Dual head placement machines are commonly used in industry for placing components on circuit cards with great speed and accuracy. This thesis evaluates a novel approach for prescribing process plans for circuit card assembly on dual head placement machines. Process planning involves assigning component types to heads and to feeder slots associated with each head and prescribing appropriate sequences of picking, placing and nozzle-changing steps. The approach decomposes these decisions into four inter-related problems: P1, P2, P3 and P4. This thesis reviews this approach; presents a new heuristic to address P1; a method to facilitate P2 and P3 solutions; a method to control nozzle changes in P4; tests approaches to P1, P2, P3 and P4; and presents a thorough analysis of computational results to evaluate the efficacy of the approach which aims to balance workloads on machine heads to maximize assembly line throughput.

The focus of this research is the line assignment problem in printed circuit card assembly systems. The line assignment problem involves the allocation of circuit card types to an appropriate assembly line among a set of assembly lines with the objective of reducing the total assembly time. These circuit cards are to be assembled in a manufacturing facility, capable of simultaneously producing a wide variety of printed circuit cards in different production volumes. A set of component types is required for each printed circuit card. The objective is to assign the circuit cards to the assembly line such that the total assembly time, which includes the setup time as well as the processing time required for all card types in a set, is minimized. The focus of this research is to develop an algorithmic strategy for addressing this problem in electronics assembly. This problem involves considering several interrelated decision problems such as assigning printed circuit cards to assembly lines, grouping circuit cards into families to reduce the number of setups, and assigning component types to machines to balance workload. The line assignment models are formulated as large scale mixed integer programming problems and are solved using a branch-and-bound algorithm, supplemented by techniques for improving the solution time. The models and solution approaches are demonstrated using industry representative data sets and can serve as useful decision support tools for process planning engineers. / Master of Science