Shift work, fatigue, and human error: An empirical analysis of railway traffic control

ABSTRACT

Fatigue is a major contributor to transportation accidents. Shift workers are particularly prone to fatigue, and organizations increasingly rely on fatigue risk models to evaluate work schedules. In this article, the authors empirically examine the relationship between fatigue risk and human errors in railway traffic control. Despite their safety-critical role, research on railway traffic controller fatigue has remained limited. The authors evaluate the predictive validity of a commonly used fatigue risk tool (the Risk Index) and investigate the effect of additional risk factors (age, gender, part-time work, and day of week). In close cooperation with Belgian railways, the authors analyze a unique full-year data set, containing more than 11,000 work shifts. By adopting a Tobit regression for censored data, the authors account for work shifts with zero error occurrence. Results validate the applied fatigue risk model under real-world circumstances and reveal risk predictors above and beyond shift schedule design: significant day-of-week effects are observed. The probability of making at least one error is highest on Saturdays (+6% compared to Mondays), and lowest on Tuesdays, Wednesdays, and Thursdays. Thus, results suggest that safe work schedule design should also take into account the day of the week and not exclusively rely on fatigue risk scores.

KEYWORDS:

• railway
• traffic control
• human factor
• fatigue
• Tobit
• day-of-week

Acknowledgments

Special thanks go to the management and experts of the Traffic Operations Department, to the Methods department, as well as to the Performance Data Division and the Business Analytics Team.

Funding

The authors would like to thank Infrabel for funding this research and providing the necessary railway data and expertise. It is to be noted that the views expressed in this article are those of the authors and do not necessarily reflect the opinions of Infrabel.

Notes

1. Infrabel is the government-owned corporation that runs the Belgian railway infrastructure and is one of the key players in the Belgian mobility sector. Its core activities are asset management (building, maintaining and renewing the infrastructure) and traffic control.

2. The server-based Business Intelligence software is QlikView (Qlik Software Company) and is linked with the rostering and traffic control databases through a direct connection.

3. The European Rail Research Advisory Council (ERRAC) is a joint initiative of the European Commission and European railway industry stakeholders.

4. The Bennett and Passmore (1984) coal miner injury analysis being an early and notable exception.

5. Briefly explained, $\text{E}(y|x)$ can be written as the product $\text{P}(y>0|x)\text{E}\left(y\text{|}y>0,x\right);$ i.e. the probability of being above zero, multiplied by the expected value of the subpopulation above zero. Also, the derivative of a product of two functions $\text{f}\left(x\right)\text{g}\left(x\right)$ is given by $\left[\text{f}\left(x\right)\text{g}\left(x\right)\right]\text{'}=\text{f}\left(x\right)\text{′g}\left(x\right)+\text{ f}\left(x\right)\text{g}\left(x\right)\text{'}.$