Designing Cellular Manufacturing System in Presence of Alternative Process Plans

Technical Summary: Designing Cellular Manufacturing Systems in the Presence of Alternative Process Plans

Problem Definition
The paper addresses the design and operational management of Cellular Manufacturing Systems (CMS), specifically focusing on the "generalized grouping problem." In traditional CMS design, it is often assumed that each part has a single process plan and that each operation can only be performed on a specific machine. However, in practical manufacturing environments, parts frequently possess multiple alternative process plans, and individual operations within those plans can often be executed on various machine types. The core challenge is to simultaneously determine the optimal grouping of machines into cells, the selection of process plans for each part, and the assignment of operations to specific machines, while considering constraints such as machine capacity, cell size limits, and budgetary restrictions. The study distinguishes between two scenarios: the design level (setting up a new system) and the operational level (reorganizing an existing system).

Methodology
The authors propose four distinct Integer Programming (IP) formulations to address these scenarios, moving beyond the sequential or hierarchical approaches often found in prior literature. The models integrate machine flexibility and alternative process plans directly into a unified optimization framework.

1. Formulation I (Operational Level): Designed for reorganizing existing systems. The objective is to minimize total inter-cell and intra-cell material handling movements. This is achieved by maximizing the assignment of consecutive operations to the same cell and, where possible, the same machine. The objective function is initially quadratic (counting co-assignments) and is linearized using the Glover and Woolsey (1974) method to facilitate solution by standard solvers.
2. Formulation II (Design Level): Based on a modification of Rajamani et al. (1990), this model aims to minimize total machine investment costs. It enforces a constraint of zero inter-cell movement (strictly disjoint cells), requiring the acquisition of sufficient machine copies to keep all operations for a part within a single cell.
3. Formulation III (Design Level): A modification of Logendran et al. (1994), this model minimizes the sum of amortized machine costs and annual operating costs. Like Formulation II, it enforces disjoint cells but shifts the objective from pure investment to total lifecycle cost. The authors note that their simultaneous approach avoids the potential sub-optimality of the hierarchical methods used in the original Logendran model.
4. Formulation IV (Design Level): This formulation seeks to minimize inter-cell and intra-cell movements while adhering to a fixed budgetary limit on total machine investment. It addresses the trade-off between the high investment costs of disjoint cells (Formulation II) and the high movement costs of minimal machine duplication.

Key Results
The study validates the formulations using a numerical example involving four part types, multiple process plans, and three machine types.

• Formulation I successfully reorganized the system to minimize movements, resulting in two inter-cell and two intra-cell movements given the constraint of available machine copies.
• Formulation II achieved a zero inter-cell movement solution with a total investment cost of 600, requiring specific machine duplication to maintain disjoint cells.
• Formulation III yielded a similar configuration to Formulation II regarding machine assignment but calculated a total objective value of 930 (600 amortized cost + 330 operating cost). The authors highlight that the hierarchical approach used in the original Logendran model (1994) resulted in a solution that could not form disjoint cells under the same constraints, leading to higher movement costs, thereby demonstrating the superiority of the simultaneous optimization approach.
• Formulation IV demonstrated significant flexibility. When the investment budget was set at 600, it matched Formulation II's results. However, when the budget was tightened to 550, the model adjusted the machine-cell configuration (increasing machine copies in one cell and reducing in another) to minimize movements within the new financial constraint, resulting in a different, feasible solution that Formulation II could not provide under the same budget.

Significance and Claims
The paper claims that its primary contribution lies in developing a unified framework that simultaneously optimizes cell formation and operational routing decisions, rather than treating them as sequential stages. By explicitly modeling alternative process plans and machine flexibility, the proposed formulations ensure a more globally optimal configuration for generalized grouping problems.

The authors emphasize that minimizing material handling costs (both inter-cell and intra-cell) is a more direct and effective objective function than maximizing similarity coefficients, which may fail to produce disjoint groups. Furthermore, the study provides a robust decision-support tool for practitioners by offering different formulations tailored to specific managerial priorities:

• Formulation I for operational reorganization without new capital.
• Formulation II for new system design with a focus on minimizing initial investment.
• Formulation III for long-term strategic planning considering lifecycle costs.
• Formulation IV for balancing movement efficiency against strict budgetary limits.

The authors modestly acknowledge that while the models are effective for the examples provided, they are NP-hard and computationally expensive as problem sizes increase. Consequently, they suggest that future research should focus on applying modern heuristic methods (such as tabu search, simulated annealing, and genetic algorithms) to solve larger-scale instances of these problems.