optimal installation angle and number of blades.

[0020] Furthermore, step S3 is specifically as follows:

[0021] The centrifugal diffuser geometry is set according to the optimal installation angle and number of blades obtained in step S2, and its model-level aerodynamic performance is calculated through flow field analysis to determine whether the aerodynamic performance is within the first preset requirement. If so, step S4 is executed. If not, the process returns to step S2 and increases the number of samples or adjusts the optimization target to redetermine the optimal installation angle and number of blades until the calculation result is within the first preset requirement.

[0022] Furthermore, step S4 is specifically as follows:

[0023] One to five sections along the blade height are selected as the surfaces to be parameterized. The two-dimensional blade profiles of different sections are expressed in the form of superimposed thickness distribution of the mid-camber angle. The blade profile is parameterized by changing the distribution of the mid-camber angle along the chord line, and then the control parameters of the blade profile are determined.

[0024] Furthermore, in step S4, the specific method for achieving blade profile parameterization by changing the distribution of the camber angle along the chord line is:

[0025] The Bezier curve is used to fit the camber angle distribution curve. The order of the Bezier curve is determined based on the trade-off between fitting accuracy and the number of design variables. The camber angle distribution is then changed by changing the positions of the Bezier curve control points to achieve blade parameterization.

[0026] Furthermore, in step S4, the control parameters of the blade profile are selected as the ordinate values ​​of the control points of the Bezier curve, and the variation range of the control parameters is controlled within 20%.

[0027] Furthermore, step S5 is specifically as follows:

[0028] S5.1 Based on the control parameters determined in step S4, Latin hypercube experimental design is used to arrange experimental samples with the ordinate values ​​of the Bezier curve as design variables, and six performance parameters of the control parameter experimental samples are calculated, including the efficiency near the gasping point, the total pressure ratio near the gasping point, the efficiency at the design point, the total pressure ratio at the design point, the efficiency near the blocking point, and the total pressure ratio near the blocking point;

[0029] S5.2 Based on the six performance parameters of the control parameter experimental sample, establish the corresponding response surface model of the control parameters and performance parameters;

[0030] S5.3 Based on the response surface model of control parameters and performance parameters, a multi-objective genetic algorithm is used to find the optimal solution and determine the optimal blade shape.

[0031] Furthermore, step S6 is specifically as follows:

[0032] Based on the optimal blade profile obtained in step S5, the centrifugal diffuser geometry in step S3 is updated to obtain a new centrifugal diffuser geometry. The model-level aerodynamic performance of the new centrifugal diffuser geometry is then calculated through flow field analysis to determine whether the aerodynamic performance is within the second preset requirement. If so, step S7 is executed. If not, the process returns to step S5 and redetermines the optimal blade profile until the calculation result is within the second preset requirement.

[0033] Compared with the prior art, the present invention has the following beneficial effects:

[0034] (1) The present invention adopts a two-step optimization design process for the numerical design of the diffuser, that is, optimizing the installation angle and the number of blades, the main influencing factors, based on sensitivity analysis, and then fine-tuning the blade profile. Compared with the traditional diffuser design method, the diffuser is optimized, so that the designed diffuser model has higher level efficiency and a wider range of operating conditions;

[0035] (2) The present invention designs a two-step optimization process, which can adjust and optimize the optimization parameters of the two steps respectively, reduce the sample calculation amount, and make the optimization process easier to control and manage. That is, the technical solution is easy to standardize and program, and the design process has low dependence on the designer's experience. BRIEF DESCRIPTION OF THE DRAWINGS

[0036] Figure 1 This is a flow chart of a numerical design method for a centrifugal diffuser according to the present invention;

[0037] Figure 2 Schematic diagram of the cross section of the blade profile along the blade height direction in step S4 of the embodiment of the numerical design method for a centrifugal diffuser of the present invention;

[0038] Figure 3 Schematic diagram of the control angle distribution of the fourth-order Bezier curve in step S4 of the embodiment of the numerical design method for a centrifugal diffuser of the present invention. DETAILED DESCRIPTION

[0039] In order to make the purpose, technical solutions and advantages of the present invention more clear, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. It should be noted that the description is for explanation rather than limitation of the present invention.

[0040] See attached Figure 1 As shown, this embodiment provides a numerical design method for a centrifugal diffuser, and the specific steps are as follows:

[0041] S1: Initial Centrifugal Diffuser Design

[0042] The initial design of the centrifugal diffuser is performed using traditional design methods, such as using one-dimensional or two-dimensional theoretical design methods to determine the inlet diameter, outlet diameter, number of blades, installation angle and blade profile of the centrifugal diffuser.

[0043] S2: Multi-objective optimization of installation angle and number of blades

[0044] The design of the centrifugal diffuser aims to improve the efficiency of the model stage and widen the operating range. Therefore, the efficiency and total pressure ratio of the three operating points near the surge, design, and near the blockage are considered during the design optimization process. A total of six parameters are considered, including the efficiency near the surge point, the total pressure ratio near the surge point, the efficiency at the design point, the total pressure ratio at the design point, the efficiency near the blockage point, and the total pressure ratio near the blockage point.

[0045] According to the specific situation, the appropriate combination of 6 parameters is used as the goal and constraint of the design optimization problem, and the global optimization method is used for optimization. The specific steps are as follows:

[0046] ① Using the installation angle and the number of blades as design variables, the Latin hypercube test was used to design the experimental sample and calculate the six performance parameters of the sample;

[0047] ②Based on the sample data, the response surface model of design variables and performance parameters is established as follows:

[0048] y(x)=f(x) T β+z(x)

[0049] Where x is the input vector, which consists of the installation angle and the number of blades; f(x) is the regression basis function vector; β is the regression coefficient vector, which is estimated by the least squares method; f(x) T β describes the overall trend of the response variable; z(x) is a zero-mean Gaussian random process, and y(x) is the output vector, which is the six performance parameters mentioned above.

[0050] ③ A multi-objective genetic algorithm is used to find the optimal solution and determine the optimal installation angle and number of blades. The specific optimization process is as follows:

[0051] Step 1: Randomly generate an initialized population, determine the population size, crossover probability, mutation probability, and maximum number of iterations, and use the response surface model of the design variables and performance parameters in step 2 above to calculate the objective function of each individual;

[0052] Step 2: The individuals in the population are layered according to the Pareto dominance relationship. The fitness priority of the individuals in each layer decreases in turn. Then, in each layer, the distance between each individual and its adjacent solutions in the target space is calculated as the crowding index. Individuals with a high crowding index are sparsely distributed in the target space and have a higher retention value.

[0053] Step 3: Randomly select two individuals, give priority to the individual with a higher non-dominated level, and if the levels are the same, select the individual with a higher congestion level. The current non-dominated solution is directly retained to the next generation to avoid the loss of excellent solutions.

[0054] Step 4: Recombining the parent individuals to generate offspring individuals, then slightly perturbing the offspring individuals to enhance population diversity, merging the parent and offspring individuals to form a temporary population, and using the response surface model of the design variables and performance parameters in step ② above to calculate the objective function of the population individuals;

[0055] Step 5: Merge the temporary population with the current population, recalculate the non-dominated sort and crowding degree, select individuals according to the non-dominated level and crowding degree, and form a new generation population;

[0056] Step 6: When the maximum number of iterations is reached or the population converges, the optimization process terminates, otherwise it returns to step 2.

[0057] S3: Aerodynamic verification

[0058] The diffuser geometry is set according to the optimal installation angle and number of blades obtained in step S2 above. Then, the flow field analysis is performed through CED to calculate the model-level aerodynamic performance and verify the rationality of the geometry. If the calculation result meets the first preset requirement, return to step S2 above and re-optimize. If they are consistent, the subsequent steps are carried out based on this geometry.

[0059] S4: Diffuser Blade Parameterization

[0060] During the design process, the blade profile needs to be parameterized to achieve an accurate description of the blade profile with as few parameters as possible:

[0061] ①Determine the profile

[0062] Select 1-5 sections along the blade height as parameterized surfaces, as shown in the attached figure. Figure 2 Taking the figure as an example, section 1 is applicable to the case of a two-dimensional blade whose blade shape does not change along the blade height. As the number of sections increases, the blade shape changes more in the blade height direction, and the possibility of obtaining the optimization target is greater. However, the design variables will also increase accordingly, and the optimization difficulty will increase.

[0063] ② Surface parameterization

[0064] The two-dimensional blade profiles of different cross sections are expressed in the form of superimposed thickness distribution of the mid-camber angle, and the distribution of the mid-camber and thickness along the chord are given respectively. The suction and pressure surface profiles of the blade can be obtained by adding the mid-camber position coordinates and the thickness value at the same chord position.

[0065] In order to ensure the strength of the blade and keep the blade thickness distribution unchanged, the blade profile is designed by changing the angle distribution along the chord line. Specifically, the Bezier curve is used to fit the mid-arc angle distribution curve. The order of the Bezier curve is determined based on the trade-off between fitting accuracy and the number of design variables, which is generally 4-7. Then, by changing the position of the Bezier curve control point and the angle distribution, the blade profile is parameterized.

[0066] The definition of the Bezier curve is as follows:

[0067]

[0068] Where n is the order of the curve; x(t) is any point on the Bezier curve; t∈(0,1) is a parameter variable corresponding to the discrete data point fitted on the mid-arc angle distribution curve; B i,n is the Bernstein basis function; is the number of combinations; P i The control vertices of the Bezier curve.

[0069] ③Determine design variables

[0070] Through the parameterization process of the previous step ②, it is easy to know that a set of Bezier curve control points corresponds to an angle distribution curve; Figure 3 Taking the fourth-order Bezier curve shown in as an example, during the optimization process, the vertical coordinate of the control point of the Bezier curve is taken as the control parameter, and the variation range of the control parameter is usually controlled within 20%.

[0071] S5: Multi-objective optimization of blade profile

[0072] ① Based on the control parameters obtained in step S4 above, Latin hypercube experimental design is used to arrange the experimental samples and calculate the six performance parameters of the samples;

[0073] ② Based on the sample data, a response surface model of the control parameters and performance parameters is established, where the input vector x is the ordinate of the control point of the Bezier curve;

[0074] ③ Use a multi-objective genetic algorithm to find the optimal solution. The specific steps are as in step S2 above to determine the optimal blade shape.

[0075] S6: Pneumatic and geometric verification

[0076] Analyze the optimal diffuser geometry to determine the geometric reliability; and perform model-level flow field analysis to determine the final aerodynamic performance of the model level.

[0077] Specifically, the diffuser geometry in step S3 is set according to the optimal blade profile obtained in step S5. The rationality of the geometry is determined by observing whether the geometry is distorted and performing strength calculation verification. At the same time, the flow field analysis is performed on the model level, and the aerodynamic performance of the model level is calculated. If the calculation result meets the second preset requirement, the process returns to step S5 and searches for the optimal solution again. If the result meets the second preset requirement, the subsequent steps are performed based on this geometry.

[0078] S7: Output Diffuser Geometry

[0079] The optimal diffuser geometry obtained in step S6 is output, and the design is completed.

[0080] Simulation test results of the technical solution of the present invention:

[0081] This invention was applied to the design of a centrifugal compressor diffuser. Simulation with ANSYS CFX software showed that, compared with traditional methods, the model-level variable efficiency and total pressure ratio increased by 2%-4% within a stable operating range. This design process has a high degree of automation and relies less on the designer's experience without reducing strength.

[0082] The above are only preferred embodiments of the present invention and are not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modifications, equivalent replacements, improvements, etc. made within the technical solution of the present invention should be included in the protection scope of the present invention. In addition, it should be noted that the drawings are only for example and are not drawn according to the conditions of equal scale, and should not be used as a limitation on the actual scope of protection required by the present invention.

Claims

1. A numerical design method for a centrifugal diffuser, characterized in that: The following steps are involved: S1: Performing an initial design of the centrifugal diffuser according to existing design methods, determining the inlet diameter, outlet diameter, number of blades, installation angle, and blade profile of the centrifugal diffuser, and obtaining the initial design parameters of the centrifugal diffuser; S2: Based on the initial design parameters of the centrifugal diffuser, the optimal installation angle and number of blades are determined using a global optimization method; S3: Preliminarily setting the geometry of the centrifugal diffuser according to the optimal installation angle and number of blades, and determining whether its aerodynamic performance meets the first preset requirement. If so, executing step S4; if not, returning to step S2 and re-determining the optimal installation angle and number of blades; S4: parameterize the blade profile and determine the control parameters of the blade profile; S5: Optimizing the blade profile parameters using a global optimization method based on the control parameters in step S4 to determine the optimal blade profile; S6: updating the centrifugal diffuser geometry in step S3 according to the optimal blade profile to obtain a new centrifugal diffuser geometry, and determining whether the aerodynamic performance of the new centrifugal diffuser geometry meets the second preset requirement. If so, executing step S7; if not, returning to step S5 and re-determining the optimal blade profile. S7: Output the new centrifugal diffuser geometry and complete the centrifugal diffuser numerical design.

2. A numerical design method for a centrifugal diffuser according to claim 1, characterized in that: Step S2 is specifically as follows: S2.1, using the installation angle and the number of blades as design variables and, based on the initial design parameters of the centrifugal diffuser determined in step S1, using a Latin hypercube test to determine an experimental sample with the installation angle and the number of blades as design variables, and calculating six performance parameters of the experimental sample, specifically including near-surge point efficiency, near-surge point total pressure ratio, design point efficiency, design point total pressure ratio, near-blocking point efficiency, and near-blocking point total pressure ratio; S2.2 Based on the experimental sample data, the response surface model of the corresponding design variables and performance parameters is established as follows: y(x)=f(x) T β+z(x) Where x is the input vector, which consists of the installation angle and the number of blades; f(x) is the regression basis function vector; β is the regression coefficient vector; f(x) T β is the overall trend of the response variable; z(x) is a zero-mean Gaussian random process, and y(x) is the output vector, which is the six performance parameters mentioned above. S2.3 Based on the response surface model of design variables and performance parameters, a multi-objective genetic algorithm is used to find the optimal solution to obtain the optimal installation angle and number of blades.

3. A numerical design method for a centrifugal diffuser according to claim 2, characterized in that: Step S3 is specifically as follows: The centrifugal diffuser geometry is set according to the optimal installation angle and number of blades obtained in step S2, and its model-level aerodynamic performance is calculated through flow field analysis to determine whether the aerodynamic performance is within the first preset requirement. If so, step S4 is executed. If not, the process returns to step S2 and increases the number of samples or adjusts the optimization target to redetermine the optimal installation angle and number of blades until the calculation result is within the first preset requirement.

4. The numerical design method for a centrifugal diffuser according to claim 1, characterized in that: Step S4 is specifically as follows: One to five sections along the blade height are selected as the surfaces to be parameterized. The two-dimensional blade profiles of different sections are expressed in the form of superimposed thickness distribution of the mid-camber angle. The blade profile is parameterized by changing the distribution of the mid-camber angle along the chord line, and then the control parameters of the blade profile are determined.

5. A numerical design method for a centrifugal diffuser according to claim 4, characterized in that: In step S4, the specific method of implementing blade profile parameterization by changing the distribution of the camber angle along the chord line is: The Bezier curve is used to fit the camber angle distribution curve. The order of the Bezier curve is determined based on the trade-off between fitting accuracy and the number of design variables. The camber angle distribution is then changed by changing the positions of the Bezier curve control points to achieve blade parameterization.

6. The numerical design method for a centrifugal diffuser according to claim 5, characterized in that: In step S4, the control parameters of the blade profile are selected from the ordinate values ​​of the control points of the Bezier curve, and the variation range of the control parameters is controlled within 20%.

7. A numerical design method for a centrifugal diffuser according to claim 6, characterized in that: Step S5 is specifically as follows: S5.1 Based on the control parameters determined in step S4, Latin hypercube experimental design is used to arrange experimental samples with the ordinate values ​​of the Bezier curve as design variables, and six performance parameters of the control parameter experimental samples are calculated, including the efficiency near the gasping point, the total pressure ratio near the gasping point, the efficiency at the design point, the total pressure ratio at the design point, the efficiency near the blocking point, and the total pressure ratio near the blocking point; S5.2 Based on the six performance parameters of the control parameter experimental sample, establish the corresponding response surface model of the control parameters and performance parameters; S5.3 Based on the response surface model of control parameters and performance parameters, a multi-objective genetic algorithm is used to find the optimal solution and determine the optimal blade shape.

8. The numerical design method for a centrifugal diffuser according to claim 1, characterized in that: Step S6 is specifically as follows: Based on the optimal blade profile obtained in step S5, the centrifugal diffuser geometry in step S3 is updated to obtain a new centrifugal diffuser geometry. The model-level aerodynamic performance of the new centrifugal diffuser geometry is then calculated through flow field analysis to determine whether the aerodynamic performance is within the second preset requirement. If so, step S7 is executed. If not, the process returns to step S5 and redetermines the optimal blade profile until the calculation result is within the second preset requirement.

Images