Ansys Design Experiments Usinf AutoDOE
Ansys Design Experiments Usinf AutoDOE
Ansys Design Experiments Usinf AutoDOE
Abstract
This paper attempts to provide a systematic procedure for the combined use of Structural Analysis
methods and Statistical methods for Design Optimization using Design of Experiments paradigm. Specific
benefits of this methodology and its applicability to similar cases in mechanical engineering design are
highlighted.
For the purposes of illustrating this approach, a Pressure Vessel is chosen. The design problem is
defined as Minimization of Hoop stress level & Weight of the structure
Introduction
In this paper, a Pressure Vessel is chosen for the purpose of illustration of this methodology. The
Pressure Vessel is subjected to an internal pressure of 0.7 bar. The design intent is to keep the stress
levels & weight of the vessel low by varying its geometrical parameters.
The Design of Experiments portion of the work is carried out using AutoDOE, a product from QuEST.
Methodology
A basic premise in this methodology is that the engineering design can be modeled and the behavior of
the system can be studied based on a statistical model to arrive at a robust design. An understanding of
ANSYS macro using Ansys-APDL is a prerequisite required for the design engineer to use this
methodology
ANSYS macro using Ansys-APDL method for the Pressure Vessel model is written as the input file to
AutoDOE that has a seamless integration for file handling, Pre-processing and Post processing. This
includes monitoring and execution of runs on ANSYS. The Post processing is carried out within the
AutoDOE tool and only requires guidelines for interpretation of the results rather than the statistical
background for arriving at optimized designs
Design Parameters:
Geometry Parameters
Minimum
Maximum
1000
1400
12
20
Thickness of jacket(TJ)
10
AutoDOE
Factors
Response
Choice of Exp
Define
Models
Generate Run
Store Runs
Set
Conditions
Model
Updates
Response
Results
Effects Plots
Response
Surfaces
Extract Results
Geometrical Modeling
Finite Element Mesh Generation
Finite Element Analysis
Post Processing
ANSYS
Results and Discussions:
AutoDOE Optimized Factor Settings:
Radius of cylinder (RC)
= 1000
= 20
= 2
Statistical
Plots
Polynomial
Equation
Curve fitting
Errors
Statistical
Analysis
Optimization
Varying Factors
Response
Behavior
What if
Questions
Sophisticated DOE designs to arrive at Robust Designs are possible using this methodology viz.,
D-Optimal and Fractional Factorial Design of Experiments that
d. Reduce the FEA runs and thereby increasing the productivity of the Analysts and Designers.
e. Contribution of each of the variables to the design can be easily studied using this methodology,
which was not possible until now.
f.
Integration of the analysis tool (ANSYS)i and statistical tool (AutoDOE)ii allows the designer to arrive at
optimized design in a very short time thereby reducing the design cycle time. It also allows the designer to
judge the level of contributions of variables on the design response.
References:
AutoDOE, Release July 2002, Quality Engineering & Software Technologies, 24655 Southfield
Road, Southfield, MI 48075,
i
ii