Data Needs for Predictive Building Movement Models

“Available Data and Input into Models,” Rita Fahy, National Fire Protection Association, Massachusetts

In order to better understand human behavior in fire, to enhance the effectiveness and completeness of evacuation models, and to provide better information for the users of evacuation models, additional study is needed in a range of areas. We need more data on all the time components of behavior, particularly those that are not a simple matter of speed and distance; data on the variability of those time components; and data or models on the factors driving behavior choices and the variability in time to perform certain actions. Some of the more specific areas will be listed and described.

“A Comprehensive Review of 28 Evacuation Models,” Erica Kuligowski, NIST

To aid with the difficult task of choosing an appropriate model, a comprehensive model review of 28 past and current egress models has recently been completed. This model review provides information on model purpose, availability, modeling method, model structure and perspective, methods for simulating movement and behavior, output, use of fire data, use of visualization and CAD drawings, etc. The model review organizes the evacuation programs into three basic categories that aim to describe the models’ level of sophistication in simulating behavior of the occupants. These categories are movement models (no behavioral capabilities), partial-behavioral models (implicit behavior is simulated), and behavioral models (occupant decision-making and behavior is simulated).

“Simulex: Simulated People Have Needs Too,” Peter Thompson, Integrated Environmental Solutions, Ltd., Glasgow, UK

This presentation will address a number of keys areas of evacuation modeling and data collection, drawing on the knowledge and experiences of the author. The following areas will be discussed: 1. Detailed human-movement data that were collected in order to create the evacuation model Simulex. 2. Use of the computer program to model situations where data are not available. 3. Required areas of concentration for future data collection efforts. 4. Notes about some current data collection activities in the UK.

“Uncertainty in Egress Models and Data: Investigation of Dominant Parameters and Extent of Their Impact on Predicted Outcomes - Current Status,” James Lord (presenting) and Brian Meacham, Arup Fire

Computer egress modeling is becoming a common tool in the building design industry. Models can provide insight into the movement of people through buildings, and sometimes provide a visual tool that is useful for presentation of a design to architects, clients, and authorities. The reality of egress modeling is that current methods of calculation must somehow account for a degree of human behavior that is not necessarily predictable. Most egress models attempt this through use of correlations based on available data, or through the addition of safety factors to the model results. When using an egress model in building design, there are many uncertain variables. However, as discussed by Fahy at the National Research Council Workshop to Identify Innovative Research Needs to Foster Improved Fire Safety in the United States, there is a severe lack of data for use in predicting evacuation times from buildings, and for the data that do exist, there has been little or no identification or assessment of uncertainty and variability, or of the impact of the uncertainty or variability on the predictive capability of egress models. Notarianni and others have discussed the importance of identifying and addressing uncertainty, as the failure to do so can lead to misapplication of models, and of the results obtained from the models as used in design and performance evaluations.

To begin addressing the above concerns, a three-year research program is underway, funded by a grant from the National Institute of Standards and Technology, Building and Fire Research Laboratory (Grant 60NANB2D0138), that aims to improve the predictive capabilities of egress models. The primary goals of this work are to: 1. Understand sources of uncertainty and variability in egress models. 2. Apply and refine a method of uncertainty analysis to computer egress modeling. 3. Identify “cross-over” variables that may have an impact on the results of the egress model that is significant enough to cause a change in an engineer’s design of a building. 4. Provide building engineers with guidance in the appropriate use of computer egress models.

This presentation will provide a brief overview of the current status of the research program, including the methodology and initial findings.

“Estimating Evacuation Time Components: Lessons from Nuclear Power Plants, Hurricanes and the First World Trade Center Bombing,” Michael Lindell, Hazards Reduction and Recovery Center, Texas A&M University

Data from the first WTC bombing show that this ambiguous situation elicited an orderly process of information seeking that tended to delay evacuation (Aguirre, Wenger, & Vigo, 1997; Prater, Wenger & Lindell, 1997; Wenger, Aguirre & Vigo, no date). Consistent with emergent norm theory, the information seeking (milling) process was influenced by pre-existing social relationships. Moreover, though there was a widespread and increasing perception of danger and some of the conditions for panic existed, the evacuation was orderly. This similarity in occupant behavior to that displayed by community residents in other types of disasters suggests that building evacuations in response to threats or acts can also be defined by a series of stages—detection/warning, psychological preparation, logistical preparation, and protective action selection/implementation (Lindell & Perry, 2004). There has been a considerable amount of research that has studied the relationship between detection/warning and evacuation. However, there has been very little research to date that has attempted to characterize household preparation times or the variables that account for differences among households in their preparation times. The available studies have found few, if any, reliable predictors of this evacuation time component (Aguirre, Wenger, & Vigo, 1997; Lindell & Perry, 1987; Lindell & Perry, 1992; Lindell & Perry, 2004; Lindell, Prater, Sanderson, Lee, Zhang, Mohite & Hwang, 2001; Lu, Lindell & Prater, 2004; Sorensen, 1991; Tierney, Lindell & Perry, 2001). However, further research is needed to determine the applicability of existing research to building evacuations. This is because social units within a building are defined more ambiguously than households within a community, employers can exercise more control over employees’ threat responses than public officials can exercise over community residents, and occupants’ perceptions of alternative protective actions in buildings are likely to be different from residents’ perceptions of the available protective actions in communities.

“Evacuation Time from a Single Family Home,” Guylène Proulx, National Research Council, Canada

This presentation identifies several factors that can affect evacuation time from a typical single family home in the event of a fire. Through a review of the literature, evacuation time is found to depend on the characteristics of the occupant(s), the characteristics of the building, the fire detection system in place, the location, cause, and time of the fire. Recent studies performed by NIST suggest that less then 3 minutes might be available for the safe evacuation from homes. This available safe egress time is of great concern since the require time to evacuate all occupants from a home fire might be much longer than 3 minutes in many fire scenarios.

Session Summary

After the session, the discussion addressed several themes that serve as an effective summary. These fall under four main categories: 1) Data needs for prediction and training 2) A repository for information and information sharing 3) Evacuation models, and  4) A discussion/working group to continue efforts after the workshop is over .

From the first set of presentations in this session, several data needs were outlined for both prediction methods and for training purposes. A few examples of data needs expressed were the evacuation of disabled occupants, flows on different types of stairs, behavior of occupants during evacuation, pre-evacuation delay times, and toxic effect on occupants (movement and decision making). After being presented with information from various types of disciplines which study behavior, participants in the data needs session were asking the following question: Are we dealing with a complete lack of data or a lack of known data? It became increasingly clear that other disciplines outside of fire research, such as human factors, sociology, psychology, and others, could contribute their wealth of knowledge to the study of evacuation from fires. Another related question that originated from the data needs session was the following: How much do we need to understand about evacuation from a building to make sure that the building is safe?

Another theme of the discussion from the data needs session revolved around the need for a repository of data on human behavior and a way to share this type of data among colleagues. NIST was looked to as a provider of this type of repository.

The third theme discussed in the data needs session was the topic of evacuation models. Many questions were posed relating to the assumptions of the models, their validation techniques, and whether they provided the appropriate information to users. Also, it was established that many current evacuation models neglect to include important aspects of an evacuation such as social/group movement, disabled movement and use of evacuation devices, and the effect of staff and other occupant roles on evacuation procedures. Among the model developers in the session, it was suggested that there be a standard for data format in all models so that the model packages can talk to one another. If this was the case, users could potentially create the “super-model” that can incorporate smoke and fire effects, people movement, behavioral aspects (if such a model existed), and building sensors to provide information to the occupants.

The third theme also addressed the issues of uncertainty in evacuation modeling. With the lack of data in some areas of evacuation and the choice of different types of data (i.e. movement data) for evacuation model inputs, a user is faced with difficult choices to make when providing inputs to the model. Work is currently being done in this area with different models to identify the inputs that have the largest impact on the results from the model, as well as to provide guidance to model users. With this information, the users can spend time collecting accurate data for the inputs that are the most significant to the individual model.

Lastly, a suggestion was made during the data needs session to develop a working group that would continue discussion and strive to eventually develop a research agenda in the area of human behavior in fire. An email discussion digest is in the works for those participants who are interested in continuing discussion from the workshop.

The session chairs of all three sessions were asked to summarize all session discussion answering the following four questions.

The current state of research in the field of fire evacuation is that several disciplines are primarily working independently on various projects related to the field. However, collaboration is lacking between these disciplines. It was clearly shown that sociology, psychology, engineering, human factors, computer science, and others could collaborate in several areas related to data needs, since much of the data are scarce and/or outdated.

The presentations and discussion established that there were several gaps in the data used for prediction methods. Also, there were gaps in the actual model inputs, simulation techniques, and users’ understanding of the models’ assumptions and limitations.

To bridge this gap, knowledge transfer between the disciplines would be a step in the right direction. Work in other areas would help to fill in gaps within data and help to increase the overall understanding of human behavior. Also, additional research in needed areas as well as information sharing among researchers would be important to developing a better understanding of occupant evacuation from buildings. Along with additional research, an understanding of how specific and accurate our predictions methods have to be is essential. Lastly, an institution, such as NIST needs to step forward and volunteer to be the central repository for evacuation data and information sharing among research disciplines.


Graphic Rule

Privacy Statement/Security Notice  |  Disclaimer  |  FOIA

NIST is an agency of the U.S. Commerce Department's Technology Administration.

 BFRL Logo