Tailspin Analysis with Patran 2010

Yesterday, I watched a small aircraft go into a tailspin. It was quite a colorful sight!

Before you accuse me for being heartless, let me explain. The plane in question was a digital model. I was looking at analysis results in Patran 2010, the latest release of MSC Software’s flagship pre-/post-processing package. The product managers showed me how they might simulate the effects of a load applied to the plane’s fuselage.

The latest release of Patran is marked by a new Windows ribbon-bar user interface (UI). In the new environment, related functions—geometry editing tools, preference settings, load configuration fields, and display settings—are neatly folded into tabs. According to MSC Software, if you’re not quite ready to embrace the new look, you can switch back and forth between the classic UI and the new one—in the same session, if desired.

To me, one of the most impressive features of Patran is its ability to assemble a single 3D model out of data sets housed in different directories. In the exercise recorded in the video clip below, the fuselage, structural ribs, and windows existed as separate 3D data sets. They were stitched together using Patran, which can be done using a dialog box that lets you specify how each part is supposed to be connected to the others. In this workflow, if a subcontractor makes changes to the dimensions of the windows, you don’t need to recreate a new 3D assembly to rerun analysis. You may simply reference the new version of the windows and go on with analysis.

You can also use Patran to recognize features from models with parametric history, so if you need to, you can edit the holes, blends, extrusions, and other common features found in your model. This helps you de-feature, or clean up models with unnecessary details that could increase your analysis time.

In the refined UI, if you wish to examine certain nodes closely—for instance, to obtain the distance between two points—you’ll notice that every selection you make with a mouse pointer automatically populates the side panel with relevant numeric data (such as node location).

In analysis, meshing is always a delicate balance. If you create a mesh network that’s too coarse, you risk getting inaccurate results. But if your mesh network is too fine, the amount of calculation required to solve them will put a strain on your CPU. That’s where Patran’s mesh on mesh function might come in handy.

Essentially, you subdivide your model into a fine layer of meshes. Then you let the software create a less granular layer of mesh based on the previous layer. Afterward, you delete the first layer. With this approach, you usually get a manageable network of subdivisions that conform to transitions (like rounded edges  and borders of holes) better than standard meshing.

Please keep in mind that the clip below is created from a WebEx demonstration, where I’m on the receiving end. Therefore, the lack of fluidity in certain animation sequences is not a reflection of the software’s performance but the unavoidable effects of streaming a graphics-heavy presentation over the web.

Oh, by the way, no real human life was ever in jeopardy at any point during this simulated structural mishap.