Degree Type


Date of Award


Degree Name

Doctor of Philosophy


Industrial and Manufacturing Systems Engineering

First Advisor

Gary A. Mirka


Work-rest scheduling is one tool available to ergonomists to reduce the risk of injury to workers performing physically demanding work tasks. Accurate assessment of the changes in these physical stresses during the work task and during the recovery period is necessary to establish sound work rest schedules. Existing work-rest scheduling models are either based entirely on empirical data without the utilization of a theoretical model of the underlying human physiology or they have relied on abstract optimization modeling techniques that have but been shown to lack robustness to realistic work scenarios. In the current study a new work-rest scheduling model has been developed. This new model makes use of an established modeling technique in the inventory management literature and adapts this technique to the modeling of the development of work stress during work and recovery from work stress during resting periods. This study required the development of time-dependent work stress profiles through the collection of data from human subjects performing simulated work activities. These work stress profile data then formed the database that was utilized by the theoretical model. This new work-rest model is shown to be able to generate optimal work-rest schedules under a variety of work conditions.

Two experiments with four participants each were conducted. Experiment I was dedicated to generate a series of regression equations for tracking the changes in physiological variables during the performance and recovery of two typical low back fatiguing exertions. These two tasks included an isometric trunk flexion task and a repetitive lifting task. As participants performed these tasks, heart rate (to establish cardiovascular stress), median frequency of the erector spinae muscles (to establish muscle fatigue stress), and sway speed (to establish changes in level of whole body stability) were collected and tracked overt time. From these empirical data, regression equations were developed that predicted these three variables as a function of time. Experiment II was conducted to test the validity of these equations by using different participants and performing three different categories of task protocols and then assessing the quality of the predictions of the regression equations developed in Experiment I. Results from Experiment II demonstrated good performances in predicting the changes in heart rate, median frequency and sway speed (with mean absolute percentage error of 5.5%, 11.7% and 15.0% respectively). These results indicate that the more objective measures (heart rate and median frequency) were relatively more stable and were subsequently included in the inventory control theory modeling phase of this study.

The next phase of the study was to use these data in an established inventory management optimization modeling technique. The adaptation of this modeling technique for this application required a translation of inventory variables into physiology variables: production rate, demand rate, set-up cost, holding cost were translated into rate of fatigue development, rate of fatigue recovery, cost of pausing from work, and cost due to increased risk of injury caused by keeping certain level of physiological stresses, respectively. The optimal work-rest schedule was generated by minimizing the total cost, as in the traditional use of the model. These results generated the appropriate duration of a physically demanding work task as well as the appropriate duration of the subsequent rest breaks. Finally, a sensitivity analysis on the values of the inputs to the model was conducted. Results of sensitivity analysis showed that all three cost factors (wage cost, setup cost and holding cost) have clear and reasonable influences on the selection of optimal work-rest schedule. As the wage costs and setup costs increased the number and duration of scheduled breaks decreased. As the holding cost (cost of injury) increased, the number of breaks increased and the duration of the work time increased. This analysis also revealed that the suggested work duration was fairly robust relative to the hourly wage of the worker, but was sensitive to the setup costs and the predicted cost of injury. The suggested recovery duration was most influenced by the hourly wage of the worker.

The resulting work-rest scheduling model can be adapted to meet the needs of a particular work scenario. The requirements are to develop the development and recovery profiles for the given work scenario. Future work may entail the development of a system to predict the form of these equations without the need to do a full data collection using human subjects. For a given work scenario, the work-rest scheduling tool developed through this research can provide the minimum cost schedule.

Copyright Owner

Xiaopeng Ning



Date Available


File Format


File Size

162 pages