CAD/CAE Integration Framework with Layered Software Architecture

5 downloads 80870 Views 293KB Size Report
Keyword: CAD, CAE, integration, layered software architecture ... Computer Aided Engineering (CAE) is a disciplinary that makes ... repair and mesh generation processes [3]. (2) Using .... automatic algorithms are provided to alleviate the user.
CAD/CAE Integration Framework with Layered Software Architecture † Bing-wan CAO, Jian-jun CHEN ‡ , Zheng-ge HUANG, Yao ZHENG (Center for Engineering and Scientific Computation, Zhejiang University, China) {caobingwan, chenjj, zhenggehuang, yao.zheng}@zju.edu.cn

Abstract By introducing an open-sourced CAD abstraction component Common Geometry Module (CGM) as the integration middleware, a three-layer scheme is implemented to integrate our CAE software High End Digital Prototyping (HEDP) with mainstream CAD software. The major advantages of a layered software design are two-folded for CAE software developers. One is to provide unified interfaces with the differences of currently available CAD products hided. The other is to encapsulate common services relevant to geometry in the integration middleware. Both advantages could help alleviate development burdens greatly. Moreover, this paper will introduce three important functionalities for CAE systems built on the integration middleware, i.e. data transfer, defeaturing and virtual geometry operations with some practical examples. The above work consists of our preliminary efforts for the research of a third-party CAD/CAE integration middleware neutral for any CAD and CAE products. Keyword: CAD, CAE, integration, layered software architecture, mesh generation

1. Introduction Computer Aided Engineering (CAE) is a disciplinary that makes performance analysis of products with the help of computer systems. The HEDP (High End Digital Prototyping) [1, 2] is a typical CAE system for modeling and solution of many physical phenomena which could be reduced to groups of Partial Differential Equations (PDEs). Geometry modeling, mesh generation, solution, and visualization consist of four major parts of an analysis flow using the HEDP. When the problem domain is not too complex, the built-in geometry builder [2] could fulfill the modeling task. However, for complex geometries, users commonly have to appeal for aid of commercial CAD products. Therefore, a common problem for the HEDP and other CAE software is their integration with CAD software. Practically, three ways are usually adopted for CAD data † ‡

access in CAE systems: (1) Using neutral files. There are many popular neutral file standards, such as IGES, STEP, and STL. These standards are public and well understood. But information losses are usually unavoidable in the writing/reading processes, which will trouble the subsequent geometry repair and mesh generation processes [3]. (2) Using CAD kernels. Commercial CAD kernels such as ACIS, ParaSolid, and Granite, and open-sourced products such as OpenCascade could be used to support CAD data access, with the help of data access components under extra commercial licenses, such as InterOp for ACIS. Information losses still exist, but usually less serious, especially when the CAD system is built on the same kernel as well. (3) Using private plug-ins of CAD systems. Commercial CAD systems usually have their own plugins for data access, such as CAA of CATIA. Using them to access CAD data could guarantee the integrity and validity of data at most. However, the disadvantage is the need of different data access functionalities for different CAD systems. A more competitive strategy is to provide an integration middleware, in which developers could implement either of the above ways or their combination to access data from CAD systems. Such a kind of flexibility is useful for developers to adapt data access ways to changes of technologies or markets. Developers could also encapsulate common CAE services relevant to geometry in the middleware, thus to make the development of CAE services easier. In industry communities, commercial middleware have appeared, such as Simulation Modeling Suite of Simmetrix [4] and CAPRI of CADNexus [5]. In research communities, Common Geometry Module (CGM) has been used in the CUBIT and SciDAC projects [6]. In this paper, our preliminary efforts for the integration of the HEDP and CAD software are presented. By using CGM, we design a three-layer CAD/CAE integration framework. In the current version, downstream CAD data access services are provided by InterOp of ACIS kernel and neutral file read/write programs developed by ourselves. Many CAE services built on the CGM have been completed, three of which, data transfer, defeaturing,

This work was supported by the National Science Fund under grant Number 10872182. Corresponding author.

_____________________________

978-1-4244-3701-6/09/$25.00 ©2009 IEEE



and virtual geometry operations will be demonstrated with some practical examples.

2. Integration framework Figure 1 illustrates the three-layer framework designed for the integration of the HEDP and CAD systems. CGM is adopted as the CAD/CAE integration middleware, the core data structure of which is a Boundary Representation (B-Rep) for geometry and topology information, and a set of wrap interfaces for geometry and topology operations. Furthermore, CGM includes many useful infrastructures to support the development of advanced CAE services, such as its support to non-manifold geometry, facet geometry and virtual geometry [6]. Downstream is the CAD infrastructure connected with CGM in three ways, i.e. neutral files, CAD kernels and development plug-ins. Upstream are CAE functionalities classified into three major types, i.e. geometry functionalities, mesh

functionalities and other functionalities such as the definition of boundary conditions and solution parameters. Of course, above these functionalities is a Graphic User Interfaces (GUIs) to help users perform these operations interactively.

In the layered framework, CGM decouples the CAD and CAE systems so that the changes happened in one side don’t affect the other side. For example, when a new CAD file format needs to be supported, we just add a data access module in the middleware, but leave the CAE side as unchanged. Notice that not all downstream data access ways are implemented in the current version of the HEDP. As shown in Figure 2, two ways are selected in the HEDP currently. One is using ACIS and InterOp, where an ACIS Bridge module is integrated to connect CGM and ACIS kernel. The other is using neutral file read/write modules to connect CGM and CAD systems with neutral file as the bridge.

Graphic User Interfaces

Geom. Functionalities CAE Functionalities

Mesh Functionalities

Other. Functionalities

Geom. Validation

Geom. Repair

Mesh Generation

Mesh Smoothing

Boundary Condition Definition

Geom. Modelling

Geom. Data I/0

Remeshing

Mesh Sizing Defintion

Solution Parameter Definition

CAD/CAE Integration Middleware

Material Definition

Common Geometry Module (CGM)

Develop Plug-ins of CAD Software

Mainstream CAD Kernels ACIS

ParaSolid

Granite

OpenCASCADE

CAD Infrastructure Mainstream CAD software CATIA

Unigraphics

Pro/E

SolidWorks

CAD Neutral File Standards IGES

STEP

STL

Figure 1. CAD/CAE Integration framework based on layered software architecture



Figure 3. Example for data transfer and topology rebuilding Figure 2. Current version of framework supported by the HEDP

integration

4. Defeaturing

3. Topology rebuilding

When transfer a model into CAE systems, just reading geometric information is not enough. A more important and also more difficult work is to rebuild the topology of the model correctly [7, 8], to bind different levels of geometric parts into integrity that can be recognized by meshing and other CAE algorithms. One obstacle for topology rebuilding is the inconsistent definitions of tolerances in CAD and CAE system, which are usually the major inducements of gaps and overlaps in the transferred models, and thus make the generation of correct topology information impossible. Figure 3(C) gives an example of errors in topology rebuilding due to inconsistence of tolerances, where gaps and overlaps are found near the inner hole for a transferred CAD model (Figure 3(A)). The corrected model is illustrated in Figure 3(D), and Figure 3(B) gives the surface mesh of the corrected model.



Once the data transfer and topology rebuilding are done, the models may have correct geometry and topology. However, they may be over detailed. Some unnecessary features for CAE analysis, such as short edge, small faces/holes and slivers, may exist and trouble the subsequent meshing process [3, 7, 9]. It is commonly a challenge for the robustness of meshing algorithms. Even if the algorithms is fortunately robust enough, unimportant features for analysis must be cleared to avoid unnecessarily large meshes generated in the local regions around these features, and thus to decrease the magnitude of solution models. In the HEDP, some practically useful defeaturing functionalities are realized, such as the recognition and cleaning of short edges, small cracks, and small faces. Figures 4(A) and 4(B) compare a model with holes and chamfers and the one with these features removed. The geometry adaptive surface meshes of the two models, with the same global mesh sizing strategy applied in them, are given in Figures 4(C) and 4(D), respectively. The numbers of triangular facets included in the two meshes are 27,733 and 3,558, respectively, where almost one order of decrement in mesh magnitudes is achieved by defeaturing.

Figure 4. The small feature’s effect on the mesh generation

Figure 5. Virtual operation on a sheet from a car

6. Mesh generation 5. Virtual geometry

Kraftcheck [10] discussed virtual geometry operations systematically, and presented some showcases like edge merge, face merge, crack filling, and so on [10, 11]. Compared to real operations, virtual operations only change the topology of a model, but keep its geometry unchanged. Therefore, virtual operations are much faster and easier to implement, and cause no information losses, and their reversals are rather more convenient.

Figure 5 shows the effect of the face merge tool integrated in the HEDP for a sheet, which is implemented in a virtual way. As it shows, five faces are merged into one, making the topology of the sheet simpler. However, geometry definitions of the two models are consistent in the merged regions, which could guarantee that mesh points generated in the simplified model locate in the original model precisely. In many problems, high geometric fidelity of mesh models is useful for achieving a reliable simulation results.

Mesh generation is the pre-requisite for solutions of PDEs, such as the Finite Element Method (FEM) and the Finite Volume Method (FVM). Water-tight models are often required for most meshing algorithms, which could be achieved by data transfer and transform algorithms. In the HEDP, a down-top mesh generation framework is designed, i.e. low dimensional entities are meshed first before the meshing of high dimensional entities containing them, in order to keep one copy of mesh structure generated in the shared entities. Integrated in the mesh generation framework are many popular meshing algorithms, where unstructured mesh generation programs are currently dominated. All mesh tasks are performed in a visual and interactive environment with the help of userfriendly GUIs, and many useful interactive tools and automatic algorithms are provided to alleviate the user labors [1, 2, 12-14]. Figure 6 presents the surface mesh of an F22 aircraft generated in the HEDP. The model is built in a CAD system and imported into the HEDP through a neutral file. Grid sources are utilized to control the mesh sizing and generate the well graded surface meshes.



Figure 6. Visual representation of the geometry, surface mesh and mesh sources for an F22 aircraft

7. Conclusions The integration with CAD systems is the key problem for the development of a CAE system. In this paper we suggest a three-layer software framework and demonstrate a partial realization of the framework in the current

References 1.



Y. Zheng, N.P. Weatherill, E.A. Turner-Smith, (2000) “An interactive geometry utility environment for multidisciplinary computational engineering”, International Journal of Numerical Methods in Engineering.

version of the HEDP, a CAE product developed by ourselves for the solutions of PDEs. The core of the framework is an integration middleware neutral to any CAD or CAE products. CGM is demonstrated to be a good candidate to acting as this role.

2.

L.J. Xie, Y. Zheng, J.J. Chen, J.F. Zou, “Enabling Technologies in the Problem Solving Environment HEDP”, Communications in Computational Physics, 2008, 4(5): 1170-1193.

3.

G. Butlin, C. Stops, “CAD Data Repair”, 5th International Meshing Roundtable, SANDIA95-2130, Sandia National Laboratories. Pittsburgh, USA, October 1996.

4.

The Simulation Modeling Suite. URL: http://www.simmetrix.com/products/SimulationModelingS uite/main.html

5.

R. Haimes, G. Follen, “Computational Analysis Programming Interface”, 6th International Conference on Numerical Grid Generation in Computational Field Simulations. University of Greenwich, 1998.

6.

T.J. Tautges, “The Common Geometry Module (CGM): A Generic, Extensible Geometry Interface”, 9th International Meshing Roundtable, New Orleans, LA, USA, 2000, pp.337-348. M.W. Beall, J. Walsh, S.S. Mark, “Accessing CAD Geometry For Mesh Generation”, 12th International Meshing Roundtable. Santa Fe: Sandia National Laboratories, 2003: 33-42.

7.

8.

N.A. Petersson, K.K. Chand, “Detecting translation errors in CAD surfaces and preparing geometries for mesh generation”, 10th International Meshing Roundtable, 2001.

9.

M. Jones, M. Price, G. Butlin, “Geometry management support for auto-meshing”, 4th International Meshing Roundtable, Sandia National Laboratories, New Mexico, 1995, pp. 153–164.

10.

J. Kraftcheck, “Virtual Geometry: A Mechanism for Modification of CAD Model Topology for Improved Meshability”, Master’s Thesis, University of WisconsinMadison, 2000.

11.

A. Sheffer, T. Blacker, J. Clements, M. Bercovier, (1997) “Virtual Topology Operators for Meshing”, 6th International Meshing Roundtable, 49–65.

12.

Y. Zheng, R. W. Lewis, T.G. David, “Three-Dimensional Unstructured Mesh Generation: Part 1. Fundamental Aspects of Triangulation and Point Creation”, Computer Methods in Applied Mechanics and Engineering, Vol. 134, No. 3/4, 249-268 (1996).

13.

Y. Zheng, R. W. Lewis, T.G. David, “Three-Dimensional Unstructured Mesh Generation: Part 2. Surface Meshes”, Computer Methods in Applied Mechanics and Engineering, Vol. 134, No. 3/4, 269-284 (1996).

14.

R. W. Lewis, Y. Zheng, T.G. David, “Three-Dimensional Unstructured Mesh Generation: Part 3. Volume Meshes”, Computer Methods in Applied Mechanics and Engineering, Vol. 134, No. 3/4, 285-310 (1996).



Suggest Documents