Safety can be an expensive aspect of industrial operations unless efforts are made to enhance and optimize health and safety programs to reduce the long-term cost associated with health and safety related incidents and damage. The objective of a health and safety program is to minimize or prevent loss to humans, the environment, property and profits due to incidents (OSHA 2006). These programs are implemented by applying human resource time to preventive intervention activities that are expected to prevent or minimize loss (OSHA 2006). The National Safety Council estimates the cost of workplace injury in the year 2004 to be $142.2 billion (NSC 2005). This cost is expected to rise (NSC 2005) due to increases in medical and legal fees unless optimization efforts and enhancements of health and safety programs that reduce the likelihood of incidents taking place. One step towards achieving this objective would be to quantify and analyze intervention activity and incidents for an existing health and safety program.
Using Neural Networks, which is a form of artificial intelligence, the researchers attempt to determine and identify a relationship between safety intervention activity and the incident rate. Once the relationship has been established it will then allow the analyst to use it as a forecasting tool to predict future incident rates given the level of safety intervention activities. In this study incidents recorded were comprised of physical injuries to workers as well as spills and equipment failure.
This research is a continuation of the previous work by Haight et al. (2001) and Iyer et al. (2004 & 2005) which focused on quantifying safety intervention activities with the incident rate. It is based on the relationship between four safety intervention factors which are considered inputs and the incident rate is the only output. Figure 1 is a graphical representation of the model established by Haight et al. (2001) that lays the foundation for quantifying safety intervention activities with the incident rate. Table 1 provides an example of a data sheet used during the data collection phase from the forestry division of a power company. This forestry division of the power company provided the setting and the context for this research.
Figure 1: Representation of the Safety and Health Program-Mathematical Model (adapted from Haight et al. 2001) (available in full paper).
Table 1: An example of the data sheet used during data collection showing hourly data used in the study. Note, this is an example only (adapted from Haight et al. 2001, Iyer et al. 2004) (available in full paper).
In Table 1 the safety intervention variables are located on the left side represented by Factors A, B, C, and D. The farthest column on the right Totals represents the sum of input levels of the various safety intervention factors. The input is the man hours allotted to each of the twenty safety intervention activities during a one week period. For further explanation of the data gathering process please refer to the methodology section.