Customer-Centered Design Sampling for CAD Products Using Spatial Simulated Annealing

CAD’17

Okayama, Japan

Erkan Gunpinar¹, Shahroz Khan¹, Masaki Moriguchi²

1 Istanbul Technical University

2 Meiji University

​

Research Objective

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

2

CAD'17, Okayama, Japan

• Technique that can sample and generate variety of design options for a CAD product based on the customer preferences.
• Sampling technique called Spatial Simulated Annealing-Design Sampler (SSA-DS) is proposed to obtain different design variations of a CAD product.

SSA-DS

Design parameters

Parametric bounds

Previous Works

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

3

CAD'17, Okayama, Japan

• Based on LHS for design selections.
• Randomly distributed designs.
• Mutation operator – to moved designs to feasible region.

Fuerle and Sienz, 2011

Chen et. al. 2013

• Used spatial simulated annealing approach based on the minimization of shortest distance criterion.

Fuerle and Sienz (2011)

Chen et al. (2013)

Drawbacks

• Not applicable for the high dimensional sampling problems.
• CAD samplingis not possible with these techniques.

​

Customer Centered Design

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

4

CAD'17, Okayama, Japan

• A Process in which customer/user of a product is involved throughout its design phase.
• Why customer centered design ?
• Customer satisfaction
• Easy-to-use product
• Improved user productivity
• Low expenditure on technical support
• Increase market share

​

CAD Design Sampling

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

5

CAD'17, Okayama, Japan

• CAD model parameterization (design parameters).
• Relationships between the design parameters (design constraints).
• Range of the design parameters (lower and upper bounds).
• Design parameters and their bounds form the design space.
• Each design parameter represents a dimension in the design space.

Parameterized yacht hull design

Sampled Designs

Work Flow of Methodology

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

6

CAD'17, Okayama, Japan

SSA-DS

Learn Customer preferences from initial designs via one-to-one interviews

Translate customer preferences as geometric constraints

Initial Designs

Design parameters

Parametric bounds

SSA-DS

Customer-Centered Designs

Design parameters

Parametric bounds

Number of designs (N)

Number of designs (N)

Inputs

Output

SSA Design Sampler (SSA-DS)

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

7

CAD'17, Okayama, Japan

Cost Function

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

8

CAD'17, Okayama, Japan

• Cost function is define based on the space-filling property.
• Design should be uniformly distributed over the design space.
• Space-filling is achieved using the Audze and Eglais‘ (2011) technique.

With Space-filling

Without Space-filling

Experimental Study

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

9

CAD'17, Okayama, Japan

• A wine glass (without base) is selected as a test model.
• Cubic Bezier curves are utilized to generate upper and lower region.
• 3D surfaces are generated by performing lofting operation.

Results –Initial designs

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

10

CAD'17, Okayama, Japan

Results –Learning perfernces

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

11

CAD'17, Okayama, Japan

• A participant was selected and recognized as customer.
• Each design was shown to the participant during an one-to-one interview.
• Participant preferences were quantified and converted into design constraints.

​

Conclusions

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

12

CAD'17, Okayama, Japan

Future Works

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

13

CAD'17, Okayama, Japan

• Current research will try to further improve the space filling property of the SSA-DS algorithm.
• More test models and participants will be involved in order to validate the performance.
• Participants response can be used to obtain the critical design parameters.

Acknowledgment

This work was supported by The Scientific and Technological Research Council of Turkey (Project Number: 315M077)

Erkan Gunpinar, Shahroz Khan, Masaki Moriguchi

14

CAD'17, Okayama, Japan