By Karl Fridolf, WSP Sverige AB
Daniel Nilsson, Håkan Frantzich, Enrico Ronchi and Silvia Arias, Lund University
In 2015, the Swedish Transport Administration, which is responsible for the design, construction and maintenance of most Swedish road and rail tunnels, initiated a research project regarding movement of people during fire evacuation in smoke-filled environments. The project resulted in a recommendation for how to represent peoples’ walking speed in smoke in quantitative life safety verifications. It is based on currently available literature and empirical data (“state of the art”) on human behaviour and movement in smoke-filled environments. As of SFPE’s 12th International Conference on Performance-Based Codes and Fire Safety Design Methods in Honolulu earlier this year, information about this recommendation, and how to apply it from a practical perspective, is now also available in English.
Regardless of the technique adopted to quantitatively assess peoples’ ability to evacuate a building safely during a fire, information about the movement of people in general, and in smoke in particular, is a pre-requisite to using any such technique. Because one especially important variable is information about peoples’ walking speed, research about peoples’ walking speed in different settings has been ongoing for a long time. However, the majority of these empirical studies has been carried out during “normal” conditions, i.e., in everyday situations, and almost exclusively in smoke-free environments. This is despite the fact that other early related research in the field of human behaviour in fire demonstrated that people tend to evacuate through smoke, and that the behaviour of people in smoke-filled environments tends to deviate from that with no smoke.
The consequence is a lack of reliable and valid correlations for predicting peoples’ walking speed in smoke, which may propagate through Required Safe Egress Time (RSET) analyses and affect the design and verification of the fire and life safety performance of buildings in general and underground infrastructures in particular. The Swedish Transport Administration, which is responsible for the design, construction and maintenance of most Swedish road and rail tunnels, initiated a research project on this topic in 2015. The purpose was to make an inventory of, investigate and describe the current knowledge base about movement of people during fire evacuation in smoke-filled environments. Another goal was to summarize the current knowledge base, and to describe and recommend how it can be used in practical application. The latter rendered a general recommendation that can be used by fire safety/protection engineers, and is the focus of this article.
The resulting recommendation depends on the treatment of uncertainties in the analysis. Three suggested methods describe how peoples’ walking speed can be represented in life safety verifications, in both smoke-free and smoke-filled environments. The differences between the methods is linked to the quality of the verification. Method 1 is easier to apply than method 3, but is likely to lead to a more-conservative result in the form of a longer RSET. Method 3 is likely to lead to shorter RSETs and a more-detailed representation of the evacuation phase, but does, on the other hand, require more effort and longer assessment times.
All three methods are based on the relationship between walking speed and visibility, as opposed to extinction coefficient (typically used previously). The reason is that it is believed that people adjust their speeds during evacuation in smoke based on what they can see in the smoke-filled environment, independent of if what they see is emitted or reflected light. The latter is, however, of importance to perceived visibility in smoke. As an example, a light-emitting lamp can be seen from a greater distance in smoke than a reflecting emergency evacuation sign.
Independent of method applied, two threshold values hold a central function:
- The visibility level at which people in general can be expected to start reducing their walking speed. Based on the review of the literature, a thorough analysis and assessment of the collected data and simple movement experiments conducted at Lund University, a visibility level corresponding to 3 meters represents this threshold value.
- The visibility level at which people in general can be assumed to be walking with their slowest speed, as well as that particular speed. Based on findings during the review of the literature, people in general walk at their slowest speed in smoke when the situation is similar to movement in complete darkness. In such a situation, people in general can be expected to walk at about 0,2 m/s. Thus, peoples’ walking speed is never represented by a value lower than 0,2 m/s in the recommendation below.
Method 1: Representation is identical for all individualsAll individuals in the life safety verification are assumed to be walking at the same speed. Practically speaking, this is treated as follows:
- Visibility levels > 3 meters — Peoples’ walking speed is represented by 1 m/s, a level at which only ~10 % can be expected to walk more slowly.
- Visibility levels ≤ 3 meters — Peoples’ walking speed is represented by a relative reduction of 0,34 m/s per meter visibility, down to the minimum speed of 0,2 m/s.
Method 2: Representation is almost identical for all individuals
The individuals are divided into three categories based on their walking speed in smoke-free conditions: average, slow and very slow. Individuals in one group are assumed to be walking at the same speed. For this method, the designer has to decide on the proportions of average, slow and very slow walkers in their verification. Practically speaking, method 2 means that:
- Visibility levels > 3 meters —
- Category average — Peoples’ walking speed is represented by 1,35 m/s (~50% can be expected to move faster than this).
- Category slow — Peoples’ walking speed is represented by 1,10 m/s (~85% can be expected to move faster than this).
- Category very slow — Peoples’ walking speed is represented by 0,85 m/s (~97, % can be expected to move faster than this).
- Visibility levels ≤ 3 meters — Peoples’ walking speed is represented by a relative reduction of 0,34 m/s per meter visibility down to the minimum speed of 0,2 m/s.
The correlation can then be described by the following equation (where w is the walking speed [m/s] and v the visibility [m]), and by Figure 2:
·
Method 3: Representation is done individually
Each individual’s walking speed in smoke-free conditions is randomised and then assumed to reduce linearly as in method 1 and 2 in smoke. Practically, method 3 means that:
- Visibility levels > 3 meters — Peoples’ walking speed is represented by a randomised value from a normal distribution with mean 1,35 m/s and standard deviation 0,25 m/s.
- Visibility levels ≤ 3 meters — Peoples’ walking speed is represented by a relative reduction of 0,34 m/s per meter visibility, down to the minimum speed of 0,2 m/s.
The correlation can be described by the following equation (where w is the walking speed [m/s], wsmoke-free is the walking speed [m/s] in smoke-free conditions and v the visibility [m]), and by Figure 3:
The recommendation presented in this article is based on the currently available literature and data on human behaviour and movement in smoke-filled environments, and builds on the information, data and conclusions that have been presented in a range of studies. The reader must be aware of the fact that representing walking speed in smoke is a complex task with many uncertainties, and the recommendation suggested in this extended abstract does not take all aspects of movement-related parameters and variables into account.
In addition, the correlation is limited to the independent variable visibility and dependent variable walking speed. It is likely that other independent variables will have an effect on walking speed in smoke-free environments, smoke-filled environments or in both. This has not been possible to cover in the proposed recommendation.
Another important aspect to keep in mind is that the recommendation is based on a selected number of empirical studies; more specifically, studies that for reliability and validity reasons:
- share a similar research method,
- share a similar data collection technique and
- has been performed in a more or less similar test environment.
All of these have been carried out in simple corridors or tunnels. In other words, these represent environments in which exit choice and other options are quite limited. To use the recommendation for representing peoples’ walking speed in smoke in buildings where many exit choices are available is therefore not suggested. In these environments, the walking speed in smoke can be expected to be lower than what the recommendation suggests.
For a full account of the technical details on which these conclusions are based, see the full technical report (in Swedish). A summary of the report conclusions, with an emphasis on the recommendation for practical application, is also presented in English in an extended abstract for SFPE’s 12th International Conference on Performance-based Codes and Fire Safety Design Methods in Honolulu earlier this year.
Acknowledgements
This article was prepared by Dr. Karl Fridolf at WSP Sverige AB, who led the research project referred to in the text (when the project was initiated, he was employed at RISE, formerly the SP Technical Research Institute of Sweden). The project group included Dr. Daniel Nilsson, Dr. Håkan Frantzich, Dr. Enrico Ronchi and Silvia Arias at the Division of Fire Safety Engineering at Lund University. All authors would like to extend their thanks to the committee members in general of the Swedish Transport Administration’s research and innovation portfolio 4 on robust and reliable infrastructure, and chairperson Jan Ekström in particular. In addition, the authors would like to thank Henric Modig, who acted as the Swedish Transport Administration’s administrator/supervisor during the project work.
Karl Fridolf is with WSP Sverige AB, Daniel Nilsson, Håkan Frantzich, Enrico Ronchi and Silvia Arias, is with Lund University.