A closeup of the mesh and weld area shows a natural mismatch in element size and location that makes traditional weld-modeling methods difficult. Manually placing elements to connect panels can easily consume 30% of the total modeling time. VW participated in the design, development, and testing of the element and has decided to standardize their weld analyses on it.
Spot-weld information from a CAD system is usually a location in space. In the drawing, it's the weld grid GS. The sketch shows a quadrilateral element connected to a triangular element by a spot-weld grid (GAGB) placed by the new weld technology. Spot welds can be placed anywhere within overlapping shell elements.
A newly invented FEA weld element in MSC.Nastran lets engineers place spot-weld elements more efficiently and accurately than previously possible. The element eliminates the need for matching meshes on sheet-metal models, and it lets engineers optimize the size of the weld. Additional technology in the solver improves the accuracy of results. Automotive engineers at Volkswagen report that the development trims 30% off the time spent modeling welds.
Conventional spot-weld modeling techniques require a similar mesh on the two or three sheets to be analyzed so nodes on meshed sheets line up. When the meshes differ, the parts have to be remeshed, usually with manual, time-consuming procedures. In addition, analysts use rigid elements to connect or "weld" nodes. "This assumes the regions between the two sheets are infinitely rigid," says Omar Ibrahim, director of Nastran development with MSC.Software, Santa Ana, Calif. "But they are not. There is a flexibility between two welded sheets and FEA's job is to simulate it."
Another modeling technique for spot welds is to use a stubby beam which leads to an ill-conditioned matrix. "For instance, think of a matrix having a lot of numbers between one and 10," suggests Ibrahim. "Then insert a single value from a rigid weld of 10,000. The numbers are too widely divergent. This would be an ill-conditioned matrix and it's difficult to interpret results from one." The new weld element in MSC.Nastran 2001 solves these drawbacks because it models the flexibility of the connection with appropriate physical properties.
Eliminating the similar-mesh requirement solves another problem as well. "Now, analysts need not remesh anything," says Ibrahim. "Just locate the spot weld, and elements that should be connected are." The element connects shell elements at any point, not just at nodes. And should meshed parts come from different sources, remeshing is avoided because a mesh of quads can be connected to one of the triangles. What's more, weld assemblies can be modeled more simply and quickly using spot welds located in the CAD system.
"Several automotive companies have written their own weld-modeling programs for seam welding, but for designers to make best use of results, values must be expressed in force/unit length. Home-grown codes can't generate those figures easily," says Ibrahim. Such outputs will be standard in the next release of MSC.Nastran. An extension of the weld element will allow modeling seam welds. For these welds, analysts need only to define the beginning and end points and the software will generate the elements and connections in-between.
Furthermore, the FEA software models spot welds with point-topoint and patch-to-patch connections. A patch is a surface described by a sequence of nodes. When the spot-weld area is as large as an element, it is more accurate to connect the spot weld to two surface patches rather than to two nodes.
Engineers can also define weld diameters. "It is easier now to run case studies with different diameters," says Ibrahim. The capability lets users optimize on the weld diameter by making it a design variable. The element works for all weld height-to-diameter ratios including zero height. Users can specify both the physical properties of the element, such as diameter of the weld, and the material properties, such as Young's modulus.