Abstract—This paper reports on the performance of deliberately unbalanced, reliable, non-automated and assembly lines that merge, whose workstations differ in terms of their mean operation times. Simulations are carried out on 5- and 8-station lines with 1, 2 and 4 buffer capacity units, % degrees of line imbalance of 2, 5 and 12, and 24 different patterns of means imbalance. Data on two performance measures, namely throughput and average buffer level were gathered, statistically analysed and compared to a merging balanced line counterpart. It was found that the best configurations are a balanced line arrangement and a monotone decreasing order for each of the parallel merging lines, with the first generally resulting in a lower throughput and the second leading to a lower average buffer level than those of a balanced line. Keywords—Average buffer level, merging lines, simulation, throughput; unbalanced. I. INTRODUCTION NPACED asynchronous parallel merge lines are high volume stochastic serial queuing systems. With no form of mechanical pacing, workers along the line are free to work at their own pace. Provisions are usually made for keeping partly finished work-in-process inventories between stations so that when work is completed at one station, the item is transferred to a storage location, called a “buffer”. Fig. 1 shows a typical merging assembly line comprised of a series of parallel workstations and buffers, and a final merge or assembly station. In the design of an unpaced merging serial assembly line, one of the main issues to be considered if efficiency is to be enhanced is where to place operators who work at different speeds. The focus of research in this area has for the most part concentrated on how best to achieve a “balanced” line, where the operator average service time at each workstation is the same, as this type of design has been perceived as leading to the best efficiency. In real life, however, processing times have been shown by [1] to be non-identical at different workstations, even in automated lines. In unpaced lines, this effect is exacerbated since operators at each station can vary according to their physical capacity, their motivation, or due to task complexity, or just simply that the amount of work along the line cannot be distributed evenly. These differences in mean operation times (MTs) were shown by [2] to lead to blocking and starving along the line, with a resulting impact S. Shaaban is with ESC La Rochelle, La Rochelle, 17024, France (phone +33 (0) 546 51 77 00; e-mail: shaabans@esc-larochelle.fr). T. McNamara and S. Hudson are with the ESC Rennes School of Business, Rennes, 35065, France (e-mail: tom.mcnamara@esc-rennes.com, sarah.hudson@esc-rennes.com). on average buffer level (ABL) and throughput (TR). In view of this, the allocation of the operators along the line becomes an important consideration, and the issue of how to improve performance in an unbalanced parallel merging line is one that needs to be investigated. The structure of this article is as follows. First, the relevant literature is reviewed in Section II. Next, the motivation and objectives of the study are presented in Section III. The methodology and experimental design are discussed in Section IV, with the simulation results and analysis given in Section V. The results are then summarised and discussed in Sections VI and VII. II. LITERATURE REVIEW One of the early findings in the study of mean time imbalance suggested that placing workstations with higher average processing times at both ends of the line could lead to improved performance in terms of throughput (TR), compared to the balanced line. This effect was termed the “bowl phenomenon” by [3]. There has been continued interest in testing this phenomenon over the years, which show varying degrees of support for this conclusion, with most recently a move towards looking at patterns of imbalance rather than focusing purely on the bowl phenomenon, as argued by [2], [4]-[9]. Of more immediate relevance to the investigation performed in this paper is work that concerns production lines (or assembly lines) that merge. Similarly, to the literature on serial production lines, it is assumed that balancing merging lines yields superior performance, as put forth by [10]. However, there is some, albeit sparse, evidence that unbalanced configurations can also perform well, or sometimes better than the balanced equivalent. Below we provide a brief overview of some of the main findings. Reference [11] did an early simulation comparison study between a system comprised of two serial lines operating in parallel versus a production line composed of an equal number of stages in which each stage had two stations that merged. Both systems were subject to failure and had normally distributed processing times. In general, it was found that a configuration of a series of merging stations provided better performance than that of parallel serial lines. Reference [12] developed a simulation model for the evaluation of the performance of production systems, with the stated goal being to bring them into balance. Their model was capable of analyzing merging lines subject to stochastic processing times. The Operating Behaviour of Unbalanced Unpaced Merging Assembly Lines S. Shaaban, T. McNamara, S. Hudson U World Academy of Science, Engineering and Technology International Journal of Economics and Management Engineering Vol:10, No:6, 2016 1893 International Scholarly and Scientific Research & Innovation 10(6) 2016 scholar.waset.org/1307-6892/10004616 International Science Index, Economics and Management Engineering Vol:10, No:6, 2016 waset.org/Publication/10004616