Design the Future 2: Enabling Design Re-use through Predictive CAD

Lead Research Organisation: University of Strathclyde
Department Name: Design Manufacture and Engineering Man

Abstract

Engineering Design work typically consists of reusing, configuring, and assembling of existing components, solutions and knowledge. It has been suggested that more than 75% of design activity comprises reuse of previously existing knowledge.

However in spite of the importance of design reuse activities researchers have estimated that 69% of companies have no systematic approaches to preventing the "reinvention of the wheel". The major issue for supporting design re-use is providing solutions that partially re-use previous designs to satisfy new requirements. Although 3D Search technologies that aim to create "a Google for 3D shapes" have been increasing in capability and speed for over a decade they have not found widespread application and have been referred to as "a solution looking for a problem"! This project is motivated by the belief that, with a new type of user interface, 3D search could be the solutions to the design reuse problem.

The system this research is aiming to produce is analogous to the text message systems of mobile phones. On mobile phones 'Predictive text' systems complete words or phrases by matching fragments against dictionaries or phrases used in previous messages. Similarly a 'predictive CAD' system would complete 3D models using 'shape search' technology to interactively match partial CAD features against component databases. In this way the system would prompt the users with fragments of 3D components that complete, or extend, geometry added by the user. Such a system could potential increase design productivity by making the reuse of established designs an efficient part of engineering design.

Although feature based retrieval of components from databases of 3D components has been demonstrated by many researchers so far the systems reported have been relatively slow and unable to be components of an interactive design system. However recent breakthroughs in sub-graph matching algorithms have enabled the emergence of a new generation of shape retrieval algorithms, which coupled with multi-core hardware, are now fast enough to support interactive, predictive design interfaces. This proposal aims to investigate the hypothesis that a "Predictive CAD" system would allow engineers to more effectively design new components that incorporate established, or standard, functional or manufacturing geometries. This would find commercial applications within large or distributed engineering organizations.

This project is an example of how data mining could potentially be employed to increase design productivity because even small engineering companies will have many hundreds of megabytes of CAD data that a "Predictive CAD" system would effectively pattern match against.

Planned Impact

The PI is an established academic who has a track record of close collaboration with industry to ensure the commercial impact of funded research. Software resulting from previous EPSRC research is incorporated in Pathtrace's EdgeCAM system and the published algorithms from other projects have been incorporated in the commercial products of two other UK companies. He has a consistent record of publishing the results of EPSRC funded work in high impact journals and has an H index of 25 and a total of 2065 citations.

Similarly, the CI is an established academic who has a track record of close collaboration with industry to ensure the commercial impact of funded research. His research on reliability growth has been adopted for international standards IEC 61164 and forms part of BS5760 and his DTI-funded REMM programme (1998-2004) of work resulted in an industrial impact case for the past REF, where the Business School, was ranked third in the UK.

Resources for Impact
Funding for conference participation and travel to do on-site demonstrations of the prototype system (as part of the evaluation process) have been requested and will ensure that the results are widely disseminated.

The research has the potential to impact on the following non-academic beneficiaries:
1) Engineering Businesses: engaged in the design of engineering products will potentially benefit from increased design productivity and the reduction in manufacturing costs associated with reuse. Beyond the immediate impact on the project's collaborators participation in AFRC's industrial events will facilitate dissemination across the HVM Catapult network.

2) CAD/CAM software developers: will potentially benefit from the creation of a user interface that directly mines the big-data of a company's product catalogue to deliver a step change in the levels of reuse. As a result of the sustained support of industrial IT by EPSRC in 1980s the UK is home to over 30 world class software developers focused on the Mechanical CAD/CAM market. These range from the large development groups of multinationals (e.g. SolidWorks, AutoDesk, Delcam) through mid-size suppliers of software components (e.g. Lightworks, Pathtrace, Theorem Solutions) to small new companies offering novel technologies (Shapespace, Anarkik3D). Collaborator Shapespace Ltd already supplies component software to the Mechanical CAD market and so will provide a pathway to the exploitation of any results.

The research also has the potential for social-economic impact:
3) Circular Economy: Beyond the immediate economic impacts there are societal benefits associate with increased levels of reuse as a contribution to the emerging circular economy. Every design that is reused not only saves design time but is also associated with reduced carbon and energy costs through-out the product life cycle. For example new tooling (e.g. moulds, jigs and fixtures) does not have to be created and 'economies of scale' will allow larger orders from established supply chains with lower transport costs. Consequently this work can contribute to, and be disseminated through, the 'Design for Remanufacture' theme of the 'Scottish Institute for Remanufacturing' of which the PI is co-director.

Applications and exploitation Protection and sharing of IP relating to the Predictive CAD system (i.e. algorithms and code) will be discussed and agreed at the start of the project with a formal collaboration agreement between all partners.

Communications and engagement Direct engagement with industry professionals involved in the project's steering committee will ensure effective dissemination to the different IP stakeholder groups identified in this proposal. The project's work plan has been designed to ensure that the project aims and results are communicated effectively to all stakeholders. Part of the steering committee meetings will be devoted to ensuring the demonstration and dissemination to stakeholder representatives.

Publications

10 25 50
publication icon
Vasantha G (2020) A Probabilistic Design Reuse Index for Engineering Designs in Journal of Mechanical Design

 
Description We have discovered that a number of data mining techniques (i.e. Item sets, point processes and Neural Networks) that can be used to identify patterns of feature occurrence within large CAD databases. We have also explored the possibilities of using N-Grams and Bayesian networks but so far with less success. Although the UKRI funding has now come to an end the project has been awarded a COVID extension and work will continue until the end of May 2021 to validate findings (via case studies), disseminate results (via publications) and increase impact (via collaboration with the National Manufacturing Institute of Scotland).
Exploitation Route They might increase the productivity of engineering designers by enabling better reuse of previous designs and manufacturing engineers in the accuracy of their estimates of production costs.
Sectors Aerospace, Defence and Marine,Creative Economy,Manufacturing, including Industrial Biotechology

 
Title Data for: "A Probabilistic Design Reuse Index for Engineering Designs" 
Description Dataset to demonstrate Proposed Probabilistic Measure of Design Reuse This file contains data collected from a furniture company with an analysis procedure to demonstrate the probabilistic measure of design reuse. Please read our paper "A Probabilistic Design Reuse Index for Engineering Designs" before understanding this dataset. The file contains a MS Excel file with three sheets. The details of each sheet are given below: Sheets 1: Furniture data - This sheet contains all the data collected for chairs type furniture. Each unique component part is represented by the category instance and component occurrences recorded with reference to a specific named product. Sheet 2: Full Dataset Chi-square score - This sheet illustrates the calculation of a probabilistic measure of design reuse for the chair products. The sheet transposes the component occurrence given in sheet one with subsequent calculation of Chi-Square distribution score. Please refer Equation 6 in the above mentioned paper. The subsequent rows demonstrates the total summation of above calculated values for each component and calculation of Chi-square value. The formula used to calculate the score can be noticed in Formula bar for each cell. Sheet 3: Group wise Chi-square score - This sheet illustrates the calculation of a probabilistic measure of design reuse for each component type. The procedure is same as illustrated in the above point. 
Type Of Material Database/Collection of data 
Year Produced 2020 
Provided To Others? Yes  
URL https://pureportal.strath.ac.uk/en/datasets/f931bfd7-1090-408e-8823-e4604b485713
 
Title Data for: "A Probabilistic Measure of Design Reuse" 
Description "This dataset contains data collected from a furniture company and analysis procedure to demonstrate our newly proposed probabilistic measure of design reuse. Please read our ASME conference paper 2018 titled Â"A Probabilistic Measure of Design ReuseÂ" before understanding this dataset. The dataset contains a MS Excel file with four sheets. The details of each sheet are illustrated below: Sheet 1: Double bed data Â- This sheet contains all the furniture data collected for Double-bed type furniture. Each component is represented by the respective classification codes and component occurrences with reference to a specific bed product. Sheet 2: Component classification Â- This sheet contains transformed data from Sheet 1. The component classification and respective occurrences are grouped in column wise. The sheet also provides the number of component options in a group frequency and the total components in a group. Sheet 3: Full Dataset Chi-square score Â- This sheet illustrates the calculation of a probabilistic measure of design reuse for Double bed product type. The sheet expands Sheet 2 with subsequent calculation of Chi-Square distribution score. Please refer Equation 7 in the above mentioned paper. The subsequent rows demonstrates the total summation of above calculated values for each component and calculation of Chi-square value. The formula used to calculate the score can be noticed in Formula bar for each cell. Sheet 4: Group wise Chi-square score Â- This sheet illustrates the calculation of a probabilistic measure of design reuse for each component type. The procedure is same as illustrated in the above point. " 
Type Of Material Database/Collection of data 
Year Produced 2018 
Provided To Others? Yes  
Impact