Component allocation to balance workload in printed circuit card assembly systems

IIE Transactions, April, 1997 by J.C. Ammons, M. Carlyle, L. Cranmer, G. DePuy, K. Ellis, L.F. McGinnis, C.A. Tovey, H. Xu

1. Introduction

The predominant delivery mechanism for electronic function across the entire spectrum of current products is the printed circuit card assembly (PCCA). Thus the production of PCCAs is a critical process required for many different types of manufactured product. Although large numbers of electronic assembly systems with common technologies and processes can be found worldwide, a universal approach for production planning has not emerged owing to the dynamic technological and economic environment associated with this field.

Driven by the goal of smaller, denser cards with greater functionality and reliability, the technologies of electronic components and printed circuit cards have changed a great deal over the past two decades. Correspondingly, assembly technologies have evolved from essentially manual operations to completely automated ones. Manual placement cannot be accomplished reliably for the newest device technologies, forcing the design and operation of automated assembly systems. Today's automated assembly equipment is quite expensive; it requires a large initial capital investment and costly operational support. This expense, combined with pressures from product proliferation, shorter life cycles, quality concerns, delivery responsiveness, and other production costs, forces the PCCA manufacturer to strive for assembly processes that maximize output while minimizing both the work-in-process inventory and the process overhead associated with process engineers and planners.

As product designs become more complex, and assembly technologies more automated, the amount of information that must be processed to optimize PCCA production is becoming prohibitive for effective manual planning. The ideal process planning approach is therefore an effective 'automated' one that provides the production planner with plans that 'optimize' the production floor relative to system objectives, yet require only a small amount of time on the part of the planner.

Associated with the interactions between process planning, production planning, and scheduling, there are a number of related problems that must be solved. At the highest level is the problem of grouping, associated with the selection of machine groups and card families and the assignment of families to groups. Given the grouping decisions, allocation is performed to assign component types to machines when a machine group contains multiple machines. At the lowest level, the arrangement and sequencing problem is solved to stage component feeders on each machine and sequence the placement operations for each machine and PCCA. Staging capacity is often limited by the available number of feeder slots, or 'channels', used to attach feeders to machines. Fig. 1 illustrates the hierarchical relationship between these decisions. The interrelationship of component allocation with grouping, feeder arrangement, and placement sequencing makes proper process planning and component allocation for electronic assembly systems an extremely complicated task. For more background on the interrelations of these decision problems, along with a more complete overview of printed circuit card assembly, the reader is referred to McGinnis et al. (1992) and the references cited therein.

In this paper an approach is described for allocating electronic component types between two or more placement or insertion machines in an effort to balance each card type's combined placement and setup times across the machines. The machines are connected by a conveyor, precluding the accumulation of work-in-process. The machines may be different, and there may be some component types that can be placed by more than one of the machines. In the remainder of this section, an overview of the component allocation problem is given, and the relevant literature is examined. In Section 2 a corresponding integer programming (IP) formulation of the problem is developed. Two alternative solution approaches are presented: a list-processing-based heuristic for a simple version of the problem, and a linear-programming-based branch-and-bound procedure. Each approach is validated with an industrial case study with results that improve production throughput by up to 10%. The paper concludes with insights and extensions.

1.1. Problem overview

The assembly of PCCAs consists of attaching electronic components to printed circuit cards. There are many different types of technology for both components and printed circuit cards. Similarly, there is a tremendous diversity in the machines used for placing/inserting components on cards. An exhaustive characterization of all technologies and equipment is not required to specify the component allocation problem, although pertinent features are noted as follows (see McGinnis et al. (1992) for greater detail).

In general, assembly operations are performed on a panel, which may consist of a single card or several individual cards. A panel containing multiple cards can present interesting problems for the arrangement and sequencing decisions. With respect to component allocation, however, the terms card and panel can be used more or less interchangeably.