Quick start

1. Pick a scenario in the Setup panel on the left. The hall resets to that scenario’s start time.
2. Press Run in the bar at the bottom (or the Space bar). Press it again to pause.
3. Choose a speed. 60× turns an hour of the day into one minute. 1× is real time, useful for watching a single doorway or queue.
4. Watch the Live panel on the right. When a scenario ends, a summary appears at the top of the hall.
5. Change one setting, such as blocking an exit, then run again with the same run number to compare like with like.

On a phone or narrow window the two panels are hidden. Open them with Setup and Live in the bottom bar.

Scenarios

Typical day

Runs from 08:30 to 18:45. Early arrivals wait in the forecourt until the doors open at 09:00. Arrivals peak around 09:30 to 10:00, lunch peaks between 11:45 and 14:15, and the closing call is at 17:45. The run ends when the hall is empty.

Start of day

Starts at 08:40 with a crowd already gathering outside both main entrances. The doors open at 09:00, and you see the surge through the doors and into the info desks and coffee kiosks. The run ends at 10:45.

End of day

Starts at 17:15 with the hall as full as it would be on a typical day at that time. The closing call is at 17:30, after which everyone heads for the two main entrances. The emergency exits are not used. The summary tells you how long the hall took to clear.

Evacuation

Starts one minute before the alarm, with the hall filled exactly as it would be on a typical day at the chosen time: 10:30 (morning peak), 12:45 (lunch rush, with people in queues and at tables) or 15:30 (afternoon). When the alarm sounds, everyone uses the nearest open exit, including the emergency exits. The summary gives the total evacuation time.

You can also press Sound the alarm now at any moment in any scenario, for example in the middle of the start-of-day surge.

Setting up a run

Changing the scenario, visitor numbers, visitor mix, staff or run number resets the hall so the new settings apply from the start. Blocking exits and the alarm reaction time take effect straight away without a reset.

Crowd

Visitors today

The total number of visitors who come during the whole day, from 500 to 5,000. Not everyone is inside at once: on a typical day the peak is about three quarters of this figure. Exhibitor staff are added on top.

Average alarm reaction

The average delay between the alarm sounding and a person starting to move, sometimes called pre-movement time. Real people finish a conversation, gather their things or wait to see what others do. Each person gets their own delay around this average (see How the model works).

Run number

Fixes the random choices behind a run: who arrives when, which persona they are, which stands they like. The same run number with the same settings gives the same result every time. New run picks a fresh number so you can check whether a result holds up with a different crowd.

Visitor mix

The five sliders set the relative share of each visitor persona. They don’t have to add up to 100; the percentages beside them show the resulting shares. The presets give quick starting points:

• Balanced: a general industry show.
• Trade day: mostly buyers and press, with more staff on each stand.
• Public day: mostly browsers and students.
• Many with reduced mobility: about a third of visitors move slowly. Use it to test the evacuation plan against a harder case.

Exhibitor staff sets how many staff work on each of the 30 stands, from 0 to 4.

Exits

Each green sign is one exit: the two main entrances (Main W and Main E) on the south side, and six emergency exits, W1 and W2 on the west wall, E1 and E2 on the east wall, N1 and N2 on the north wall. Select one to block it, for example to represent smoke, a fire near the door or a locked gate. It turns red in the panel and in the hall. Everyone’s routes are recalculated at once, including people already evacuating, who turn round and head for the next nearest exit. Select it again to reopen it. Blocked exits stay blocked when you reset, so you can compare scenarios with the same exits closed.

View

Crowd density heatmap

Colours the floor by how many people are standing in each 2 m square: green is comfortable, yellow about 1 person per m², orange about 2, red about 3, and deep magenta 4 or more.

Labels and signs

Shows or hides the floating labels over stands, counters and exits.

Colour people by

Activity colours people by what they are doing: walking, at a stand, queueing, eating or resting, waiting outside, going home, reacting to the alarm, or evacuating. Persona colours them by visitor type. The key at the bottom of the Live panel always matches the current choice.

Camera

Angled gives a three-quarter view from the entrances. Plan looks straight down, which is best for judging distances and the heatmap.

Moving around the hall

• Rotate: drag with the left mouse button, or one finger.
• Pan: drag with the right mouse button, or two fingers.
• Zoom: scroll, or pinch.
• Hover over a person to see their persona, what they are doing (for example “Queueing at Grill, position 4”), their walking pace and, for buyers, how many shortlisted stands they still have to see.
• Hover over a stand to see how many people are there now and how many visits it has had.
• Hover over a counter to see its queue, the longest queue so far, how many it has served and how many people gave up because the queue was full.
• Space runs or pauses the simulation when you’re not typing in a field.

The headline stands, marked with a star, are A1, C3, C4 and E6. Row A is nearest the entrances and row E nearest the food court.

Reading the live figures

Clock, status and timeline

The bottom bar shows the simulated time, a one-line status (for example how long until the doors open, or how many are still inside since the alarm) and a timeline of the scenario with its key moments marked.

Headline figures

Normally: people inside the hall (including staff), people waiting in the forecourt, people in queues, people eating at tables, visitors who have arrived so far and visitors who have gone home. After an alarm these switch to: still inside, evacuated, not yet moving, heading to exits, time since the alarm (or total evacuation time once everyone is out), and how many were inside when the alarm sounded.

Busiest 2 m square

The highest density anywhere right now, in people per m², with the peak for this run. The coloured square uses the heatmap colours.

Who is inside

How many of each persona are in the hall. During an evacuation it becomes a table showing, for each persona, how many are still inside, their average time to get out so far and the time the last of them got out. Once everyone is out, the slowest group is highlighted.

Chart

Normally shows people inside, people queueing and people outside over the day, with dashed lines at the doors opening and the closing call. After an alarm it zooms to the evacuation, showing people still inside minute by minute.

Queues

The current length of each queue against its limit. A bar turns red when the queue is full and new arrivals are turning away.

People out through each exit

How many people have left by each exit during this run. Uneven bars show which exits do the work and which are hardly used.

Busiest stands now

The five stands with the most visitors at this moment. Staff are not counted.

Personas

Every visitor has one persona, which changes their walking pace, which stands they choose, how long they stay, how likely they are to want food, coffee or the toilets, and how quickly they react to an alarm.

Each person’s pace varies a little around their persona’s figure, so no two walk at exactly the same speed.

How the model works

This is an agent-based simulation: every person is modelled individually, makes their own decisions and moves step by step. The patterns you see, such as queues spilling into aisles or congestion at doors, are not scripted. They come out of thousands of individual choices and movements.

The hall

The hall is 100 m wide and 72 m deep. It holds 30 island stands of 9 × 5 m in five rows. Each stand has a solid core with a 1 m carpet edge that visitors step onto. There are three coffee kiosks, three food counters with 48 tables and 288 seats, two info desks and two toilet blocks. The two main entrances are 6 m wide and the six emergency exits are 2.4 m wide. A forecourt outside the south wall holds arriving and waiting visitors.

Finding the way

The floor is divided into a grid of 0.5 m squares, each marked as open or blocked by walls, stands, counters and tables. For every destination (each stand, counter, toilet block, the seating, the main exits and the open emergency exits) the model calculates the shortest walking distance from every open square to that destination, going around obstacles. That gives 43 route maps. A person heading somewhere looks at the squares around them and steps towards the ones closest to their destination, which naturally finds the shortest way around stands and through doors. For the final few metres to a queue spot or a seat, people walk straight there if the way is clear, and follow the route map if something is in the way.

Walking and crowding

Each person has a preferred walking pace from their persona. They change speed and direction smoothly, taking about half a second to respond. People keep a minimum spacing of about 0.56 m between centres: when two get closer, they push apart, with people who are standing still (at a stand, seated or in a queue) giving way less than people walking. People also slow down as the area around them gets busier. The model counts everyone within a 3 × 3 m area and applies a standard pedestrian speed–density relationship (Weidmann’s): walking is unaffected below about 0.6 people per m², drops to roughly 80% of normal at 1 person per m² and about 45% at 2, and falls to a slow shuffle at 4 or more.

Time moves in steps of a quarter of a second. At high speeds the model takes up to 10 steps for each frame drawn. On a slower device it may not keep up, so the clock can run a little slower than the speed you chose; the results are the same either way.

Daily routine and choices

Whenever someone finishes an activity, they decide what to do next, in this order:

1. Go home, if they have reached their planned leaving time.
2. Get lunch, between 11:45 and 14:15, if they haven’t yet. The chance rises as the lunch period goes on.
3. Get coffee. This is most likely before 11:00 and between 14:30 and about 16:15, and very unlikely within 90 minutes of their last one.
4. Visit an info desk. This is likely as a first stop (very likely for press) and rare afterwards.
5. Use the toilets, which is more likely in the hour after lunch.
6. Rest at a table, for people with reduced mobility.
7. Otherwise visit a stand.

Each step is a chance, adjusted by persona, not a certainty, so people’s days differ. When choosing a stand, people weigh its popularity against the walking distance and strongly avoid stands they have already visited. A stand 22 m away is half as attractive as an identical one next door. Some decisions are strolls: the person walks further afield and only glances at the stand for a few seconds before moving on, which keeps the aisles busy as they are in real halls. Info desks and coffee are chosen with a preference for nearby counters with shorter queues.

Queues and seats

Each counter has a real line of people, spaced 0.55 m apart and extending away from the counter. A person joins when they get close to the end of the line. If the queue is already at its limit (13 to 18 people depending on the counter), they give up and come back later: about 10 minutes later for food, 20 minutes later for coffee. The person at the front is served in a random time around the counter’s average: about 22 seconds at coffee, 11 seconds at food (several tills) and 13 to 16 seconds at the info desks. After buying food, people look for the nearest free seat with a little randomness. If none is free, they eat standing up nearby. Lunch at a table lasts 10 to 24 minutes; a rest lasts 4 to 9.

Arrivals and departures

Before a run starts, the model writes a timetable for every visitor: when they arrive and when they plan to leave. On a typical day about 18% of visitors arrive before the doors open and wait outside, most arrive between 09:00 and about 10:30, and a steady trickle continues into the afternoon. Arrival times and lengths of stay depend on persona; for example students come later and leave earlier. Many visitors plan to stay until the closing call. Arrivals use one of the two main entrances at random (an open one, if one is blocked).

At the closing call, each person gives themselves a short random delay to finish what they are doing, averaging about 4 minutes and never more than 14. Staff take 5 to 20 minutes to close up. Everyone then walks out through the nearest main entrance.

Evacuation

When the alarm sounds, anyone still in the forecourt is treated as already safe, arrivals stop, queues and seats are abandoned, and people leave the toilets. Each person then waits a reaction time before moving. It is the average reaction time you set, multiplied by a random factor (usually between a quarter and twice the average, occasionally more), by their persona’s factor and by their situation: people in the toilets take 35% longer, people seated at tables 20% longer, people at stands 5% longer, and people already walking 15% less. No one waits more than six times the average. Once moving, people walk about 15% faster than normal and head for the open exit with the shortest walking distance from where they are. They count as out when they pass through an emergency exit door or walk clear of a main entrance into the forecourt.

Measuring density

Density is measured on a grid of 2 m squares by counting the people standing in each, divided by its 4 m² area, and smoothed slightly over time so the heatmap doesn’t flicker. As a rough guide from crowd-safety practice: below about 1 person per m² people move freely; around 2, movement becomes noticeably restricted; sustained densities of 4 or more, especially in moving crowds, are where dangerous crowd pressure becomes possible.

Randomness and run numbers

All random choices (arrival times, personas, preferences, service times, reaction times) come from a sequence started by the run number. The same run number and settings always reproduce the same run, which lets you change one thing and see its effect without noise from a different crowd. Because any single run is one possible day, it’s worth checking important results with a few different run numbers.

Rehearsal ideas

• Which exits matter most? Run the 12:45 evacuation and note the time. Then block each exit in turn with the same run number and compare. The exit whose loss adds most time is the one to protect.
• Worst plausible case: combine the lunch-rush evacuation, the “Many with reduced mobility” preset, 4,000 to 5,000 visitors, a 90 to 120 second reaction time and a blocked main entrance.
• Value of training and announcements: compare evacuation times with average reaction times of 20, 45 and 90 seconds. Clear announcements and trained stewards shorten real reaction times.
• Opening the doors: in Start of day, use the heatmap to see how dense the forecourt gets and whether the lobby info desk and coffee queues block the paths from the entrances.
• Lunch capacity: in the typical day, watch whether the food queues fill (red bars) and how many people turn away. Hover a counter to see the totals.
• Closing time: in End of day, see how long the hall takes to clear through the two main entrances alone, and whether one entrance takes far more people than the other.
• Headline stands: use the plan view and heatmap to check whether crowds at the headline stands (A1, C3, C4, E6) spill into the main aisles.

Limits of the model

This is a planning and rehearsal aid for comparing scenarios. It is not a validated evacuation model and should not be used on its own to show compliance with fire or safety regulations. For that, use specialist evacuation software and the advice of a fire engineer.

• The layout is a generic hall, not your venue. Door widths, distances and counter positions all affect the results.
• The figures for walking pace, service times, reaction times and visitor behaviour are reasonable assumptions, not measurements from your event. Absolute times should be treated as rough; differences between runs are more meaningful.
• Flow through doors comes from people’s spacing and speed rather than being set to a measured flow rate, so door capacity is approximate.
• Everyone knows every exit and heads for the nearest by walking distance. Real people often head for the way they came in, follow others, or avoid unfamiliar doors, which usually makes evacuations slower and less even.
• People move individually. Groups and families who stay together, wheelchair users needing extra width, staff guiding visitors, and people going back for belongings or others are not modelled.
• There is no smoke, fire spread, reduced visibility, stairs, lifts or outside conditions. A blocked exit is simply closed.
• Crowd pressure is not modelled as a physical force, so the model shows where densities become dangerous but not what happens to people in a crush.