Single Simulation
• Single Solve
– Solves a single simulation with
only a single physics
– Engineers are interested in
solution robustness, speed,
accuracy, ease of use and Doesn’t account for
accuracy, ease of use and
engineering results
Doesn’t account for
physics interactions
“Virtual Prototyping” typically requires looking
at multiple physics, This single physics
approach is often not enough.
Single Multiphysics Solve
– Solves a single simulation
involving multiple physics
– Engineers are still interested in
solution robustness, speed,
accuracy, ease of use and
engineering resultsDoesn’t provide
direction for design engineering results
– And the ease and power of the
physics coupling
direction for design
improvement
How can I improve performance?
Can I reduce weight or cost?
What is limiting performance?
Is this a robust design?
“What If” Study
?
??
• User adjusts inputs and investigates
results
• Builds on previous expectations,
adds requirement of easy and robust
parametric updates and comparative
reports
Solutions compared,
but design is not well
understood and no reports
understood and no
optimum is found
Need a more scientific and automated way to
decide which points to solve
Need a way to interpolate between these points
Design Exploration• Scientific methods to explore the
design space fully
• Allows user for: sensitivity and
correlation, DOE and response surface
technology, mesh morphing, charting
and reporting
• For complex solutions, user can take
Response
Surface
Provides design
understanding, But ….
Optimize a design with selected inputs and goals
• For complex solutions, user can take
advantage of robust efficient &
affordable ANSYS distributed solver
Technologies.
understanding, But ….
Arriving at a set of Optimal
design solutions• Optimization
– Searches the design space for
optimal candidates, given user-
defined goals and priorities
– Adds requirements for: advanced
optimization algorithms to
Solutions may be too
sensitive to input optimization algorithms to
efficiently search for candidates,
comparative reporting
Real-world inputs typically have some variation
and may require a more “robust design” goal
sensitive to input
variability
• Taking the variation of inputs into account, and
seeking a design with a probabilistic goal
• RDO => Min standard deviation of the results
Robust Design
Input
distribution
Output
distribution
• RDO => Min standard deviation of the results
• Six Sigma => Optimal design within a safe domain
• In order to arrive at Robust design, User needs to
- Specify probabilistic parameters
- Use probabilistic optimization algorithms
Robust Design optimizes design variables to achieve a
particular probabilistic level such as Six Sigma, which
translates into 3.4 failures in one million parts.
Robust Design optimizes design variables to achieve a
particular probabilistic level such as Six Sigma, which
translates into 3.4 failures in one million parts.
The Path to Robust DesignRobust Design
“What if”
Design Exploration
•DOE, Response
Surfaces, Correlation,
Optimization
•Algorithms
Robust Design
•Probabilistic
Algorithms
•Adjoint solver
methods
Single Physics
Solution
•Accuracy,
robustness, speed…
Multiphysics
Solution
•Integration
Platform
“What if”
Study
•Parametric
Platform
Surfaces, Correlation,
Sensitivity, etc.
Optimization tools at ANSYS
– ANSYS DesignXplorer
• Unified Workbench
solution
– ANSYS Fluent
• Has built-in morphing and
Baseline Design
• Has built-in morphing and
optimization tools
• Has an adjoint solver
– ANSYS MAPDL
• DX VT
– And more
Optimized Design
ANSYS DesignXplorerIntegral with Workbench
• Parametric multiphysics
modeling with automated
updates
• Bi-directional CAD, RSM,
scripting, reporting and
more...
Integral with Workbench
• Parametric multiphysics
modeling with automated
updates
• Bi-directional CAD, RSM,
scripting, reporting and
more...more...more...
ANSYS DesignXplorer
DesignXplorer is everything
under this Parameter bar…
• Low cost & easy to use!
• It drives Workbench
DesignXplorer is everything
under this Parameter bar…
• Low cost & easy to use!
• It drives Workbench ANSYS
• Improves the ROI!• Improves the ROI!
DX
ANSYS Workbench
Solvers
ANSYS DesignXplorer Design of
Experiments
With little more effort than for a single
run, you can use DesignXplorer to
create a DOE and run many variations.
With little more effort than for a single
run, you can use DesignXplorer to
create a DOE and run many variations.
Correlation Matrix
Understand how your parameters are
correlated/influenced by other parameters!
Understand how your parameters are
correlated/influenced by other parameters!
Sensitivity
Understand which
parameters your
design is most
sensitive to!
Understand which
parameters your
design is most
sensitive to!
Response Surface
Understand the
sensitivities of the output
parameters (results) wrt
the input parameters.
Understand the
sensitivities of the output
parameters (results) wrt
the input parameters.3D Response
2D Slices Response
Goal-Driven Optimization
Use an optimization algorithm
or screening to understand
tradeoffs or discover optimal
design candidates!
Use an optimization algorithm
or screening to understand
tradeoffs or discover optimal
design candidates!
Robustness EvaluationInput parameters have variation!
Make sure your
design is robust!
Six Sigma, TQM
Make sure your
design is robust!
Six Sigma, TQM
Output
parameters
vary also!
Understand how your
performance will vary
with your design
tolerances?
Understand how your
performance will vary
with your design
tolerances?
Predict how
many parts will
likely fail?
Predict how
many parts will
likely fail?
Understand which
inputs require the
greatest control?
Understand which
inputs require the
greatest control?
• Engineers face numerous obstacles that prevent them from using
optimization fully
• Our plans are largely based on removing those obstacles so our
customers can get more out of simulation
Challenges
Do you use optimization
software?
Matlab
16%
18.6%41.9%
21.8%
The Path to Robust DesignRobust Design
OptimizationOptimization
Robust DesignRobust Design
• Build on the solid foundation of Workbench as
a platform for parametric simulation
• Develop DX as an optimization platform
• Build up to probabilistic optimization
Single PhysicsSingle Physics
MultiphysicsMultiphysics
“What if” Study“What if” Study
DesignDesign ExplorationExploration
OptimizationOptimization