The Importance of Input Parameters in Computational Modelling - Evacuation Simulations
Share |




The Importance of Input Parameters in Computational Modelling - Evacuation Simulations

By Olaf Albano Pérez


Modern assembly buildings tend to be larger and have an increased complexity and capacity compared to their predecessors. High occupancy buildings have also created a growing need to adequately address fire safety concerns. The logical way to address these concerns is through the use of fire safety engineering (FSE) and a performance-based design (PBD) approach.

An evacuation analysis is often a key part of the PBD process [1]. To perform such an analysis, engineers can use a wide range of engineering tools, from hand calculation to advanced methods. When the space to be analysed is particularly complex, it is necessary to use advanced evacuation modelling to conduct an evacuation analysis.

Currently, there are many types of simulation software available (60+) [5], each one with its own specific characteristics. However, regardless of the software used, the user (a fire safety engineer) needs to establish specific modelling parameters based on his/her engineering judgement, which can have a significant impact on the overall results.

When performing an evacuation analysis, the following fundamental question can be asked: what is the specific weight certain modelling parameters will have on the calculation of the evacuation time?

A parametric analysis was undertaken to analyze the influence of several critical parameters on total evacuation time.

Design Scenario

The building selected for the analysis was a football stadium. In this type of building various characteristics coincide that are suitable for the analysis developed in this research.

  1. The building is symmetrical with the same number of egress paths and population at opposite sides.
  2. The building has high occupation density.
  3. There are merge of flows in terraces, steers and corridors.
  4. There are confluence of wide and narrow corridors.
  5. There are stairs in the model.

Numerical Model of Evacuation

To undertake the analysis the evacuation model Pathfinder was used [2], [3], [4].



Pathfinder is an agent-based egress simulator. An agent-based model (ABM) is one of a class of computational models for simulating the actions and interactions of autonomous agents (in this case occupants).

These occupants have their own goals and behaviours and they interact with the environment and with the other occupants.

Pathfinder provides two modes for the occupant motion: an SFPE mode and a steering mode.

The SFPE mode implements the concepts shown in the SFPE Handbook of Fire Protection Engineering [6].

The Steering Mode is based on the idea of inverse steering behaviours. Craig Reynolds first presented the concept of steering behaviours [7]. Later Heni Ben Amor [8] refined the steering behaviours concept into inverse steering behaviours to solve some issues detected in the combination of simple steering behaviours to produce more complex behaviours.

In this study, the steering mode was used. In this mode each occupant has a behaviour which is assigned in the user interface. This behaviour dictates a series of goals that the occupant must achieve. In order to achieve such goals, for example egress out of a room, each occupant must plan the best path for moving toward their destination. Pathfinder uses a procedure named as locally quickest [3], [9] to establish the best path.

Once a local target has been chosen through path planning, it is necessary to establish a path. Pathfinder uses the A star pathfinder algorithm [3], the triangulated navigation mesh and a procedure known as string pulling [10] to establish this path.

Now the occupants must follow their path, interacting with other occupants and the environment. Pathfinder uses inverse steering behaviours [3], [7] and collision handling [3] to control how the occupant follows their path.


Input parameters

As previously mentioned, the impact that the variation of different input parameters have on the calculation of the evacuation time for a football stadium has been assessed.

The Table below summarizes the input parameters evaluated in this study.



In order to get accurate results for the analysis, the input parameters named as cost were also handled properly establishing adequately high and low values for each parameter.

The Table below summarizes the cost parameters and the value established in the research.



Parametric Study

A Parametric analysis was undertaken in order to analyse the influence that some input parameters have on the total evacuation time. The impact that each of the parameters have on the time required to evacuate the building was determined by a sensitivity analysis. Each variable was analysed varying its value from a minimum to a maximum, while all other variables remained constant with their default value. The table below summarizes the analysis carried out and the results obtained in the analysis.




The analysis undertaken has shown that input parameters such as occupant speed and shoulder width, have a very strong influence on evacuation time.

In the following graphs, the results of the sensitivity analysis for both walking speed (Ve) and shoulder width (Si) are shown.



On the left side in Graph 1, it can be seen that there is a strong dependence between the time needed to evacuate the building and the maximum walking speed established by the user.

For the established scenario, a variation in the maximum walking speed of the occupant from its default value (1.19 m/s) to 1.1 m/s produced a variation of 25% in the evacuation time of the building.

On the right side of Graph 1, it can be seen that there is a lesser dependence between the time needed to evacuate the building and the Shoulder width established by the user of the Model.

For the established scenario, a variation in the shoulder width from 60 cm to 65 cm produced a variation of 13% in the evacuation time of the building.

In the following graphs, the results of the sensitivity analysis for persist time (Pt) and collision response time (crt) can be observed.



Persist time and collision response time are involved in the processes used by Pathfinder to control how the occupants follow their paths interacting with the other occupants, to prevent occupants movement conflicts with another occupants, etc. Therefore, the value established in these parameters do not significantly affect the free movement of the occupants through the corridors.

As can be seen neither of the parameters have any significant effect on the evacuation time.


Based on the analysis undertaken for this particular assembly building chosen (a football stadium), it can be seen that the influence of the parameters on the total evacuation time can vary from almost negligible to very high.

The following table shows the overall results of the analysis, outlining the sensitivity of each parameters to the overall evacuation time.



Olaf Albano Pérez is with JVVA Fire & Risk


  1. R.L.P. Custer, B.J. Meacham, “Introduction to Performance-Based Fire Safety,” Society of Fire Protection Engineers, Gaithersburg, MD, and National Fire Protection Association, Quincy, MA, 1997.
  2. Thunderhead Engineering, Pathfinder 2016 User Manual.
  3. Thunderhead Engineering, Pathfinder 2016 Technical Reference.
  4. Thunderhead Engineering, Pathfinder 2016 Verification and Validation.
  6. Nelson, H. E., and Mowrer, F. W. “Emergency Movement,” SFPE Handbook of Fire Protection Engineering, 2008.
  7. Craig W.Reynolds, “Steering behaviors for autonomous character,” Proceedings of the Game Developers Conference 1999, San Francisco, California, 1999.
  8. Heni Ben Amor, Jan Murray y Oliver Obst, “Fast, Neat, and Under Control: Arbitrating Between Steering Behaviors,” AI Game Programming Wisdom 3, ed. S. Rabin. 2003.
  9. Charles Thornton, Richard O'Konski, Bryan Klein, Brian Hardeman, Daniel Swenson, “New Wayfinding Techniques in Pathfinder and Supporting Research,” Reference document Thunderhead, Download from
  10. Johnson Geraint, “Smoothing a navigation mesh path,” AI Game Programming Wisdom, 2006.
  11. Olaf Albano Pérez, “Sensitivity analysis for modelling parameters used for advanced evacuation simulations – How important are the modelling parameters when conducting evacuation modelling?” SFPE Performance-Based Design Conference, Warsaw, Poland, 2016.
  12. Olaf Albano Pérez, “Análisis de sensibilidad paramétrico de variables clave en el modelado de escenarios de evacuación y análisis exhaustivo del modelo de simulación computacional Pathfinder,” Master Thesis, Master in Fire Safety, ICAI University, Spain, 2015.

    © SFPE® | All Rights Reserved
    Privacy Policy