• International Journal of Technology (IJTech)
  • Vol 9, No 4 (2018)

Genetic Algorithm-based Multi-criteria Approach to Product Modularization

Genetic Algorithm-based Multi-criteria Approach to Product Modularization

Title: Genetic Algorithm-based Multi-criteria Approach to Product Modularization
Binay Kumar, Ritesh Kumar Singh, Surendra Kumar

Corresponding email:


Published at : 25 Jul 2018
Volume : IJtech Vol 9, No 4 (2018)
DOI : https://doi.org/10.14716/ijtech.v9i4.819

Cite this article as:
Kumar, B., Singh, R.K., Kumar, S., 2018. Genetic Algorithm-based Multi-criteria Approach to Product Modularization. International Journal of Technology. Volume 9(4), pp. 775-786

818
Downloads
Binay Kumar - Department of production engineering, Birla Institute of technology, mesra, Ranchi, India
-
Ritesh Kumar Singh birla institute of technology, mesra, ranchi, India
Surendra Kumar Ex. Professor, birla institute of technology, mesra, ranchi, India
Email to Corresponding Author

Abstract
Genetic Algorithm-based Multi-criteria Approach to Product Modularization

Modularization is one of the key strategies for increasing responsiveness to customers.  In modular product architecture a wide variety of product configurations can be generated by altering a limited number of modules and components. Product modules are identified by grouping highly coupled components in the same module. A Design Structure Matrix (DSM) is a compact presentation of the interaction between the components. In this paper, Analytical Hierarchy Process (AHP) and Genetic Algorithm (GA)-based methodology is proposed for the clustering of highly coupled DSM components in modules. Multi-criteria DSMs are proposed, which are aggregated by using weights generated by AHP. A genetic algorithm is designed to change the order of components in DSM and to bring highly coupled interactions near the diagonal. An illustrative case study is also made to validate the proposed algorithm. Two large sized and two small size modules are identified by selecting high density clusters around the diagonal. The clustered DSM also shows independent components and loose coupling between the two modules.

Analytical hierarchy process; Clustering; DSM; Genetic algorithm; Modularization

Introduction

In the current competitive and volatile market, the challenges of managing a variety of products is increasing exponentially. Consequently, companies are looking for strategies to improve their competitiveness in terms of cost of production and lead time in the demand uncertainty environment. One of such concept, which satisfies customer needs by providing a range of products at optimal cost, without delay and without sacrificing the value of the product, is “Product Modularity” (Partanen & Haapasalo, 2004).   This can be defined as a basic building block of a product, which performs a specific task. Modules are physical structures that have a one-to-one correspondence with functional structures (Ulrich & Tung, 1991).  A product is usually made of two or more building blocks or modules, which interact through conversion or transmission of energy, and physically interface with one or more physical components.

Many researchers have proposed different methods for the modularization of products. Pimmler and Eppinger (1994) developed a system engineering technique by using design decomposition for complex interaction between design components. Stone (1997) proposed a new method for component clustering for the development of new products using modular design.

Salhieh and Kamrani (1999) constructed a framework for the integration of components by using matching modular strategies, while Dahmus et al. (2001) presented a new approach for the development of modular product families, in which the modules are interchangeable. Kusiak and Huang (1996) developed a model and solution for modularity-based systems. Chen et al. (2011) proposed GA based panning system that employs building information model (BIM), object sequencing matrix (OSM), to obtain an optimal crew assignment under resource and workspace constraints. Nasruddin et al. (2018) applied a GA based optimization technique for optimizing total energy destruction and total annual cost of a geothermal power plant.

Kreng and Lee (2004) used a grouping genetic algorithm to create modular product design, while Nepal (2005) presented a fuzzy logic-based approach to product modularization. Lee (2010) proposed an NSGA-based methodology for the formation of modules using strategic factors; however, they did not consider important factors which define physical proximity, energy and information, and material-based interactions. Rogers et al. (2006) proposed a genetic algorithm-based method for the clustering of a DSM square matrix. However, their model only addresses a single matrix and does not consider the clustering of multiple matrices for mapping the interactions between components with respect to different factors.  As the number of components increases, the complexity of the problem of identification of appropriate modules increases. In such a scenario, many of the existing methods either become very time consuming or inefficient. A genetic algorithm is a proven method for quickly finding near optimal solutions to complex problems.

This paper aims to develop an AHP, genetic algorithm and DSM-based methodology for developing product modules. The overall methodology is divided into three parts. In the first, DSM is used to populate the product architecture and identify the degree of interaction between the various components of the product. The component interactions are quantified in terms of factors such as spatiality, energy, information and material, depending upon the degree of interaction between factors. In the second part, the importance of each of the factors with respect to the formation of the product module is enumerated using the analytical hierarchy process (AHP). The weights of the factors are used to aggregate all the matrices into one matrix. Finally, a genetic algorithm (GA)-based method is proposed for the clustering of the DSM. The proposed method partitions the product into a set of modules, in which interactions within individual modules are maximized, and those outside them are minimized. The proposed method for developing product modules is then verified using an example of a real product.

Conclusion

For complex design projects, due to mass customization and shorter design times, the execution of design cycles has become a very difficult task. Product architecture selection directly affects the ease or difficulty of design. Adoption of integrated architecture leads to efficient product functionality, but exponentially increases the difficulty of design and manufacturing tasks. However, by adopting modular architecture a firm can rapidly and cost effectively introduce products that meet the varied customer demand.

Most of the available methods do not consider multiple DSMs for mapping the relationship between components. In DSM, off-diagonal elements show strong coupling between two elements. In this paper, a Genetic Algorithm (GA) and Analytical Hierarchy Process (AHP)- based methodology is propose for the development of modules (chunks). In the proposed method, multiple DSMs represent the interaction between components with respect to different factors, such as spatiality and energy.

A case study of portable drills has been made. The interaction between two corresponding elements has been measured on a scale of 1 to 4, as proposed by Pimmler and Eppinger (1994). Using AHP, the weights of the factors are determined, and an aggregate DSM is obtained by adding together all the weighted DSMs. The penalty concept is used to calculate the fitness value. Each cell with a value greater than the threshold value ? is multiplied by its distance from the diagonal. The problem is defined in terms of the minimization of the fitness value or the overall penalty. By using a heuristic mutation method, the GA rearranges the order of appearance of the components in the aggregated DSM to bring higher value cells near the diagonal. By rearranging the order of components, the GA tries to minimize the overall penalty. After 100 to 150 generations, negligible variation in the fitness value was observed..  Finally, clustering was achieved by keeping adjacent high-density items along the diagonal in the same cluster. This method is best suited for cases in which the number of elements is large, and it significantly reduces the time required for modularization.

References

Choubey, A.M., 2007. Platform Based Approach for Economic Production of a Product Family. Ph.D. Dissertation, Kansas State University, USA

Cheng, R., Gen, M., 1994. Crossover on intensive search and traveling salesman problem. Computers & Industrial Engineering, Volume 27(1-4), pp. 485–488.

Chen, Y., Feng, C., Wang, Y., Wu, H., 2011. Using BIM Model and Genetic Algorithms to Optimize The Crew Assignment for Construction Project Planning. International Journal of Technology. Volume 2(3), pp.179–187

Dahmus, J.B., Gonzalez-zugasti, J.P., Otto, K.N., 2001. Modular Product Architecture. Design Studies, Volume 22(5), pp. 409424

Kusiak, A., Huang, C.-C., 1996. Development of Modular Products. IEEE Transactions on Components, Packaging, and Manufacturing Technology: Part A, Volume 19(4), pp. 523538

Kreng, V.B., Lee, T.P., 2004. Modular Product Design with Grouping Genetic Algorithm—A Case Study. Computers & Industrial Engineering, Volume 46(3), pp. 443460

Lee, Michael , 2010. Product Modularity: A Multi-objective Configuration Approach. Ph.D. Dissertation, Loughborough University, UK

Nepal, B., 2005. An Integrated Framework for Modular Product Architecture. Ph.D. Dissertation, ETD Collection for Wayne State University, USA

Nasruddin, Nasution, S., Aisyah, N., Surachman, A., Wibowo, A.S., 2018. Exergy Analysis and Exergoeconomic Optimization of a Binary Cycle System using a Multi Objective Genetic Algorithm. International Journal of Technology. Volume 9(2), pp. 275–286

Pimmler, T.U., Eppinger, S.D., 1994. Integration Analysis of Product Decompositions. ASME Design Theory and Methodology Conference, Minneapolis, MN, USA

Partanen, J., Haapasalo, H., 2004. Fast Production for Order Fulfillment: Implementing Mass Customization in Electronics Industry. International Journal of Production Economics, Volume 90(2), pp. 213222

Rogers, J.L., John, J.K., Bilardo Jr., V.J., 2006. Development of a Genetic Algorithm to Automate Clustering of a Dependency Structure Matrix. National Aeronautics and Space Administration, NASA/TM-2006-214279

Saaty, T.L., 1980. The Analytic Hierarchy Process. McGraw hill, New York, USA

Steward, D.V., 1981. The Design Structure System: A Method for Managing the Design of Complex Systems. IEEE Transactions on Engineering Management, Volume EM-28(3)

Salhieh, S.M., Kamrani, A.K., 1999. Macro Level Product Development using Design for Modularity. Robotics and Computer-Integrated Manufacturing, Volume 15(4), pp. 319329

Stone, R.B.., 1997. Towards a Theory of Modular Design. Ph.D. Dissertation, The University of Texas at Austin, USA

Ulrich, K., Tung, K., 1991. Fundamentals of Product Modularity. American Society of Mechanical Engineers, Volume 39, pp. 7379

Yassine, A.A., Sharman, D.M., 2004. Characterizing Complex Product Architectures. Systems Engineering 7, Volume 1, pp. 35–60