• Introduction

  • The purpose of this study is to analyse the predictive capabilities of the Simulex model, used to simulate the movement of people in evacuation simulations. Other evacuation models used within the fire engineering community, i.e. Firewind WayOut and simple hand flow calculations, provide quick and easy access to a reasonable estimate for a required movement time for egress in a building. This study will help to reveal whether the additional data used within the Simulex methodology aids the user in reaching a more accurate overall estimate. This will be done by carrying out a number of evacuation scenarios and comparing the results collected using the Firewind WayOut model and hand calculations. A multi storey hotel tower will be used to carry out the study. The outcome of the study will help to calibrate the components of the human behaviour in the Simulex model, as it is suggested that Simulex "enables you to simulate occupant behaviour in the event of a building evacuation" (IES, Simulex - simulation of occupant evacuation).

    A considerable amount of study has been carried out on all aspects of human evacuation from emergency situations, and the affects of human behaviour on evacuation times can be seen as a major factor in terms of life safety. The majority of movement models to date take into account little consideration of the behavioural aspects of the occupants under emergency and focus their work on the flow of occupants. An evaluation of the results gathered in this study will help to show whether Simulex takes occupants-occupants interaction into account.

    Studies carried out in the past have revealed that occupant evacuation times are highly dependent on their perceived threat of the fire event. "Appearance, proximity, propagation, time, and toxic gases of the fire threat also tend to predispose the individual to a higher level of behavioral activity, again depending upon the individual's perception of these threat variables. Thus, occupants located in close proximity to a developing fire, and with clear sensual links with smoke and heat, are likely to react more speedily than those who are reacting solely on alarm signals" (John L. Bryan, Human Behavior and Fire).

    The importance of such an analysis tool is becoming essential in building design as regulation moves to a more performance based system.

    The purpose of this dissertation is to outline the methodology used within the Simulex model. The outputs determined by each of the models can then be compared along with the hand calculation work carried out. A sensitivity analysis will be performed for the Simulex model and this will help provide a clear evaluation of its predictive potential.

  • Aim

  • To evaluate the predictive capabilities of the Simulex movement model by carrying out both sensitivity and comparative analysis from results gained using the Firewind WayOut movement models and simple hand flow calculations. To gauge the effectiveness of the additional methodological approach taken by Simulex in gaining an overall more accurate estimate.

  • Objectives

  • Carry out a literature review of papers available which cover all aspects of building evacuation.
  • Estimate crowd densities for use as input assumptions for Simulex, Firewind WayOut and hand calculations.
  • Evaluate the model outputs and make a comparison between the methodologies adopted by each of the models. Conclusions should be formed on the basis of this evaluation.
  • Carry out a sensitive analysis of both the Simulex and WayOut models. This can be achieved by altering the user input data to see how this effects the overall evacuation times.
  • Gauge the predictive capabilities of each of the models in terms of how all aspects of building evacuation are taken into account.
  • Investigate how the Simulex model attempts to deal with merging behaviours in a staircase. This will be carried out in reference to the information gained from previous work detailed in the literature review
  • Provide concluding statements with reference to the results gained using the Simulex model. This should include an insight into whether the results gained using this method provide a more accurate estimate of the likely real life evacuation time.
  • Methodology

  • Carry out a literature review of the existing information available which relates to building evacuation and evacuation modeling.

  • Factors to consider include a detailed evaluation of all aspects which affect the evacuation procedures of occupants' i.e. affects of alarms, pre-movement times, human behaviour, crowd dynamics, and travel times.
  • A study will be required relating to the current scope of movement models used within the fire engineering community.
  • The methodologies used within the Simulex and the Firewind WayOut models will be studied and form part of the literature review. This will highlight all the differences and similarities between the methodologies incorporated into the tools. This information will be of significant importance when analysing the output data and forming any conclusions.
  • Choose a suitable building design which can be used to carry out the study. The chosen building has been selected as the multi-storey hotel tower, Shibboleth project. Further information of the building will be provided further on in the text.
  • CAD drawings of the Shibboleth hotel tower are required as this design will form the basis of the study.

  • A collection of CAD drawings showing the Shibboleth floor plans will be used as a base to creating these geometries.
  • The CAD drawings will be stripped down (removing inanimate objects i.e. furniture etc) to reveal only boundary layers i.e. walls, floors, etc. These barriers are those in which occupants are unable to pass through.
  • A sensitivity analysis of Simulex and Firewind WayOut models will be carried out. This requires some factor of validation data to be collected by varying single point of input data and analysing how greatly they affect the overall results.

  • By independently altering all the required input data, it is possible to monitor the effect each of the inputs has on the model outputs.
  • This analysis will be carried out for each of the models.
  • Run a mock evacuation using the Simulex model. To achieve this all CAD drawing will be turned into DXF files and inputted into the model. Staircases and floor plans can be linked and occupants will be added relevant to the room sizes as uses (Occupant loading will be calculated for the building and agents will be calculated and added accordingly).
  • Run a mock evacuation using the Firewind WayOut model. The CAD drawings will again be used to measure all lengths and areas in the building. The occupant loading will be kept similar to those used in the Simulex calculation.
  • Carry out hand calculations for the building. The process which will be followed is provided in the SFPE Handbook (SFPE Handbook, Section 3, Chapter 14; Emergency Movement). All input data used will be maintained from the previous work carried out in the computer models.

  • This enables a conclusion to be reached as to whether the innovative tool can be used under the performance based regulatory system and form part of a successful fire engineered solution. Gaining a clear understanding of such information will allow the user to evaluate the results in a more efficient manner.
  • An exhaustive analysis of the output data produced by both movement models will be carried out by the author.

  • A conclusion will then be reached as to whether the FDS+Evac model can provide similar data as the movement model selected as the comparative tool.
  • As this Simulex model is used presently within the fire engineering community to perform evacuation analysis on a number of real projects, it can then be assumed that such a tool supports the engineers performance based design solution.
  • The results will also provide the evidence which will be required to evaluate whether the FDS+Evac model takes into account the threat perceived by the occupants in close proximity to a fire event, and incorporates this into the evacuation time for these occupants.

    Scope and Limitations

  • Only two models are being reviewed in the study. The university has both Firewind WayOut and Simulex available for use at present and no other models were available at the time this study was carried out.
  • The Simulex model is not used extensively through the course at the university and as a result the user had limited experience in operating it at the time of the study.
  • The modelling work, i.e. measurements of lengths and areas, was carried out entirely from the drawing provided; no site visits etc were made to the building.
  • Only one scenario has been run to carry out the study, a greater timescale for the work would have allowed a more exhaustive study, i.e. greater test cases, to be carried out.
  • As the methodologies in each of the models vary slightly, it was only possible to minimise the extent to which occupant characteristics varied, but it was not possible to eliminate it altogether.

    Literature Review

  • Regulatory Perspective

  • As architects, designers and engineers continue to push the boundaries of building design, the regulatory system in Scotland continues to move towards a more performance based system. This system allows all parties involved in the design stage a far greater amount of freedom, i.e. promote innovation and limit the impact of regulation (S. Kipp, 1999), when ensuring a building design meets the requirements of the relevant codes. Professionals working within the built environment are now able to incorporate much more of their experience and judgement when developing a design than when following the outdated prescriptive approach, which were conceived for 'typical' buildings. As a result of this, a number of tools have been developed within each discipline which allows each innovative design to be exhaustively tested, ensuring an adequate level of safety is provided before they are incorporated into any building design.

    For a fire engineer, many of these tools require computational technologies to perform a number of these tasks. Fire modelling is becoming more and more involved in the design stage of many large and complex projects all over the world. A number of models are available, varying in complexity, to carry out any necessary analysis within a number of complex spaces. They allow engineers to evaluate many fire safety related features of a building design before they are finalised, and ensure that any areas of issues with the design can be resolve before a project reaches the construction phase, as altering designs at this point can be extremely expensive and time consuming for all parties.

    In the UK, the current emphasis for escape design sets out to limit the distance and therefore time in which occupants are subjected to surrounding which will increase the risk of alarm or injury. The current timeframe in which occupants should have to travel from their place or origin and reach a place of safety is 2 minutes 30 seconds. This time had been calculated as a factor of the maximum allowable travel distance and the average walking speed of an occupant. Storey exit widths are sized assuming a specific flow of 80 persons/minute/metre clear width and a flow time of 2.5 minutes (Boyce et al, 2009).

    The time which is required to clear a floor is an important factor which must be considered to achieve an effective fire safety engineered design. The functional standards allow an engineer to carry out comparative analysis between the required safe egress time (RSET) and the available safe egress time (ASET). A building is deemed to provide an acceptable solution if the time required for egress is less than the time available before conditions are judged untenable by some factor of safety. This requirement is subject to an exhaustive analysis being carried out by a suitable professional, on all aspects of the design which will affect occupant egress.

  • Human Behaviour in Fires

  • A lot of research has been undertaken within the fire engineering community to gain as much understanding as possible of the factors affecting human behaviour when occupants are faced with emergency evacuation procedures in the built environment.

    John L. Bryan has covered a lot of work studying person-fire interaction and how occupant awareness can affect pre-movement times D. Canter has done a lot of work in gathering data from a number of sources to paint a clearer picture of the evacuation process. E. R. Galea covered a study dealing with human behaviour during evacuation of the world trade centre attack in 2001. Jonathan D Sime has produced work dealing with peoples ability to way find in a building, his work has shown that it may be more effective to incorporate escape routes into the general circulation routes as this will increase occupant familiarity with evacuation routes. Lars Benthorn provided an insight into how people evaluate information and subsequently choose their escape path. There are many more professionals who have done excellent work in analysing human behaviour in emergency situations and all the information collected is useful as it can then be incorporated into the design of evacuation tools.

    Building evacuation takes on a number of stages and involves a timeframe from the incipient stage of a fire right through until the last occupant has reached a place of safety. Human behaviour can affect both pre-movement and movement times, therefore it is essential to have a clear understanding of how to adapt an evacuation design to maximise its potential in life safety terms.

    The time to evacuate a building is a combination of several stages, these stages are:

    The time taken for each of these stages of the evacuation process is dependent on the occupant's response and behaviour.

    Figure 1:

    Factors involved in assessing the total escape time. (CIBSE Guide E: Fire safety engineering design approaches, 4-7).

    Pre-movement Time Distribution

    The pre-movement time of a building is the time for occupants to react to the alarm signal and begin their evacuation process. There are many factors which can affect the pre-movement times of occupants and these will be highlighted later in this text. In multi storey, multiple use occupancies, such as the one selected as part of the study, it can be assumed that not all occupants will have comparable pre-movement times, and for this reason it is good practice to study the appropriate time distribution curves in order to provide an accurate account of an expected pre-movement time in a building simulation.

    Purser et al, 1999,

    suggest from their work that

    "

    Once the first few occupants have begun to move, the pre-movement times for the remainder of the occupants in an enclosure tend to follow a logarithmic–normal frequency time distribution".

    The shape of the above curves follow a typical pre-movement tome distribution following what has been observed historically; the initial delay of start up highlights the time taken for the first of the occupants to make the preliminary movements towards their chosen exit. This is followed by a rapid increase in frequency as the majority of others tend to initiate their travel phase. The long tail of the curve illustrates the last remaining occupants who will begin their travel period which will signify the end of the total pre-movement phase of the evacuation process.

    The above distributions are fit well for open plan occupancies where occupants have a clear view of the majority of other persons in the premises. In a building hosting a large number of enclosures, it can be assumed that the time distribution will be far wider than shown in the above diagram. This is due to the limited visibility which would be available for occupants in such a premises; the herding effect as occupants will be reduced as they would have less chance of grouping together and following the actions of the first occupants who move.

    Purser et al, 1999

    , suggests that a

    range of 20-30 minutes

    would be more suitable for a multi occupancy building with sleeping risk (such as the Shibboleth hotel tower used to carry out the study).

    Many different factors will influence how a person will react and the decisions they make will determine their evacuation process.

    "It can be very difficult to obtain real evacuation behavior; real evacuations may be undertaken by people who are unaware of the actual urgency to escape. They may perceive the alarm as a drill" (Jake Pauls, 2003)

    People are often unaware that the alarm they hear is not a false one and so they will proceed to evacuate as they see fit to do so. Stopping to gather up personal belongings or only beginning to evacuate when others around them do. People have both reaction times and pre-movement times, reaction time is the time taken to perceive the alarm and decide to take action; and the pre-movement time is the time that elapses while the occupant is preparing to leave.

    L. Benthorn (1999): "People usually choose to leave a building the same way they came in, even if this is a poorer alternative than other available. Within the field of behavioural science, it is pointed out that people often choose the known before the unknown, which would explain the above behaviour."

    Occupants in a building will tend to head for the exit them came in through not only are they familiar with this exit it but it will lead them to a place they will recognise. This is particularly true for those people who are not familiar with their surroundings. People will continue to do this and follow the crowd until they are either faced with the fire or are given further information. It has been suggested that incorporating evacuation routes wherever possible into the main circulation routes at the design stage will aim to optimise the effectiveness of the evacuation strategy. This is due to the fact that occupants tend to use a familiar route.

    The occupant characteristics that should be considered in performing an evacuation analysis are listed below:

  • Population numbers and Density

  • The maximum potential load should be used to give a conservative estimation. The number of people using a building or space and their distribution will greatly affect the travel and flow speeds speed of occupants.

  • Familiarity

  • A person's familiarity and regular use of the building and its systems may cause them to respond differently. Competent users of the building will have prior knowledge of the nearest escape routes and they may have had the opportunity to have participated in drills. Those unfamiliar with the building will rely upon the knowledge of staff and the clarity of signage available, and may be less responsive to warning systems.

  • Distribution and Activities

  • Distribution will impact on movement speeds and density will impact on the ability to communicate instructions. Activities people are involved in will affect their initial response. Those who are dedicated to a task within a building will not necessarily be able stop their job on activation of the alarm system.

  • Level of Alertness

  • The commitment of people to their activity or their interaction with others can affect their awareness. A premise which holds a sleeping risk for occupants can be expected to have a delayed response time.

  • Physical and Mental Ability

  • Some occupants may rely entirely on assistance, disabled; those with a hearing disability or those with a visual disability may require special means of notification.

  • Level of Mobility

  • Affected by the age of occupants, age can influence the ability of an individual to independently make their way along an exit route and reach a place of safety within an acceptable timescale. It may also reduce an occupant's ability to withstand exposure to smoke and other harmful bi-products of fire.

  • Social Affiliation

  • Behaviour will be strongly influenced with the interaction between occupants. Groups of people who have a social connection (i.e. parent and child who are separated within premises at the time of the fire event) will try and regroup before making their way to an exit. The time spend undertaking such an act may increase the level of risk for these occupants. Groups of evacuees try to stay together and the slowest member of the group influences their speed.

  • Role and Responsibility

  • Sufficiently, well-trained and authoritative staff will shorten the pre-movement phase of an evacuation process. An effective management plan followed by all members of staff will ensure this is provided within premises.

  • Location

  • Can influence a person's choice of exit and the time to notification. Travel distances will be affected by location.

  • Commitment

  • Those who are committed to their activity will be reluctant to respond to an alarm, especially if it means their task is to be started again.

  • Responsiveness

  • The extent to which a person is likely to respond to alarms, those who have previous experience of emergency situations may be less likely to respond quickly as they are aware of the most appropriate action to take.

  • The Panic Theory

  • "When people, attempting to escape from a burning building pile up at a single exit, their behaviour appears highly irrational to someone who learns after the panic that other exits were available. To the actor in the situation who does not recognise the existence of these alternatives, attempting to fight his way to the only exit available may seem a very logical choice as opposed to burning to death." (Turner and Killian 1957)

    The concept of panic is attributed to occupant's lack of knowledge about a fires existence before a fire reaches a size where it can seriously hamper the ease in which evacuees are able to escape. This can be due to a problem with the detection and alarm system installed within premises, or the lack of information available to occupants as they try and make their way to the relevant escape routes.

    The theory of panic is not an easy thing to define, yet a set of definitions are presented below:

    "A sudden and excessive feeling of alarm or fear, usually affecting a body of persons, originating in some real or supposed danger, vaguely apprehended, and leading to extravagant and injudicious efforts to secure safety". (John L. Bryan 1984)

    "A fear-induced flight behavior which is nonrational, nonadaptive, and nonsocial, which serves to reduce the escape possibilities of the group as a whole", (Kentucky State Police, 1977).

    "In the stress of a fire, people often act inappropriately and rarely panic or behave irrationally. Such behavior, to a large extent, is due to the fact that information initially available to people regarding the possible existence of a fire and its size and location is often ambiguous or inadequate." (Ramachandran, 1990.)

  • Affect of Alarm on Pedestrian Movement

  • The type of detection and alarm system in a building can greatly affect the way in which occupants despond to the emergency signal, and this is turn will affect the response time of occupants. The level of information that occupants are provided with in the early stages of evacuation can influence their decision to evacuate. It has been common practice to use traditional ringing sounders within non-domestic premises in recent years. One drawback of using this form of alarm signal is that occupants are not being provided with any informative information regarding the fire event. Evacuees could benefit from a system which would inform them of a fires location and lets them know which evacuation route is the safest in terms of their location in the building. This is a difficult system to integrate into a building as fires are extremely unreliable and information is specific to a single fire scenario.

    "Sounders themselves are not the most informative method of warning system; they convey little information and have been proven ineffective" (Bob Choppen, 2003).

    Voice alarm systems are largely becoming a more acceptable mode of informing occupants of a fire occurrence in modern buildings. Large premises which are designed to cater mainly for the general public will benefit greatest from a voice alarm system. Occupants are fuelled with much more information of the emergency event than in the past using traditional alarm signals. Voice messages can convey a greater deal of information to the occupants. John L Bryan concluded from his research that the use of voice alarms/public announcements with an alarm bell was the most effective way of warning occupants.

    Ramachandran in his review of the research on human behaviour in fires in the UK since 1969 summarized the effectiveness of alarm bells as awareness cues: "The response to fire alarm bells and sounders tends to be less than optimum. There is usually skepticism as to whether the noise indicated a fire alarm and if so, is the alarm merely a system test or drill?

    A lack of panic is attributed to a number of factors including:

    fire and smoke.

    eliminating the chance for queuing to occur, i.e. little competition for similar exits by

    occupants.

    Human Stress Model. (University coursework notes, Evacuation Systems Design model; Powerpoint Presentation namely Human Behaviour in Fire (Slide 48/51), Dr. Iain Sanderson, 2008).

    Evacuation Modeling

    Evacuation models can help engineers prove that tenable conditions will be available to occupants for the timescale required for all occupants to reach a place of safety, which an element of safety built in. The total time for occupants for occupants from the time of detection and alarm, to the time for the last occupant to reach a place of safety, is called the Required Safe Egress Time (RSET). This is traditionally compared with the time from fire ignition until tenable limits are exceeded, and conditions have reached a level where humans will be unable to continue their process of escape. This time is called the Available Safe Egress Time (ASET). As long as RSET > ASET by some factor of safety, a building is deemed to provide an adequate level of safety for all occupants to escape in an emergency situation.

    Pedestrian movement models have typically fallen into two categories, one category dealt independently with movement and the other tried to connect both movement and human behaviour.

    S. Gwynne (1999) highlights the main approaches available of computer analysis models: "Computer based analysis of evacuation can be performed using one of three different approaches, namely optimization, simulation and risk assessment. Furthermore, within each approach different means of representing the enclosure, the population and the behaviour of the population are possible". Movement models can be categorised in a number of forms; Ball bearing, Optimisation, Simulation or Risk Assessment models.

  • Ball bearing / Gaseous

  • This example of movement model treats its subjects as inanimate objects. Sometimes referred to as 'environmental determinism', subjects are unthinking individuals who respond only to external stimuli, thus human behaviour it not taken into account. Occupants are assumed to begin their evacuation instantly, with no regard to the time taken for detection, alarm and pre-movement times. Factors effecting occupant movement therefore only include physical considerations of the occupants and their surroundings (i.e. crowd densities, exit widths and travel speeds). Individual occupants are merged into units and their movement treats their "egress on masse" (S. Gwynne, 1999). A good example of a model which employs this type of methodology is Firewind, with its WayOut tool.

  • Optimisation

  • This form of pedestrian movement model deals with large crowds of people at the same time. Evacuees are treated as homogeneous groups, thus there are no independent characteristics for a particular individual. People are uniformly distributed; all exits will be equally shared. One of the best examples of this form of model is EVACNET.

    6.5.3 Simulation

    These models try and take into account not only the physical characteristics of the space, but also consider some representation of human behaviour in emergency scenarios. They attempt to produce as an output the path and decisions taken my individuals during the evacuation process. Examples of this type of model include Simulex and buildingEXODUS.

    6.5.4 Risk Assessment Model

    These models are an attempt to identify hazards associated with the evacuation of a building, be it due to the occupants or the building, and attempt to quantify the resultant risk. An example of this type of model would be Crisp, and WayOut.

    6.5.6 Enclosure Representation

    Enclosure representation of the geometries created within a computer models can take on two forms; fine and course networks. Enclosures are subdivided into a number of zones which are interconnected with neighbouring zones, and the characteristics of each of these affect the parameters found in each on the adjoining cells. The detail and size of each of these zones determines which category a model shall fall into.

    NODE ARC NODEOne or more arcs connecting 2 nodes are called a Path. (John M Watts 1987).

    Definition of a network model is given by John M Watts (1987), "A network models is a graphical representation of routes by which objects or energy may move from one point to another".

    6.5.6.1 Fine Networks

    Models using this method divide the entire floor space of the enclosure into a selection of shapes or nodes. The size and shape of these nodes will vary for different models. The node is connected to its neighbouring node by an arc. Paths of individuals are tracked over time.

    Examples of such models include Bgraf, Egress, buildngExodus, Magnetmodel, Simulex and Vegas.

    6.5.6.2 Coarse Networks

    Models following this form of enclosure representation do not allow individual occupants to be followed independently of other within the group. Single nodes represent large spaces, such as rooms and corridors. As evacuees moved from space to space, users will be unaware of their position in each node. A coarse network does not provide information regarding person-wall, person-person and person-obstacle interactions within each room.

    Examples of Crisp, E-scape, Evacsim, Evacnet, Paxport and Takahacji's model.

    "In the Coarse network approach…the geometry is defined in terms of partitions derived from the actual structure." (S. Gwynne et al, 1999).

    The type of model should be carefully chosen with the regard to the complexity of the output data required. Coarse networks are evidently a lot quicker to process output results than their fine network counterparts, and although they provide a less extensive outlook of a space, their outputs are generally still accurate enough for most evacuation modelling purposes.

    6.5.7 Population Models

    6.5.7.1 Individual

    This form takes account of personal attributes of occupants. Individual characteristics can be prescribed to each occupant within the model, either randomly or user described. These attributes determine an occupants individual ability to move and make decisions, thus it allows each person to have their independent characteristics. As a result of providing a user with extensive input options, individual population models require a longer time to process all the data. Examples of individual population models include Egress, buildingExodus, Simulex and Vegas.

    6.5.7.2 Global

    Models which pursue this form do not allow individual characteristics to be given to every occupant; instead it generates the population as a homogeneous group. As a result, outputs from this type of model will allow the user to know of the number of successful evacuees, but it does not provide information on individual occupants. Benefits of this system include simpler output data for the user to use and a shorter processing time required to achieve model outputs. Some drawbacks of using this mode include that the models lack the intricate detail required for some complex problems. Due to the grouping of occupants and areas, there is a difficulty in estimating many of the tenability criteria in specific points within an enclosure. This is because fires starting in single points within a space will not uniformly distribute the harmful products of combustion around a room. The grouping of occupants means that those with evacuation times in the proportionate extremes are included in terms of the overall population. This means that those with limited mobility i.e. the elderly, wheelchair users and young children, will increase the average evacuation times of all occupants and thus may produce inaccurate output data for the majority of the occupants. Examples of global population models include Donegans's Entropy, Exit 89, Evacnet, Takahashi's Model and WayOut.

    6.5.8 Behavioural Perspective

    6.5.8.1 No Behaviour Rules

    Those models that use no behavioural rules rely completely on the physical movement of the occupants to determine the movement during evacuation. The decisions made are done solely on the basis of physical influences. Evacnet is a good example of this form of model.

    6.5.8.2 Implicit Behaviour Rules

    Such models do not declare behavioural rules but assume them to be implicitly represented through the use of physical methods. They rely on secondary data which incorporates physiological and sociological influences. Secondary data is that which is gathered from other sources, and this form relies entirely on the accuracy of this form of data. Examples include exit 89, Paxport, Simulex and WayOut.

    6.5.8.3 Rule Based Behavioural System

    These models are stochastic; decisions that are taken by occupants are done so according to a set of predefined set of rules, which can be defined by the model user. Models using this form of behavioural perspective include Bygraf, Crisp, E-scape, Evacism, and buildingExodus.

    6.5.8.4 Artificial Intelligence Based Systems

    Models, in which individuals are designed to mimic real human intelligence, include Donegan's Entropy, Egress and Vegas.

    6.6 Simulex

    Simulex is a partial behaviour evacuation model for use with computers to estimate the expected evacuation of occupants for a number of complex geometrical spaces.

    "The computer model SIMULEX combines spatial analysis with the escape movement of large numbers of individuals in a building, and is intended for eventual use as a design tool" (P.A. Thompson, E.W. Marchant, 1995).

    The geometries are created by inputting a number of CAD drawings, in DXF format, which form the basis of the model. These firstly have to be stripped down to reveal only bare obstructions i.e. walls and other immovable barriers. Final exits are defined by the user, which allows the model to calculate travel distances for all exit paths throughout the entire building using a distance contouring method. "Each person heads towards an exit by taking a direction which is at right angles to the contours shown on the chosen distance map" (Thompson, 1995). This contouring process allows Simulex to plot the minimum travel distance to an exit from anywhere within the building thus optimising the efficiency of the evacuation process for the occupants. This can be modified however by the user to simulate an exit being unavailable.

    The building occupants, namely the agents, can be characterised either as individuals or as part of groups before being placed anywhere within the model. "Each person is assigned a normal, unimpeded walking speed" (Thompson, 1995). Agents placed within the model have a range of travel speeds, 0.8-1.7 m/s (PA. Thompson, 1995). The range of values allows the user to alter each agents travel speed which corresponds to young, old, fit or mobility impaired occupants. "The walking velocity for a person is dependant on the forward linear distance (proximity) to people ahead". The inter-person distance is calculated by measuring the distance between the centre points of each of the agents' bodies, which is illustrated in figure 6, below. This should allow the Simulex model to provide a more accurate estimate of an evacuation time than a number of other movement models (e.g. Firewind WayOut) as a range of occupant types can be processed at the same time, thus providing a more realistic simulation.

    The range of travel speeds also allows occupant speed to be varied in relation how densely populated the crowd they are travelling in becomes. Such a value will fluctuate as agents pass in close vicinity and through doorways etc. "Walking speeds are reduced as people get closer together" (Thompson, 1995). The above speed density curve illustrates the speed/density curves representing a number of researchers work. Used within the Simulex model is the curve of Ando et al (1988), which is shown in Figure 5, above.

    Simulex attempts to take into account some factors of escape movement involved in evacuation scenarios. They attempt to take into account the path and decisions made by the agents as they make their way to a place of safety. Although it is not yet fully feasible to couple an evacuation movement model with an extensive range of human behaviour qualities, it is however possible to measure a number of physical aspects of crowd behaviour into an evacuation tool, "these basic parameters include speed fluctuations; crowd flow behaviour, travel distances and overall evacuation time, based upon certain assumptions" (P.A. Thompson, E.W. Marchant, 1995).

    It adopts a fine network to carry out the analysis, a 0.2m X 0.2m special mesh. Models using this method divide the entire floor space of the enclosure into a selection of shapes or nodes. The size and shape of the nodes will vary for different models. The node is connected to its neighbouring node by and Arc (see Figure 1, above).

    The Simulex model adopts implicit behaviour rules to simulate occupant behaviour. This form does not declare behavioural rules but assume them to be implicitly represented though the use of physical methods.

    The ability to individually characterise the agents in the model allows an extensive range of both social and physical parameters to be analysed. This means that social interactions can be modelled for groups of individuals i.e. group dynamics can be altered in reference to the perceived relationship between the agents within the group. It is assumed that as agents get closer together; their initial travel speed will decrease accordingly.

    The Simulex model is unable to provide a pre-movement time for occupants; these can be selected using sound engineering judgement of close analysis of the problem. Relevant information on such times is crucial in designing life safety systems into premises and can be collected from BS9999, CIBSE Guide E and the SFPE Handbook, and it will be necessary to add this time onto the travel time to produce an accurate total evacuation time for each scenario. Simulex at present is also unable to take into account the occupants familiarity with a building, and this will also be judge with inspection of experience gained dealing with building evacuation.

    A three circle representation of the human body, shown below, has been incorporated into the Simulex model, which allows each agent to have rotational degrees of freedom.

    This proves a more lifelike representation of the human body shape than the single circle shape found in other models and should increase the chance of output data being similar to that found in real life situations. "Overtaking, body rotation, sideways stepping and small degrees of back-stepping are all accommodated" (Thompson, 1995). Allowing each agent to be individually characterised provides the user with useful information regarding major issues with a real design, as it is assumed that in real life in the majority of cases a good mix of occupant types will be found at any one time. Walking speeds of each individual person can also be measured allowing each individual to independently change their rate of speed as they pass obstacles and other obstructions. Each agent is also able to 'decide' on their own travel speeds and direction of travel.

    Simulex works by calculating a distance map which calculates the total distance to an exit for every point in the special mesh. Every calculation carried out in this way shows the shortest route to an exit, i.e. the shortest travel distance.

    A visual display of the evacuation can be examined by the user, which allows an exhaustive analysis of each of the occupant's movements from start to finish. Close inspection of individual agents, or crowds or agents as they make their way along a chosen exit path is therefore easily obtainable. This can aid the user when evaluating the output data received from Simulex. Also, hazard areas can be highlighted by the use of the visual display tool, i.e. bottlenecking.

    The individual characteristics of the users within the models can also be altered to analyse how this affects the evacuation times in each of the scenarios. Occupant characteristics will be altered to change the size and shape of each evacuee, which in turn will adjust the occupant travel speeds and interaction with their neighbouring subjects.

    The agent-agent interaction between can also be monitored using Simulex as the travel speeds will vary in relation to crowd densities and the resultant travel speeds.

    Simulex uses data which was collected from a variety of test cases, in which a building evacuation was carried out and recorded using perspective image analysis of custom designed software.

    This body representation has been developed by the work of Langston et al, and gives a more realistic interpretation of a real life body shape, inclusion of rotational degrees of freedom.

    6.7 Firewind WayOut

    The Firewind WayOut program is an evacuation modelling tool which allows the user to simulate the egress time required for a variety of spaces, including single and multiple rooms and multi storey spaces.

    The Required Safe Egress Time (RSET) for a space can be taken as the total time which is required from fire initiation to the last occupant reaching a place of safety. The RSET can therefore be split into a number of sub factors; time to detection and alarm, pre-movement times, and the evacuation/movement time. Firewind WayOut calculates solely with the evacuation time required, information regarding the other factors should be gained using the relevant means. The pre-movement times of occupants can then be inputted into the program and this will be added to the movement time to give the user an estimate of time to alarm until the last occupants have reached a place of safety.

    The theory behind the model has been taken from the work by Shestopal and Grubits, (1994). This work utilised the experimental data gathered from Predtechenskii and Milinski, (1978), and later Russian research implemented in Russian standard GOST 12.1.004-91.

    The model is based on a non-linear flow algorithm utilizing an experimentally obtained speed – density dependence by Predtechenskii & Mininskii. WayOut works on the assumption that foot traffic flow is dependent on the flow density for any given route type. The model includes a trend of the pedestrian flow to jump into the maximum-density mode when the flow intensity reaches a critical value. (Victor O. Shestopal, 2007).

    The user is given the opportunity to switch between the two options. Both methodologies carry slight differences in their approach to the movement problem, and it is good practice for the user to consider both methods when carrying out evacuation analysis.

    Although both figure 5 and 6 show differences in relation to the flow rates, both follow a consistent profile in terms of the shape of the graph. Both V.M. Predtechenskii, A.I. Millinskii (Moscow, 1969), as well as work carried out by Fruin (New York, 1971) and Pauls (Boston, 1980) have produced information regarding the flow of traffic and its relationship with crows density:

    "If the population density is less than about 0.54 persons/m2 of exit route 1.85 m2/person, individuals will move at their own pace, independent of the speed of others. If the population density exceeds about 3.8 persons/m2, no movement will take place until enough of the crowd has passed from the crowded area to reduce the density. Between the density limits of 0.54 and 3.8 persons/m2 the relationship between speed and density can be considered as a linear function." (SFPE Handbook, Hazard Calculations, Section 3, Pg. 370).

    Firewind WayOut focuses mainly on the movement of occupants and the physical factors which will affect the flow efficiency of crowds i.e. exit route efficiency, travel distances, width of passage and the degree of difficulty with reference to the entire exit system. (Firewind Manual, Pg. 85).

    Both horizontal (corridor flows) and vertical (ascending and descending staircase flows) are considered by the model. Staircases have been given a standard tread/riser ratio and this cannot be altered by the user.

    6.8 Effects of Merging Flows in a Staircase

    A significant factor in allowing an engineer to accurately estimate the evacuation time of a building is to be able to "quantify merging flows and the factors that influence evacuee merging behaviour in an emergency situation". (Karen E Boyce, David Purser, Jim Shields, 2009). The ability to design an efficient evacuation system will maximise the potential for occupants to reach a place of safety within an acceptable timeframe i.e. before conditions within the enclosure becomes untenable. Merging flow calculations can help an engineer to achieve such a goal.

    Merging flows occur where two streams of occupants meet, typically at the floor entrances to staircases. Those occupants making their way down the staircase will merge with those who are leaving a lower floor and entering the stairs. It is important to examine which stream will likely take precedence over the other as this can largely affect clearance time of floors and subsequent evacuation times from unprotected regions of a building. A good understanding of how to achieve an effective evacuation system includes an ability to maximise the efficiency of merging streams. As mentioned earlier in the text, Boyce et al, 2009, highlight the underlying principle of current fire design guidance in the UK; "the storey exit widths are sized assuming a specific flow of 80 persons/minute/metre clear width and a flow time of 2.5 minutes". Another important factor which must not be ignored when estimating floor clearance times is the merging capabilities of the floor occupants and the occupants already within the staircase. Data collected in the work carried out by Purser and Utting (1996), suggests that for buildings designed with simultaneous evacuation in mind, according to the prescriptive guidance, "the rates of floor clearance halve at each successively higher floor if merge ratios of approximately 50:50 occur once the stair becomes congested, and this could potentially lead to floor clearance times significantly greater than the 2.5 minute target flow time on upper floors." This highlights the importance of understanding the factors which may influence merging and deference behaviour in evacuation situations. In a multi storey building for example, the halving of floor clearance on each upper floor could pose much greater problems for those on the topmost floors than would be witnessed on lower floors.

    Experimental studies carried out by Hukugo et al (1985), emphasised the importance of which stream establishes itself first: "if both streams arrive at the merge at the same time, the bias was in favour of the floor (average 60%); otherwise approximately 50:50 merging occurred". This collection of data signifies that on a balanced plane of occupants reaching the mergence point, the stream flowing from the floor to the staircase would take a slight precedence over those occupants making their way down the staircase. This form of merging could develop into a significant problem for those on upper floors than first noted if merging flows were not taken into suitable consideration. Work produced by Takeichi et al (2006), takes this idea further gathering experimental data on merging stream flows, in which "the geometry of the door/landing/stair interface can be an influencing factor" (Boyce et al, 2009). The concluding points of the experimental identify that "a greater flow onto the landing (is achievable) when the landing door was located adjacent to the incoming stair". Although the study was limited in terms of occupant numbers and the experiment data was collected through staged scenarios, the results gained were consistent with those gained in the work carried out by Gelea, (2008). Its findings also worked to conclude that an evacuation process could be run more efficiently if the floor/stair connection is "adjacent to the incoming stair".

  • Shibboleth Hotel Tower

  • The Shibboleth hotel tower is the chosen building for the study. All CAD drawing have been provided through staff at the University for use in the study.

  • Building Description

  • The hotel tower consists of an innovative design which hosts a 212 room hotel which incorporates a conference centre, a health and fitness facility, in addition to the provision of 1900m2 of commercial office accommodation arranged over levels 3 and 4. The internal arrangement is characterised by an atrium interconnecting all habitable floors which hosts a scenic lift. The atrium void is reduced in size on levels 1, 0 and -1, which permits a scenic lift to travel past a feature wall and a similar device is introduced between levels 14 and 18. The principle entrance level (01) is a double height space with a gallery (02) which provides lounge and waiting spaces together with hotel administration offices. Guest room accommodation is located on levels 5 to 13 each of which is served by two exit stairways linked by an exit access corridor which provides views into and across the atrium containing three scenic lifts. The atrium extends from the principle entrance level (01) to the topmost guest accommodation level (13). Level 14 is occupied by the main plant room and a kitchen is located at level 15 to serve the restaurants which are located on levels 16, 17 and 18. These levels comprise a single volume containing 2 mezzanine levels which houses 3 restaurants. Level 19 is reserved for plant. Ground floor has 6 individual means of escape to a place of safety, each evenly distributed around the perimeter of the floor.

  • Occupant Numbers

  • Floor Level

    Occupancy Type

    Area

    (m

    2

    )

    Sleeping Risk

    Occupant Load

    Response Time (s)

    Response Time Distribution (+/-) (s)

    Level 00

    Foyer, Reception, Offices, Ancillary

    1190

    No

    550

    120

    60

    Level 01

    Mezzanine Admin Area

    593

    No

    212

    60

    30

    Level 02

    Open Plan Office

    882

    No

    100

    60

    30

    Level 03

    Open Plan Office

    968

    No

    116

    60

    30

    Level 04

    Bedrooms, Suites

    1250

    Yes

    66

    1800

    900

    Level 05

    Bedrooms1, Lounge2

    1250

    Yes

    196

    18001, 1802

    9001,

    902

    Level 06

    Bedrooms

    1250

    Yes

    66

    1800

    900

    Level 07

    Bedrooms

    1300

    Yes

    70

    1800

    900

    Level 08

    Bedrooms

    1310

    Yes

    70

    1800

    900

    Level 09

    Bedrooms

    1240

    Yes

    67

    1800

    900

    Level 10

    Bedrooms

    1100

    Yes

    65

    1800

    900

    Level 11

    Bedrooms

    1045

    Yes

    60

    1800

    900

    Level 12

    Bedrooms

    880

    Yes

    54

    1800

    900

    Level 13

    Plant Floor

    665

    No

    3

    60

    30

    Level 14

    Kitchens

    495

    No

    59

    120

    60

    Level 15

    Restaurant, Lounge, Bar

    567

    No

    225

    120

    60

    Level 16

    Restaurant Coffee Shop

    241

    No

    64

    120

    60

    Level 17

    Restaurant Asian Cuisine

    160

    No

    63

    120

    60

    Level 18

    Plant Floor

    85

    No

    3

    60

    30

    Total

    16471

    2109

    Table 1:

    Floor by Floor Occupant Load Factors with other user defined evacuation characteristics

    Simulex Scenario Data

    Simulex was used to simulate the estimated evacuation time for the entire building with a worst case occupancy loading, which meant all areas of the building were at capacity. To produce as fair a study as possible, the occupant loading in each of the enclosures and floors remained constant over the course of the study. The same building floor plans, together with consistent numbers of links and exits, were maintained over the scope of the study to ensure any confliction in the results was due to the variance in the occupant characteristics and the selection of choice of distribution curve i.e. the available user variables under analysis. All exit routes (e.g. door widths, stair widths and corridor widths) also remained constant throughout the course of the study.

    When carrying out a sensitivity analysis of the choice of distribution curves, all occupant characteristics were grouped in relation to the room use in which they were located i.e. offices were given "office staff" characteristics; those within the hotel rooms were given "hotel" characteristics and so on. Some of the room uses for this building were not available in the Simulex program, (e.g. "plant room" and "restaurant"), therefore a best fit characteristic from those which were available was chosen by the user, and each has be justified as to their selection in each of the individual scenario descriptions provided below.

    In addition to this, all occupants were also given suitable response times in relation to their activity. Such information was gathered from the CIBSE Guide E document (second edition 2003). Response times were selected from table 4.5 "suggested guide values of pre-movement times" and the chosen information has been illustrated in Table 1, above. Agent response times were kept constant throughout each of the scenarios as this allowed a fair study to be carried out.

    When carrying out a sensitivity analysis of the "occupant type characteristics" variable, the distribution curve remained similar as a normal distribution curve. This meant that the only varying user input was the occupant characteristics. A number of scenarios were carried out and the occupant's physical characteristics were altered from by selecting "all male", "all female", "all children" and "all elderly" from the characteristics menu.

    A single "default distance map" was used in all of the studies; allowing occupants to utilize all of the available exits. In evacuation analysis, particularly for life safety, is it good practice to carry out an evacuation simulation where one of more of the exits is unavailable, as this could represent a real life worst case scenario. As this study is analysing the predictive capabilities of the Simulex movement mode in comparison to both Firewind WayOut and SFPE Hand calculations, in order to carry out a fair study all exits must remain available for occupants. Both WayOut and hand calculations methodology work on the basis that all exits are available, the default distance map in Simulex allowed the exit efficiency to remain constant over the course of the study. It has been noted though that creating different distance maps where exits are deemed unavailable is a useful tool in carrying out a more realistic evacuation analysis of a space. It is possible to execute the removal of exits in WayOut but it remains time consuming as a new evacuation file will require to be created each time the exit layout is altered.

    To simplify, the main parameters under analysis were the distribution curves and the occupant characteristics. Below is a description of each of the scenarios which were carried out for this study:

    8.1 Simulex Scenario 1

    Occupants were characterised in relation to the room use in which they were located. The first group of scenarios (scenario 1, 2 and 3) were carried out in order to analyse the effect the distribution times have on an evacuation. Scenario 1 was carried out using a normal distribution. The following table provides information regarding the chosen input data for the Simulex model.

    Floor Number

    Occupant Load

    Characteristics

    Response Times (s)

    Level 00 Foyer, Reception, Offices, Ancillary

    550

    Shoppers

    120; +/-60

    Level 01 Mezzanine Admin Area

    212

    Office Staff

    60; +/-30

    Level 02 Open Plan Office

    100

    Office Staff

    60; +/-30

    Level 03 Open Plan Office

    116

    Office Staff

    60; +/-30

    Level 04 Bedrooms, Suites

    66

    Hotel

    1800; +/-900

    Level 05 Bedrooms1, Lounge2

    196:

    (1392, 571)

    Hotel

    1800; +/-9001, 120; +/-602

    Level 06 Bedrooms

    66

    Hotel

    1800; +/-900

    Level 07 Bedrooms

    70

    Hotel

    1800; +/-900

    Level 08 Bedrooms

    70

    Hotel

    1800; +/-900

    Level 09 Bedrooms

    67

    Hotel

    1800; +/-900

    Level 10 Bedrooms

    65

    Hotel

    1800; +/-900

    Level 11 Bedrooms

    60

    Hotel

    1800; +/-900

    Level 12 Bedrooms

    54

    Hotel

    1800; +/-900

    Level 13 Plant Floor

    3

    All Male

    60; +/-30

    Level 14 Kitchens

    59

    Office Staff

    120; +/-60

    Level 15 Restaurant, Lounge, Bar

    225

    Shoppers

    120; +/-60

    Level 16 Restaurant Coffee Shop

    64

    Shoppers

    120; +/-60

    Level 17 Restaurant Asian Cuisine

    63

    Shoppers

    120; +/-60

    +18 Plant Floor

    3

    All Male

    60; +/- 30

    Table 2:

    Input data for Scenarios 1, 2 and 3.

    As mentioned above, there were a small number of room uses in the chosen building which were not available for selection in the range of occupant types provided within the Simulex model. This meant the user had to select a character type as a best fit for the room use, and these are justified below.

    Shoppers:

    were used as a replacement for the restaurant and ground floor levels of the building. Using the above, from the Simulex user guide manual as a reference, it was concluded that the percentages of both male, female and children would be consistent with the typical occupant type found within a shopping mall. On ground floor level is could also be acceptable to use office staff for a number of areas but for the purposes of this study a constant agent characteristic type was maintained for the duration of the study.

    Office Staff:

    was used in the kitchens floor as this represented a good mix of both male and female agents, but as expected in such an area, no children or elderly occupants would be present. Also, it is assumed that staff would have a reasonable awareness of the building layout.

    All Male:

    agent was used in both plant floors of the building. It is assumed that staff working within such an area would demonstrate movement traits similar to a fully independent person. It is also assumed that no elderly and children agents would be located in such floors.

    8.2 Simulex Scenario 2

    Occupant numbers, locations, characteristics and response times remained the same as the first scenario. As a result the above, Table 2 can be used to show the input data carried forward into this scenario. The variable in this scenario was the distribution curve which was selected as "triangular distribution".

    8.3 Simulex Scenario 3

    Occupant numbers, locations, characteristics and response times again remained the same as the first and second scenario. As a result the above Table 2 can be used to show the input data carried forward into this scenario. The variable in this scenario was the distribution curve which was selected as "random distribution".

    8.4 Simulex Scenario 4

    In order to carry out a sensitivity analysis of the people characteristics available in the model, a set of scenarios was required to be analysed which takes only such variables into account. As a result, the distribution curve for the remaining scenarios was kept constantly set at the default "normal distribution". As before, all occupant numbers, locations and response times remained the same through the entire course of the study. The occupant characteristics for the entire building population will be selected as "all male" for this scenario.

    8.5 Simulex Scenario 5

    To ensure an effective sensitivity analysis of the occupant characteristics options is carried out the remaining scenarios are kept the same other than the alternating of occupant types. As a result, the distribution curve for the remaining scenarios was kept constantly set at the default "normal distribution". As before, all occupant numbers, locations and response times remained the same through the entire course of the study. The occupant characteristics for the entire building population will be selected as "all female" for this scenario.

    8.6 Simulex Scenario 6

    As stated above all user specified parameters will be maintained with the exception of the occupant type, which for this scenario will be selected as "all children".

    8.7 Simulex Scenario 7

    Again, as stated above all user specified parameters will be maintained with the exception of the occupant type, which for this scenario will be selected as "all elderly".

    The second set of scenarios (4-7) should provide an insight into how responsive the different occupant types are to the evacuation scenario.

    It is expected that a building full of fit and healthy males should provide a shorter evacuation time than what should be gained when analysing data for both children and elderly agents. This is due to the parameters included in the methodology of Simulex which sets free moving occupant travel speeds in relation to the characteristics of the agent.

    Is it however assumed that as the occupants queue up at doorways and subsequently located in an area of high occupant density that, such data e.g. unimpeded travel speeds become less significant as the speed of travel is subsequently slowed down no matter what the agent characteristic has been set at.

  • Observations from Simulex Playback Files

  • The playback files in Simulex allow the user to step frame by frame either forwards or backwards during the evacuation process. This is beneficial as it allows the user to investigate with greater detail, the occurrences at points of interest. One major area of interest in relation to this study is to observe the way in which agents interact with each other at merging points in the building e.g. stairways and floor exits. The earlier literature review provided an insight into the findings of previous studies and historical data as to how occupants interacted in such areas. The following text describes the observations which could be made when reviewing the scenarios carried out in the study. A conclusion can then be reached as to whether the methodology adopted by Simulex in relation to merging flows shows a consistency with the information available from other investigations. This will allow the reader to gauge how effective the model is at simulating the merging phenomenon and therefore it's provide an insight into its general predictive capabilities.

    Observation 1: Beginning the Evacuation process.

    As the simulation begins, those who make their way towards the staircase on upper floors are able to travel at their normal unimpeded walking speed, as occupant density is low. As they reach the vicinity of other agents travelling along a similar exit path the occupant density increases and therefore the speed of the agent decreases accordingly.

    Observation 2: Floor Exits.

    Occupant density increases as more and more agents reach the doorway to the staircase, and as the exit width decreases, the flow rate of occupants reduces (i.e. more occupants are reaching the doorway than passing through) thus creating a build up of agents at the doorway. Crowd shuffling becomes a factor as the agent attempt to attain their personal space. Travel speeds become extremely limited and as highlighted at an earlier stage, all agents, no matter which characteristics have been set to them, can only move as fast as the crowd they are adjoined to.

    Observation 3: Merging Flows at Staircase entry.

    When occupant loading in the staircase is low, agents passing through the doorway and into the staircase are able to travel at a constant velocity.

    After reviewing the data using the playback mode in the Simulex model, it seems that at the early stages of the merging process the merging flow of the agents seems to favour those coming from the floor onto the staircase. Evidence of this is provided on staircase one in the first few minutes of escape, where occupant's joining the staircase from first floor level merges with those already making their way down the staircase from upper floors. There is a build up of occupants behind the merging flow which slows and causes a crowd to form from first to third levels. As a similar effect is happening at the second floor merge point, this adds to the effect and it can be concluded that agent above this level are being effected by two separate merging streams. This should be highlighted as a potential issue in staircase design especially for multi-storey buildings as each floor merge point will increase this factor i.e. occupants on topmost floors will take a lot longer to enter the protected zone of the staircase than those located on lower floor levels. Below the merging flow, agents move at a relatively constant pace and the crowd density is limited. A similar effect can be witnessed on all staircases when merging flows are present (e.g. staircase 4 at around 2-3 minutes). These merging flows are present and differing timeframes due to the range of response time's specified to the agents groups on different floors.

    The flow rate of those making their way down the stairs exceeds the flow rate entering the staircase from first floor level and this has to be taken into account, yet it seems like the merging flow at this point favours those entering the stair enclosure.

    It is observed that the agents coming down the stair slows to an extent that occupant density behind those merging increases and agent travel speed decreases accordingly.

    Observation 4: Agent Interaction on Staircases

    A travel speed on staircases has been given in the Simulex User guide 12: Walking Speeds, as 50% of the horizontal walking velocity for the individual. As every agent is individually accounted for by Simulex, this means that those groups with slower travel speeds will also have slower travel speeds in staircases.

    Groups of such agents (e.g. elderly, children) on a staircase will resultantly slow down the agents behind as overtaking rates are far less effective in such areas due to the limited exit width.

  • Firewind WayOut Scenario Data

  • In order to provide a comparative analysis and ultimately gain an insight into the predictive potential of the Simulex evacuation movement model, a further evacuation simulation will be carried out using the Firewind WayOut model. To provide as fair a study as possible, some of the input data will be kept similar to that used in the Simulex work. Occupant loading will be kept consistent with previous scenarios and the sizes of the stair widths and door widths will remain the same. The methodology adopted by the WayOut model only allows one exit path to be measured at the same time. For this reason, the input data provided by the user was set out for one staircase. This was set up from a final exit on the ground floor and a staircase which covered all floors of the building. As only one staircase is being monitored in the WayOut program, to provide a fair study, the occupant loading as part of the input data had to be altered accordingly. It has been assumed that including 50% of the occupants on each floor represents a reasonable comparison with the occupant numbers using each staircase in the Simulex program. It is not possible to ensure this is an exact match and so it has been noted that this could affect the evacuation times when analysing the output data. The start time box in the WayOut program allows the user to input a response time relevant to the activity being carried out by those on each floor. These have been kept the same as those used in the Simulex program. It is not possible to select a distribution curve for the occupants in WayOut and so it is assumed that all occupants assigned with the same response time will begin their evacuation process at the same time. This again is a minor difference that the input data included in Simulex and therefore has been noted as a possible factor if differing output data is collected.

    Firewind WayOut allows the user to alter only a limited number of factors which will affect the evacuation process.

  • Justification of Input Data for Firewind WayOut

  • Choice of Evacuation Route

  • It is important to ensure that all occupants are taken into account when simulating an evacuation scenario. For the purposes of Firewind WayOut, occupants on the ground floor have been omitted from the evacuation modeling as it is assumed that all occupants in this area will reach a place of safety in a reasonable timeframe before occupants on upper floors have reached the exit floor. It has been noted that occupants on the ground floor would affect queuing times at the main exits at early stages of evacuation but this should not affect overall evacuation times.

    Due to the limitations of the Firewind WayOut model, it is not possible to create the exact geometries which have been gained from the CAD drawings as used in the Simulex software. For this reason occupants will not be placed around the floors as they are shown in the drawings. Each floor will be represented by a square room large enough to hold the number of occupants on the floor and as a result this is where the agent will initiate their egress. Again, this will mean that travel times to the staircases will be shorter than in Simulex but it is assumed that the queuing times as agents gather outside the staircases will eliminate this from being a factor in the overall evacuation time.

  • WayOut Scenario 1:

  • Firewind WayOut does not allow the user to choose occupant characteristics in terms of room use. The user is able to select summer, mid-season or winter dress code from the Dress menu. In this scenario the dress code has been selected as mid-season. To carry out a suitable comparison between the two model outputs, some of the input data had to remain the same. Due to the limitations of the program, only one staircase can be modelled at the same time. Therefore the number of occupants, as mentioned above, will be taken as 50% of each of the floor occupant loads (0.5 X occupant load from table 2, above). All response times have been taken from table 2 above and maintained over both models. Door widths have been retained at 1.3m and similarly stair widths have also been kept at the same value as in the Simulex model, 1.2m. For this scenario, as well as the first set of three), option one has been used in reference to the flow density curve options which are discussed in greater detail within the literature review. Option 1 is the experimental data gathered by the scientists V.M. Predtechenskii & A.I. Milinskii (1969).

  • WayOut Scenario 2:

  • All user data which is used for scenario 2 is similar to that user for scenario 1, apart from the dress code being changed from mid-season to summer dress.

  • WayOut Scenario 3:

  • Again, all user data which is used for scenario 2 is similar to that user for scenario 1, apart from the dress code being changed from mid-season to winter dress.

    The above set of scenarios will provide an adequate view as to how the choice of dress code affects the occupants' ability to escape.

  • WayOut Scenario 4:

  • The next set of scenarios aims to test the sensitivity of the two options of flow density curves. The data collected from these scenarios can be compared to the results gathered in the previous set of scenarios. For this reason scenario 4 will set dress code as mid season. All other input values mentioned above will remain similar.

    Option 2

    has been chosen for this set of scenarios, which is the

    Russian Standard (GOST) 12.1.004-91.

  • WayOut Scenario 5:

  • Dress code will be set at summer dress; all other values will remain constant.

  • WayOut Scenario 6:

  • Dress code will be set at winter dress; all other values will remain constant.

  • Data for Hand Calculations

  • To form part of this study, hand calculations will also be included as a way of further analysing the results gained by Simulex. These hand calculations have been selected from the SFPE Handbook, Section 3; Chapter 14: Emergency Movement. The calculation procedure has been included as an Annex (B) at the back of this text. The values used to carry out the calculation have been selected to as far as possible correspond with the input values used within both Simulex and WayOut. Occupant numbers and exit widths are consistent with the values used in the model work. The hand calculations do not take into consideration all of the factors that the models include, therefore additional data (e.g. response times) will be added to the calculated travel time in order to allow a comparison between all evacuation times to be impartial.

  • References

  • IES (Integrated Environmental Solutions Ltd). Simulex - simulation of occupant evacuation.

    IES VE Simulex Virtual Environment, User Guide, Version 5.0.

    Bryan. John, L. Human Behavior and Fire, Chapter 1. Section 4; 4-7. "A Review of the Methodologies Used in Evacuation Modelling"

    S. Gwynne, E. R. Galea, M. Owen, P. J. Lawrence, L. Filippidis 'Fire and Materials pg. 383 – 388', 1999.

    Sime, Jonathan D, "Human Behaviour in Fires Summary Report", 'Research Report No. 45', 1992.

    John L. Bryan. "Human Behaviour in Fire – Chapter 1", 'Section 4 – Human Behaviour in Fire Emergencies',

    Boyce Karen, Purser David, Shields Jim, experimental studies to investigate merging behaviour in a staircase, University of Ulster, UK, Hartford Environmental Research, UK; Human Behaviour in Fire Symposium, 2009.

    Purser D A and Bensilum M Quantification of escape behaviour during experimental evacuations (Garston: Building Research Establishment) (1999)

    D. Canter, "Studies of Human Behaviour in Fire – Empirical Results and Their Implications for Education and Design", 'BRE Research Establishment Report', 1985.

    Lars Benthorn, Hakan Frantzich, "Fire Alarm in a Public Building: How Do People Evaluate Information and Choose Evacuation Exit", 'Department of Fire Engineering, Lund', June 1996.

    Yang Lizhong, Fang Weifeng, Fan Weicheng, "Modeling Occupant Evacuation using Cellular Automata – Effect of Human Behavior and Building Characteristics on Evacuation", 'State Key Laboratory of Fire Science University of Science and Technology of China', 2003.

    H. A. MacLenna, M. A. Regan and R. Ware. August 1999. An Engineering Model for the Estimation of Occupant Pre-movement and or Response Times and the Probability of their Occurrence. Fire and Materials.

    Jonathan D Sime, Human Behaviour in Fires Summary Report, Research Report 45, Portsmouth Polytechnic, 1992.

    E. R. Galea, Ph.D. An Analysis of Human Behavior during Evacuation, Fire Protection Engineering, Fall 2005

    Sime J D and Kimura M, 1988. The timing of escape; exit choice behaviour in fires and building evacuations. Safety in the built environment. London: E and F N Spon.

    Ramachandran, 1990. G. Human Behaviour in Fires - A Review of Research in the United Kingdom. Fire Technology, Vol. 26, No. 2., pg. 149–155.

    Proulx, G., and Sime, J. D, 1991. To Prevent Panic in an Underground Emergency - Why Not Tell People the Truth? Fire Safety Science - Proceedings of the 3rd International Symposium, Elsevier Applied Science. pg. 843–852.

    Ramachandran, G, 1990. Human Behaviour in Fires - A Review of Research in the United Kingdom. Fire Technology, Vol. 26, No. 2, pg. 149–155.

    Ramachandran, G. 1991. Informative Fire Warning Systems. Fire Technology, Vol. 27, No. 1 pg. 66–81

    Lars Benthorn

    ,

    Fire alarm in a public building: How do people evaluate information and choose evacuation exit? Department of Fire Safety Engineering, Lund University; June, 1996

    P.A. Thompson, E.W. Marchant, Computer and fluid modelling of evacuation, Safety Science 18, 1995

    L. Benthorn, 1999. Managing evacuating people from facilities during a fire emergency. Pg. 325-330.

    Cable, E. A. January 1994. Cry Wolf Syndrome: Radical Changes Solve the False Alarm Problem. Department of Veterans Affairs.

    Proulx, G., and Sime, J. D, 1991. To Prevent Panic in an Underground Emergency - Why Not Tell People the Truth? Fire Safety Science - Proceedings of the 3rd International Symposium, Elsevier Applied Science. pg. 843–852.

    Ramachandran, G, 1990. Human Behaviour in Fires - A Review of Research in the United Kingdom. Fire Technology, Vol. 26, No. 2, pg. 149–155.

    Ando, K., Ota, H. and Oki, T., 1988. Forecasting the flow of people, (In Japanese). R.R.R. Railway Research Review, (45)8: 8-14.

    Ramachandran, G. 1991. Informative Fire Warning Systems. Fire Technology, Vol. 27, No. 1 pg. 66–81

    Shestopal, V.O. and Grubits, S.J., Evacuation Model for Merging Traffic Flows in Multi-room and multi storey buildings. For merging flows in multi room and multi storey buildings. Fire Safety Science – Proceedings of the 4-th International Symposium, International Association for Fire Safety Science, Ottawa, Ont., 1994, p. 625-632

    Predtechenskii, V. M. and Milinskii, A. I., Planning for foot traffic flow in buildings, Amerind Publ., 1978.

    Victor O. Shestopal, Computer Models for Fire and Smoke, 2007

    GOST 12.1.004-91. Occupational safety standards system. Fire safety. General requirements. (Russian).

    Cable, E. A. January 1994. Cry Wolf Syndrome: Radical Changes Solve the False Alarm Problem. Department of Veterans Affairs.

    V.M. Predtechenskii and A.I. Milinskii, Planning for Foot Traffic in Buildings (translated from the Russian), Stroizdat Publishers, Moscow (1969).

    J.J. Fruin, Pedestrian Planning Design, Metropolitan Association of Urban Designers and Environmental Planners, Inc., New York (1971).

    J.L. Pauls, "Effective-Width Model for Evacuation Flow in Buildings," in Proceedings, Engineering Applications Workshop, Society of Fire Protection Engineers, Boston (1980).

    Galea, E.R., Sharp, G., Lawrence, P.J., Investigating the Representation of Merging Behaviour at the Floor-Stair Interface in Computer Simulations of Multi-Floor Building Evacuations. Journal of Fire Protection Engineering, 18 pp 291-315, 2008.

    Takeichi, N., Yoshida, Y., Sano, T., Kimura T., Watanabe, H. And Oymiya, Y., Characteristics of merging occupants in a staircase, In Fire Safety Science, Proceedings of the Eighth international Symposium, Gottuk, D.T and Latimer, B.Y. (Eds), International Association of Fire Safety Science, London, 2006, pp 591-598.

    Boyce1 Karen E, Purser David, Shields Jim, Experimental Studies to Investigate Merging Behaviour in a Staircase, University of Ulster, UK, 2Hartford Environmental Research, UK

    Websites:

    http://www.vtt.fi/proj/fdsevac/?lang=en

    http://www.iesve.com/Software/VE-Pro/Egress/Simulex

    http://angel.elte.hu/panic/

    http://fseg.gre.ac.uk/exodus/exodus_faq.html#one

    http://fseg.gre.ac.uk/exodus/

    Bibliography:

    VTT Technical Research Centre of Finland.

    Korhonen, Timo; Hostikka, Simo; Heliövaara, Simo; Ehtamo, Harri. FDS+Evac: Modelling Social Interactions in Fire Evacuation. Proceedings of 7th International Conference on Performance-Based Codes and Fire Safety Design Methods. Auckland, New Zealand, 16 - 18 Apr. 2008. SFPE (Bethesda, MD, USA, 2008), pp. 241-250.

    Korhonen, Timo; Hostikka, Simo; , Heliövaara, Simo; Ehtamo, Harri; Matikainen, Katri. Integration of an Agent Based Evacuation Simulation and the State-of-the-Art Fire Simulation. Proceedings of the 7th Asia-Oceania Symposium on Fire Science & Technology. Hong Kong, 20 - 22 Sept. 2007.

    Heliövaara, Simo. Computational Models for Human Behavior in Fire Evacuations. M.Sc. Thesis, Department of Engineering Physics and Mathematics, Helsinki University of Technology, 2007.

    Matikainen, Katri. KÄYTTÄYTYMINEN UHKATILANTEESSA – Poistumisreitin valintaan vaikuttavat sosiaalipsykologiset tekijät tulipalossa. Pro Gradu -tutkielma, Valtiotieteellinen tiedekunta, Helsingin yliopisto, 2007.

    Korhonen, Timo; Hostikka, Simo; Heliövaara, Simo; Ehtamo, Harri; Matikainen, Katri. FDS+Evac: Evacuation Module for Fire Dynamics Simulator. Proceedings of the Interflam2007: 11th International Conference on Fire Science and Engineering. London, UK, 3 - 5 Sept. 2007. Interscience Communications Limited (London, UK, 2007), pp. 1443-1448.

    Korhonen, Timo; Hostikka, Simo; Heliövaara, Simo; Ehtamo, Harri. An Agent Based Fire Evacuation Model. 4th International Conference on Pedestrian and Evacuation Dynamics. Wuppertal, Germany, 27 - 29 Feb. 2008.

    Heliövaara, Simo; Ehtamo, Harri; Korhonen, Timo; Hostikka, Simo. Modeling Evacuees' Exit Selection with Best-Response Dynamics. 4th International Conference on Pedestrian and Evacuation Dynamics. Wuppertal, Germany, 27 - 29 Feb. 2008.

    Kimura, M., and Sime, J. D, 1989. Exit Choice Behaviour during the Evacuation of Two Lecture Theatres. Fire Safety Science— Proceedings of the 2nd International Symposium, Hemisphere Publishing Corporation pg. 541–550.

    Kimura, M., and Sime, J. D, 1989. Exit Choice Behaviour during the Evacuation of Two Lecture Theatres. Fire Safety Science— Proceedings of the 2nd International Symposium, Hemisphere Publishing Corporation pg. 541–550.

  • Annex A: Stair Length Calculation

  • The following calculation method shows how the stair lengths were calculated in order to provide Simulex and Firewind WayOut with the appropriate input data to run an accurate evacuation simulation.

    Useful Information

    Floor to Floor Height = 4250mm

    Acceptable slope angle of staircases = 30°

    Simple trigonometry will allow a calculation for the true length of staircase required in order for occupants to travel between floors. The above figure shows the ratios for a simple triangular shape, similar to a section drawing of the staircase. This can be adapted to calculate the actual stair length required between floors.

    Using the rules of trigonometry is can be taken that:

    Substituting the valid numbers and rearranging the formula allow the hypotenuse length can therefore be calculated as:

    X 4250

    Each floor step will have two flights of stairs connected with a landing to carry the distance due to the restriction on the angle of the staircase.

    As a result, an additional length must be added to the calculated slope length (7.36m) to take account of the travel distances on the landings. It is important that such factors are carried into the calculation as in a multi-storey building such as the one used in the study the landing distance can become a sizeable portion of the overall travel distance and will therefore affect travel times.

    Taking the centre line of the landing (assuming this to give an average), an additional 2.4m (as stair lengths remain constant is can be assumed that 4 X 0.6m can be used as the centre line path guide) of travel distance should be added to the slope length (7.36m) to give a total of 9.76m. For this purposes of this exercise the length had been rounded to 10m for each level rise.

    The total travel distance for occupants in the staircases between floors has been calculated as 10m.

  • Annex B: Hand Calculations for Estimated Time of Building Evacuation

  • The following calculation methodology has been taken from the SFPE Handbook of Fire protection Engineering and should provide an accurate estimation of the estimated evacuation time for the hotel tower. All assumptions have been stated, as any results which are collected from using a calculation method are heavily dependent on the assumptions which are made by the user.

    Step 1

    It has been assumed that the main controlling factor will either be the stairways or the doors which discharge from them. It is also assumed that queuing will occur in such areas, thus the specific flow used in calculation will be maximum specific flow. For this calculation, all occupants are assumed to start egress at the same time. This is unlikely in reality but will produce an accurate estimate which will ultimately be given a factor of safety. The assumption will also be that occupants use all exit facilities in the optimum balance which is also unlikely in reality but again will be reasonable for the purpose undertaken here.

    Step 2

    From Table 3-14.1, Boundary layer widths;

    Assume width of stair as 1.2m

    Less boundary layer 0.15m x 2 = 0.3m

    This gives an

    effective stair width = 0.9m

    Assume width of door as 1.3m

    Less boundary layer 0.15m x 2 = 0.3m

    Leaves an

    effective door width of 1.0m

    From table 3-14.5 (for 6.5" riser, 12" tread)

    Maximum specific flow (stairways) =

    F

    sm

    = 1.09

    persons/s/m effective width

    Specific flow equals maximum specific flow (as noted above in step 1)

    Therefore using equation 6: the flow from each stairway is limited to;

    Fc = Fs x We Fc = calculated flow

    Fs = specific flow

    We = effective width

    Fc = 1.09 x 0.9 = 0.98 persons per second

    =

    58 persons per minute for the stairway

    Step 3

    From table 3-14.5 The

    maximum specific flow through a door

    is 1.3 persons/s/m of effective width.

    = 0.91 persons per second

    =

    54 persons per minute for a door

    (gained following similar process as step 2)

    Since flow capacity of door is greater than the flow capacity of the stairway which serves it: flow is controlled by the stairways at the rate of 54 persons/stairway exit door/minute.

    Step 4

    Estimated Speed of movement;

    Using equation 3

    S = k – akD S = speed along the line of travel

    D = density in persons per unit of area

    k = 1.16, constant from 3-14.2 (for 6.5" riser, 12" tread)

    a = 0.266

    S = 1.16 – (0.266 X 1.16 X 1.9)

    = 0.574 ms-1 (X60)

    =

    34.44 metres/minute

    Using conversion factor from table 3-14.3: travel distance between floors is:

    Floor to floor height = 3.7m (taken from section drawings)

    Conversion factor (for 6.5" riser, 12" tread) = 2.08

    Floor to Floor height X conversion factor = 3.7 X 2.08 = 7.7m on the slope of the stairs plus an additional 5m travel on the landings

    Therefore total travel distance between floors = 7.7 + 5 =

    12.7m

    Therefore the travel time for a person moving with the flow = travel distance / travel speed

    = 12.7 / 34.42

    =

    0.37 minutes per floor

    Step 5

    Estimate building evacuation time with only one escape stair available

    If all occupants start to evacuation at a simultaneous same time then the stairway can discharge 58persons/minute (above from step 3)

    The population above the hotel tower (2109) will require

    2109 / 58 =

    36.7 minutes

    to pass through the exit.

    This is a worst case scenario and it reality more than one exit will be available for occupants to reach a place of safety. The next set of calculation data show the estimated evacuation time required when two stairs are available for occupants to use in the egress.

    Evacuation of floors 2 – 18, Two Stairs Available, No Pre-movement Time.

    Step 1

    It has been assumed that the main controlling factor will either be the stairways or the doors which discharge from them. It is also assumed that queuing will occur in such areas, thus the specific flow used in calculation will be maximum specific flow. For this calculation, all occupants are assumed to start egress at the same time. This is unlikely in reality but will produce an accurate estimate which will ultimately be given a factor of safety. The assumption will also be that occupants use all exit facilities in the optimum balance which is also unlikely in reality but again will be reasonable for the purpose undertaken here.

    Step 2

    From Table 3-14.1, Boundary layer widths;

    Assume width of stair as 1.2m

    Less boundary layer 0.15m x 2 = 0.3m

    This gives an

    effective stair width = 0.9m

    Assume width of door as 1.0m

    Less boundary layer 0.15m x 2 = 0.3m

    Leaves an

    effective door width of 0.7m

    From table 3-14.5 (for 6.5" riser, 12" tread)

    Maximum specific flow (stairways) =

    F

    sm

    = 1.09

    persons/s/m effective width

    Specific flow equals maximum specific flow (as noted above in step 1)

    Therefore using equation 6: the flow from each stairway is limited to;

    Fc = Fs x We Fc = calculated flow

    Fs = specific flow

    We = effective width

    Fc = 1.09 x 0.9 = 0.98 persons per second

    =

    58 persons per minute for the stairway

    Step 3

    From table 3-14.5 The

    maximum specific flow through a door

    is 1.3 persons/s/m of effective width.

    = 0.91 persons per second

    =

    54 persons per minute for a door

    (gained following similar process as step 2)

    Since flow capacity of door is greater than the flow capacity of the stairway which serves it: flow is controlled by the stairways at the rate of 54 persons/stairway exit door/minute.

    Step 4

    Estimated Speed of movement;

    Using equation 3

    S = k – akD S = speed along the line of travel

    D = density in persons per unit of area

    k = 1.16, constant from 3-14.2 (for 6.5" riser, 12" tread)

    a = 0.266

    S = 1.16 – (0.266 X 1.16 X 1.9)

    = 0.574 ms-1 (X60)

    =

    34.44 metres/minute

    Using conversion factor from table 3-14.3: travel distance between floors is:

    Floor to floor height = 3.7m (taken from section drawings)

    conversion factor (for 6.5" riser, 12" tread) = 2.08

    Floor to Floor height X conversion factor = 3.7 X 2.08 = 7.7m on the slope of the stairs plus an additional 5m travel on the landings

    Therefore total travel distance between floors = 7.7 + 5 =

    12.7m

    Therefore the travel time for a person moving with the flow = travel distance / travel speed

    = 12.7 / 34.42

    =

    0.37 minutes per floor

    Step 5

    Estimate building evacuation time with two available staircases.

    If all occupants begin to evacuation at a similar time each of the two available stairways can discharge 58 persons per minute. (Calculated above in from step 3)

    The population above the ground floor (2542) will require:

    2542 / 116 (58 persons x 2 stairs) =

    21.9 minutes

    to pass through the exit.

    Request Removal

    If you are the original writer of this and no longer wish to have the published on the UK Essays website then please click on the link below to request removal:


    More from UK Essays