CN112084724A - A CFD Local Fast Calculation Method - Google Patents

Abstract

A CFD local rapid calculation method relates to a local rapid calculation method in computational fluid dynamics. The method solves the problems of wasting computing resources, limiting the speed of optimization iteration and the like in the prior art, and provides a CFD local rapid computing method, which comprises the following steps: mapping the CFD calculation result of the original example which is calculated to a new example which is not calculated; partitioning: partitioning the new example, and separating the part of the new example, which has a difference in geometric structure with the original example, to obtain partitioned examples; example processing: processing the obtained blocking calculation example to obtain a blocking calculation example for independent calculation; and (3) calculating: and calling a block calculation example obtained by CFD software calculation to finish CFD local quick calculation. The invention splits a large and complex complete example, calculates only the required local part, greatly reduces the calculation resource, accelerates the single iterative calculation speed, ensures that the local optimization can obtain the timely feedback, and greatly reduces the iterative period.

Description

A CFD Local Fast Calculation Method

technical field

The present invention relates to a local fast calculation method in computational fluid dynamics.

Background technique

Computational Fluid Dynamics (CFD) is based on fluid mechanics and computer science. It starts from computational methods and uses the fast computing power of computers to obtain approximate solutions of fluid governing equations. It is an important part of studying fluid mechanics related theories and engineering applications technical means.

Because problems in practical engineering cannot be analyzed and solved, and the cost of making models for experiments is high, CFD is usually relied on for numerical simulation in the process of product design. The accuracy and speed of CFD calculation results directly determine the design iteration cycle. length. Therefore, how to improve the calculation speed and accuracy of CFD has always been the focus of research.

The product structure in actual engineering is often extremely complex. In order to ensure the accuracy, it is necessary to use a sufficiently fine physical model and a sufficiently dense grid for numerical simulation, so the calculation speed is greatly reduced. If some calculation problems are encountered, the design iteration cycle is still to be extended; conversely, if the model is simplified, the computation speed can be improved, but the accuracy will suffer. Especially when performing local optimization iterations, even if the model changes very little, it still takes the same long numerical simulation time, which is a waste of time.

In addition, because the required computing resources are often very large, in order to save resources, the optimization of different parts of the product requires a unified iteration cycle and unified calculation, which makes the local optimization work mutually restrictive and affects work efficiency.

Therefore, when the product is optimized for local details, the traditional calculation method has seriously wasted computing resources, which limits the speed of optimization iteration, and requires new flexible and fast calculation methods for local areas.

SUMMARY OF THE INVENTION

In order to solve the problems of wasting computing resources and limiting the speed of optimization iteration in the prior art, the present invention provides a local fast calculation method for CFD, which is mainly aimed at the complex original models and calculation examples that have completed CFD calculation, and occurs in the local geometric structure of the CFD calculation. After modification, the local fast calculation of CFD for new models and examples is realized.

A CFD local fast calculation method, which is realized by the following steps:

Step 1, mapping;

Map the CFD calculation results of the original case that has completed the calculation to the new case that has not been calculated;

Step 2, block;

Perform block processing on the new calculation example obtained in step 1, and separate the part of the new calculation example that is different from the geometric structure of the original calculation example to obtain the divided calculation example;

Step 3, calculation example processing;

Process the block calculation example obtained in step 2 to obtain a block calculation example for independent calculation;

Step four, calculation;

The block calculation example obtained in step 3 of the CFD software calculation is called to complete the local fast calculation of CFD.

Beneficial effects of the present invention: the present invention splits the huge and complex complete calculation example, and only calculates the required part, greatly reduces the computing resources, speeds up the calculation speed of a single iteration, enables the local optimization to get timely feedback, and greatly reduces the iteration cycle;

The present invention enables the local optimization work to be performed more flexibly and in a decentralized manner, and avoids the situation that a plurality of local optimization works are mutually restricted when they are performed synchronously;

The invention applies the method of mapping interpolation to provide a reasonable initial field for CFD calculation, which can speed up calculation convergence and reduce the number of iterations;

The invention can realize fast self-defined calculation area division, and is convenient for users to adjust the divided calculation area, so as to take into account the calculation speed and accuracy.

Description of drawings

FIG. 1 is a schematic diagram of a CFD local fast calculation method according to the present invention.

Detailed ways

DETAILED DESCRIPTION OF THE PREFERRED EMBODIMENTS 1. This embodiment is described with reference to FIG. 1, a CFD local fast calculation method, which is realized by the following steps:

1. Mapping;

The interpolation mapping tool is used to interpolate and map the CFD calculation results of the original calculation example that has completed the calculation to the new calculation example that has not been calculated, and the user-defined tool is used to assign reasonable values to the grid areas that have not been mapped, so as to block the subsequent blocks. The calculation of the example provides accurate boundary conditions and reasonable initial fields.

Since there is only a slight geometric difference between the old and new studies, the calculation results are almost the same in areas far from this geometric difference, so the new study can directly use the calculation results of the original study in most areas, and these areas do not need to be further processed. calculate. Therefore, there are areas that need to be calculated and areas that do not need to be calculated in the new study.

2. Block processing;

New studies are binned in parallel using the Parallel Binning tool, in which a custom binning method is used to precisely locate the local geometry that differs from the original study geometry, and The structure is separated from the full structure, thereby separating the corresponding local meshes and corresponding physics into block studies that cannot yet be calculated.

In this embodiment, before parallel computing, the grid and physical field of the calculation example are divided into multiple parallel sub-calculations, and each computer thread only needs to calculate a single parallel sub-calculation, so the parallel computing efficiency is higher than that of a single core calculate. However, since parallel substudies need to pass data through the newly created boundaries, they cannot be computed independently.

Based on the above principles, the new calculation example is processed in parallel blocks, and it is divided into the area that needs to be calculated and the area that does not need to be calculated. The former is called a block calculation example. At this time, there is no difference between the block study and the parallel sub-study. The boundary conditions are related to other parallel sub-studies. Due to the lack of physical properties and other documents, independent calculations cannot be performed.

3. Calculation processing;

Migrate the meshes and physical fields of the block study that cannot be calculated to an independent study file, modify its boundary conditions, and supplement the necessary files such as acceleration of gravity, material properties, and turbulence models to make it a block that can be independently calculated. Examples.

4. Calculation;

Call the CFD software to calculate the block calculation example, get the calculation result, and complete the local fast calculation of CFD.

Specific embodiment 2, this embodiment is an application example of a CFD local fast calculation method described in specific embodiment 1:

This embodiment takes CFD simulation calculation of turbine guide vanes as an example. After the calculation of the original example A is completed, modify its local geometry (delete an air film hole at the back of the leaf) to generate a new example B. The calculation steps of the CFD part of the new example B are as follows:

1. Interpolate and map the calculation results of Example A to Example B. In this process, the tool mapFields of the OpenFOAM software is called, and a custom tool is used to rationalize the physical fields corresponding to the grid areas that cannot be mapped in Example B. Assignment, so as to provide accurate boundary conditions and reasonable initial field for the calculation of the block calculation example;

2. Perform parallel block processing on Case B. In this process, the tool decomposePar of the OpenFOAM software is called, and based on this tool, automatic custom block is realized, so that the occurrence of Case A and Case B can be precisely located. Modified blade back region and separate this local geometry from the full structure, thereby separating the corresponding local mesh and corresponding physics into a block study C that cannot be calculated yet;

3. Transplant example C, modify its boundary conditions, and supplement necessary files such as material thermophysical properties and turbulence model parameters. Example C can perform independent calculations, and this process is mainly realized through automated scripts;

4. Calculation example C, complete the local fast calculation of CFD, this process calls OpenFOAM software.

In the actual operation of the calculation method described in this embodiment, it is only necessary to fill in the calculation example positions of calculation example A and calculation example B and the parameters required by the custom block into the file, and only need to run the script that has been written. All the above-mentioned processes can be realized, and the local fast calculation of CFD can be realized.

In this implementation manner, the number of grids in calculation example B is about 13.15 million. If the traditional calculation method is used, the memory occupied by the calculation is about 10G, and the single-step iteration time is about 62.5s; For example C, the number of grids is about 1.45 million, the memory occupied by the calculation is about 1.2G, and the single-step iteration time is about 4.6s. It can be seen that the method of this embodiment has obvious advantages over the traditional method.

The technical features of the above-described embodiments can be combined arbitrarily. For the sake of brevity, all possible combinations of the technical features in the above-described embodiments are not described. However, as long as there is no contradiction between the combinations of these technical features, All should be regarded as the scope described in this specification.

The above-mentioned embodiments only represent several embodiments of the present invention, and the descriptions thereof are specific and detailed, but should not be construed as a limitation on the scope of the invention patent. It should be pointed out that for those skilled in the art, without departing from the concept of the present invention, several modifications and improvements can be made, which all belong to the protection scope of the present invention. Therefore, the protection scope of the patent of the present invention shall be subject to the appended claims.

Claims (6)

1. A CFD local rapid calculation method is characterized in that: the method is realized by the following steps:

step one, mapping;

mapping the CFD calculation result of the original example which is calculated to a new example which is not calculated;

step two, partitioning;

carrying out blocking processing on the new example obtained in the step one, and separating the part of the new example, which has a difference with the geometric structure of the original example, to obtain a blocking example;

step three, example processing;

processing the blocking calculation example obtained in the step two to obtain a blocking calculation example for independent calculation;

step four, calculating;

and calling the block calculation example obtained in the CFD software calculation step three to finish the CFD local quick calculation.

2. The CFD local fast calculation method according to claim 1, wherein: in the first step, an interpolation mapping tool is adopted to perform interpolation mapping on a CFD calculation result of an original example which is subjected to calculation to a new example which is not subjected to calculation, and a user-defined tool is adopted to perform assignment on a grid area which is not subjected to mapping, so that accurate boundary conditions and a physical field are provided for calculation of a block example.

3. The CFD local fast calculation method according to claim 1, wherein: and in the second step, parallel computing is adopted, the new algorithm is subjected to parallel block processing, and the corresponding local grids and the corresponding physical field are separated into the non-computable block algorithm.

4. The CFD local fast calculation method according to claim 1, wherein: and the example processing process of the step three is realized through an automatic script, and the boundary conditions and the material physical properties of the block examples are modified by adopting the transplanting block examples to obtain the independently calculated block examples.

5. The CFD local fast calculation method according to claim 4, wherein: the material properties include gravitational acceleration, material properties, and turbulence models.

6. The CFD local fast calculation method according to claim 1, wherein: and in the fourth step, the OpenFOAM software is called to calculate the block calculation example, and the CFD local calculation is completed.

Images