This website uses cookies to store information on your computer. Some of these cookies are used for visitor analysis, others are essential to making our site function properly and improve the user experience. By using this site, you consent to the placement of these cookies. Click Accept to consent and dismiss this message or Deny to leave this website. Read our Privacy Statement for more.
FPE eXTRA Issue 54, June 2020

Survey of Occupant Load Density in Retail Buildings

By: Gianluca De Sanctis, Michael Moos, and Christian Aumayer

This paper summarizes a study conducted by DeSanctis et al.  It reviews the major components of the study and includes a selection of references used within the study. Readers are referred to the full study report for more information.

Background

The risk of human losses in fires can be reduced by an accurate design of the means of egress. The design of the means of egress and the proof of a safe evacuation of the occupants by an egress analysis implies the use of a reasonable estimation of the number of occupants to be evacuated. Therefore, occupant load is a crucial parameter to ensure safety in buildings. Numerous occupant load surveys have been conducted in the past for different building uses. Most of these studies are based on walk-through investigations or assessments of building capacity, e.g. in residential buildings, hotels, hospitals, school buildings, theatres, cinemas, etc.

In contrast to this, the number of persons in retail stores is usually not influenced by the occupant capacity of a store. The number of persons is highly influenced by individual choice and the current demand for store products. This leads to a high temporal variability of the number of retail store occupants. Therefore, walk-through investigations are not suitable for surveying retail buildings. A common approach is to assess the occupant load density during (supposed) peak sales days of the year. Then, the derived occupant load density is used as a design value. The drawback of this approach is that the variability in occupant load density remains unknown, since only one point in time, the one with a presumed maximum load, is assessed. Finally, it remains unclear whether such a design value is appropriate for representing the occupant load of retail occupancies during the rest of the year, since fires do not necessarily occur on-peak. Therefore, it is inadvisable to use this assessed value for the occupant load in probabilistic approaches used for risk assessment. Only survey methods that include the number of persons at a given time can be used for a proper survey of the occupant load density.

The study summarized in this paper describes a systematic survey methodology to assess the occupant load densities in retail buildings and a statistical analysis of data from a survey of Swiss retail stores conducted using this methodology. The study was initiated by Espace.mobilité, which is an interest group made up of leading Swiss retail companies. Its members are competitors on the market, but partners in more fundamental issues such as spatial planning, environmental protection, mobility, and building regulations. The findings of this study are part of a larger project in Switzerland aiming at a revision of the prescriptive requirements for the means of egress in Swiss retail buildings.

Origin and Use of Codified Values

Spearpoint & Hopkin[1] reviewed different normative specifications for the occupant load density in retail stores in different guidelines. They found that occupant load densities are interpreted very differently, with densities for the main floor of retail stores varying between 0.1 pers./m2 and 0.5 pers./m2. To some extent the differences may be explained by cultural differences, but it is more likely that there are discrepancies in jurisdictional practices.

The origin of these values cannot be accurately traced. Certain densities can be tracked back to an extensive survey by Courtney et al.[2] from 1935. In this survey, the occupant load densities for retail stores were estimated based on an interview with a sales manager who was asked to approximate the maximal occupant load density for his or her retail store. The densities estimated by Courtney et al. can still be found in various guidelines.

However, it should be noted that most of these occupant load densities are intended to be used within a prescriptive regulation for design of the means of egress. It is possible that, depending on the design format of the prescriptive code, smaller densities may nevertheless lead to wider exits — and therefore to a higher level of safety. This could apply, for instance, when regulations contain a larger codified required exit width per occupant. It is therefore difficult to make a comprehensive international comparison on the level of safety when assessing only the occupant load density.

Performance-based design approaches, however, aim to generate a more realistic picture of the fire risk in buildings. For performance-based design approaches, the design values for occupant load densities are often borrowed from prescriptive regulations. Nevertheless, these values usually do not represent a realistic occupant load scenario. Providing design values for performance-based design approaches that are related to realistic conditions is therefore very important when applying a performance-based design approach.

Assessment Methods for Occupant Load Density

The occupant load density [d] can be defined using the ratio of the number of persons [p] that are present in a compartment and its net floor area [af] and leads to 𝑑 = 𝑝 / 𝑎𝑓 Here, the occupant load density is measured in persons per square meter [pers./m2]. The number of persons in a room p [pers.] varies in time. Hence, p can be represented as a random process P(t). Usually, the number of persons in a room cannot be counted directly and must be derived by other measurements. Two methods can be applied:

Method A: The occupant load is derived based on the counting of arrivals NA(t) and departures ND(t). The number of persons present in a store P(t) at any time [t] can be assessed by: 𝑃(𝑡) = 𝑁𝐴 (𝑡) − 𝑁𝐷(𝑡). In order to use this method, systems must be used that are able to distinguish between arriving and departing persons. Alternatively, the systems can be applied at locations where the flow of people is unidirectional.

Method B: The occupant load is derived based on the counting of arrivals (or departures) and the measurement or information available regarding the length of time customers spend at the store (dwell time). In order to use this method, systems can be used that a) count the persons arriving or leaving and b) measure dwell time for each occupant, e.g. by tracking systems.

In the survey conducted in this study, Method A was used as the basis for determining the occupant load. To count the customers in retail stores, optical sensors have been judged to be best suited due to the high accuracy and cost-value ratio.

Survey of Occupant Load Density in Swiss Retail Stores

The aim of the survey was to provide a comprehensive figure of the occupant load density in Swiss retail stores. Ninety six (96) stores with different retail types participated, including supermarkets, malls, department stores, electronic shops, hardware shops, clothing shops, furniture shops, and sports shops. The participating stores were drawn from the basic population of Swiss stores to form a representative sample and the measurements taken using a set of specific requirements. Care was taken to ensure that all characteristics of the stores that could have an influence on the occupant load density were included, e.g.: type or category of use of the store; size of sales areas; location (branches in rural areas, in the agglomeration, and in urban areas); user frequency (highly frequented and normally frequented retail stores); floor level; etc. The retail stores were divided into 13 different retail types. Almost 75% of retail stores are supermarkets, i.e. sales outlets with a strong focus on food sales. Therefore, a particular focus of the study was on supermarkets. Other non-food stores account for about 25% of the stores. The data survey lasted over a measurement period of one year and included, among others, special offer sales, highly frequented Sunday sales, or Christmas and Easter sales that usually have a higher customer frequency.

Survey Data Analysis and Results

The data was analyzed using two statistical approaches: a) an evaluation of quantile values per store; and b) a group evaluation by retail type. The impact of retail type, floor level, sales area, store location, and dwell time were assessed. For consideration for performance-based design, the data is represented as probability distributions, e.g. a gamma or a lognormal distributions were illustrated as the best fit as a function of retail type. Thus, the study not only provides values for the occupant load factors, but also the associated probability of occurrence. Further, the study illustrated that the key factor influencing occupant load is retail type. It also provides comment on the use and limitations of the data as a basis for prescriptive regulations and performance-based design approaches.

Conclusions

The densities show considerable differences for the design values in several codes and guidelines. Therefore, across several countries, the values from fire protection standards and guidelines may be set too high overall. However, for highly frequented supermarkets at important public traffic hubs, the values might be set too low according to some standards. A store’s retail type has a big impact on the occupant load density and should be considered when setting design values for fire protection standards and guidelines. The study proposes a differentiation between food stores (supermarkets), non-food stores (including department stores and malls), and highly frequented supermarkets located at important public traffic hubs.

In some current fire protection standards and guidelines, the occupant load density depends on the floor level. The data does not provide an obvious trend that would justify assigning either more or less persons to a specific floor level. This contradicts the current use of codified values for performance-based design approaches. It should be noted that the occupant load depends mainly on the store location and the allocation of its products and not necessarily on the floor level.

The data shows a high variability of the occupant load density and the probability distribution is heavily right skewed. Therefore, the simultaneity of a high occupant load density and a (rare) severe fire might be very unlikely. This should be considered in the design of the means of egress when applying performance-based design approaches.

An appropriate consideration of the distribution of the occupant load is only possible when applying probabilistic or risk-based approaches. The role of the variability of the occupant load density in retail stores in relation to other uncertainties in evacuation analysis might be crucial and should be the focus of future research.

References:

  1. [1] M. Spearpoint, C. Hopkin, “A Review of Current and Historical Occupant Load Factors for Mercantile Occupancies,” J. Phys. Conf. Ser. 1107 (2018) 072005. https://doi.org/10.1088/1742-6596/1107/7/072005.
  2. [2] J. Courtney, H. Houghton, G. Thompson, Design and Construction of Building Exits, United States Government Printing Office, Washington, DC, 1935.
  3. [3] L.T. Wong, “Occupant load factor in local residential old high-rise buildings,” Int. J. Eng. Perform.-Based Fire Codes. Volume 6 (2004) p.197-201.
  4. [4] M. Angerd, “Är utrymningsschablonerna vid brandteknisk dimensionering säkra?” Department of Fire Safety Engineering, Lund University, 1999.
  5. [5] D. Charters, D. McGrail, N. Fajemirokun, Y. Wang, N. Townsend, P. Holborn, “Preliminary analysis of the number of occupants, fire growth, detection times and pre-movement times for probabilistic risk assessment.” Proc. 7th Int. Symp. On Fire Safety Science, Worcester, MA, USA, 2002.
  6. [6] G. De Sanctis, J. Kohler, M. Fontana, “Probabilistic assessment of the occupant load density in retail buildings,” Fire Saf. J. 69 (2014) 1–11 https://doi.org/10.1016/j.firesaf.2014.07.002.
  7. [7] J.D.C. Little, “A Proof for the Queuing Formula: L= λ W,” Oper. Res. 9 (1961) 383–387. https://doi.org/10.2307/167570.

Gianluca De Sanctis is with EBP Schweiz AG, Switzerland.

Michael Moos is with ASE (Analysis Simulation Engineering) AG, Switzerland.

Christian Aumayer is with Migros Genossenschaftsbund, Switzerland.

Corporate 100 Visionaries

About Us

© 2019 SFPE | All Rights Reserved
Privacy Policy