- Email: [email protected]
Balancing Assembly and Transfer Lines
European Journal of Operational Research 168 (2006) 663–665 www.elsevier.com/locate/ejor
Guest editorial
Balancing Assembly and Transfer Lines
The first idea of a Feature Issue on Line Design and Balancing appeared in April 2001 at the Conference MOSIMÕ01 ‘‘Design, Analysis and Management of Industrial Systems’’. This conference has been sponsorized by EURO. Later, in July 2002, an invited session on line balancing has been organized at the World Congress of the IFAC. These two events are at the origin of this Feature Issue. For this Issue, more than 30 papers were submitted from the most known research teams in the domain. After per review process, 15 best papers were selected. Professor Armin Scholl is invited for a state-of-the-art contribution. So, this Issue is a photography of the present state of advanced discrete optimization methods for decision aid in production lines design. The most interesting problems and the most promising approaches are explained and developed in this Issue. The proposed methods are complementary and they open several interesting research perspectives. Marc Peeters and Zeger Degraeve in ‘‘An LPbased lower bound for the simple assembly line balancing problem’’ propose a new lower bound for the classical simple assemble line balancing problem (SALBP); this bound is based on LPrelaxation and Danzig–Wolfe decomposition. Computational results and analysis of the quality of the new lower bound are presented. Alexandre Dolgui, Nikolai Guschinsky and Genrikh Levin in the paper entitled ‘‘A special case
of transfer lines balancing by graph approach’’ deal with a new line balancing problem so-called transfer line balancing problem (TLBP). Usually, the line balancing problems are considered in an assembly environment. In this paper, the authors showed a line balancing problem in machining/ processing environment. The specificity of this problem is the existence of blocks of parallel operations at the workstations (multi-tools spindle heads for machining). The block time is equal to the maximal time of its operations. The operational time of each workstation is equal to maximal time of its blocks. The authors propose an approach to solve a cost-oriented version of TLBP, by reduction of initial problem to a constrained shortest path one. Matthias Amen in the paper ‘‘Cost-oriented assembly line balancing: model formulations, solution difficulty, upper and lower bounds’’ deals with a cost-oriented assembly line balancing problem. The author proposes a mathematical model of the problem and new uppers and lower bounds for some problemÕs parameters. These results may be used to solve the cost-oriented SALBP by a branch-and-bound algorithm. The paper ‘‘An optimal piecewise-linear program of the U-line balancing problem with stochastic task times’’ by Timothy L. Urban and Wen-Chyuan Chiang presents an U-shaped assembly line balancing problem. The authors analyze a case with tasks having independent Gauss distributed durations and formulate a stochastic model
0377-2217/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ejor.2004.07.021
664
Guest editorial / European Journal of Operational Research 168 (2006) 663–665
of the problem and propose a procedure to obtain a lower bound for the number of workstations. Yuri N. Sotskov, Alexandre Dolgui and MarieClaude Portmann in the paper ‘‘Stability analysis of an optimal balance for an assembly line with fixed cycle time’’ consider a SALBP with non-classical assumptions under consideration. Namely, the durations of the ‘‘manual’’ operations may have small variations. The problem is to calculate the stability radius of an optimal line balance. The authors show that this stability radius can be calculated by a polynomial-time algorithm. Nguyen Van Hop in ‘‘Heuristic solution for fuzzy mixed-model line balancing problem’’ addresses to mixed-model line balancing problem with fuzzy processing time. A mixed model of fuzzy ALBP is formulated and a heuristic algorithm to solve it is proposed. The main idea is to arrange the jobs in a sequence and to allocate them to workstations by an exchange procedure. Gregory Levitin, Jacob Rubinovitz and Boris Shnits in ‘‘A genetic algorithm for robotic assembly line balancing’’ consider robotic assembly lines where the durations of works depend on the characteristics of robot types and only one robot can be assigned to each workstation. The objective function is to minimize the cycle time of a line with a given number of workstations by assigning robots and tasks to workstations. To solve the problem the authors propose a genetic algorithm (GA). The results of the GA are improved by a local optimization (hill climbing) work-piece exchange procedure. The paper ‘‘Balancing assembly lines with Tabu search’’ by Sophie D. Lapierre, Angel Ruiz, and Patrick Soriano propose a new Tabu search algorithm to solve an assembly line balancing problem. The main idea is to incorporate an intensification/ diversification framework and to redefine the solution space in order to check infeasible solutions. The extended computational experiments are carried out and the advantages of proposed procedures are discussed. The paper ‘‘An endosymbiotic evolutionary algorithm for the integration of balancing and sequencing in mixed-model U-lines’’ by Yeo Keun Kim, Jae Yun Kim and Yeongho Kim deals with the problem of balancing and sequencing in an
U-shaped line. The authors propose a new evolutionary algorithm. This algorithm is based on using auxiliary information for the modeling of the evolution process. This approach allows decreasing the number of operations required to solve the initial problem and to improve the quality of the obtained solutions. Ronald G. Askin and Jiaqiong Chen in the paper ‘‘Dynamic task assignment for throughput maximization with worksharing’’ study a dynamic assembly line balancing problem. The case is considered where task sequences are known but the workforce is partially cross-trained and some tasks can alternate between workstations. The objective is to define a tradeoff between the cost of workin-process (WIP) inventory and the cost of crosstraining. A simulation model and various heuristic rules that attempt to approximate an optimal solution are explained. John J. Bartholdi III, Donald D. Eisenstein and Yun Fong Lim in the paper ‘‘Bucket brigades on in-tree assembly networks’’ consider a network of assembly lines. The goal is to balance each assembly line such that all lines produce with the same rate. The authors present an adaptation of ‘‘bucket brigade’’ protocol which allows obtaining a balance and synchronization. Stefan Bock, Otto Rosenberg and Thomas van Brackel in the paper ‘‘Controlling mixed-model assembly lines in real-time by using distributed systems’’ study an assembly line which produces several variants of products. Unforeseen disturbances of the production process can compromise its planned execution, therefore some on-line decisions is necessary. The authors formulate an integer programming model. A two phases approach is proposed. At first phase, an initial solution is generated by a well-known approach. At second phase, a set of heuristics is used to solution improvement. These heuristics can be executed in a distributed computational system. The authors present a comparison with a simulated annealing algorithm. In the paper ‘‘Job sequencing and material flow control of an assembly line’’ by Subhash C. Sarin, Michael P. Greco and Ezey M. Dar-El, the problem of loading of products (flow of material) on a assembly line is considered. The number of items is kept the same, and a new item is loaded only
Guest editorial / European Journal of Operational Research 168 (2006) 663–665
when an item leaves the line. Therefore, the problem is to determine an optimal cyclic sequence in order to achieve optimum throughput with minimum WIP. A lower bound of the WIP level is proposed. A new policy of product release is developed. The paper contains some comparisons of the developed approach with other product release policies and with other sequencing heuristics. The paper ‘‘Optimal allocation of work in assembly lines for lots with homogenous learning’’ by Yuval Cohen, Gad Vitner and Subhash C. Sarin deals with the problem of line balancing for multi-model assembly lines. There is no buffer permitted in between the stations, and the line operates under the same learning rate for all stations. The authors show that in the presence of learning, the optimal decision requires imbalanced allocation of work to stations. The level of savings in the optimal makespan value due to the imbalanced loading of work over the balanced loading case are demonstrated as a function of the value of the learning constant, number of stations on the line as well as lot size. Finally, Alfred J.D. Lambert in the paper ‘‘Generation of assembly graphs by systematic
665
analysis of assembly structures’’ deals with a well-known problem in the domain of the line balancing. This problem is called assembly sequence planning. If a product can be assembled by several ways, then assembly sequences having the maximal number of parallel operations should be selected, because they allow to obtain more flexibility and minimize the assembly lead-time. In the paper a new approach for AND/OR graph generation based on subassembly detection is presented. I would like to thank the authors for their contributions and the referees for the time they put in reviewing all papers. Thanks to Professor Jacques Teghem, Editor-in-chief of the EJOR, for his support and the possibility he gives to publish this Issue. Professor Alexandre Dolgui Director of the Division for Industrial Engineering and Computer Sciences, Ecole Nationale Supe´rieure des Mines de Saint Etienne, France E-mail address: [email protected]. Available online 28 August 2004
Guest editorial
Balancing Assembly and Transfer Lines
The first idea of a Feature Issue on Line Design and Balancing appeared in April 2001 at the Conference MOSIMÕ01 ‘‘Design, Analysis and Management of Industrial Systems’’. This conference has been sponsorized by EURO. Later, in July 2002, an invited session on line balancing has been organized at the World Congress of the IFAC. These two events are at the origin of this Feature Issue. For this Issue, more than 30 papers were submitted from the most known research teams in the domain. After per review process, 15 best papers were selected. Professor Armin Scholl is invited for a state-of-the-art contribution. So, this Issue is a photography of the present state of advanced discrete optimization methods for decision aid in production lines design. The most interesting problems and the most promising approaches are explained and developed in this Issue. The proposed methods are complementary and they open several interesting research perspectives. Marc Peeters and Zeger Degraeve in ‘‘An LPbased lower bound for the simple assembly line balancing problem’’ propose a new lower bound for the classical simple assemble line balancing problem (SALBP); this bound is based on LPrelaxation and Danzig–Wolfe decomposition. Computational results and analysis of the quality of the new lower bound are presented. Alexandre Dolgui, Nikolai Guschinsky and Genrikh Levin in the paper entitled ‘‘A special case
of transfer lines balancing by graph approach’’ deal with a new line balancing problem so-called transfer line balancing problem (TLBP). Usually, the line balancing problems are considered in an assembly environment. In this paper, the authors showed a line balancing problem in machining/ processing environment. The specificity of this problem is the existence of blocks of parallel operations at the workstations (multi-tools spindle heads for machining). The block time is equal to the maximal time of its operations. The operational time of each workstation is equal to maximal time of its blocks. The authors propose an approach to solve a cost-oriented version of TLBP, by reduction of initial problem to a constrained shortest path one. Matthias Amen in the paper ‘‘Cost-oriented assembly line balancing: model formulations, solution difficulty, upper and lower bounds’’ deals with a cost-oriented assembly line balancing problem. The author proposes a mathematical model of the problem and new uppers and lower bounds for some problemÕs parameters. These results may be used to solve the cost-oriented SALBP by a branch-and-bound algorithm. The paper ‘‘An optimal piecewise-linear program of the U-line balancing problem with stochastic task times’’ by Timothy L. Urban and Wen-Chyuan Chiang presents an U-shaped assembly line balancing problem. The authors analyze a case with tasks having independent Gauss distributed durations and formulate a stochastic model
0377-2217/$ - see front matter Ó 2004 Elsevier B.V. All rights reserved. doi:10.1016/j.ejor.2004.07.021
664
Guest editorial / European Journal of Operational Research 168 (2006) 663–665
of the problem and propose a procedure to obtain a lower bound for the number of workstations. Yuri N. Sotskov, Alexandre Dolgui and MarieClaude Portmann in the paper ‘‘Stability analysis of an optimal balance for an assembly line with fixed cycle time’’ consider a SALBP with non-classical assumptions under consideration. Namely, the durations of the ‘‘manual’’ operations may have small variations. The problem is to calculate the stability radius of an optimal line balance. The authors show that this stability radius can be calculated by a polynomial-time algorithm. Nguyen Van Hop in ‘‘Heuristic solution for fuzzy mixed-model line balancing problem’’ addresses to mixed-model line balancing problem with fuzzy processing time. A mixed model of fuzzy ALBP is formulated and a heuristic algorithm to solve it is proposed. The main idea is to arrange the jobs in a sequence and to allocate them to workstations by an exchange procedure. Gregory Levitin, Jacob Rubinovitz and Boris Shnits in ‘‘A genetic algorithm for robotic assembly line balancing’’ consider robotic assembly lines where the durations of works depend on the characteristics of robot types and only one robot can be assigned to each workstation. The objective function is to minimize the cycle time of a line with a given number of workstations by assigning robots and tasks to workstations. To solve the problem the authors propose a genetic algorithm (GA). The results of the GA are improved by a local optimization (hill climbing) work-piece exchange procedure. The paper ‘‘Balancing assembly lines with Tabu search’’ by Sophie D. Lapierre, Angel Ruiz, and Patrick Soriano propose a new Tabu search algorithm to solve an assembly line balancing problem. The main idea is to incorporate an intensification/ diversification framework and to redefine the solution space in order to check infeasible solutions. The extended computational experiments are carried out and the advantages of proposed procedures are discussed. The paper ‘‘An endosymbiotic evolutionary algorithm for the integration of balancing and sequencing in mixed-model U-lines’’ by Yeo Keun Kim, Jae Yun Kim and Yeongho Kim deals with the problem of balancing and sequencing in an
U-shaped line. The authors propose a new evolutionary algorithm. This algorithm is based on using auxiliary information for the modeling of the evolution process. This approach allows decreasing the number of operations required to solve the initial problem and to improve the quality of the obtained solutions. Ronald G. Askin and Jiaqiong Chen in the paper ‘‘Dynamic task assignment for throughput maximization with worksharing’’ study a dynamic assembly line balancing problem. The case is considered where task sequences are known but the workforce is partially cross-trained and some tasks can alternate between workstations. The objective is to define a tradeoff between the cost of workin-process (WIP) inventory and the cost of crosstraining. A simulation model and various heuristic rules that attempt to approximate an optimal solution are explained. John J. Bartholdi III, Donald D. Eisenstein and Yun Fong Lim in the paper ‘‘Bucket brigades on in-tree assembly networks’’ consider a network of assembly lines. The goal is to balance each assembly line such that all lines produce with the same rate. The authors present an adaptation of ‘‘bucket brigade’’ protocol which allows obtaining a balance and synchronization. Stefan Bock, Otto Rosenberg and Thomas van Brackel in the paper ‘‘Controlling mixed-model assembly lines in real-time by using distributed systems’’ study an assembly line which produces several variants of products. Unforeseen disturbances of the production process can compromise its planned execution, therefore some on-line decisions is necessary. The authors formulate an integer programming model. A two phases approach is proposed. At first phase, an initial solution is generated by a well-known approach. At second phase, a set of heuristics is used to solution improvement. These heuristics can be executed in a distributed computational system. The authors present a comparison with a simulated annealing algorithm. In the paper ‘‘Job sequencing and material flow control of an assembly line’’ by Subhash C. Sarin, Michael P. Greco and Ezey M. Dar-El, the problem of loading of products (flow of material) on a assembly line is considered. The number of items is kept the same, and a new item is loaded only
Guest editorial / European Journal of Operational Research 168 (2006) 663–665
when an item leaves the line. Therefore, the problem is to determine an optimal cyclic sequence in order to achieve optimum throughput with minimum WIP. A lower bound of the WIP level is proposed. A new policy of product release is developed. The paper contains some comparisons of the developed approach with other product release policies and with other sequencing heuristics. The paper ‘‘Optimal allocation of work in assembly lines for lots with homogenous learning’’ by Yuval Cohen, Gad Vitner and Subhash C. Sarin deals with the problem of line balancing for multi-model assembly lines. There is no buffer permitted in between the stations, and the line operates under the same learning rate for all stations. The authors show that in the presence of learning, the optimal decision requires imbalanced allocation of work to stations. The level of savings in the optimal makespan value due to the imbalanced loading of work over the balanced loading case are demonstrated as a function of the value of the learning constant, number of stations on the line as well as lot size. Finally, Alfred J.D. Lambert in the paper ‘‘Generation of assembly graphs by systematic
665
analysis of assembly structures’’ deals with a well-known problem in the domain of the line balancing. This problem is called assembly sequence planning. If a product can be assembled by several ways, then assembly sequences having the maximal number of parallel operations should be selected, because they allow to obtain more flexibility and minimize the assembly lead-time. In the paper a new approach for AND/OR graph generation based on subassembly detection is presented. I would like to thank the authors for their contributions and the referees for the time they put in reviewing all papers. Thanks to Professor Jacques Teghem, Editor-in-chief of the EJOR, for his support and the possibility he gives to publish this Issue. Professor Alexandre Dolgui Director of the Division for Industrial Engineering and Computer Sciences, Ecole Nationale Supe´rieure des Mines de Saint Etienne, France E-mail address: [email protected]. Available online 28 August 2004