Summary Roofing work has been ranked as one of the most hazardous jobs, with a fatality rate of 29 per 100,000 workers (Toscano, 1997). Roofers are exposed to approximately six times more risk of serious and fatal accidents than the average worker, with falls being the fatal events in 75% of cases (Ruser, 1995). The injuries caused by these falls are usually serious, requiring long periods of treatment and convalescence resulting in significant medical costs. The international regulations are clear; the Occupational Safety and Health Administration (OSHA) stipulates that workers exposed to a risk of falling 1.8 metres must be protected (OSHA, 1998) while the Safety Code for the construction industry (S-2.1, r.6) and the Regulation respecting occupational health and safety (S-2.1, r.19.01) stipulate that workers exposed to a risk of falling 3 metres must be protected (S 2.1, r.6, 2001; S-2.1, r.19.01, 2001). On a floor, a guardrail is the most appropriate collective means of fall protection (Lan, 2005). It allows mobility and exempts the worker from wearing a harness and avoids the installation of anchors for the workers' lanyards. The installation of anchors with a resistance of 18 kN is always problematic, particularly for light metal structures, temporary and/or obsolescent buildings.Currently, roofers use prefabricated guardrails made of metal structural members, installed on-site around flat roofs. According to the Association des Maîtres couvreurs du Québec (AMCQ, association of master roofers of Québec), these guardrails are practical and well suited to the work carried out on flat roofs, but have never been the subject of verification or tests to determine whether they are safe and comply with the Safety Code for the construction industry (S-2.1, r.6). At the request of the AMCQ, the current study aims to verify, by calculations and tests, whether the three models of metal guardrails most used by roofers are safe and meet the requirements of sections 3.8.2 and 3.8.3 of S 2.1, r.6 to ensure that roofers have appropriate protection against falls from heights.The main objectives of the study are therefore to:Verify whether the three models of prefabricated metal guardrails most used by roofers, installed on-site around flat roofs, are safe and meet the requirements of sections 3.8.2 and 3.8.3 of the Quebec Safety Code for the construction industry to ensure that roofers have appropriate protection against falls;Validate the anchoring parameters for flat roofs and/or to define these parameters;Verify the capacity of these three guardrails to retain a 100-kg wooden torso having a speed of 2 m/s;Study the main characteristics of the most realistic impact between the worker and the guardrail.The procedures followed included: Review of the literature on guardrails relating to normative and regulatory aspects (Québec, Canada and United States), consultation of existing design or user guides, mainly that of the Construction Safety Association of Ontario (CSAO) and test protocol for validating guardrails; Survey by the AMCQ of its members to determine the three types of guardrails most used by roofers on flat roofs; Meeting with the manufacturer(s) of these three guardrails to collect the geometric and mechanical characteristics of the guardrails, including: the dimensions of the post, the top plate, the characteristic span of a typical section of guardrail, the guardrail anchor(s), and the method of assembly of each guardrail;On-site observations of these three guardrails to identify the detailed installation conditions on the work sites during roof repair work in order to reproduce the same installation conditions in the laboratory; Reconstruction in the laboratory of a typical roof for a light industrial metal structure made of two 20' x 16' sections at the Hydro-Québec laboratory of the École Polytechnique according to the established rules and procedures for metal structures with a sandwich type parapet, which, according to the industry, is one of the most commonly constructed parapets and the least resistant; Strength tests on the three guardrails to verify experimentally the resistance of the components of the guardrail (tubular section of the post, weld, base plate, top plate); Tests on the guardrails installed as on construction sites on the reconstructed roof to check their compliance with the Quebec Safety Code for the construction industry according to the IRSST's test protocol;Dynamic tests to verify the capacity of the guardrails to stop the fall of a wooden torso at a speed of 2 m/s;Analysis of the results in order to make recommendations.The main results of this study showed that:Installed on the block of wood and on the new sandwich parapet, the three guardrails meet the requirements of sections 3.8.2 and 3.8.3 of the Quebec Safety Code for the construction industry, S-2.1, r.6; as a result, they comply with the Quebec Safety Code for the construction industry.The different guardrail assemblies on the block of wood and on the parapet in the directions parallel and perpendicular to the joists validated and defined the anchor parameters for guardrails on flat roofs.In the static tests on a single post anchored to the block of wood, we observed plasticization of the steel at the base of the post under a horizontal load less than the prescribed load of 900 N for the guardrail manufactured by Alcor and Innovation Malenfant Inc. In all the other static tests, the posts of all the guardrails remained elastic. Even if the load case on the single post is the most critical load case, it must be taken into account because it is a failure mode that can occur on a worksite.In the dynamic tests where the guardrails were anchored to the parapet, all the guardrails resisted the impact of the 100-kg wooden torso. The wooden torso remained within the perimeter of the roof, and the posts and top plates of the guardrails suffered only slight damage during the dynamic tests. The horizontal deformation on impact varied between 172 mm and 286 mm, depending on the manufacturer and test configuration. The residual deformation after impact varied between 1 mm and 76 mm, depending on the manufacturer and test configuration. The three guardrails were therefore able to retain a 100-kg wooden torso with a speed of 2 m/s without releasing the load.These results led to the following recommendations:The limited number of dynamic tests did not allow sufficient experimental data to be collected to 1) define the most severe realistic impact of a person on a guardrail by taking into consideration the deformability of the human body; 2) translate this impact into mechanical terms (use of biomechanics); and 3) translate this impact in terms of forces on the guardrails (the mechanical characteristics of the guardrail and the human body must be taken into consideration). As a result, we recommend a subsequent study to better define the dynamics of the fall arrest by a guardrail because the dynamic tests with the 100-kg wooden torso are very stringent tests and do not reflect the reality of the impact of a worker on the top plate.In the static tests on a single post anchored to the block of wood, we observed plasticization of the steel at the base of the post under a horizontal load less than the prescribed load of 900 N for guardrails manufactured by Alcor and Innovation Malenfant Inc. In all the other static tests, the posts of all the guardrails remained elastic. Even though the load case on the single post is the most critical load case, it must be taken into account because it is a failure mode that can occur on a worksite. As a result, we recommend that the manufacturers Alcor and Innovation Malenfant Inc. choose for the post a slightly more resistant member for new guardrails. Besides the increased resistance, the guardrails will have a longer service life.All of the results are valid for the test conditions described in the document.Installed on the block of wood and on the new parapet, the guardrails meet the requirements of S-2.1, r.6. Since the resistance of the parapet/guardrail assembly depends mainly on the resistance of the parapet, we strongly recommend that the parapet's condition be verified before the guardrails are installed. To do this, we can use a rule of thumb that consists of shaking the guardrail to evaluate its resistance.