This pilot study aimed to compare the predicted recovery time of workers of a self-paced, standardized, asynchronous indexing assembly line with the observed recovery time. To achieve this goal, a field observation and event-based motion study of 7 workstations in an automotive assembly line was conducted. Results show that, on average, there was no significant difference between the predicted recovery time and the observed recovery time (p = 0.839). However, results for most workstations (6 out of 7) analyzed individually demonstrated a significant difference (p < 0.05).
Aims: This pilot study aimed to compare workers’ recovery time as predicted by a predetermined time standard with the observed recovery time on an asynchronous indexing assembly line. A motion study of 7 assembly line workers was conducted toward this end. Motivation: In a standardized assembly line, each workstation has a specific list of instructions a worker must perform during each work cycle. There are 2 main purposes for using standardization: (1) to ensure product quality and (2) to ensure the work is completed within the allotted time. Along with process standardization comes time standardization. It is used to estimate the time required for completing assigned tasks before they are performed. This time estimate is then used to set the assembly line standard time which also includes a time allowance meant to accommodate some work time variances and provide time for worker recovery to prevent fatigue. Studies suggesting how to manage cycle-to-cycle variations have been conducted for a long time with some studies suggesting self-pacing as a way to improve productivity and quality and reduce waste (idle time) (Conrad, 1954, 1955a, 1955b; Davis, 1965; Lin, Drury, and Kim, 2001; Bosh, Mathiassen, Visser, Looze, and Dieën, 2011) while others suggest caution when using self-pacing because workers may not take advantage of being able to control their pace and recover but instead work continuously, resulting in a shorter cycle time with a high % duty cycle (Dempsey, Mathiassen, Jackson, and O’Brian, 2010). A study by Carrasquillo, Armstrong, and Hu (2017) evaluated different assembly line pacing methods and found similar results. They proposed an assembly line where workers may have the work object in their station for a minimum time but can keep it in the station longer if needed without stopping the line may provide the benefits of self-pacing while reducing its drawbacks. An asynchronous indexing line with buffer space between each of the stations works similarly to the proposed line. In this study, an asynchronous indexing assembly line (self-paced) with buffer spaces between workstations is evaluated to determine if the observed recovery time differs from the predicted recovery time. Methods: An observational pilot study was conducted in 7 workstations of an automotive self-paced, indexing, assembly line with buffer stations to assess the difference between predicted and observed recovery time. Workers were recorded while they performed their usually assigned tasks for 25 – 35 minutes in which they completed an average of 47.3 work cycles. Event-based motion analysis of each workstation using frame-by-frame analysis of the video recordings was used to calculate the observed cycle times, the work times, and the recovery times. Results: Overall, the average cycle time was 31.1 sec (s.d. = 6.7 sec), and the average work time was 22.0 sec (s.d. 5.6 sec), which resulted in an overall recovery time of 9.1 sec (s.d. 6.3 sec). With an overall predicted recovery time of 9.6 sec (s.d. 4.3 sec), the difference between predicted and observed was not statistically significant (p = 0.839). Individually, however, 6 out of the 7 workstations had a significantly different observed versus predicted recovery time (p < 0.05). Four stations had more recovery time than predicted (1.5 sec – 3.5 sec) and 2 had less (3.9 sec, 10.4 sec). Discussion and conclusion: The results of this study show that overall, the difference between the observed recovery time and the predicted recovery time was not significant (p > 0.05) because some workstations had more recovery time and others had less recovery time than predicted, leading to a total average difference of -0.5 sec, which was not large enough to be significant. However, it is important to evaluate workstations individually because some stations may have much less recovery time than predicted. For example, station 3 had 10.4 seconds less.
