Applying generic constraint solving techniques

1
Applying generic constraint solving techniques
in providing insights into engineering
design Hiroyuki Sawada CAD Centre 30 January,
2001
2
Contents 1. Background of Research 2. Research
Approach 3. Prototype System DeCoSolver (Design
Constraint Solver) 4. Design Example Heat Pump
System Design 5. Conclusion
3
Background of Research
Design Process Process of decision making
? Introducing many design parameters
? Difficulties in gaining an insight into
underlying relationships among design
parameters including
? Critical design parameters for the performance
? Trade-off between design requirements
? Less optimal and/or inferior design solution
Aim of Research
Overcoming the above difficulties by applying
generic constraint solving techniques
4
Research Approach
(1) Formalisation of design as a process of
defining constraints and solving a design
problem using constraints
(2) Development of new constraint solving methods
based on symbolic algebra overcoming
conventional difficulties in
? Analysing incomplete design solutions
? Detecting underlying conflicts between
constraints
? Establishing explicit relationships between
design parameters
Advantages of this Approach
(1) Generic Constraint Based Approach
? Design support system is generic enough to deal
with multidisciplinary design problems
involving mechanics, electrics,
thermodynamics, hydrodynamics, etc.
(2) Rigorous Constraint Solving Methods
? All the results are guaranteed to be correct
mathematically.
5
Prototype System DeCoSolver (Design Constraint
Solver)
Product Explorer Results of Analysis
Context-Tree Is-a Hierarchy of Design
Alternatives
Solver Handler Interface to Constraint Solver
? Optimising Design Parameter Values ?
Computing Possible Design Parameter Values ?
Detecting Conflicts ? Plotting Graph / Design
Solution Space
Constraint Editor Defining Constraints
Component Library Database of commonly used
Components
6
Tc Condensation Temp. Ac Heat Transfer Area
Design Example Heat Pump System
Pd Discharging Press. k Compression Ratio
Qr Mass flow rate of Refrigerant
Te Evaporation Temp. Ae Heat Transfer Area
7
Conventional difficulties ? Loop structure of the
heat pump system ? Non-linearity of thermodynamic
properties
? Complicatedly coupled design parameter
relationships
? Difficulty in gaining insights into underlying
relationships among design parameters
? Design by trial and errors without insights
Design procedure with DeCoSolver
(1) Constructing the product model
(2) Drawing graphs between design parameters
to gain insights into underlying relationships
(3) Determining design parameter values based on
the gained insights
8
(1) Constructing the product model
9
(2) Drawing graphs between design parameters
Gained Insights (1) As Tc increases, Te and Pd
also increase almost linearly. (2) Qr and k are
almost unchanged. (3) As Tc increases, Ac
decreases non-linearly. (4) As Tc increases, Ae
increases non-linearly.
Tc Condensation Temp.
Tc Condensation Temp.
10
(3) Determining design parameter values based on
the gained insights
A small heat transfer area leads to a small
equipment.
? Total heat transfer area, Ac Ae, should be
minimised.
? Other design parameter values will be
determined.
11
Conclusion Advantages of DeCoSolver
Generic and Rigorous Constraint Solving Methods
? Incomplete design solutions in multidiscipline
can be analysed.
? All the analysis results are guaranteed to be
correct mathematically.
? Computational mistakes due to numerical errors
or computational convergence problems are
completely excluded.
? Deep and accurate insights into a design
problem as well as design solutions