Research and analysis

Fire safety: Means of escape for disabled people (executive summary)

Published 22 December 2025

Applies to England

Background

The work reported in this report was carried out by a BRE Global Project team under a Contract placed by the Department for Levelling Up, Housing and Communities. Any views expressed are not necessarily those of the Department for Levelling Up, Housing and Communities.

This final report is delivered as part of the Department for Levelling Up, Housing and Communities (DLUHC) (formerly the Ministry of Housing, Communities and Local Government (MHCLG)) project titled “Fire Safety: Means of escape for disabled people”, DLUHC Contract reference CPD/004/120/201.

Executive summary

The aims of the project were to assess the current provisions in Approved Document B (AD B) and review alternative approaches in providing means of escape for disabled people to ensure Approved Document B provides adequate guidance to meet the minimum requirements under Schedule 1 of Part B1 of the Building Regulations 2010.   

It should be noted that the edition of Approved Document B referred to throughout this project is the 2019 edition.  Where reference is made to current provisions within this report it is with reference to this edition of AD B. 

This project had the following specific objectives:

• To review types of disability or impairment, and their effect on evacuation (both for the person with an impairment and also any impact on other evacuees), plus to review dimensions of evacuation aids, and whether these might introduce any additional risks within a building.  Also, to carry out a scoping study on alternative means of provision for means of escape.
• To review the current regulatory approach and associated guidance, with an emphasis on how this works in practice from the perspective of building occupants and managers as well as building designers and fire engineers.
• To carry out a more detailed review and analysis of potential alternative means of escape, and in particular, the use of evacuation lifts, with the aim of providing evidence for their effectiveness.

The last (third) objective was revised during the course of the project, as follows:

• Research into the effectiveness of evacuation lifts, refuges, and evacuation chairs
• Further research into the behaviour of disabled people (particularly people with neurological  conditions) during an emergency evacuation
• Evaluate costs and benefits of evacuation lifts, refuges, and evacuation chairs

The scope of buildings covered by this project are specifically those to which AD B applies; with the exception of blocks of flats, care homes and specialised housing (all of which are covered in separate DLUHC projects), uncommon buildings (for example, fire engineered) and those with alternative guidance (such as hospitals and healthcare, and schools).  The scope of buildings did not exclude dwellinghouses, but these were not the focus of this project.  Other means of increasing safety, for example the provision of sprinklers or other automatic fire suppression which would reduce the need for escape, or increase the time available, were outside the scope of this project. 

The BRE Global Project team comprised BRE Global and Project Partners, Ulster University, University of Edinburgh and a Private Disability consultant. 

This work involved the active participation of a Project Technical Steering Group of industry experts.

The individual tasks of this project are summarised as follows.

Literature review

The literature review covered the prevalence of disability and current demographic trends and attempted to understand patterns of use of buildings by disabled people.  It considered experiences from real evacuation situations and other related research studies which have focused on human factors associated with the use of refuges and current options for vertical evacuation, that is, assisted escape and use of lifts.  Current understanding of factors that influence the flow dynamics of building occupants, including unimpeded walking speed, space requirements, and individual and group behaviours were also reviewed.

The literature review examined statistics on the prevalence of different types of disability collected for the UK as a whole.  Approximately 20% to 25% of the population have one or more disabilities, which are defined as having some impact on day to day living.  However, this may not necessarily impact on their ability to evacuate in an emergency. 

The most prevalent forms of disability are physical and affect mobility.  About 10% of the overall population have a mobility impairment, and 7.5% have problems with either stamina or breathing difficulties.  These percentages include people with multiple disabilities.  It has been estimated that 0.5% of the population find stairs impossible, and approximately 5% experience at least some difficulty in using stairs.  The percentage of wheelchair users in the overall population is 1.8%, of whom 1.2% are regular users.  Users of mobility scooters account for 0.5% of the overall population, but this fraction is rising. 

There is clear evidence that there are people with multiple impairments within the population of disabled people; however, the statistics published online do not provide details on the relative likelihoods of different combinations.

Statistics or estimates for the numbers of (non-disabled) people with impairments such as injuries (for example a broken leg), pregnant women, family groups incorporating young children/baby buggies, etc, were not found by the literature review (with some isolated exceptions). 

The numbers of people with different disabilities or impairments may vary with building type, such as groups with young children may be common in a supermarket, but not in an office building.  However, the literature review was unable to explore this point due to insufficient data.  There is evidence that disabled people and their carers visit (particularly) public buildings in groups, and are actively encouraged to do so (for example, particular performances for groups of disabled people, free entry to carers).

The demographics for ageing or obesity suggests that the percentage of disabled people will increase, with Public Health England predicting that in 2050, 60% of men and 50% of women will be obese. 

There is plenty of data (both experimental and from real-life evacuations) on movement speeds and flow rates, both for disabled people and non-disabled people.  A key finding from the literature was that movement speeds for disabled people are reduced, but also increased space requirements of disabled people (including space for their assistors), which make overtaking difficult if not impossible in restricted environments such as corridors, and different behaviour characteristics mean that flow dynamics of diverse occupants are very different from homogeneous occupants and from what is implicitly assumed in guidance such as AD B, BS 9999, etc.

However, there is much less data available for the impacts of other types of disability, compared to mobility, on evacuation capabilities.  There are issues of wayfinding for people with visual impairments, and the capabilities of neurodivergent people with information and sensory processing differences are almost unknown, at least within the field of human behaviour in fire.  Further research is needed in this area, particularly in the latter issues.

The literature review of real-life emergency evacuations has shown a number of problems regarding the use of refuges.  There are many people (not only wheelchair users and people accompanying them) who may wish to use a refuge, either to wait for assistance, or a place where people may wait for short periods before resuming their evacuation without waiting for assistance.  The minimum recommendation for a refuge is an area measuring 900mm x 1400mm (generally sufficient for a single wheelchair user) on a landing in a protected staircase, which may result in building occupants being unable to access a place of relative safety and be exposed to the growing fire hazard if more people want to use the same refuge simultaneously.  The number/size of refuges, and their locations, has also been questioned elsewhere, for example, focus Group meetings and MHCLG consultation responses.

Real-life experience also shows that refuge areas often have poor communications with building management for various reasons for example inadequate provisions, conflicting provisions such as loud alarms in the vicinity, or poor staff procedures/training.  The literature also suggests wider issues with management communication, irrespective of the physical means, and management generally regarding assisted escape.  Evacuation chairs are not suitable for all disabled people.  Evacuation lifts offer a potential solution; there may need to be a change in people’s behaviour to encourage their use (standard messaging is not to use lifts in the event of fire), although perhaps not that much of a change since for many disabled people it would be their preferred option.  Personal emergency evacuation plans (PEEPs) for disabled people can encourage the use of evacuation lifts.

Focus groups

Evidence was obtained directly from disabled people on their experience of the current arrangements for their protection and escape, in terms of building design, management procedures and additional aids and equipment.  Eleven user experts with various impairments or from organisations representing particular impairment groups participated in the Expert User Focus Group sessions.  The Expert User participants represented people with mobility and sensory impairments, neurological conditions, mental health conditions and dementia.  

The Expert Users were also asked how the current regulations and guidance affect the performance of the building in practice.

Two additional Focus Group meetings were arranged for a total of nineteen building professionals, including designers, fire engineers, regulators, managers and others.

Each Focus Group session was facilitated by following an agenda, guided by the list of questions and taking a number of snapshot polls to establish how participants felt about the current position regarding building design, management arrangements and AD B provisions.

There was generally a consensus among all the Focus Groups’ members regarding the requirements of disabled people, and that AD B did not go far enough to address these.  The key points that arose were:

• Further research was needed on disability and the requirements of disabled people, particularly neurodivergent people;

• AD B should be fully inclusive and not only cover obvious impairments to ensure safe, easy and dignified evacuation for disabled people;

• Better guidance was required on how to design buildings to better serve the requirements of a wider range of people;

• Too often, problems with the building design were left to management to address which resulted in more complex arrangements being necessary and often resulting in inadequate solutions;

• Staircases, steps, mezzanines are building features/challenges which make fire evacuation impossible or very difficult for older or physically disabled people;

• Refuges in general are not being designed with sufficient consideration to likely user requirements.  Existing guidance (in AD B and elsewhere) on calculating the number and sizes of refuges was considered inadequate; more research and guidance is needed;

• Evacuation chairs had many problems, and in the absence of evacuation lifts, it was felt that the requirements of people who could not or would not use evacuation chairs were not met;

• Wider use of evacuation lifts would enable more people to evacuate easily, safely and with dignity.  Some Focus Group members suggested AD B should include a “requirement” (recommendation) that all lifts should be usable in the event of an emergency.

Interviews with experts in neurological and mental health conditions

The BRE Global Project team held unstructured one to one meetings with selected individual Focus and Technical Steering Group members to discuss aspects of neurological and mental health conditions with an emphasis on the following:

• how these might affect people’s behaviour in a fire emergency evacuation;

• whether there was any data available to inform the modelling activity;

• whether the interviewees were aware of any academic literature sources that could be a starting point for further study.

Attempts were made to incorporate the findings from these interviews, regarding the behaviour of neurodivergent people, within the simple evacuation model described later.  This involved a gross simplification of an extremely complex set of issues, although the current ‘state of the art’ does not permit much more than this.

Review of sizes of wheelchairs and other movement aids

In a previous BRE Global project for the Department for Communities and Local Government (DCLG) completed in 2015, data was collected relating to the distribution of the sizes of different types of wheelchairs.  An updated review of the distribution of the sizes of different types of wheelchair and other mobility aids was carried out.

The review suggested that wheelchair size is tending to increase, compared to previous years.  One driver behind this may be the increase in obesity in the population.  This means that an increasing proportion will be too large to fit in a minimum-sized refuge space.

Review of fire hazards of mobility aids

Previous BRE Global research on mobility aids such as powered wheelchairs or mobility scooters and additional/associated fire risk was reviewed and summarised.  Some real-life fires involving mobility scooters were also described, along with appraisals of the risk from various organisations, and recommendations for the storage of mobility scooters within buildings.

The review showed that mobility scooters had the potential to cause relatively large fires.  A number of these occurred during the charging of the Lithium-ion batteries used to power the scooters.  If the fire occurred in the common parts of a residential building, for example, if a mobility scooter is left in the corridor outside the front door of an apartment, it could quickly render escape routes impassable due to smokelogging.  Further research is needed in this area to inform consideration of Lithium-iIon battery storage and charging.  

Although not specifically covered by the review, it should also be noted that scooters in corridors would cause physical obstructions even if they were not the source of the fire.

Review of international approaches

Various international approaches to regulations and guidance on means of escape for disabled people were identified, but not reviewed in detail.  Note that this list was not comprehensive and did not include all of the international regulations and guidance related to means of escape for disabled people.

Specific information about approaches to regulation and guidance on means of escape for disabled people or people with impairments in different countries, was requested via the International FORUM of Fire Research Directors.  This was intended to identify alternatives to the AD B guidance that would be worth considering.  Seven responses were received, from Austria, France, Germany, Italy, Norway, Sweden and the USA.  The information received was collated and reviewed.    

In all these countries, the national fire code includes specific regulations or guidance to make specific provisions for the means of escape for disabled people or people with impairments.  The provisions mentioned most frequently were:

• Building management (all seven countries)
• Refuges in stairwells (six countries)
• Refuges in other locations (five countries)
• Use of adjoining compartments as refuges (six countries)
• Two-way voice communication between building managers and refuge occupants (five countries)
• Evacuation aids, such as evacuation chairs (six countries)
• Evacuation lifts (five countries)

Two countries’ guidance mentioned all of the specific provisions for means of escape listed in the questionnaire.  However, the questionnaire responses did not give reasons why particular countries mentioned (or omitted mention) of particular provisions.

All seven countries specify an overall strategy for the means of escape for disabled people or people with impairments.  However, there is no uniform approach among the different regulations regarding specific strategies.  In the majority of the countries under consideration, it has been pointed out that the Fire and Rescue Services are expected to rescue disabled people (unlike in the UK, where this is the responsibility of the building managers).

All seven countries have special regulations or guidance, according to the building type (one country, however, did not give any examples of specific types).  The specific building types vary from country to country and there is no uniform approach.

Most of these countries considered that their fire safety regulations or guidance enable disabled people or people with impairments to “escape easily, safely and with dignity”.  However, as this terminology was not defined in the questionnaire, the responses may be somewhat subjective.  Note that other countries’ assertion that their codes enable people to escape easily, safely and with dignity is at odds with what the Focus Groups and literature review suggest with respect to UK guidance. 

From the international survey, it did not appear that other countries have a radically different approach to current AD B guidance for England.  The project therefore did not attempt to gather further details from other countries’ approaches.

Review of information on different technological solutions

Information on various technological solutions to fire safety and evacuation of disabled people, or people with short-term impairments, was gathered and reviewed.  The key solutions that were investigated in depth were evacuation lifts, evacuation chairs and refuge areas.  

Other solutions identified in the scoping study, including innovative means of raising the alarm, enabling wayfinding or vertical movement, were not reviewed further.  BRE Global’s view is that innovative approaches would be part of a bespoke fire-engineered solution, rather than something suitable for guidance in AD B which has to cover a broad range of common building situations.

The current UK guidance for evacuation lifts is contained in BS 9999:2017 Fire safety in the design, management and use of buildings. Code of practice, although DCLG has also published guidance (BD 2466) intended to supplement AD B.  CIBSE Guide D Transportation systems in buildings also contains much useful information on lift design and operation, although it tends to repeat BS 9999:2017 as far as guidance on evacuation lifts is concerned.  Draft prEN 81-76 Safety rules for the construction and installation of lifts - Particular applications for passengers and goods passenger lifts Part 76: Evacuation of persons with disabilities using lifts may, once approved, become the published British/European standard for such lifts in the near future, although it is not yet considered acceptable for use in the UK since it is currently a draft.  The main difference between prEN 81-76 and BS 9999:2017 is that the former introduces additional operation modes (automatic and remote control) to the driver-assisted mode assumed in the latter.

BS 9999 states that “A test evacuation, in which people who require help are assisted to a place of ultimate safety, should be carried out at least once a year and should be both horizontal and vertical”. However, many people feel the risk involved in procedures using evacuation chairs or similar aids should not be taken until an emergency demands it.  Guidance is lacking regarding the procedure for use of evacuation aids, and the transfer of people, for example from their own wheelchair, to such evacuation aids.  It is essential that evacuation assistants have prior training and experience and a good knowledge of the building, and that sufficient trained assistants will always be present when needed.  Training needs to be refreshed regularly.

Movement speed on stairs varies depending on the nature of the evacuation aid.  The space required for effective use also varies considerably; in some cases, the evacuation assistants will need to stop to allow other stair users to overtake.  If the assistants have to make repeated trips to rescue people (the alternative being provision of sufficient evacuation aids and assistants to avoid this) then fatigue is likely to reduce the travel speed; there is also the problem of contraflows as the assistants return to rescue others.

In addition to the review of the effectiveness of these measures, and any problems or issues associated with them, the cost information derived from a previous research project by BRE Global for DCLG has been included in this report for ease of reference.  These costs have not been updated to 2023 values.  In any case, further work is recommended to establish more accurate, up-to-date costs.

Review of responses to MHCLG consultation

The 92 detailed responses to the government’s consultation exercise on AD B in 2019 that were relevant to the provision of means of escape for disabled people and people with impairments were reviewed.  

Some identical responses were received from different organisations and some responses were only relevant to flats or care homes (which are outside the scope of this project).  Therefore, 56 responses were reviewed in detail.  From analysis of these, it has been identified where there are gaps in knowledge or guidance.

As with the Focus Group responses, problems with current approaches (particularly management, provision of adequate refuges and the appropriate use of suitable evacuation chairs) were highlighted, and evacuation lifts suggested as a solution to overcome these.

Review of current regulations and guidance

Current regulations and guidance were reviewed, not only for means of escape from fire (such as AD B, Fire Safety Risk Assessment Supplementary Guide Means of Escape for Disabled People (Regulatory Reform (Fire Safety) Order (RR(FS)O) Guide for disabled people), and BS 9999:2017), but also other documents including Approved Document M Access to and use of buildings (2015), and the Equality Act 2010.  

Approved Document B (2019) contains little in terms of guidance on building design or management specifically addressing means of escape for disabled people.  In paragraph 0.8, there is mention of Inclusive Design and sections on Detection and Alarm, Emergency Lighting, Refuges and Management.  In common with the approach throughout AD B, these sections refer the reader to other documents for guidance (for example British Standards for alarms and lighting, BS 9999 for design, management and use of buildings).  In order to improve AD B, it may be necessary to also revise the cited documents to produce a consistent and coherent set of guidance that covers any identified gaps.

The AD B guidance (AD B sections 3.4 to 3.9) with respect to the number and sizing of refuges is questioned in light of the literature review and Focus Groups’ feedback on current demographic trends, use of buildings, potential users of refuges and implications of inappropriate sizing.  The suggestion in AD B clause 3.4c that the number of refuge spaces does not need to equal the number of wheelchair users who may be in a building since a single refuge may be occupied by more than one person during the evacuation procedure should be questioned also in light of literature review highlighting reliance on management and resources required for assisted escape.  The idea that one user will enter a refuge as another leaves, is based on the assumption that the first user will be efficiently removed from the refuge before the other user needs to use it to remain safe, for example if the fire is on that floor.

AD B mentions that evacuation lifts may (not should) be used, and again refers to BS 9999 for further details.

The main issues with regard to AD B that have come out of the reviews above are:

• Lack of knowledge by building designers, managers and regulators, of disability and the requirements of disabled people, particularly people with neurological or mental health conditions;

• Insufficient guidance on building design to address the requirements of all users (inclusive design). This project focuses on the need to provide suitable means of escape, although the term “inclusive design” is much broader than this.

• Lack of step-free evacuation options

• The need for further research (One Technical Steering Group member believes there is already sufficient information for improved guidance on refuges without the need for further research.) and improved guidance on provision (size and numbers) of refuges;

• Problems with the use of evacuation chairs;

• Insufficient guidance on building management to address the requirements of all users;

• Use of evacuation lifts as a potential solution.

As examples of the above:

• The first two points are closely linked.  Approved Document B has a short paragraph on Inclusive Design (paragraph 0.8), but this does no more than state some very general principles.  In order to actually undertake such a design, more guidance would be needed.  There is a lack of such guidance in a single cohesive document that AD B can refer to.  At the very least, Approved Document M should be mentioned at this point.  Note that Sport England has some helpful management guidance under development, which builds on the approach taken in the RR(FS)O Guide for disabled people on PEEPs.

• AD B covers provision of refuges (paragraphs 3.4 to 3.9 and diagrams 3.1 and  3.2), but the number and/or size of these may be insufficient (the largest wheelchairs may exceed the minimum size of a refuge area, and such a refuge could certainly not accommodate two at once).  By linking the refuge size to a wheelchair, this may lead to misconceptions that refuges are only for wheelchair users.

• Provision of communication systems for refuges (paragraph 3.7) may be insufficient (as one Focus Group participant stated, “if the comms link is to the reception desk, what happens when reception closes at 5pm?”), although this may be more due to failings in building management rather than AD B per se.  However, it does raise the issue that AD B should encourage building design that does not rely on management to fix problems.

• Relying on the use of evacuation chairs presents many challenges, for example, some people are unable or unwilling to be moved in this way, and they are not suitable for some steps. Evacuation chairs are not mentioned in AD B (it would be a management decision to provide them and ensure sufficient trained operators); however, they are mentioned in BS 9999: 2017, which is referred to by AD B.

• Evacuation lifts are mentioned by AD B (paragraph 5.32, and reference to BS 9999: 2017), although the tone is permissive rather than encouraging designers to adopt them.  There are no recommendations on the number of lifts that might be needed for a particular population of building occupants, for example.  Instead, the management plan is left to make the best use of such facilities that are available. The type of evacuation lifts described in BS 9999 are suitable for the assisted evacuation of a building since the lift described needs to be controlled by a lift driver to go to the floors from which people are awaiting evacuation.  However, there may be problems if buildings do not have in place trained people who can take control of the lift and manage the evacuation of people unable to use the stairs.

• Building management is covered by a single paragraph (paragraph 0.6) and then sporadically mentioned elsewhere.  There is reference to the Regulatory Reform (Fire Safety) Order 2005, but no reference to BS 9999 in this context, despite the Standard covering design, management and use of buildings.

Simple evacuation simulation including disabled people

Modelling and analysis were undertaken to compare the effectiveness of the provision of lifts for evacuation (including upgrading existing lifts in refurbished buildings) versus assisted vertical evacuation (refuges and evacuation chairs; the current AD B approach).  This included the development and application of a simple evacuation model, “Evac-flo”.  This model has a simple network/node representation of the building geometry, where the nodes are building floors, stair flights, or lift cars.  The network of links joining the nodes represent doors and passages.  People are represented individually.  Refuges, and the use of evacuation aids such as evacuation chairs, are represented implicitly.

Additional models, Pathfinder and Elevate, were used in order to calibrate/validate aspects of the Evac-flo model.  The Pathfinder model was used to investigate the movement of a heterogenous population of building occupants (faster walkers and slower wheelchair users) within a corridor.  However, the attempts to replicate experimental data were inconclusive.  Therefore, the Evac-flo model neglects this interaction, on the assumption that the key limiting factor for evacuation time would be the flow rate through storey exits rather than within corridors before the storey exits.  People pass through storey exits in the same order in which they join the queue. A similar approach applies to movement on stairs; slower-moving people may be overtaken by faster moving people, but once a person joins a queue (for example to move to the next flight of stairs down) the order of the people leaving the queue is the same as the order in which they join.

The Elevate model was used to attempt to validate the lift movement algorithms within Evac-flo.  The agreement between the models is considered satisfactory.  However, the comparison between the two models has not considered certain kinematic factors (acceleration and jerk), nor various door opening and closing delays, which would cause the Evac-flo model to underestimate the lift round trip time and therefore might merit further work to investigate and refine the Evac-flo model if necessary (these factors were also disabled in the Elevate model in order to compare the results with those from Evac-flo).  As detailed in this report, the various simplifications in Evac-flo are likely to result in consistent underestimates of the time taken to evacuate disabled people using lifts, and the results need to be interpreted with this caveat in mind.

The Evac-flo model was used to examine over 100 different scenarios, with different Purpose Groups, building height and size, and varying populations of building occupants.  The Purpose Groups examined were 2b (Hotel), 3 (Office) and 4 (Department store).  Other Purpose Groups were initially considered but ultimately not included in the study, because the chosen groups covered a sufficiently wide range of possibilities.  For example, an exhibition hall (Purpose Group 5) could have similar layout, number of people, proportion of disabled people and design behavioural scenario as a department store (Purpose Group 4) but Purpose Group 5 would only use simultaneous evacuation whereas the department store might also have phased evacuation as an option.  Hence, by modelling Purpose Group 4, it was not necessary to model Purpose Group 5 as well.

The Evac-flo model was deliberately kept as simple as possible in order to facilitate the examination of many different scenarios.  The detailed findings have to be treated with some caution as a number of subjective assumptions had to be made during the development of the model (due to a lack of suitable data).  Further work could be undertaken in order to improve the robustness of these findings.

Evacuation times

Transit times (after passing the storey exit) for people using stairs or lifts seem to be broadly similar.  This was as expected since the capacity of the evacuation lift was designed (using simple formulae for lift and stair evacuation times) to achieve this.  One or two evacuation lifts were required depending on the stair capacity.

The evacuation time is correlated with the predicted duration of stair movement (the dominant component of evacuation time in offices and department stores) but is longer for hotels, due to the longer time required for the occupants to react to the alarm.

The effect of the ratio of building occupants per floor to the standing capacity of a stair flight (including landings, but not lobby, if present) was investigated.  It was expected that the higher the value of this ratio, the more likely stairs were to become full to capacity (and therefore people could be unable to pass the storey exit in order to reach the stairs).  However, the results do not show a clear trend, other than that higher ratios tend to be associated with higher predicted times (using simple formulae) for the passage of the storey exit anyway (which are proportional to the number of occupants using the exit). 

In the simulation, the average waiting time for people using lifts is defined as the time to move off the level the occupants were picked up from, minus the time the occupant passed the storey exit and entered the lift lobby.  This allows the definition of an equivalent waiting time for stair users, that is, the time to move to a different level on the stairs, after having passed the storey exit threshold.  

The average waiting time is correlated with the number of building occupants per stair (although a proportion of these people will use lifts).  With stairs, there is a tendency (particularly for those on higher floors) to become full and people are forced to wait until the stairs below become less crowded so that there is space to move into them.  With lifts, the control algorithm prioritises calls from the highest levels, so conversely occupants at low level may have a long wait before the lift responds to their call.

Lift behaviour

The number of trips made by the lift will be determined by its capacity and the number of people waiting to use it (which will be a fraction (about 50%) of the total number of disabled people within the building).  With offices and retail, this correlation is fairly clear, but much less so in hotels.  This is probably due to the long reaction times before hotel occupants start to move, therefore randomising the number of people who might be waiting for the lift each time it makes a trip.

For offices and retail, the number of pickup stops per round trip does not appear to be affected by building height.  The occupants react quickly so there will be queues waiting on every floor.  On some trips, the lift may be full after the first pickup, whereas on others there may be free space making a second pickup possible.  With hotels, there is a correlation with building height; this is due to the long reaction times of the occupants which make it more likely that there will be people waiting the taller the building (and hence the total number of occupants) each time the lift makes a round trip.

In the model, the average round trip time depends on the height of the building and the speed of the lift, and also its average loading (since the time to embark and disembark is proportional to the number of passengers). 

The reason that the lift does not achieve its nominal capacity is due to the average floor space requirement of disabled people being larger than the “standard body” (considerably larger in the case of wheelchair users) therefore the lift is full in terms of floor space occupied although carrying fewer people than the nominal capacity (which is based on the weight of a “standard body”, 75 kg).

The model assumed the lift would not make further stops to pick up passengers once it was full (in terms of occupied space).  If not under direct control of a driver (this is current practice as per BS 9999: 2017), this would require video observation of the lift car interior, and in the case of automatic operation, some means of analysing/interpreting the video picture (not just counting heads) to determine whether the lift was full. 

Alternatively, in automatic evacuation mode, for low-rise and “simple” systems the lift could go directly from the upper floor to the exit floor, or for high-rise and more complex situations a different weight threshold (say 20%) could be used so that the lift would only stop at an intermediate floor if the weight were less than that threshold.  However, these suggestions were not modelled in this project. 

It has been assumed that the probability of choosing the lift was constant, rather than people on higher floors being more likely to use the lifts.  Research investigating the use of lifts to evacuate all building occupants (not only disabled people) suggests that the proportion of building occupants wishing to use the lift would increase with building height.  This would increase the lift round trip time, by how much would depend on how strongly the probability depends on the building height, and how tall the building is. 

It has been suggested that lifts could be used to preferentially evacuate the fire floor and its vicinity.  This would affect the order of evacuation, but not the overall time (assuming the whole building is evacuated).  It makes sense to do this if possible (but not in a building such as a hotel, where the long times for people to wake up and respond to an alarm would make this approach very inefficient).  During the early stages of evacuation, the lift might be operating in automatic mode, so would need a suitable interface with the Building Management System to know which floor to evacuate first.

Refuges

The simulation keeps track of the number of runs where the floor area required for resting people on a stair exceeds the refuge area.  The number of people needing to stop in a refuge is expected to be proportional to the number of disabled people within the building.  If the building has lifts, it was assumed that wheelchair users will use those rather than wait in a refuge until they are rescued.

When the building has lifts, the size of the lobby in the model was sufficient to accommodate all the people waiting without becoming overloaded.  When the building does not have lifts, the model predicts that 100% of runs will see refuges on stair landings being overloaded, at least for a short period.  In some cases, the overloading might be sufficiently severe for the flow on the stairs to be impeded.  However, hotels are likely to be an exception, where long reaction times lead to very low densities of people using the stair at any time, so making it less likely that flow would be impeded by more people waiting on the landing.

The finding that refuge size may be insufficient contradicts earlier research and it is not entirely clear why. The earlier research may have underestimated the proportion of the building occupants needing to use a refuge (if it assumed they were only used by wheelchair users), or may have looked at scenarios where the stairs were less heavily loaded than in the current research.

There is a need to investigate the severity of overcrowding, and the duration.  Some indicative results have been obtained for the effect of a bottleneck on the overall flow in a single example (a 10-storey office).  In this example, provided that the stair landing can accommodate at least three times the nominal capacity of a minimum-sized AD B refuge before the flow down the stairs is impeded, the impact of overcrowding on total evacuation time would be negligible.  However, this finding is not necessarily applicable to other scenarios.

Trips and falls

The relative probability of a person tripping (and possibly falling) is assumed to be a function of their disability type, and the number of flights of stairs that they traverse.  The relative risk, divided by the number of disabled people per storey, shows a clear quadratic dependence on building height as expected.  Note the different relative risk for buildings with lifts, or with stairs only.  The latter have higher relative risk because a number of disabled people are forced to use stairs (with their risk of trips and falls) rather than the lift (with no risk).

Based on statistics, the actual risk of trips and falls during an evacuation is likely to be low.  Consequently, the impact of trips and falls has not been modelled.

Evacuation chairs

A proportion of wheelchair users are assumed to be unable to transfer to an evacuation chair.  The simulation, for this study, does not model use of alternative techniques or powered evacuation aids.  Hence, if the building does not have a lift, people who cannot evacuate without assistance and who cannot transfer to an evacuation chair are assumed (in the model) to remain indefinitely within the building refuge areas.  The number of people affected in this way is proportional to the total number of disabled people within the building.

It was assumed that evacuation chairs, which are not particularly wide, would work on all stairs.  Other evacuation equipment, such as powered stair climbers or descenders, would need greater width to operate.

It was also assumed that evacuation chairs would be sufficiently narrow for other, faster-moving, evacuees to overtake on both stairs and landings, without the evacuation chair needing to stop to allow them to pass.

It must be reiterated that in the interests of developing a simple model, many of the “real world” problems associated with evacuation chairs and other similar evacuation aids have not been included.  It should not be implied from this approach that such problems are negligible.  Rather, the decision was taken that the problems are “known”, and it would have required considerable effort (beyond available project resources) to replicate these explicitly within the model.

A number of assumptions in the model regarding the use of evacuation aids were very optimistic, for example that there would be “sufficient” aids and people trained in the use to evacuate anyone who needed such assistance, without each aid and operating team having to make multiple trips.  Fatigue and contraflow modelling would have been required without this assumption.

Conclusions from the modelling study

The modelling was intended to examine the effectiveness of evacuation lifts, refuges, and evacuation chairs.  The modelling is possibly too simplistic to enable quantitative conclusions to be drawn about the relative performance of evacuation lifts versus assisted vertical evacuation (refuges and teams of assistants with evacuation chairs and similar aids).  On the one hand, the simplified modelling of lift kinematics, and the assumption that on upper floors of high-rise buildings had the same probability of choosing lifts as people on lower floors, tended to underestimate the time required for evacuation by lift.  On the other hand, many of the complexities (such as the need for sufficient aids and staff to use them, contraflows and fatigue if insufficient staff, some people unable or unwilling to use chairs) of evacuating people using evacuation chairs could not be modelled simply.

Results from this current research suggest that refuge sizes (in buildings without lifts) may not be sufficient, due to the number of people wishing to use them.  The AD B guidance on provision of refuges (associated with stairs) should consider the number of people expected to use each stair.  In addition, the proportion of all people who need assistance to escape from a refuge (which may depend on Purpose Group) should be factored in.

If evacuation lifts are provided, the lift lobbies may serve as refuge areas.  The lobbies would tend to be larger than the minimum refuge size specified in AD B, thus reducing the potential for overcrowding.

The AD B guidance if evacuation lifts are provided should reflect the expected numbers of people using the stair or lift(s) from a given storey exit (and account for discounting should fire make one exit inaccessible).  Simple formulae, or tabulation of values, would probably suffice for most cases where AD B is likely to be applied, although further work is required to determine what the values should be.

The modelling has assumed that the current “baseline” solution is to provide refuges and use evacuation chairs or similar aids for people unable to use the stairs.  However, this should not be construed as recommending evacuation chairs per se, because they are inappropriate or do not provide a sufficiently safe evacuation method for a lot of people.  Furthermore, the model is simple and therefore does not model the complex cases where it is not possible to get someone out using a standard evacuation chair.  In the model, people in such cases are considered to remain within the building because the model cannot simulate their evacuation.  It would not be acceptable for such people to remain indefinitely within the building in real life.

Further work

Additional suggestions for further work are also included in this report.

Some recommendations for changes to AD B are made, with regard to inclusive design, management, evacuation lifts, and provision of refuges. 

The Greater London Authority London Plan 2021 is part of the statutory development plan for London and provides high-level guidance on a number of issues including Inclusive Design and Fire Safety.  One recommendation is that in all developments where lifts are installed, as a minimum at least one lift per core (or more subject to capacity assessments) should be a suitably sized fire evacuation lift suitable to be used to evacuate people who require level access from the building.  DLUHC should consider whether AD B should also include similar guidance.