Building Automation Tools for Mechanical/Structural Analysis and Pre-Processing with HyperMesh/SolidWorks/CATIA

theflyingrahul


Description:

TCL or Tool Command Language is very popular among mechanical engineers and CAE/CAM experts. Today, all the industries rely on Finite Element Analysis (FEA) results to validate their product's structural integrity. Based on the FEA results, the product design is finalized and drawings are prepared for production.

Problems faced in Finite Element Analysis

The problem with FEA is the time spent behind pre-processing of model. To perform FEA on any object, a large number of load-combinations and load-steps have to be defined in HyperMesh. This is a very time consuming process for any analyst. When such analysis is done in an industrial level, it involves a large number of elements in a single body. Also for performing correlation testing between FEM analysis and experimental results, engineers have to spend a considerable amount of time to organize and validate data. Generating load-combinations and load-steps manually caused many errors and was very time consuming as the number of combinations and cases are very large.

During my internship at Bombardier Transporatation, I had developed a tool with Python to solve this problem by automating the data cleansing and data feed process by taking input from MS Excel sheets, preparing which is less time consuming.

Magnitude of the problem

On average, creating a load combination in HyperMesh takes about 20 seconds for a pro-user with a fast enough computer. Imagine the time it takes to define 50,000-1,50,000 load combinations in the CAE software. This pre-processing task alone can take approximately 1500 days for a single person. The same applies to the process of defining load steps also. There is a probability that errors might creep in as this is a totally manual task.

Also, the HyperMesh software itself is quite buggy. Trying to modify a property/attribute of an object in HyperMesh doesn't work well. The updating has to be done at least twice via the GUI for the change to reflect. This bug hasn't been solved for years. The updating issue is not seen when scripts are used instead of the HyperMesh GUI.

Python is our savior!

This task can be automated by using TCL (Tool Control Language) script, which is the only script that is recognized by the Altair HyperWorks. This script also works on a variety of simulation softwares like Dassault Systems SolidWorks and CATIA. The Python code takes the inputs from a CSV file containing the load case combinations in a pre-set format and generates a TCL script. The TCL script when executed in HyperWorks, generates the required load cases automatically.

It might appear a little easy to build such an tool, but the problem had its own challenges...

  • Upgrading an existing prototype

  • Migrating from CSV to Excel Workbook

  • Interpreting TCL scripts (I was new to FEA then!)

  • Being choosy - PyQt or wx? / OpenPyXl or xlrd?

During this talk, I'll walk you through my experience in building this automation tool.

  • Rinsing our data and preparing it for export to TCL script

  • Getting started with PyQt

  • Building rich-UI to deliver the best experience for my fellow mechanical engineers (Yeah I made their lives easier!)

  • Error reporting (because IT people can't assist all the time!)

  • Optimizing performance for large datasets (because power is priceless...)

  • Sticking to the Zen of Python

Python for a reason...

  • Rich collection of libraries

  • Painless UI design

  • Easier to code (and for someone else to read as well)

and the obvious reason...

  • Cross Platform. 100% Guaranteed!

The company uses different operating systems. Data was fed into Excel using a linux system, pre-simulations were carried on a Mac and analysis was done on a Windows machine with HyperMesh. In order to enable the same experience across all platforms, I had to resort to Python!

Overview of the tool's logic

  • Prepare the unit load/pressure collectors in HyperMesh

  • Prepare the Excel Sheet with scaling factors for each load-case combination

  • Data Interpretation with Python and generate TCL script

  • Inject script into HyperMesh for the load combinations to be generated automatically

I would like to thank the developers of xlrd, pyinstaller, PyQt, openpyxl for their valuable libraries!

A big hug to Bo Milanovich - Pluralsight course instructor (Twitter: @PythonBo) for his Python lessons.

Prerequisites:

Core Python

Basics of TCL scripting (not mandatory, I myself learnt it only after building the tool!)

Speaker Info:

Rahul Pon is a Software Engineering Intern at Bombardier Transportation, Hyderabad. He is a pre-final year engineering student at Loyola-ICAM College of Engineering and Technology. Rahul has been programming in Python since the age of 16 and has volunteered in organizing workshops, classes and lectures for students. He is also a freelancer and develops websites and applications with Web2PY and PyQt respectively.

Speaker Links:

Find me on:

GitHub

LinkedIn

Id: 1265
Section: Scientific Computing
Type: Talks
Target Audience: Beginner
Last Updated: