Analytical performance models for automatic assembly systems

Automatic assembly systems (AASs) are finding increasing use in today's industry. In the planning and design phase of these systems, it is useful to have analytical tools that predict system performance for various operating conditions. The larger the system the greater is the need for tools based on analytical methods as cost of simulation can be prohibitive.;For asynchronous AASs, a closed tandem queueing network model is developed. Due to the non-product form nature of the AAS model, an approximate technique is developed to analyze the closed tandem network of general single-server queues. The analysis technique is based on queueing approximations for the single-server queue that require only the mean and variance of service times. First, we consider a model that ignores transport delay, blocking due to finite buffers, and operator interference. For large and balanced AASs, the approximations yield simple formulas for steady-state performance measures including production rate and station utilization. For unbalanced AASs and smaller systems, efficient iterative algorithms are developed to compute the steady-state performance measures. Comparisons with simulation results indicate that the approximation technique is reasonably accurate for a broad range of parameter values and system size.;The model is then extended to include operator/station interference. Interference time is treated as a part of the workstation service time. The new model is obtained via a synthesis of the AAS model and a machine-repairman like queueing model. This new model accurately predicts the effect of interference on system productivity. It does not require any distributional assumption on station clear times.;Next, finite buffer AAS models are examined. Transport delay is handled using delay nodes. The transport delay approximation is evaluated using examples of balanced and unbalanced systems with large buffers. Blocked time of an assembly at a workstation is viewed as an extension of its processing time at that station. The solution of an AAS model with modified station service times gives an approximate solution to the model with blocking. Solution procedure and numerical examples are presented for balanced AASs.;Finally, extensions are proposed to model system reliability and scrapping of assemblies, to generalize the interference problem, and to calculate throughput variance.