Coping With Aging in Manual Assembly Systems

 

Age-Differentiated Analyses and Mathematical Modeling for Predicting the Time Structure of Sensorimotor-Skill Acquisition for Assembly in Series Production With Numerous Product Variants

 

Challenges

Against the backdrop of the globalization of sales and procurement and the individualization of customer requirements, companies are producing their products in an increasing number of variants, while the duration of product life cycles is decreasing. As a result, employees in manufacturing and assembly areas are more frequently entrusted with new or modified work tasks. These tasks are mostly of a sensorimotor nature and have to be learned based on the individual task. This involves both an introduction to the task and practicing the task until a previously defined reference performance is reached, which can be determined, for example, by means of Methods-Time Measurement (MTM) or by time recordings according to REFA. The period of time where practice increases performance is referred to as the learning time and can be described mathematically with the aid of learning curves. Since the learning time represents an important planning variable for manufacturing companies and could previously only be predicted with great uncertainties, a model for predicting the learning time (dissertation Jeske 2013) was developed at IAW – Institut für Arbeitswissenschaft (Institute of Industrial Engineering and Ergonomics) – as part of the research project "FlexPro – Innovative Management of Flexible Production Capacity" which was funded by the BMBF – Bundesministerium für Bildung und Forschung (Federal Ministry of Education and Research) – and the ESF – Europäischer Sozialfonds für Deutschland (European Social Fund). This model is based on a statistical approach and, in addition to taking into account the various influencing factors of the employee, also takes into account the influence of various forms of representation of the work plan used and the difficulty of the work task.

Goal and Procedures

The aim of the research project was to develop a model for predicting the learning time for typical industrial series production assembly tasks. Compared to the existing research results at IAW, more in-depth and advanced statistical methods were used and additional influencing factors were taken into account to increase the prediction quality. Against the background of demographic change, the focus was placed on the influence of age on learning. In this context, particular attention was paid to how the design of work schedules, the use of methods for instructing workers, and breaks of different lengths between work tasks influence the learning of workers of different ages.

Methodology

Tests were carried out at a standardized assembly workplace in the laboratory with test subjects between 20 and 35 years of age and between 52 and 67 years of age. In all test series, the subjects were required to repeatedly assemble a technical component. Associated performance changes were quantified using the execution times and assembly errors of each repetition.
​In addition, the subjects' perceived workload was assessed using the Rating Scale Mental Effort (RSME) and the NASA Task Load Index, and a subsample was assessed using electromyography (EMG) measurements. To analyze changes in movements due to repeated task performance, movement trajectories were also recorded using motion tracking.

Results

In the test series, it was found that the execution times and error rates of both age groups were influenced differently by the presentation format of the work plan and the method used to introduce the work task. A significant influence of the duration of the break on the process parameters could not be demonstrated. The collected data was used as a basis for developing the prediction models. To begin with, existing learning curves from the literature were fitted to the data as best as possible. The learning curves according to de Jong (1960) and Jeske (2013) proved to be the most suitable for describing the collected data, depending on the experimental conditions. By introducing an additional parameter, both learning curves could be combined during the model development. To predict the learning curve parameters, regression equations were calculated for three cases: (1) age-mixed prediction, (2) young worker prediction, and (3) older worker prediction. Predictors are the characteristics of the work person, the work task, and the training methods. The results of the project can help assembly planners to estimate the profitability of a new learning method before its introduction, taking into account the composition of their workforce.