IRSST - Institut de recherche Robert-Sauvé en santé et en sécurité du travail

Experimental analysis of tools used for estimating risk associated with industrial machines


In this study, 31 qualitative tools used for estimation risks associated with industrial machines and which follow the ISO 14121-1: 2007 guidelines we re analysed by (i) comparing their risk estimation parameters and (ii) applying the differ ent tools to estimate risks associated with 20 hazardous situations (scenarios). The objective of this study wa s to theoretically compare the performances of tools in estimating risks and to evaluate whether tools estimate risks uniformly. Ideally, the risk levels obtained by different user s when applying the different tools to the same risk scenarios should be more or less similar. The risk levels obtained by the same users when applying the different tool s to the same risk scenarios at diffe rent times should al so show certain similarities. As such, any important variability in risk estimations can be attributed to flaws or biases in the tools and these can be based on parameters as well as the architectures of the tools. In order to compare tools which have different parameters, as well as different number of levels or thresholds for those parameters, it was required to set up equivalence scales for the different risk estimation parameters. By using common benchmarks, it is showed that this comparison of different risk estimation tools is possible. Theref ore, in this report, the setting up of those equivalence scales is described a nd some analysis of the different parameters used in the tools is presented. Some guidelines on how to define parameters in risk estimation tools in order to make risk estimations easier as well as independent of tools and users (i.e. repeatability of risk estimation results), are also provided. Moreover, the differences obtained in the risk estimation results when applying different risk estimation t ools to the same hazardous situations involving dangerous machines are studied by investigating (i) the influence of the types of risk estimation parameters and methods of construction of the tools, (ii) the influence of the number of levels for each parameter and (iii) the influence of the number of risk levels on the results. As such, the 31 risk estimation tools are compared by applying them to 20 hazardous situations. The results show significant differences among the tools in estimating risks associated with the same hazardous situations, i.e. risk is tool dependent. The scope of the tool and its construction or architecture seem to be one of the contributi ng factors in this variability of the results. Tools that follow the 2 configurations proposed in ISO 14121-1:2007 produce similar average risk levels but both configurations have tools that wi ll underestimate or overestimate risk associated with hazardous situations. This leads to conclude that simple t ools, which have 2 parameters, can be as effective as more detailed tools, which have 4 parameters. It was also observed that the 31 tools could be grouped as 9 low risk estimating tools, 8 intermediate risk es timating tools and 14 high risk estimating tools . Moreover, there are tools which are not appropriate for machinery risk assessment even if their scopes often state the opposite. Finally, the observations in the behaviours of the different tool s have guided the authors in pr oposing a series of construction rules for the tools in order to alleviate most of the problems associated with the variability in the risk estimations. Those recommendations can poten tially guide users of risk estimation tools when choosing, designing or using a risk estimation tool. Future works include th e validation of the most promising tools with a large sample of different users from industries. It should be mentioned that this study was done in collabo ration with the Health and Safety Laboratory (HSL) in United Kingdom and the authors would lik e to acknowledge the i nvolvement of Nicola Stacey, Nicola Healey and Simon Rice.

Additional Information

Category: Research Report
  • Yuvin Chinniah
  • François Gauthier
  • Serge Lambert
  • Florence Moulet
Research Project: 0099-6610
Online since: July 14, 2011
Format: Text