Simulation prediction and design optimization method and device for particle flow blockage in pump

Abstract

The application discloses a kind of simulation prediction and design optimization method, device for particle flow blockage in pump, the method comprises: solid particles are divided into fine particles and coarse particles according to boundary particle diameter, fine particles are combined with base liquid as pseudo-fluid, and the tail interval of coarse particle size distribution is grouped, the volume frequency and equivalent transverse occupation width of each particle size group are obtained;According to the volume fraction of coarse particle and design flow, generate particle package time sequence;In combination with the local flow velocity, residence time and narrowest flow width of each key flow area, the occupation of particle package in time window is counted, and the remaining flow proportion is calculated;Then determine the bridging probability by the remaining flow proportion distribution, identify high-risk areas, and use the corresponding flow passage geometric parameters as optimization variables, under the condition of meeting head and efficiency constraints, design optimization is carried out to reduce the risk of particle blockage in pump.

Description

Technical Field

[0001] This application relates to the field of simulation design technology, and more specifically, to a method and apparatus for simulation prediction and design optimization of particulate flow blockage in a pump. Background Technology

[0002] Pumps are commonly used in applications such as sewage discharge, river dredging, and slurry transportation to convey two-phase media containing both solid and liquid phases. In these media, the solid particles have a wide particle size distribution. While most fine particles can pass through the pump's internal flow channels with the liquid phase, a small number of larger particles at the tail end of the particle size distribution can easily form temporary bridging in localized areas such as the impeller inlet, the throat of the flow channel between the blades, the transition zone from the impeller outlet to the volute, and near the volute tongue. This rapid concentration of larger particles can drastically reduce the flow area, leading to occasional blockages, flow fluctuations, and decreased efficiency.

[0003] Existing pump anti-clogging designs mostly rely on the maximum passable particle size, a single representative particle size, or empirically amplified flow cross-section for judgment. While these methods can reflect particle throughput to some extent, they are difficult to characterize the random arrival characteristics of large particles at the tail end of the particle size distribution, and also difficult to evaluate the clogging risk caused by multiple coarse particles occupying the same local area nearly simultaneously. Although some numerical simulation methods can analyze the solid-liquid two-phase flow process, they usually involve a large amount of computation and are difficult to form a stable and efficient closed loop with the optimization of flow channel geometry parameters during the design phase. Summary of the Invention

[0004] This application provides a method and apparatus for simulating, predicting, and optimizing the design of particulate flow blockage in a pump, in order to at least solve some of the technical problems existing in the related technologies described above.

[0005] According to a first aspect of the embodiments of this application, a method for simulation prediction and design optimization of particulate flow blockage in a pump is provided, including:

[0006] Solid particles are divided into fine particles and coarse particles according to the boundary particle size. The fine particles are combined with the basic liquid to form a pseudo-fluid. The tail section of the coarse particle size distribution is divided into several particle size groups, and the volume frequency and equivalent transverse occupancy width of each particle size group are obtained.

[0007] The arrival frequency of coarse particles is calculated based on the volume fraction of coarse particles and the design flow rate. The particle pack time series is generated by sampling according to the volume frequency of each particle size group. The particle pack includes the equivalent lateral occupancy width and the entry time.

[0008] Calculate the local velocity of the pseudo-fluid in each key flow region. Use the ratio of the mainstream length to the local velocity in each key flow region as the residence time and as a time window. Calculate the time when each particle arrives at each key flow region. Calculate the sum of the equivalent lateral occupancy widths of the particle within each time window. Subtract the sum of the widths from the narrowest flow width in each key flow region to obtain the remaining flow width. Take the ratio of the remaining flow width to the narrowest flow width as the remaining flow ratio.

[0009] The bridging probability is calculated from the remaining flow ratio distribution of each key flow region. High-risk regions are identified by sorting them according to the bridging probability. The flow channel geometry parameters of the high-risk regions are used as optimization variables. Under the constraints of head and efficiency, the flow channel geometry parameters are adjusted so that the bridging probability is lower than the predetermined target value.

[0010] As an optional approach, the tail section of the coarse particle size distribution is determined as follows: the narrowest flow width is extracted for each key flow region, and the starting particle size at which the ratio of particle size to the narrowest flow width reaches a predetermined flow influence ratio is determined on the cumulative distribution curve of coarse particle size. The range of particle sizes above the starting particle size is taken as the tail section. The flow influence ratio is the lower limit value at which the ratio of the equivalent diameter of a single particle to the narrowest flow width of the key flow region reaches the point where it generates a occupancy effect on the flow cross section.

[0011] As an optional approach, the equivalent lateral occupancy width is obtained by multiplying the equivalent particle size of each particle size group by the shape factor; the equivalent particle size is the volume-weighted average particle size of the particles within the particle size group; the shape factor is the ratio of the maximum outer dimension of the particle in the flow channel cross-section direction to the equivalent sphere diameter; when there is no measured shape data, the shape factor is taken as 1.0.

[0012] As an optional approach, the step of generating a particle pack time series by sampling the volume frequency of each particle size group includes: dividing one rotation cycle of the impeller into several angular intervals according to the number of blades; for each angular interval and each particle size group, calculating the expected number of particles in the angular interval based on the total volume flow rate of coarse particles, the volume frequency of the particle size group, the single particle volume corresponding to the equivalent particle size of the particle size group, the rotation cycle, and the number of blades; sampling the actual number of particles using the expected number of particles as a parameter according to a Poisson distribution; randomly generating the entry time for each particle within the time period of the corresponding angular interval; and repeating the above process for multiple consecutive rotation cycles to obtain the particle pack time series.

[0013] As an optional approach, the calculation of bridging probability based on the remaining current ratio distribution of each key current-carrying region includes: arranging the remaining current ratios of each key current-carrying region in all sampling periods according to their values, and calculating the ratio of the number of sampling windows with remaining current ratios lower than a predetermined lower current limit to the total number of sampling windows. This ratio is used as the bridging probability of the corresponding key current-carrying region.

[0014] As an optional approach, adjusting the flow channel geometry parameters includes: generating initial sample points within the optimization variable value space; performing particle size grouping, particle pack time series generation, simultaneous occupancy statistics, and bridging probability calculation on each sample point to obtain training data; constructing a surrogate model using the training data; the inputs of the surrogate model include flow channel geometry parameters, coarse particle tail particle size distribution parameters, coarse particle volume fraction, design flow rate, and rotational speed; the outputs of the surrogate model include bridging probability prediction values ​​and head prediction values; and iteratively searching for flow channel geometry parameter combinations that satisfy a bridging probability lower than a predetermined target value using the surrogate model.

[0015] As an alternative approach, the surrogate model is constructed using the Gaussian process regression method, and during training, the mass conservation constraint and the minimum flow area constraint are embedded into the mean function prior of the Gaussian process regression model.

[0016] As an optional approach, the flow channel geometry parameters include at least two of the following: impeller inlet width, blade inlet edge thickness, blade inlet placement angle, narrowest width of the flow channel between blades, front cover profile curvature, rear cover profile curvature, and the circumferential variation rate of the flow area near the volute tongue.

[0017] As an alternative, the key flow areas include at least two of the following: the impeller inlet section, the throat section of the inter-blade flow channel, the impeller outlet to volute transition section, and the section near the volute tongue.

[0018] According to a second aspect of the embodiments of this application, a device for simulating, predicting, and optimizing the blockage of particulate flow in a pump is also provided, comprising:

[0019] The particle division and parameter determination module is used to divide solid particles into fine particles and coarse particles according to the boundary particle size. The fine particles and the basic liquid are combined into a pseudo-fluid. The tail section of the coarse particle size distribution is divided into several particle size groups to obtain the volume frequency and equivalent lateral occupancy width of each particle size group.

[0020] The particle pack sequence generation module is used to calculate the arrival frequency of coarse particles based on the volume fraction of coarse particles and the design flow rate, and to generate a particle pack time series by sampling the volume frequency of each particle size group. The particle pack includes the equivalent lateral occupancy width and the entry time.

[0021] The residual flow ratio calculation module is used to calculate the local velocity of the pseudo-fluid in each key flow region. The residence time is calculated as the ratio of the mainstream length to the local velocity in each key flow region and used as a time window. The arrival time of each particle bag in each key flow region is calculated. The sum of the equivalent lateral occupancy width of the particle bag in each time window is calculated. The residual flow width is obtained by subtracting the sum of the narrowest flow width in each key flow region from the narrowest flow width. The ratio of the residual flow width to the narrowest flow width is taken as the residual flow ratio.

[0022] The blockage risk identification and optimization module is used to calculate the bridging probability from the remaining flow ratio distribution of each key flow area, identify high-risk areas by sorting them according to the bridging probability, and use the flow channel geometry parameters of the high-risk areas as optimization variables to adjust the flow channel geometry parameters under the constraints of head and efficiency so that the bridging probability is lower than the predetermined target value.

[0023] This application incorporates fine particles into the pseudo-fluid processing and groups and calculates the equivalent lateral occupancy of coarse particles at the tail end of their size distribution. This allows the focus of clogging prediction to be placed on a small number of large particles that actually affect the flow cross-section, avoiding the bias caused by evaluating flow capacity based solely on a single representative particle size. Simultaneously, a particle pack time series is generated based on the arrival frequency of coarse particles, and the physical residence time of each key flow region is used as a statistical window. This reflects the risk of bridging caused by multiple coarse particles simultaneously occupying areas such as the impeller inlet, the throat of the flow channel between blades, the transition zone from the impeller outlet to the volute, and the vicinity of the volute tongue. By calculating the bridging probability and identifying high-risk areas through the distribution of the remaining flow proportion, the risk of occasional clogging can be transformed into comparable and constrained design indicators. Based on this, design adjustments are made using relevant flow channel geometric parameters as optimization variables, combined with constraints such as head and efficiency. This helps reduce the probability of clogging while ensuring pump hydraulic performance, providing a reliable basis for the anti-clogging design of pumps carrying solid-liquid two-phase media.

[0024] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only, and are not intended to limit this disclosure. Furthermore, no embodiment in this disclosure is required to achieve all the effects described above. Attached Figure Description

[0025] The accompanying drawings, which are incorporated in and form a part of this specification, illustrate embodiments consistent with this disclosure and, together with the description, serve to explain the principles of this disclosure.

[0026] Figure 1 A flowchart illustrating a method for simulation prediction and design optimization of particulate flow blockage in a pump, provided in an embodiment of this disclosure.

[0027] Figure 2 A flowchart illustrating the coarse and fine particle division and processing provided in this embodiment of the disclosure.

[0028] Figure 3 A flowchart illustrating the generation process of the particle pack time series provided in this embodiment of the disclosure.

[0029] Figure 4 A flowchart for generating the residual current ratio sequence provided in this embodiment of the disclosure.

[0030] Figure 5This is a schematic block diagram of a device for simulating, predicting, and optimizing the blockage of particulate flow in a pump, provided as an embodiment of this disclosure.

[0031] Figure 6 This is a schematic block diagram of an electronic device provided in an embodiment of the present disclosure. Detailed Implementation

[0032] To enable those skilled in the art to better understand the present application, the technical solutions in the embodiments of the present application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present application, and not all embodiments. Based on the embodiments in the present application, all other embodiments obtained by those of ordinary skill in the art without creative effort should fall within the scope of protection of the present application.

[0033] This disclosure provides a simulation prediction and design optimization method for particle flow blockage in pumps, applicable to pumps conveying solid-liquid two-phase media, including sewage pumps, sludge pumps, and slurry pumps. In the above-mentioned operating conditions, the conveyed medium contains solid particles with a wide particle size distribution range. Most particles can pass smoothly through the flow channel inside the pump, but there are a small number of particles at the tail end of the particle size distribution that are much larger than the median particle size. When several of these particles arrive at the same local flow area at the same time in a short period of time, they may form temporary bridging at the impeller inlet, the throat of the flow channel between blades, or the transition area of ​​the volute, causing occasional blockage.

[0034] Typically, the inputs to this method include the pump's geometric parameters, design operating parameters, basic liquid density and viscosity, solid particle density and particle size distribution data, and total volume concentration. All of the above data are obtained from conventional design data or on-site screening tests, without the need for additional sensors. Particle size distribution data can be obtained directly from screening tests, or it can be represented by the Rosin-Rammler distribution function or a piecewise frequency distribution table.

[0035] The implementation process of the method described in this application will be described in detail below with reference to specific embodiments. It should be noted that this embodiment is only used to explain this application and is not intended to limit the scope of protection of this application. Conventional adjustments or substitutions of each step by those skilled in the art without departing from the concept of this application should be included in the scope of protection of this application.

[0036] Please see Figure 1 , Figure 1 This is a flowchart of a method for simulating, predicting, and optimizing particulate flow blockage in a pump according to an embodiment of the present invention, as shown below. Figure 1 As shown, the method includes steps S1-S4:

[0037] In step S1, the solid particles are divided into fine particles and coarse particles according to the boundary particle size. The fine particles are combined with the basic liquid to form a pseudo-fluid. The tail section of the coarse particle size distribution is divided into several particle size groups, and the volume frequency and equivalent lateral occupancy width of each particle size group are obtained.

[0038] Please see Figure 2 , Figure 2 A flowchart illustrating the coarse and fine particle division and processing provided in this embodiment of the disclosure is shown below. Figure 2 As shown in box 201, the boundary particle size between coarse and fine particles is determined based on the geometric characteristics of the flow channel inside the pump and the particle size distribution. Particles with a diameter smaller than the boundary particle size are defined as fine particles, and particles with a diameter greater than or equal to the boundary particle size are defined as coarse particles. The boundary particle size is selected based on the fact that fine particles have good following properties within the pump, their movement trajectory is basically consistent with the mainstream fluid, and they do not produce a significant occupancy effect on the local flow cross-section.

[0039] In some embodiments, the boundary particle size can be taken as a certain proportion of the narrowest width of the flow channel, such as a value between 5% and 15% of the narrowest width; it can also be determined based on engineering experience or existing operating data. Since the particle size of fine particles is much smaller than the size of the flow channel, their contribution to local blockage can be ignored. Therefore, fine particles are incorporated into the fluid phase for processing, while coarse particles are kept separately as independent calculation objects.

[0040] In box 202, after the division is completed, the fine particles are regarded as suspended components uniformly dispersed in the base liquid, and merged with the base liquid to form a pseudo-fluid.

[0041] The pseudofluid density can be calculated using the mixture density formula based on the base liquid density, fine particle density, and the volume fraction of fine particles in the pseudofluid. Specifically, the pseudofluid density equals the sum of the base liquid density multiplied by the liquid volume percentage and the fine particle density multiplied by the fine particle volume percentage; the pseudofluid viscosity is calculated based on the fine particle volume fraction using a commonly used suspension viscosity model in this field. For example, when the fine particle volume fraction is low, the Einstein viscosity formula can be used, which expresses the suspension viscosity as the base liquid viscosity multiplied by a correction term linearly related to the volume fraction; when the volume fraction is high, its high-concentration correction form can be used.

[0042] In frame 203, the tail section of the coarse particle size distribution is divided into several particle size groups, and the volume frequency and equivalent lateral occupancy width of each particle size group are obtained.

[0043] Conventional design methods often use a single representative particle size to measure the throughput capacity of coarse particles, which does not adequately reflect the clogging contribution of a small number of large particles at the tail end of the particle size distribution. Therefore, this embodiment retains detailed information on the tail end of the coarse particle size distribution and refines it into several particle size groups, so that the degree of occupancy of each particle size group on the flow cross section can be statistically analyzed in subsequent steps.

[0044] Specifically, the process of determining the tail section includes: first, obtaining the cumulative particle size distribution curve of coarse particles. This curve can be directly obtained from sieving data, or it can be fitted using a Rosin-Rammler distribution or piecewise frequency distribution, which will not be elaborated here. Then, the narrowest flow width is extracted for each key flow area in the pump. On the cumulative particle size distribution curve of coarse particles, the initial particle size at which the ratio of particle size to the narrowest flow width reaches the predetermined flow influence ratio is found. The particle size range above this initial particle size is the tail section of the corresponding region. Here, the flow influence ratio is expressed as the lower limit of the ratio of the equivalent diameter of a single particle to the narrowest flow width of the key flow area. When this value is reached, the particle has a non-negligible occupancy effect on the local flow cross section.

[0045] In some embodiments, the flow effect ratio can be configured to a value between 0.15 and 0.40. When a smaller value is taken, the range of particles included in the tail section is wider, and the clogging assessment is more conservative. When a larger value is taken, only the largest few particles are considered. In practical applications, it can be calibrated based on the pump's operation clogging records or experience with similar operating conditions. For different key flow areas in the pump, since the narrowest flow width may be different, the corresponding initial particle size may also be different. Therefore, each area corresponds to a tail section. In implementation, the union of the tail sections of each area is usually taken as a unified particle size grouping range.

[0046] In one embodiment, within the tail section, the coarse particle size is further divided into several particle size groups. The division can be done using an equal-interval method, dividing the tail section's particle size range into several equal segments; or using an equal-volume-frequency method, ensuring that the volume proportion of coarse particles within each particle size group is approximately equal. Each particle size group has an equivalent particle size, which is the volume-weighted average particle size of the particles within that group. The volume frequency of each particle size group is determined by the volume proportion of coarse particles within that group in the tail section, and the sum of the volume frequencies of all particle size groups equals 1. Based on the total volume concentration and the volume distribution proportion of the tail section in the overall solid particle size distribution, the volume fraction of coarse particles participating in subsequent sampling is determined.

[0047] In one embodiment, for each particle size group, the equivalent lateral occupancy width is further calculated. This parameter represents the maximum size that the particle may actually occupy in the channel width direction. The equivalent lateral occupancy width is equal to the product of the equivalent particle size of the corresponding particle size group and the shape factor. The shape factor is defined as the ratio of the maximum outer dimension of the particle in the channel cross-section direction to its equivalent sphere diameter. For spherical particles, the shape factor is equal to 1.0; for oblong, flocculent, or fibrous particles, the shape factor is greater than 1.0. The shape factor can be obtained by estimating from the particle's length-to-minor axis ratio or approximating it by the ratio of the maximum outer dimension of the particle to the standard sieve aperture size in a sieving test. When measured data on particle shape is lacking, the shape factor is taken as 1.0, and the particle is treated as a spherical particle. In this case, the lateral occupancy estimate is based on the assumption of spherical particles. For particles that are actually oblong, flocculent, or fibrous, this estimate may be lower than their actual lateral occupancy width.

[0048] In step S2, the arrival frequency of coarse particles is calculated based on the volume fraction of coarse particles and the design flow rate. The particle pack time series is generated by sampling the volume frequency of each particle size group. The particle pack includes the equivalent lateral occupancy width and the entry time.

[0049] Please see Figure 3 , Figure 3 The flowchart for generating the particle pack time series provided in this embodiment of the disclosure is as follows: Figure 3 As shown in box 301, the coarse particle arrival frequency is calculated based on the coarse particle volume fraction and the design flow rate.

[0050] Specifically, based on the pump's design flow rate and the impeller inlet cross-sectional area, the average axial velocity of the pseudo-fluid at the impeller inlet is calculated. Based on the volume fraction of the tail coarse particles sampled and this velocity, the total volumetric flow rate of the tail coarse particles passing through the impeller inlet cross-section per unit time is calculated. For each particle size group, the total volumetric flow rate of the tail coarse particles is first allocated according to the volumetric frequency of that particle size group to obtain the particle volumetric flow rate of that particle size group. Then, the particle volumetric flow rate of that particle size group is divided by the single particle volume corresponding to the equivalent particle size of that particle size group to obtain the average arrival frequency of coarse particles in that particle size group passing through the impeller inlet. The physical meaning of this frequency is the average number of coarse particles of that particle size group entering the impeller per unit time. The expected number of particles is subsequently allocated accordingly in each angle interval and each particle size group.

[0051] In this context, particle bags are a computational representation method. Each particle bag corresponds to a coarse particle computational unit generated through sampling. Each particle bag contains the following attributes: its particle size group number, the equivalent lateral occupancy width of that particle size group, the number of particles represented by the bag, and the time it enters the impeller inlet. The reason for using particle bag representation is that by assigning the same equivalent lateral occupancy width to the sampled coarse particles according to their particle size groups, it is not necessary to build a complex morphological model for each real particle. This preserves the statistical characteristics of the tail end of the particle size distribution while keeping the computational load controllable.

[0052] In frame 302, a particle pack time series is generated by sampling the volumetric frequency of each particle size group. The particle pack includes the equivalent lateral occupancy width and entry time. Specifically, using the impeller rotation cycle as the time reference, one rotation cycle is divided into several angular intervals according to the number of blades. Each angular interval corresponds to the inlet of the flow channel between one blade on the impeller.

[0053] For each angular interval and each particle size group, the calculation and sampling process includes: First, multiplying the average arrival frequency of coarse particles in that particle size group by the rotation period and dividing by the number of blades yields the expected number of particles of that particle size group received in one rotation period within that angular interval. This expected number is then used as a parameter for a Poisson distribution for random sampling to obtain the actual number of particles of that particle size group received in the current rotation period within that angular interval. The Poisson distribution is a discrete probability distribution describing the number of independent random events occurring per unit time or unit space, with its parameter being the average number of occurrences of the event. The sampling result is a non-negative integer. When the expected number is small, the sampling result exhibits random fluctuations across different periods, reflecting the statistical characteristics of tail particles aggregating in a short time.

[0054] Each sampled particle packet enters the impeller inlet at a time that is uniformly and randomly generated within the corresponding angle range. After sampling all angle ranges and all particle size groups for one rotation cycle, the same process is repeated for subsequent consecutive rotation cycles. The total number of sampling cycles can be configured to be hundreds to thousands to obtain a sufficient statistical sample size.

[0055] After all sampling is completed, a set of particle pack time series covering multiple rotation cycles is obtained. Each particle pack in the series records the equivalent lateral occupancy width and entry time of its corresponding particle size group.

[0056] In step S3, the local velocity of the pseudo-fluid in each key flow region is calculated. The residence time is taken as the ratio of the mainstream length of each key flow region to the local velocity and used as a time window. The arrival time of each particle in each key flow region is calculated. The sum of the equivalent lateral occupancy width of the particle in each time window is counted. The remaining flow width is obtained by subtracting the sum of the widths from the narrowest flow width in each key flow region. The ratio of the remaining flow width to the narrowest flow width is taken as the remaining flow ratio.

[0057] Please see Figure 4 , Figure 4 A flowchart for generating the residual current ratio sequence provided in the embodiments of this disclosure is shown below. Figure 4 As shown in box 401, key flow areas prone to particulate clogging are identified based on the pump's geometric model.

[0058] In some embodiments, the key flow regions include at least two of the following: the impeller inlet section, the inter-blade flow channel throat section, the impeller outlet to volute transition section, and the section near the volute tongue. The impeller inlet section is the region where the inlet section is located near the leading edge of the blade; the inter-blade flow channel throat section is the region with the smallest flow area between two adjacent blades; the impeller outlet to volute transition section is the region where the flow direction and flow area change transitionally between the impeller outlet section and the volute inlet section; the section near the volute tongue is the region where the flow cross-section changes abruptly on the volute inlet side. For each key flow region, its length along the mainstream direction and its narrowest flow width are extracted.

[0059] In box 402, calculate the local velocity of the pseudofluid in each key flow region. Using the pseudofluid density and pseudofluid viscosity obtained in the previous steps as fluid property inputs, calculate the local velocity of the pseudofluid in each key flow region.

[0060] In some embodiments, a one-dimensional flow analysis method can be used for calculation. Specifically, based on the three-dimensional fluid domain geometry of the pump, a calculation section is established in each key flow region, traversing the local mainstream direction of that region. This calculation section can be a plane or a curved surface that follows the shape of the flow channel. The surface region formed by the intersection of the calculation section and the three-dimensional fluid domain is taken as the effective flow surface region. The effective flow surface region is then integrally calculated using the surface area, or the areas of each element constituting the effective flow surface region are summed in a discrete geometric model to obtain the effective flow area of ​​the corresponding calculation section. Based on the mass conservation principle, the average velocity of the calculation section is obtained by comparing the volumetric flow rate passing through the calculation section under design conditions with the effective flow area. This average velocity is the equivalent velocity used in one-dimensional flow analysis to estimate particle transport time and residence time, and does not represent a uniform actual velocity distribution on the calculation section.

[0061] In other embodiments, optionally, for regions with obvious jet or wake structures, such as the transition section from the impeller outlet to the volute, or in cases where higher computational accuracy is required, quasi-three-dimensional flow field calculation or simplified numerical simulation methods are used to obtain the local velocity distribution of the region, and the local flow velocity corresponding to each flow channel is determined based on the local velocity distribution.

[0062] In box 403, the residence time is calculated as the ratio of the mainstream length to the local velocity in each key flow region, and this ratio is used as the time window. The arrival time of each particle pack in each key flow region is then calculated.

[0063] For each particle packet generated in the aforementioned steps, based on the time it enters the impeller inlet, combined with the flow channel length between each region and the local flow velocity within each region, the time when the particle packet arrives at each key flow region is calculated sequentially. Among them, when using one-dimensional flow analysis, the transmission time of the particle from the inlet to a certain key flow region is approximately equal to the cumulative value after dividing the length of each flow channel along the main flow direction by the local flow velocity of the corresponding segment.

[0064] For each critical flow region, the residence time is defined as the ratio of the length of the region along the mainstream direction to the local flow velocity within that region. Residence time represents the average time required for a particle to travel from entering to leaving the region, reflecting the timescale of a particle's continued presence within that region. In this embodiment, this residence time also serves as a time window for determining whether multiple particles exist simultaneously within the same region. Specifically, the determination rule is as follows: when the difference in the arrival times of two or more particle packets within the same critical flow region is less than the residence time of that region, these particle packets are considered to exist simultaneously within that region, indicating a near-simultaneous occupancy event.

[0065] Therefore, the time window is automatically determined by the physical time scale of the flow within the pump. When the pump speed or flow rate increases, the local flow velocity in each region increases, the residence time shortens accordingly, and the time window also shortens; conversely, when the flow velocity decreases, the window widens. At the same time, the occupancy determination condition is thus consistent with the actual operating state of the pump, avoiding the subjectivity and limited applicability caused by manually setting fixed time thresholds.

[0066] For each critical flow region, with a defined residence time as the window width, the arrival times of each particle packet in that region are traversed along the time axis in a sliding window manner within the particle packet time series. The sliding window moves along the time axis with a predetermined step size. At each window position, all particle packets whose arrival times fall within the window range are selected, and the equivalent lateral occupancy widths of these particle packets are summed to obtain the sum of the equivalent lateral occupancy widths of the particle packets within that window. The predetermined step size remains consistent within the same pump type, the same operating condition, or the same round of optimization comparison, and is not greater than the residence time of the corresponding critical flow region to avoid missing nearly simultaneous occupancy events with arrival time intervals shorter than the residence time during the sliding window movement.

[0067] In box 404, the sum of the equivalent lateral occupancy widths of the particle packs within each time window is calculated. The remaining flow width is obtained by subtracting this sum from the narrowest flow width of each critical flow region. The ratio of the remaining flow width to the narrowest flow width is taken as the remaining flow ratio. Based on the above, the remaining flow width and the remaining flow ratio are calculated separately, where the remaining flow width is equal to the narrowest flow width of the critical flow region minus the sum of the aforementioned equivalent lateral occupancy widths.

[0068]

[0069] in, For the first The narrowest flow width in each key flow region For the first in this window The equivalent lateral space occupied by each particle packet. This is the sum of the equivalent horizontal width of all particles within the window. This is the remaining width available for the pseudo-fluid to pass through after deducting the space occupied by particles; when When the value is less than 0, it means that the particle occupancy in the window has exceeded the narrowest flow width, and the remaining flow width is counted as 0.

[0070] The residual flow ratio is equal to the ratio of the residual flow width to the narrowest flow width:

[0071]

[0072] The value ranges from 0 to 1. The smaller the value, the more severe the particle occupancy within that window period. When the particle density approaches zero, the particles almost completely block the flow cross section, forming a temporary bridge.

[0073] The above-mentioned overlay of vacancy widths adopts a direct summation method, corresponding to the case where each particle occupies a different position in the width direction of the flow channel. Optionally, a vacancy overlap coefficient can be introduced to correct the summation result. This coefficient is used to multiply the sum of the above-mentioned equivalent lateral vacancy widths. The value of this coefficient is between 0 and 1. When there is no statistical data on the lateral position of particles or flow field simulation data, the vacancy overlap coefficient is 1.0, that is, no reduction is made to the direct summation result. When there is statistical data on the lateral position of particles or simulation data, the coefficient is determined according to the overlap ratio of the lateral projection of particles within the same time window, which will not be elaborated here.

[0074] In step S4, the bridging probability is calculated from the remaining flow ratio distribution of each key flow region. High-risk regions are identified by sorting them according to the bridging probability. The flow channel geometry parameters of the high-risk regions are used as optimization variables. Under the constraints of head and efficiency, the flow channel geometry parameters are adjusted so that the bridging probability is lower than the predetermined target value.

[0075] For each critical flow region, the remaining current proportions obtained throughout all sampling periods are arranged in ascending order to form the remaining current proportion distribution for that region. The number of sampling windows with remaining current proportions lower than the predetermined lower limit is counted, and this number is divided by the total number of sampling windows in that region. The resulting ratio is the bridging probability for that critical flow region.

[0076] The predetermined lower limit of flow rate indicates that when the remaining flow rate is lower than this value, the resistance encountered by the pseudo-fluid passing through the region increases significantly, which is sufficient to cause a short-term flow interruption or particle retention. In some embodiments, this value can be configured between 0.10 and 0.30, and the specific value is determined based on the pump's allowable clogging frequency or experience with similar operating conditions. This embodiment does not impose any restrictions. The higher the value, the more stringent the judgment of clogging risk, and the higher the calculated bridging probability.

[0077] After the bridging probability of each key flow area is calculated, the areas are sorted from largest to smallest according to their bridging probability values. The areas at the top of the list have the highest risk of tail particles occupying space simultaneously, and are marked as high-risk areas, which will be the focus of subsequent optimization of flow channel geometry parameters.

[0078] The significance of the distributed sorting method is that, under the same predetermined lower limit ratio of flow, the relative blockage risk of each key flow area can be compared; in order to reduce the interference of a single threshold selection on the optimization decision, multiple values ​​can be selected within the range of the predetermined lower limit ratio of flow to repeatedly calculate the bridging probability and compare whether the risk ranking of each area is consistent; when the ranking is consistent or the change is small, the top ranked areas are identified as high-risk areas.

[0079] Optionally, when the pump is running under multiple operating conditions, the bridging probability under each operating condition can be weighted and averaged according to the frequency of occurrence of the operating condition to obtain a comprehensive bridging probability index. The frequency of occurrence of the operating condition is determined by the actual operation record of the pump or the design operating condition distribution, and this embodiment does not impose any restrictions.

[0080] In one embodiment, based on the aforementioned identified high-risk areas, flow channel geometric parameters directly associated with them are selected as optimization variables. Selectable flow channel geometric parameters include at least two of the following: impeller inlet width, blade inlet edge thickness, blade inlet angle, narrowest width of the inter-blade flow channel, front shroud profile curvature, rear shroud profile curvature, and the circumferential rate of change of the flow area near the volute tongue. The specific selection depends on the location of the high-risk area; for example, when the throat section of the inter-blade flow channel is a high-risk area, the narrowest width of the inter-blade flow channel and the blade inlet angle are the preferred optimization variables.

[0081] The optimization constraints are divided into performance constraints and manufacturing constraints. Performance constraints include a predetermined percentage of pump head not lower than the design value, a predetermined percentage of efficiency not lower than the design value, and wear rate of key flow surfaces not exceeding the limit. Manufacturing constraints include the value of each geometric parameter being limited by the casting or machining process boundary, and the minimum wall thickness of the blades meeting the structural strength requirements.

[0082] Specifically, the optimization objective is to adjust the flow channel geometry parameters to make the bridging probability in high-risk areas lower than a predetermined target value, while meeting the head and efficiency constraints. This target value can be determined according to the allowable requirements for the frequency of blockage in engineering applications, and this disclosure does not impose any restrictions.

[0083] In one embodiment, to reduce the number of times the entire process calculation is repeatedly executed during the optimization process, a proxy model may be constructed to assist in searching for the optimal parameter combination. Specifically, this includes first generating a set of initial sample points in the value space of the optimization variables using Latin hypersquare sampling or orthogonal experimental design methods; performing a complete process of particle size grouping, particle bag time series generation, simultaneous occupancy statistics and bridging probability calculation for each sample point, while simultaneously calculating the corresponding pump head and efficiency, using all sample points and their calculation results as training data.

[0084] The surrogate model can be constructed using Gaussian process regression, a nonparametric regression method based on Bayesian inference. After fitting the model to known sample points, it simultaneously provides mean predictions and variance estimates for unknown input points. The variance estimate characterizes the uncertainty of the prediction. The inputs to the surrogate model include flow channel geometry parameters, coarse particle tail size distribution parameters, the volume fraction of coarse particles sampled, design flow rate, and rotational speed. The outputs include bridging probability predictions, head predictions, and efficiency predictions.

[0085] In one embodiment, during training, mass conservation constraints and minimum flow area constraints are embedded into the mean function prior of the Gaussian process regression model. Specifically, using the univariate flow analysis relationship from the preceding steps, a baseline flow velocity that satisfies mass conservation is determined based on the design flow rate and the flow area corresponding to each candidate geometric parameter. The bridging probability and performance calculation results corresponding to this baseline flow velocity are used as the baseline prediction term of the mean function. For candidate geometric parameter combinations whose minimum flow area is lower than the predetermined manufacturing or flow lower limit, they are excluded as infeasible sample points, or a high-risk prediction value that does not meet the constraints is assigned to the mean function. In this way, the model still satisfies the basic physical constraints in regions where training samples are sparse, improving the reliability of predictions.

[0086] Using the trained surrogate model, a sequential optimization strategy is employed to iteratively search for parameter combinations. In each iteration, the surrogate model predicts the bridging probability, head, and efficiency for a batch of candidate parameter combinations. It selects the candidate point with the highest prediction uncertainty or the lowest predicted bridging probability, while also satisfying the corresponding constraints for head and efficiency. A complete calculation and performance verification are then performed on this point. The newly calculated results are added to the training dataset, and the surrogate model parameters are updated. The next iteration then begins, and the above process is repeated until convergence conditions are met, such as the change in bridging probability being less than a predetermined threshold in several consecutive iterations, or the number of iterations reaching a predetermined upper limit. Finally, the flow channel geometry parameter combination that satisfies both a bridging probability lower than a predetermined target value and the head and efficiency constraints is determined.

[0087] Optionally, after optimization, the final determined combination of geometric parameters can be re-verified to confirm that the bridge construction probability, head and efficiency of each high-risk area meet the design requirements. In some embodiments, computational fluid dynamics (CFD) simulation or solid-liquid two-phase flow numerical simulation can be used to independently verify the optimization results.

[0088] Therefore, through the above embodiments, the object of clogging prediction is changed from the traditional judgment based on the passage of a single particle to the near-simultaneous occupancy of multiple particles in the same local area at the tail of the particle size distribution. The time window for occupancy judgment is automatically determined by the physical residence time of the flow in each area of ​​the pump. The statistical characteristics of the occupancy events are obtained by Monte Carlo sampling, and high-risk areas are identified by distribution sorting. The bridging probability is transformed into a constraint for flow channel geometry optimization. This method can capture sporadic clogging risks that are difficult to detect by traditional average passage performance indicators. It is especially suitable for working conditions with long tails of particle size distribution, a low proportion of coarse particles, but serious clogging consequences. It provides a probabilistic analysis and optimization basis for the anti-clogging design of pump flow components without introducing additional detection devices.

[0089] Please see Figure 5 , Figure 5 This is a structural block diagram of a pump internal particle flow blockage simulation prediction and design optimization device provided in an embodiment of this application. Figure 5 As shown, the device includes:

[0090] The particle division and parameter determination module 501 is used to divide solid particles into fine particles and coarse particles according to the boundary particle size. The fine particles and the basic liquid are combined into a pseudo-fluid. The tail interval of the coarse particle size distribution is divided into several particle size groups to obtain the volume frequency and equivalent lateral occupancy width of each particle size group.

[0091] The particle pack sequence generation module 502 is used to calculate the arrival frequency of coarse particles based on the volume fraction of coarse particles and the design flow rate, and to generate a particle pack time series by sampling the volume frequency of each particle size group. The particle pack includes the equivalent lateral occupancy width and the entry time.

[0092] The residual flow ratio calculation module 503 is used to calculate the local velocity of the pseudo-fluid in each key flow region. The residence time is calculated as the ratio of the mainstream length to the local velocity in each key flow region and used as a time window. The arrival time of each particle bag in each key flow region is calculated. The sum of the equivalent lateral occupancy width of the particle bag in each time window is calculated. The residual flow width is obtained by subtracting the sum of the narrowest flow width in each key flow region from the narrowest flow width. The ratio of the residual flow width to the narrowest flow width is taken as the residual flow ratio.

[0093] The clogging risk identification and optimization module 504 is used to calculate the bridging probability from the remaining flow ratio distribution of each key flow area, identify high-risk areas by sorting them according to the bridging probability, and use the flow channel geometry parameters of the high-risk areas as optimization variables to adjust the flow channel geometry parameters so that the bridging probability is lower than the predetermined target value under the constraints of head and efficiency.

[0094] Each processing unit and / or module in the embodiments of this application can be implemented by an analog circuit that implements the functions described in the embodiments of this application, or by software that executes the functions described in the embodiments of this application.

[0095] Based on the same inventive concept, this application also provides an electronic device, the method corresponding to which can be the method in the foregoing embodiments, and its problem-solving principle is similar to that method. For example... Figure 6 As shown, Figure 6 This is a schematic block diagram of an electronic device provided in an embodiment of the present disclosure. The device includes: at least one processor; and a memory communicatively connected to the at least one processor; wherein the memory stores instructions executable by the at least one processor, the instructions being executed by the at least one processor to enable the at least one processor to perform the methods and / or technical solutions of the foregoing embodiments of the present application.

[0096] In particular, the methods and / or embodiments in this application can be implemented as computer software programs. For example, the embodiments disclosed in this application include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing program code for performing the methods shown in the flowchart. When the computer program is executed by a processor, it performs the functions defined in the methods of this application.

[0097] Another embodiment of this application provides a storage medium storing computer program instructions thereon, which can be executed by a processor to implement the methods and / or technical solutions of any one or more embodiments of this application.

[0098] In the above embodiments, the descriptions of each embodiment have different focuses. Parts not described in detail in a certain embodiment can be referred to in the relevant descriptions of other embodiments. The above descriptions are merely preferred embodiments of this application and explanations of the technical principles used. Those skilled in the art should understand that the scope of the invention involved in this application is not limited to the technical solutions formed by specific combinations of the above technical features, but should also cover other technical solutions formed by arbitrary combinations of the above technical features or their equivalent features without departing from the inventive concept.

Claims

1. A method for simulation prediction and design optimization of particulate flow blockage in a pump, characterized in that, include: Solid particles are divided into fine particles and coarse particles according to the boundary particle size. The fine particles are combined with the basic liquid to form a pseudo-fluid. The tail section of the coarse particle size distribution is divided into several particle size groups, and the volume frequency and equivalent transverse occupancy width of each particle size group are obtained. The arrival frequency of coarse particles is calculated based on the volume fraction of coarse particles and the design flow rate. The particle pack time series is generated by sampling according to the volume frequency of each particle size group. The particle pack includes the equivalent lateral occupancy width and the entry time. Calculate the local velocity of the pseudo-fluid in each key flow region. Use the ratio of the mainstream length to the local velocity in each key flow region as the residence time and as a time window. Calculate the time when each particle arrives at each key flow region. Calculate the sum of the equivalent lateral occupancy widths of the particle within each time window. Subtract the sum of the widths from the narrowest flow width in each key flow region to obtain the remaining flow width. Take the ratio of the remaining flow width to the narrowest flow width as the remaining flow ratio. The bridging probability is calculated from the remaining flow ratio distribution of each key flow region. High-risk regions are identified by sorting them according to the bridging probability. The flow channel geometry parameters of the high-risk regions are used as optimization variables. Under the constraints of head and efficiency, the flow channel geometry parameters are adjusted so that the bridging probability is lower than the predetermined target value.

2. The method according to claim 1, characterized in that, The tail section of the coarse particle size distribution is determined as follows: the narrowest flow width is extracted for each key flow region, and the starting particle size at which the ratio of particle size to the narrowest flow width reaches the predetermined flow influence ratio is determined on the cumulative distribution curve of coarse particle size. The range of particle sizes above the starting particle size is taken as the tail section. The flow influence ratio is the lower limit value at which the ratio of the equivalent diameter of a single particle to the narrowest flow width of the key flow region reaches the point where it generates a occupancy effect on the flow cross section.

3. The method according to claim 1, characterized in that, The equivalent lateral occupancy width is obtained by multiplying the equivalent particle size of each particle size group by the shape factor; the equivalent particle size is the volume-weighted average particle size of the particles in the particle size group; the shape factor is the ratio of the maximum outer dimension of the particle in the flow channel cross-section direction to the equivalent sphere diameter; when there is no measured shape data, the shape factor is taken as 1.

0.

4. The method according to claim 1, characterized in that, The process of generating a particle pack time series by sampling the volume frequency of each particle size group includes: dividing one rotation cycle of the impeller into several angular intervals according to the number of blades; for each angular interval and each particle size group, calculating the expected number of particles in the angular interval based on the total volume flow rate of coarse particles, the volume frequency of the particle size group, the single particle volume corresponding to the equivalent particle size of the particle size group, the rotation cycle, and the number of blades; sampling the actual number of particles according to a Poisson distribution using the expected number of particles as a parameter; randomly generating the entry time for each particle within the time period of the corresponding angular interval; and repeating the above process for multiple consecutive rotation cycles to obtain the particle pack time series.

5. The method according to claim 1, characterized in that, The calculation of bridging probability based on the remaining current ratio distribution of each key current-carrying region includes: arranging the remaining current ratios of each key current-carrying region in all sampling periods according to their values, and calculating the ratio of the number of sampling windows with remaining current ratios lower than a predetermined lower current limit to the total number of sampling windows. This ratio is used as the bridging probability of the corresponding key current-carrying region.

6. The method according to claim 1, characterized in that, The adjustment of the flow channel geometry parameters includes: generating initial sample points in the optimization variable value space; performing particle size grouping, particle pack time series generation, simultaneous occupancy statistics, and bridging probability calculation on each sample point to obtain training data; constructing a surrogate model using the training data; the input of the surrogate model includes flow channel geometry parameters, coarse particle tail particle size distribution parameters, coarse particle volume fraction, design flow rate, and rotational speed; the output of the surrogate model includes bridging probability prediction and head prediction; and iteratively searching for flow channel geometry parameter combinations that satisfy the bridging probability being lower than a predetermined target value using the surrogate model.

7. The method according to claim 6, characterized in that, The surrogate model is constructed using the Gaussian process regression method. During training, the mass conservation constraint and the minimum flow area constraint are embedded into the mean function prior of the Gaussian process regression model.

8. The method according to claim 1, characterized in that, The flow channel geometry parameters include at least two of the following: impeller inlet width, blade inlet edge thickness, blade inlet placement angle, narrowest width of the flow channel between blades, front cover profile curvature, rear cover profile curvature, and the circumferential variation rate of the flow area near the volute tongue.

9. The method according to claim 1, characterized in that, The critical flow areas include at least two of the following: the impeller inlet section, the throat section of the inter-blade flow channel, the impeller outlet to volute transition section, and the section near the volute tongue.

10. A device for simulation prediction and design optimization of particulate flow blockage in a pump, characterized in that, include: The particle division and parameter determination module is used to divide solid particles into fine particles and coarse particles according to the boundary particle size. The fine particles and the basic liquid are combined into a pseudo-fluid. The tail section of the coarse particle size distribution is divided into several particle size groups to obtain the volume frequency and equivalent lateral occupancy width of each particle size group. The particle pack sequence generation module is used to calculate the arrival frequency of coarse particles based on the volume fraction of coarse particles and the design flow rate, and to generate a particle pack time series by sampling the volume frequency of each particle size group. The particle pack includes the equivalent lateral occupancy width and the entry time. The residual flow ratio calculation module is used to calculate the local velocity of the pseudo-fluid in each key flow region. The residence time is calculated as the ratio of the mainstream length to the local velocity in each key flow region and used as a time window. The arrival time of each particle bag in each key flow region is calculated. The sum of the equivalent lateral occupancy width of the particle bag in each time window is calculated. The residual flow width is obtained by subtracting the sum of the narrowest flow width in each key flow region from the narrowest flow width. The ratio of the residual flow width to the narrowest flow width is taken as the residual flow ratio. The blockage risk identification and optimization module is used to calculate the bridging probability from the remaining flow ratio distribution of each key flow area, identify high-risk areas by sorting them according to the bridging probability, and use the flow channel geometry parameters of the high-risk areas as optimization variables to adjust the flow channel geometry parameters under the constraints of head and efficiency so that the bridging probability is lower than the predetermined target value.

Images