Compressor numerical simulation method based on blade geometric volume force model

Abstract

The invention relates to the field of turbine fluid dynamics, and discloses a gas compressor numerical simulation method based on a blade geometric volume force model, which comprises the following steps: S1, constructing a volume force source item based on normal volume force, parallel volume force and a correction function; s2, introducing a blade blockage factor, and constructing a volume force theoretical model; s3, geometric information of the blade surface is obtained, a mid-plane normal vector and a blocking factor are extracted, and a volume force model suitable for the adjustable guide blade is constructed; s4, extracting model parameters; s5, substituting the extracted model parameters into the volume force model, and carrying out iterative calculation until a convergence condition is met; in the iterative calculation process, the volume force source item is updated. According to the invention, the influence of the change of the adjustable stationary blade mounting angle on the performance of the gas compressor can be quickly and accurately simulated.

Description

Technical Field

[0001] The present invention relates to the field of impeller fluid dynamics, in particular to a method for quickly estimating the performance of an aviation axial compressor under variable installation angle conditions of guide vanes / stators, and to a compressor numerical simulation method based on a blade geometric volume force model. Background Art

[0002] As one of the important components of aircraft engines, the axial compressor is very important in its development process to estimate the performance under design and non-design conditions. However, since design variables such as flow path and blade parameters often need to be iterated many times, in order to shorten the development cycle, it is necessary to be able to quickly estimate the performance of the compressor and finally meet the design requirements. Over the years, researchers have established many relatively mature rapid estimation methods, including parallel compressor model, excitation disk model, volume force model, etc. The parallel compressor model can estimate the influence of total pressure, total temperature and their combined distortion, and the concept is clear, but the assumption adopted by this method that the compressor characteristics are only related to the local flow coefficient does not hold when the circumferential velocity and pressure gradient are large. The excitation disk is broadly defined as a plane that causes a sudden change in flow parameters. Accordingly, the excitation disk model does not consider the flow process inside the blade row, but only establishes the aerodynamic parameter correlation of the airflow before and after passing through the excitation disk, so it is difficult to characterize the three-dimensional flow inside the compressor. Although the URANS (Unsteady RANS) method is an important means to explore the distorted inlet problem, the computational resources consumed by this method are almost unacceptable in the compressor design stage. Compared with the above methods, the volume force model can significantly reduce the amount of mesh by using the volume force source term instead of the real blade. While retaining the key three-dimensional flow characteristics, the amount of calculation can be reduced by about 2 to 3 orders of magnitude compared with URANS. With the continuous development of compressor theory and computer technology, the rapid prediction method of compressor performance shows a development trend from low dimension to high dimension and from low precision to high precision. As a quasi-three-dimensional low-order method, the volume force model can significantly reduce the amount of calculation while retaining the key three-dimensional flow characteristics, and quickly estimate the performance of the compressor. The volume force model can significantly reduce the amount of mesh by using the source term instead of the real blade. While retaining the key three-dimensional flow characteristics, the amount of calculation can be reduced by about 2 orders of magnitude compared with the full-circle unsteady numerical calculation (URANS). After years of development, the volume force model has been widely used in the propagation of nonlinear small disturbances, calculation of compressor stability and stall process, performance estimation of compressor under distorted inlet flow conditions, aerodynamic benefit research of BLI layout, numerical simulation of inlet / compressor integration, engine external flow field noise prediction, nacelle lip optimization, and flow estimation of compressor windmill state. Although the existing volume force model has a high prediction accuracy, it is still necessary to improve its applicability to different blade geometries so that it can be quickly applied to different compressor objects. In addition, with the continuous improvement of the integration of aircraft and engines, the resulting intake distortion problem will threaten the stable operation of the compressor. As an effective stabilization technology, adjustable guide vanes can broaden the stable operating range of the compressor by adjusting the guide vane installation angle, while effectively reducing the rotor inlet swirl intensity and improving the sensitivity of the fan rotor to intake distortion. Therefore, it is widely used.At present, there are still deficiencies in the modeling of adjustable guide vanes, and the existing body force model is difficult to simulate the adjustment process of the blade installation angle. Summary of the invention

[0003] The purpose of the present disclosure is to provide a compressor numerical simulation method based on a blade geometric body force model to address the deficiencies in the prior art, which can quickly and relatively accurately simulate the impact of changes in the installation angle of adjustable stator blades on compressor performance.

[0004] The present disclosure proposes a compressor numerical simulation method based on a blade geometric body force model, which includes the following steps:

[0005] S1, construct the body force source term based on the normal body force, parallel body force and correction function;

[0006] S2, introduce the blade blocking factor and construct the body force theoretical model;

[0007] S3, obtain the geometric information of the blade surface, extract the mid-surface normal vector and the blocking factor; and construct a body force model suitable for the adjustable guide vane;

[0008] S4, extract model parameters;

[0009] S5, inputting the model parameters into the body force model of the adjustable guide vane constructed in step S3 for iterative calculation until the convergence condition is met; during the iterative calculation process, the body force source term is updated.

[0010] In the compressor numerical simulation method based on the blade geometric body force model as described above, optionally, in step S2, a blocking factor is used to represent the influence of blade thickness so as to reduce the flow prediction deviation; the expression of the blocking factor is:

[0011]

[0012] Where b is the blocking factor, B is the number of blades, θ PS and θ SS are the circumferential coordinates of the blade pressure surface and suction surface, respectively.

[0013] As described above, in the compressor numerical simulation method based on the blade geometric body force model, optionally, a blade blocking factor is introduced to construct a body force theoretical model as follows:

[0014]

[0015] in, is the momentum source term, is the circumferential velocity at the local radial position, where the momentum source term Can be decomposed into normal volume force and parallel body forces ρ represents the density of the fluid; t is the time; represents the fluid velocity vector; b is the blockage factor; p is the fluid static pressure; e t represents internal energy; h t represents the total enthalpy.

[0016] As described above, in the compressor numerical simulation method based on the blade geometric body force model, optionally, in step S1, the formula of the correction function is:

[0017]

[0018] Where δ is the local deviation angle, and c1 and c2 are constants.

[0019] In the compressor numerical simulation method based on the blade geometric body force model as described above, optionally, in step S1, the normal body force is determined by the following conditions:

[0020]

[0021] in, is the relative speed, is the normal body force, n θ is the blade mid-surface normal vector Tangential component in cylindrical coordinates.

[0022] In the compressor numerical simulation method based on the blade geometric body force model as described above, optionally, the parallel body force calculation formula is:

[0023]

[0024] in, is the pressure gradient along the axial direction, U tip and w are the wheel rim speed and axial speed respectively, K p1 and K p2 are the two coefficients of the loss term respectively.

[0025] In the compressor numerical simulation method based on the blade geometric body force model as described above, optionally, the step of extracting model parameters includes:

[0026] S41, obtaining geometric information of the pressure surface and the suction surface of the blade;

[0027] S42, transforming the geometric information of the blade pressure surface and the geometric information of the blade suction surface into a cylindrical coordinate system;

[0028] S43, interpolating geometric information of the blade pressure surface and geometric information of the blade suction surface to obtain a smooth surface;

[0029] S44, calculate the blade mid-surface and the clogging factor;

[0030] S45, interpolation of blade mid-surface and clogging factor respectively;

[0031] S46, calculating the blade mid-surface normal vector;

[0032] S47, calculating the gradient of the blade blocking factor.

[0033] As described above, in the compressor numerical simulation method based on the blade geometric body force model, optionally, in step S43, the method for interpolating the geometric information of the blade pressure surface and the geometric information of the blade suction surface is to interpolate the mid-surface of the blade.

[0034] As described above, in the compressor numerical simulation method based on the blade geometric body force model, optionally, step S3 includes:

[0035] S31, reading the coordinate data of the blade pressure surface and the blade suction surface, and obtaining the distribution of the mid-surface normal vector and the blockage factor through coordinate system transformation, RBF interpolation and central difference;

[0036] S32, calculate the related items of normal volume force and blockage factor according to local geometric information and flow parameters; add parallel volume force as needed, and save all the finally calculated source items.

[0037] As described above, in the compressor numerical simulation method based on the blade geometric body force model, optionally, step S3 further includes:

[0038] S33, obtaining the installation angle of the blade at the current moment, and if the installation angle changes, rotating the coordinates of the pressure surface, suction surface, leading edge and trailing edge of the blade to a given angle;

[0039] S34, recalculating the distribution of the blade mid-surface normal vector and the blocking factor;

[0040] S35, determining the grid cells located inside the body force region at the current moment and marking them;

[0041] S36, setting all grid unit source items outside the body force region as the blades rotate to 0; so as to construct a body force model suitable for the adjustable guide vanes.

[0042] Compared with the prior art, the present disclosure first proposes a theoretical modeling method for the blade geometry volume force model, divides the force exerted by the blade on the fluid into normal volume force and parallel volume force, and analyzes the construction and correction method of the normal volume force. Then, a calculation method for extracting the blade mid-surface normal vector and the blockage factor distribution from the blade geometry data is derived, and a method for calculating the normal volume force and the parallel volume force is given. Finally, the calculation and addition process of all five volume force source items is realized based on UDF. The blockage factor considering the influence of blade thickness on the channel area is added to the volume force source item, and a local deviation angle correction function is proposed for the normal volume force, and finally a blade geometry volume force model is established. This method inherits the advantages of the volume force model with fast calculation speed and low consumption of computing resources. At the same time, the volume force model based on blade geometry can more accurately simulate the influence of the change of the guide vane installation angle on the compressor operating point, and can respond to the influence of the incoming flow swirl. In addition, the established blade geometry volume force model can quickly and more accurately simulate the influence of the change of the adjustable stator installation angle on the compressor performance.

[0043] It is to be understood that both the foregoing general description and the following detailed description are exemplary, and are intended to provide further explanation of the technology as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0044] Figure 1 It is a schematic diagram of the body force model;

[0045] Figure 2 It is a schematic diagram of the blade blocking factor, which represents the blocking effect of the blade thickness on the fluid;

[0046] Figure 3 is a schematic diagram of the local deviation angle δ;

[0047] Figure 4 It is a flow chart for extracting mid-surface normal vector and blocking factor;

[0048] Figure 5 It is a schematic diagram of the blade installation angle and body force area changing with time;

[0049] Figure 6 It is the flowchart of the DEFINE_ADJUST function;

[0050] Figure 7 is the computational domain and computational grid of the body force model;

[0051] Figure 8 It is a comparison of the pressure ratio characteristics of a single rotor at 88% of the design speed;

[0052] Fig. 9 It is the spanwise parameter distribution of a single rotor at the operating point of 10.9 kg / s;

[0053] Fig.10 It is the comparison of total pressure ratio characteristics under different VIGV installation angles;

[0054] Fig.11 It is the comparison of spanwise parameter distribution under different installation angles of VIVG;

[0055] Fig.12 It is the working point change during the constant back pressure VIGV installation angle adjustment process. DETAILED DESCRIPTION

[0056] The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present disclosure, but should not be construed as limiting the present disclosure.

[0057] Most of the existing volume force models construct source items based on a fixed single geometric model, which requires the construction of empirical parameters through a large amount of test data or the extraction of force sources through a large amount of numerical calculations. In the early stages of compressor design, repeated iterations of "geometric design-performance estimation-optimization and modification" are required. Therefore, the application of volume force models at this stage requires that they have universality for different compressor designs and as little preliminary preparation as possible.

[0058] At the same time, as the integration of aircraft and engines continues to increase, the resulting intake distortion problem will threaten the stable operation of the compressor. As an effective stabilization technology, adjustable guide vanes can broaden the stable operating range of the compressor by adjusting the guide vane installation angle, while effectively reducing the rotor inlet swirl intensity and improving the fan rotor's sensitivity to intake distortion. Therefore, they are widely used. At present, there are still deficiencies in the modeling of adjustable guide vanes, and the existing volume force model is difficult to simulate the adjustment process of the blade installation angle.

[0059] The above two deficiencies lead to limitations in the existing volume force model in the study of problems such as different geometric design iterations and dynamic changes in guide vane / stator installation angles, which to a certain extent restricts its wide application.

[0060] To solve the above problems, the present disclosure proposes the following solutions.

[0061] S1, construct the body force source term based on the normal body force, parallel body force and correction function.

[0062] For different volume force models, the normal force construction method and the required data input are also different. This paper chooses to use the normal force formula proposed by Hall to build the blade geometric volume force model, that is:

[0063]

[0064] In the formula, f n , W, r are vectors respectively Size, relative speed The size of the local radius, δ is the local deviation angle, which is defined as follows Figure 3 As shown, n θ is the blade mid-surface normal vector The reason for using the above formula is that: ① When there is swirl at the inlet, the compressor operating point will be on different isospeed lines. The volume force model that uses the local flow coefficient as the index on the isospeed line is difficult or unable to reflect the impact of swirl on the compressor. The local deviation angle δ and relative speed in the normal force formula are It will change with the tangential velocity and can reflect the influence of swirl. ② The normal force formula is based on the theoretical solution of the flat plate lift coefficient (c1 = 2πδ) and does not contain coefficients that need to be corrected. The normal force can be calculated by the blade mid-surface normal vector and local flow parameters. Therefore, it can be directly applied to different blade geometries and is also suitable for variable installation angles of adjustable guide vanes / static vanes. The direction of the normal force is determined by the following conditions:

[0065]

[0066] Among them, condition 1 can be obtained through the definition of normal force, and the meaning of condition 2 is Coplanar, condition 3 makes the normal force have the characteristics of negative feedback. When |δ| increases, f n increases and decreases |δ|, that is, f n The effect is to tend to make |δ|=0. This negative feedback characteristic helps the calculation converge.

[0067] Compared with the stator, the flow inside the compressor rotor is more complicated. Directly using equation (1) in the rotor area often results in large deviations. Therefore, this paper introduces a δ correction function to improve the prediction accuracy of the body force model, namely:

[0068]

[0069] In actual calculation, the δ correction function is only used in the rotor domain, that is, the local deviation angle δ is replaced by δ′ given by equation (3), and the δ correction function is not used in the body force region corresponding to the stator. For the current research object, c1 and c2 are taken as 0.117 and 50.0, respectively, and results that are relatively consistent with the experiment and RANS can be obtained.

[0070] The parallel volume force is related to entropy increase, and its direction is opposite to the direction of the local relative velocity. When no parallel force is added, no entropy increase is generated when the fluid flows through the blade domain, and the isentropic efficiency is 1. The model in this paper is for adjustable guide vanes, and a row of VIGV (Variable Inlet Guide Vanes) is added in front of the rotor of the transonic compressor for model verification. Compared with other blade rows, the total pressure loss of VIGV is generally smaller, and the total pressure drop caused by lower outlet kinetic energy is not large; for VIGV, the loss coefficient is not the most important factor affecting the performance of the entire compressor, and the outlet swirl angle is crucial to the operation of the subsequent stages. Therefore, no parallel volume force is added when the volume force model is applied to VIGV.

[0071] For the remaining blade rows (including rotors and stators), the parallel body force formula is given:

[0072]

[0073] In the formula, is the pressure gradient along the axial direction, U tip and w are the wheel rim speed and axial speed respectively (the axis of rotation selected in this paper is the Z axis), K p1 and K p2 are the two coefficients of the loss term respectively.

[0074] In formula (4), the parallel body force consists of two parts: The form is similar to the resistance coefficient, which is a constant K. p1 The size of the resistance; Related to the degree of diffusion, the increase of adverse pressure gradient and the decrease of flow rate (decrease of axial velocity) will cause Increased accordingly The entropy increase caused by the load will also increase, which is consistent with the actual working conditions of the compressor. However, it should be noted that the modeling of parallel body forces has a strong empirical dependence, and the values ​​of the loss terms and loss coefficients used need to be adjusted accordingly for different compressor designs.

[0075] S2, introduce the blade blocking factor and construct the body force theoretical model.

[0076] The body force model generally contains the assumption of local axisymmetry. By locally responding to the flow parameters, the body force source term can produce a non-axisymmetric distribution. Inside the body force region, the pressure rise and deflection effect of the actual blade on the fluid is realized by adding the body force source term to the Euler equation, that is, the following equation (5):

[0077]

[0078] in, is the momentum source term, is the circumferential velocity at the local radial position. The momentum source term in equation (5) is Can be decomposed into normal volume force and parallel body forces like Figure 1 As shown. Among them, Perpendicular to the direction of the local relative velocity, it deflects the fluid and does work without generating entropy increase, i.e., there is no viscous body force; The direction of is opposite to the local relative velocity and is related to viscous loss and entropy increase, that is, viscous volume force. and Both do work on the fluid, and the corresponding energy source term is the right side of the energy equation item.

[0079] like Figure 2 As shown in Figure 2, the blade thickness will reduce the actual flow area, but the actual blade is not included in the body force area. If the influence of blade thickness is not considered, there will be a significant flow prediction deviation, especially the blockage flow. For this reason, it is necessary to introduce the blade blockage factor b given by formula 2:

[0080]

[0081] Where B is the number of blades, θ PS and θ SS are the circumferential coordinates of the blade pressure surface and suction surface, respectively. After adding the blockage factor b, the blockage factor on the left side of the control equation is as follows:

[0082]

[0083] The effect of channel contraction and expansion is obtained by adding In one dimension, this term is similar to the lateral pressure change caused by the contraction and expansion of the pipe. b is the source term on the right side of the control equation as follows:

[0084]

[0085] Compared with equation (7), this can keep the solved control equation unchanged, that is, the left side of the equation remains the same as equation 1, avoiding scaling the mesh in the body force area or changing the solver itself. Therefore, this paper uses equation (8) to add the related terms of the blocking factor b.

[0086] S3, obtain the geometric information of the blade surface, extract the mid-surface normal vector and the blockage factor; and construct a volume force model suitable for the adjustable guide vane.

[0087] This step S3 includes,

[0088] S31, reading the coordinate data of the blade pressure surface and the blade suction surface, and obtaining the distribution of the mid-surface normal vector and the blockage factor through coordinate system transformation, RBF interpolation and central difference;

[0089] S32, calculate the related items of normal volume force and blockage factor according to local geometric information and flow parameters; add parallel volume force as needed, and save all the finally calculated source items.

[0090] The steps to implement the adjustable guide vane are:

[0091] S33, obtain the installation angle of the blade at the current moment. If the installation angle changes, rotate the coordinates of the blade pressure surface, suction surface, leading edge and trailing edge to a given angle; S34, recalculate the distribution of the blade mid-surface normal vector and the blockage factor; S35, determine the grid cells located inside the body force region at the current moment and mark them; S36, set all the grid cell source items outside the body force region as the blade rotates to 0. It can be seen that this method is essentially applying the current body force model to the adjustable guide vanes at different installation angles multiple times. Since the blade geometry body force model established in this paper can be directly applied to different blade geometries and does not require repeated correction, it is expected to have good applicability to adjustable guide vanes.

[0092] Specifically, for the case of variable blade installation angle, there are two problems after introducing blade rotation: ① After the blade rotates, the normal vector and blockage factor distribution of the blade mid-surface will change, and ② the volume force area corresponding to the blade row will change. This section will show the solutions to these two problems.

[0093] The assumption that the blades of the body force model are infinite and infinitely thin means that the rotation of the blade only needs to be performed once, and then the calculated blade mid-surface normal vector is rotated circumferentially to cover the entire body force area. Among them, the rotation of the blade around the axis of rotation can be converted into the rotation of any point on the blade around the given axis of rotation, that is, rotating point A (a0, a1, a2) around the given axis of rotation BC (n0t + s0, n1t + s1, n2t + s2), t∈R by an angle of θ to point A', and calculating the coordinates of the rotated point A', which are calculated by the following formula:

[0094]

[0095] After rotating all the data points on PS and SS, the blade data at another installation angle is obtained. At this time, after extracting the mid-surface normal vector and blockage factor distribution from the rotated blade data again, the volume force source term can be calculated using the new mid-surface normal vector and blockage factor. As the blade rotates, the blockage factor of the part that was originally located in the volume force area and had a blockage factor less than 1 but is located outside the volume force area after rotation will become 1. Although the blockage factor of this part of the area will change, it is now outside the volume force area, so there is no need to add source terms, and all source terms can be set to 0 directly. For the part that is still in the volume force area after rotation, the impact of the change in blockage factor needs to be considered. Taking the continuity equation as an example, since the spatial distribution of the blockage factor b will change with the blade installation angle at this time, In fact, it is no longer 0, and an additional term will be generated on the right side of the equation, namely the following formula:

[0096]

[0097] In addition to the continuity equation, additional terms will be generated in the momentum and energy equations, and the blades will exchange energy with the fluid during the installation angle adjustment process. However, considering that the rotation speed of the adjustable guide vanes is generally much smaller than the rotor speed, this paper assumes that the installation angle adjustment process is quasi-steady state and can be ignored. The impact on the source term. Still taking the continuity equation as an example, Figure 4 As shown, the blade cross section is simplified to a rectangle with a large aspect ratio, and the blade is assumed to be a straight blade. At a certain spanwise position, the blockage factor b in the body force region is equal (except for the axial interval corresponding to the four vertices of the rectangle), the blade cascade pitch is τ, and the rectangle width is l0. At t = 0s, the blade installation angle ξ = 0°, and the angular velocity of the blade rotation is ω ξ Then at time t, a point in the body force region That is the following formula:

[0098]

[0099] Comparing the two mass source terms again, since the blade cross section is simplified to a rectangle, in most of the axial intervals in the body force region, the partial derivatives of the blocking factor b in the z direction and in the θ direction are both 0. In order to reasonably estimate Assuming that the blocking factor decreases linearly from 1 to the minimum value and then increases linearly to 1, then at time t, the total volume force area The size can be estimated using the following formula:

[0100]

[0101] If the guide vane rotates 10° in 0.1s and the cascade density is assumed to be about 1, then according to equations (11) and (12), we have:

[0102]

[0103] according to Figure 4 , the change of body force area caused by the change of installation angle still needs to be considered. In this regard, the method adopted in this paper is: when rotating the pressure surface and suction surface coordinates of the blade, the leading and trailing edge coordinates of the blade are rotated synchronously, and then the position of the leading and trailing edges of the blade after projection to the meridian plane is calculated, and finally the curved quadrilateral surrounded by the four curves of the rim, hub, leading edge and trailing edge on the meridian plane is determined. The space area covered by the quadrilateral when it rotates around the compressor shaft is the body force area.

[0104] S4, extract model parameters;

[0105] This step mainly includes the following specific steps: S41, obtaining the geometric information of the blade pressure surface and the geometric information of the blade suction surface; S42, transforming the geometric information of the blade pressure surface and the geometric information of the blade suction surface into a cylindrical coordinate system; S43, interpolating the geometric information of the blade pressure surface and the geometric information of the blade suction surface to obtain a smooth surface; S44, calculating the blade mid-surface and the blocking factor; S45, interpolating the blade mid-surface and the blocking factor respectively; S46, calculating the blade mid-surface normal vector; S47, calculating the gradient of the blade blocking factor.

[0106] The key to obtaining the input parameters of the body force model is to extract the mid-surface normal vector from the geometric information of the blade PS (Pressure Surface) and SS (Suction Surface) and the distribution of the blockage factor b. Assuming that the blade PS and SS are given in the form of a series of three-dimensional Cartesian coordinate points, in order to obtain the distribution of the mid-surface normal vector and the blockage factor, it is first necessary to convert the coordinates from the Cartesian coordinate system to the cylindrical coordinate system, and interpolate these discrete points into two smooth surfaces. The reason for interpolation is that the given PS and SS data point distribution and the unit center of the volume force calculation grid are generally not matched, and the original coordinate data may be arranged in disorder, so it is necessary to interpolate the geometric information to the grid unit center by interpolation. There are two reasons for the coordinate system transformation: ① According to the blade mid-surface calculation formula (14) below,

[0107]

[0108] Circumferential coordinate of blade mid-surface θ MSThat is, it is the arithmetic mean of the circumferential coordinates of PS and SS. According to formula (2), the calculation of the blade blocking factor b also depends on the circumferential coordinates. ② In order to facilitate the subsequent calculation of the normal vector of the mid-surface and the gradient of the blocking factor b, taking the mid-surface of the blade as an example, when calculating the surface normal vector, it is necessary to first express it as θ MS =θ MS (z, r), and then use partial derivatives to represent the normal vector; if Cartesian coordinates are used, in some special cases the surface can only be represented as f(x, y, z) = 0, which will bring unnecessary difficulties to the calculation of the normal vector. Similarly, the calculation of the blocking factor gradient will also encounter similar problems. When interpolating the surface, it is not to directly interpolate the data points (interpolation nodes) on the PS and SS surfaces separately, but to interpolate the two quantities generated by their linear superposition, that is, formula (15):

[0109]

[0110] Among them, θ1 is the mid-surface, and θ2 can be directly used to calculate the blade blockage factor. This formula also implies a condition: when the circumferential coordinates of PS and SS are linearly combined, their axial coordinates z and radial coordinates r need to be consistent. If the original PS and SS data points cannot meet this condition, an additional interpolation is required to make the interpolation nodes on the two surfaces have the same z and r coordinate values. This article assumes that the original data meets this condition. In the selection of interpolation methods, the arrangement of a series of interpolation nodes on PS and SS is not a uniform rectangular lattice. According to RBF (Radial Basis Function) interpolation, multi-dimensional interpolation can be performed on disordered scattered points. This paper chooses to use two-dimensional RBF interpolation. The independent variables during interpolation are the axial and radial coordinates, and the function value of the interpolation node is the circumferential coordinate value θ. The distribution of the blade mid-surface and the blockage factor can be expressed as a function of (z, r), and finally written as the following formula:

[0111]

[0112] When performing CFD calculations, the control equations discretized by the solver are not in the cylindrical coordinate system, but in the Cartesian coordinate system. Therefore, in the calculation of the surface normal vector When converting the surface normal vector to the Cartesian coordinate system, it is necessary to pay attention to converting the surface normal vector to the Cartesian coordinate system. For a point (z, r, θ) on the mid-surface, the blade mid-surface normal vector Obtained by the following formula:

[0113]

[0114] Since the mid-surface has been represented as an interpolation surface, the partial derivative in equation (17) can be directly calculated by central difference. After completing the calculation of the partial derivative and vector normalization, the blade mid-surface normal vector can be obtained. distribution.

[0115] The second formula of formula (16) can be used to calculate the blocking factor b at any position in the blade domain. However, when adding the blocking factor source term, it is also necessary to use That is, the gradient of b. By definition, the gradient of b in the circumferential direction is 0. The partial derivatives in the radial and circumferential directions can be calculated by central difference, which is the same as the mid-surface normal vector. is also the gradient in Cartesian coordinates, Calculated by the following formula:

[0116]

[0117] In summary, the blade mid-surface normal vector is extracted from the blade PS and SS data points and blocking factor b The steps are as follows Figure 4 As shown, it includes: ① First, the coordinates of PS and SS are converted to the cylindrical coordinate system to obtain each interpolation node (z i ,r i ) at (θ 1,i ,θ 2,i );②By θ 2,i Calculate the blocking factor b for each interpolation node i ; ③ Perform RBF interpolation on the blade mid-surface θ1 and the blocking factor b respectively; ④ Calculate the blade mid-surface normal vector ⑤Calculate the gradient of the blade blockage factor b.

[0118] S5, inputting the extracted model parameters into the body force model of the adjustable guide vane constructed in step S3 for iterative calculation until the convergence condition is met; during the iterative calculation process, updating the body force source term.

[0119] Specifically, in one implementation, the Fluent 19.0 software platform is used to carry out the calculation of the volume force model, and the blade geometry volume force model is implemented by writing a Fluent UDF. The Fluent macros used in the currently constructed model include the DEFINE_ADJUST function for calculating mass, momentum, and energy source terms, 5 DEFINE_SOURCE functions for returning source term values, and the DEFINE_ON_DEMAND function for resetting data. In the currently constructed blade geometry volume force model, the calculation of the 5 volume force source terms is completed by DEFINE_ADJUST. The DEFINE_SOURCE function used to return mass, momentum, and energy source terms adds the source terms to the right side of the control equation and does not involve mathematical calculations, only returning the source term values.

[0120] The DEFINE_ADJUST function is the core of the entire UDF code. Figure 6 A simplified flowchart of the function is given. The function body can be roughly divided into two parts: ① Read the PS and SS coordinate data of the blade, and finally extract the PS and SS through a series of processes such as coordinate system transformation, RBF interpolation, and central difference. and b distribution, and save it in UDM; ② According to the local geometric information ( and b) and flow parameters, calculate the related terms of normal body force and blockage factor, add parallel body force as needed, and finally save all calculated source terms in UDM and return them to the solver through DEFINE_SOURCE. Figure 5 It can be seen that the first part is only executed once and is not repeated with the iteration of the solver, while the second part calculates and updates the volume force source term at each iteration step.

[0121] This paper uses the calculation results of a single-stage transonic axial flow compressor to demonstrate the simulation effect based on the blade geometry volume force model.

[0122] (1) Compared with conventional numerical calculation methods, this calculation method can greatly reduce the consumption of computing resources and significantly shorten the calculation time while ensuring accurate acquisition of compressor characteristics and study of flow field characteristics. Figure 7The mesh for the volume force model calculation is given. Compared with URANS calculation, there is no need to retain the actual structure of the blade during volume force calculation. The axial range between the leading and trailing edges is used as the blade domain for applying volume force, so the mesh structure is greatly simplified. The number of axial mesh nodes in the volume force area corresponding to the four rows of blades is 31, 35, 31, 31, respectively, the number of spanwise nodes is 35, and the number of circumferential nodes is 160. The entire circumference contains a total of 1.42 million meshes, which is about one-tenth of the URANS calculation mesh. It takes about 60 minutes for a single-channel RANS calculation to fully converge on a PC, while the volume force model only takes about 5 minutes, which greatly saves calculation time.

[0123] (2) This calculation method can accurately obtain the compressor flow field characteristics and flow field parameter distribution at multiple operating points under uniform intake conditions, and the volume force model has a high spatial resolution. First, it is compared with the single-channel RANS calculation results to verify the calculation efficiency and accuracy of this calculation method in solving the conventional compressor flow field and characteristics. Figure 8 The pressure ratio characteristics of a single rotor compressor at 88% of the design speed calculated by different calculation methods are shown in the figure. The horizontal and vertical axes are flow rate and inlet and outlet total pressure ratio, respectively. The results show that after adding the local deviation angle correction function and loss term, the deviation between the volume force model based on blade geometry and the experimental and RANS results is significantly reduced. Fig. 9 The distribution of total temperature, total pressure and absolute swirl angle along the span direction of the rotor upstream and downstream is given. After adding the correction function and loss term, the span-wise parameter distribution based on the geometric body force model is highly consistent with RANS.

[0124] (3) The model proposed in this disclosure has the ability to simulate variable guide vane / stator installation angles, and can especially well characterize the impact of dynamic changes on compressor performance and stability margin. A total of five VIGV installation angles are set, namely -8°, -4°, 0°, 4°, and 8°, and the guide vanes are straight blades. Fig.10 The calculation results of the pressure ratio of RANS and volume force models under five installation angles are given, and the total characteristics are extracted at the outlet of the calculation domain. It can be seen that the deviation of the pressure ratio characteristic curves of the two is within an acceptable range (taking the installation angle of 0° as an example, the pressure ratio is 1.377 when the RANS flow rate is 11.92kg / s, and the pressure ratio is 1.340 when the volume force model flow rate is 11.94kg / s, and the relative error of the pressure ratio is 2.687%); in addition, the volume force model correctly reflects the influence of the change of the guide vane installation angle on the compressor pressure ratio characteristics, which is reflected in two aspects: ① When only one set of grids is used, the volume force model constructed in this paper can rotate the guide vane according to the different installation angles and calculate the corresponding source terms; ② The current volume force model can capture the influence of the incoming flow swirl on the compressor characteristics. For VIGV, compared with the loss coefficient, the influence of the absolute swirl angle of the blade outlet on the compressor characteristics is more important, Fig.11(a) shows the spanwise distribution of total pressure before and after the VIGV. Fig.11 (b) shows the spanwise distribution of the absolute swirl angle at the VIGV outlet. The total pressure distribution of the volume force model in the figure is consistent with expectations. Fig.12 The calculation result of the variable guide vane installation angle of the current VIGV body force model is shown in Figure 2. The guide vane adjustment rule is set to linearly transition from +8° to -8°, the time is 1s, and the time step of the unsteady calculation is set to 0.002s. Fig.12 It can be seen that the dotted line represents the adjustment process of the guide vane installation angle. Under constant back pressure, the dotted line passes through the back pressure point corresponding to the steady-state calculation, which is consistent with expectations. The above results all show that the established blade geometry volume force model can quickly and accurately simulate the influence of the change of the adjustable stator installation angle on the compressor performance.

[0125] The above describes in detail the structure, features and effects of the present disclosure based on the embodiments shown in the drawings. The above is only a preferred embodiment of the present disclosure, but the present disclosure is not limited to the scope of implementation as shown in the drawings. Any changes made in accordance with the concept of the present disclosure, or modifications to equivalent embodiments with equivalent changes, which still do not exceed the spirit covered by the description and drawings, should be within the protection scope of the present disclosure.

Claims

1. A compressor numerical simulation method based on a blade geometric body force model, characterized in that: The following steps are included: S1, construct the body force source term based on the normal body force, parallel body force and correction function; S2, introduce the blade blocking factor and construct the body force theoretical model; S3, obtain the geometric information of the blade surface, extract the mid-surface normal vector and the blocking factor; and construct a body force model suitable for the adjustable guide vane; S4, extract model parameters; S5, inputting the model parameters into the body force model of the adjustable guide vane constructed in step S3 for iterative calculation until the convergence condition is met; during the iterative calculation process, the body force source term is updated.

2. The compressor numerical simulation method based on blade geometric body force model according to claim 1, characterized in that: In step S2, the blockage factor is used to represent the influence of the blade thickness to reduce the flow prediction deviation; the expression of the blockage factor is: Where b is the blocking factor, B is the number of blades, θ Ps and θ ss are the circumferential coordinates of the blade pressure surface and suction surface, respectively.

3. The compressor numerical simulation method based on blade geometric body force model according to claim 2 is characterized in that: Introducing the blade blocking factor, the theoretical model of body force is constructed as follows: in, is the momentum source term, is the circumferential velocity at the local radial position, where the momentum source term Can be decomposed into normal volume force and parallel body forces ρ represents the density of the fluid; t is the time; represents the fluid velocity vector; b is the blockage factor; p is the fluid static pressure; e t represents internal energy; h t represents the total enthalpy.

4. The compressor numerical simulation method based on blade geometric body force model according to claim 3 is characterized in that: In step S1, the formula of the correction function is: Where δ is the local deviation angle, and c1 and c2 are constants.

5. The compressor numerical simulation method based on blade geometric body force model according to claim 1, characterized in that: In step S1, the normal body force is determined by the following conditions, in, is the relative speed, is the normal body force, n θ is the blade mid-surface normal vector Tangential component in cylindrical coordinates.

6. The compressor numerical simulation method based on blade geometric body force model according to claim 5 is characterized in that: The formula for calculating parallel body force is: in, is the pressure gradient along the axial direction, U tio and w are the wheel rim speed and axial speed respectively, K p1 and K p2 are the two coefficients of the loss term respectively.

7. The compressor numerical simulation method based on blade geometric body force model according to claim 1, characterized in that: The steps to extract model parameters include, S41, obtaining geometric information of the pressure surface and the suction surface of the blade; S42, transforming the geometric information of the blade pressure surface and the geometric information of the blade suction surface into a cylindrical coordinate system; S43, interpolating geometric information of the blade pressure surface and geometric information of the blade suction surface to obtain a smooth surface; S44, calculate the blade mid-surface and the clogging factor; S45, interpolation of blade mid-surface and clogging factor respectively; S46, calculating the blade mid-surface normal vector; S47, calculating the gradient of the blade blocking factor.

8. The compressor numerical simulation method based on blade geometric body force model according to claim 1, characterized in that: In step S43, the method of interpolating the geometric information of the blade pressure surface and the geometric information of the blade suction surface is to interpolate the mid-surface of the blade.

9. The compressor numerical simulation method based on blade geometric body force model according to claim 1, characterized in that: Step S3 comprises, S31, reading the coordinate data of the blade pressure surface and the blade suction surface, and obtaining the distribution of the mid-surface normal vector and the blockage factor through coordinate system transformation, RBF interpolation and central difference; S32, calculate the related items of normal volume force and blockage factor according to local geometric information and flow parameters; add parallel volume force as needed, and save all the finally calculated source items.

10. The compressor numerical simulation method based on blade geometric body force model according to claim 9, characterized in that: Step S3 also includes, S33, obtaining the installation angle of the blade at the current moment, and if the installation angle changes, rotating the coordinates of the pressure surface, suction surface, leading edge and trailing edge of the blade to a given angle; S34, recalculating the distribution of the blade mid-surface normal vector and the blocking factor; S35, determining the grid cells located inside the body force region at the current moment and marking them; S36, setting all grid unit source items outside the body force region as the blade rotates to 0; In order to construct a body force model suitable for adjustable guide vanes.

Images