Assembly System Reconfiguration Planning Using Genetic Algorithm

Author(s):  
A. Bryan ◽  
S. J. Hu ◽  
Y. Koren

Due to increased competition, the rate at which manufacturers introduce new product families to the market is increasing. However, the cost of changing manufacturing facilities to produce new product families can outweigh the benefits obtained from increased revenue. Reconfigurable Manufacturing Systems (RMSs) have been proposed as a cost effective strategy for manufacturing product families. Although methods for measuring RMS scalability and convertibility exist, there is a lack of methods for obtaining reconfiguration plans for assembly systems. This paper introduces assembly system reconfiguration planning (ASRP) as method to obtain reconfiguration plans for assembly systems. A genetic algorithm is developed for solving the ASRP problem.

Author(s):  
A. Bryan ◽  
S. J. Hu ◽  
Y. Koren

The need to cost effectively introduce new generations of product families within ever decreasing time frames have led manufacturers to seek product development strategies with a multigenerational outlook. Co-evolution of product families and assembly systems is a methodology that leads to the simultaneous design of several generations of product families and reconfigurable assembly systems that optimize life cycle costs. Two strategies that are necessary for the implementation of the co-evolution of product families and assembly systems methodology are: (1) The concurrent design of product families and assembly systems and (2) Assembly system reconfiguration planning (ASRP). ASRP is used for the determination of the assembly system reconfiguration plans that minimize the cost of producing several generations of product families. More specifically, the objective of ASRP is to minimize the net present cost of producing successive generations of products. This paper introduces a method for finding optimum solutions to the ASRP problem. The solution methodology involves the generation of a staged network of assembly system plans for all the generations that the product family is expected to be produced. Each stage in the network represents a generation that the product family is produced, while each state within a stage represents a potential assembly system configuration. A novel algorithm for generating the states (i.e. assembly system configurations) within each generation is also introduced. A dynamic program is used to find the cost minimizing path through the network. An example is used to demonstrate the implementation of the ASRP methodology.


Author(s):  
April Bryan ◽  
S. Jack Hu ◽  
Yoram Koren

Decreasing product life cycles and reduced product development times have led to a need for new strategies for coping with the rapid rate of product family design changes. In this paper, assembly system reconfiguration planning (ASRP) is introduced as a method for cost effectively designing several generations of assembly systems in order to produce a product family that gradually evolves over time. In the ASRP approach, the possible assembly systems for each generation are first considered and then the sequence of assembly system configurations that minimize the life cycle cost of the process are selected. A nonlinear integer optimization formulation is developed for finding the cost minimizing assembly system reconfiguration plan using the ASRP approach. Dynamic programming and genetic algorithm are used to solve the optimization problem. Simulation results indicate that the ASRP approach leads to the minimum life cycle costs of the assembly system, and the relative cost of reconfiguration and production have an impact on the assembly system reconfiguration plan selected. Comparison of the results of the dynamic program and genetic algorithm indicate that the dynamic program is more computationally efficient for small problems and genetic algorithm is preferred for larger problems.


Author(s):  
A. Bryan ◽  
S. J. Hu ◽  
Y. Koren

In order to gain competitive advantage, manufacturers require cost effective methods for developing a variety of products within short time periods. Product families, reconfigurable assembly systems and concurrent engineering are frequently used to achieve this desired cost effective and rapid supply of product variety. The independent development of methodologies for product family design and assembly system design has led to a sequential approach to the design of product families and assembly systems. However, the designs of product families and assembly systems are interdependent and efficiencies can be gained through their concurrent design. There are no quantitative concurrent engineering techniques that address the problem of the concurrent design of product families and assembly systems. In this paper, a non-linear integer programming formulation for the concurrent design of a product family and assembly system is introduced. The problem is solved with a genetic algorithm. An example is used to demonstrate the advantage of the concurrent approach to product family and assembly system design over the existing sequential methodology.


Author(s):  
Jian Liu ◽  
Derek M. Yip-Hoi ◽  
Wencai Wang ◽  
Li Tang

Manufactures are adopting Reconfigurable Manufacturing Systems (RMS) to better cope with frequently changing market conditions, which place tremendous demands on a system’s flexibility as well as its cost-effectiveness. Considerable efforts have been devoted to the development of necessary tools for the system level design and performance improvement, resulting in approaches to designing a single RMS. In this paper, a methodology for cost-effective reconfiguration planning for multi-module-multi-product RMS’s that best reflect the market demand changes is proposed. Formulated as an optimization procedure, reconfiguration planning is defined as the best reallocation of part families to production modules in an RMS and the best rebalancing of the whole system and each individual module to achieve minimum related cost and simultaneously satisfy the market demand. A Genetic Algorithm (GA) approach is proposed to overcome the computational difficulties caused by the problem complexity. Effectiveness of the proposed methodology is demonstrated with a case study.


Author(s):  
A M Farid

In recent years, many design approaches have been developed for automated manufacturing systems in the fields of reconfigurable manufacturing systems (RMSs), holonic manufacturing systems (HMSs), and multiagent systems (MASs). One of the principle reasons for these developments has been to enhance the reconfigurability of a manufacturing system, allowing it to adapt readily to changes over time. However, to date reconfigurability assessment has been limited. Hence the efficacy of these design approaches remains inconclusive. This paper is the second of two in this issue to address reconfigurability measurement. Specifically, ‘reconfiguration ease’ has often been qualitatively argued to depend on the system's modularity. For this purpose, this paper develops modularity measures in a three-step approach. Firstly, the nature of typical manufacturing system interfaces is discussed. Next, the qualitative understanding underlying existing modularity measures is distilled. Finally, these understandings are synthesized for a manufacturing system context. This approach forms the second of two pillars that together lay the foundation for an integrated reconfigurability measurement process described elsewhere.


Author(s):  
A M Farid ◽  
D C McFarlane

In recent years, many design approaches have been developed for automated manufacturing systems in the fields of reconfigurable manufacturing systems (RMSs), holonic manufacturing systems (HMSs), and multi-agent systems (MASs). One of the principle reasons for these developments has been to enhance the reconfigurability of a manufacturing system, allowing it to adapt readily to changes over time. However, to date, reconfigurability assessment has been limited. Hence, the efficacy of these design approaches remains inconclusive. This paper is the first of two in this issue to address reconfigurability measurement. Specifically, it seeks to address ‘reconfiguration potential’ by analogy. Mechanical degrees of freedom have been used in the field of mechanics as a means of determining the independent directions of motion of a mechanical system. By analogy, manufacturing degrees of freedom can be used to determine independent ways of production. Furthermore, manufacturing degrees of freedom can be classified into their production and product varieties. This paper specifically focuses on the former to measure the product-independent aspects of manufacturing system ‘reconfiguration potential’. This approach will be added to complementary work on the measurement of ‘reconfiguration ease’ so as to form an integrated reconfigurability measurement process described elsewhere [1—5].


2011 ◽  
Vol 110-116 ◽  
pp. 1442-1446 ◽  
Author(s):  
Dodla Srikanth ◽  
M. S. Kulkarni

Reconfigurable Manufacturing Systems (RMS) have the potential to emerge as a cost effective solution that will help manufacturing organizations to stay competitive in an environment where product mix changes frequently and product life cycles are getting shorter. Reconfigurable Manufacturing Systems can achieve this as they are designed for quick changes in its configuration, as well as software and hardware components. This not only helps in accommodating production capacity but also production of new variety of products and introduction of new product within part family. However, the configurations have a significant impact on the productivity and reliability of the machines and the manufacturing system. In the present paper, the main objective is to present a framework consisting of Maintenance plan to be followed for the reconfigured machine, Reliability of the reconfigured machine, Quality of the product obtained. This framework can form as the basic idea and a link between maintenance, reliability and quality issues.


Sign in / Sign up

Export Citation Format

Share Document