Crowd Estimates

“Everyone is entitled to his own opinion, but not to his own facts.” ― Daniel Patrick Moynihan

Estimating crowd numbers, evaluating crowd density and understanding safety limits for event capacity is an essential requirement for crowd safety and risk analysis. Where crowd counting can be very useful (after the event) we focus on a dynamic estimation process (using the RAMP analysis techniques). this allows for real-time predictive estimation and risk mitigation for places of public assembly.

Over the last three decades the same types of questions are asked during our short courses and workshops 

"I don’t know how many people might be coming to this event so what’s the best way to plan if we don’t have the numbers?"

 I would only need to ask four questions before I could develop a crowd dynamics plan and risk assessment for the site

1. What direction will crowds approach and depart the area?

       Route question (transportation hubs to and from event site)

2. What is your event capacity?

        How much Area do you have? How will the area be used?

3. How quickly will these spaces fill?

       What crowd Movements do we expect? Over what period of time?

4. What type of crowd are you expecting?

      What type of People/Profile, are you expecting - what do you know of the crowd demographics?

With the above information it is then relatively straightforward to look at where, when and how congestion might build up, where queues may be expected and where you need to deploy stewards and security to manage the crowds.  

Getting crowd number right is essential for crowd safety planning but can create heated debate after the event (always with only one side providing evidence based analysis, the other, emotional speculation). 

See this page for some background information on our 2017 Inauguration count.


When we were asked to evaluate the crowds at the Eagles Victory Parade (Feb 2018) we started with the RAMP analysis. Were were we expecting crowd to accumulate, what areas were available, how would this crowd build up over time, what do we know about the nature/behaviour of this type of crowd. I’ve included a few screen shots from the teams report and analysis of the crowds. The team were 

Jim Aughney, Bert Bruyninckx, Benjamin Nigel Cowcill, Eric Kant, Ise Murphy, Jade Margaret Patel, Mark Whitten, Marcel Altenburg and G. Keith Still

Three Policing events (one event commander from Notting Hill Carnival), a Marathon organiser, IT/Simulations experts, and event safety experts. All were assignent to sub-teams and specific sections of the route. 

The route was 8km long. 

First pass approximations are used to evaluate the potential numbers (before the event).

A route of 8,000m x 2 (both sides of the road) with people standing shoulder to shoulder, along the entire route (a person is just over 1/2 metre wide, woul host 16,000 linear metres (both sides) and therfore 32,000 people. 

If the crowd were stacked 5 deep, that would be 160,000 people

If the crowd were stacked 10 deep, that would be 320,000 people

if the crowd were stacked 20 deep, that would be 640,000 people

But, only if this were a average of 20 deep along the entire route, packed shoulder to shoulder, 4 people per suare metre, 20 people deep, boths sides of the route. 

Note: Consider how averages work. A 20m section packed 40 people deep is 20 x 40 = 800 people, typical of intersections which is the same as a 160m metre section packed 5 deep (160 x 5 = 800 people). 

This gives us our first pass approximation for the route. We now can then check each linear segment (10m lengths) and evalaute the depth, during the event - we had continuous helicopter footage from the TV/Media channels). One team (we had 7 people counting and Marcel/Keith overseeing the process and checking the numerical results).

We measures both linear length and density (many areas were less that fully packed) - see the table below (part of the analysis). Both sides of the route were evaluated (from the continuous helicopter monitoring the parade and ground images).

The next segment was again evaluated using a first pass approximation measuring the available ground area, then evaluating how this space filled over time, and finally density checks along the route as the parade passed. 

We anticipated large crowds in this area, and could see the barrier plan, policing and crowd maangement as the crowd built up (various webcams around the area provide continuous monitoring of the site during crowd build up - before the parade).

27797945 10209391017567237 3973165429702916709 o

The third team were monitoring Franklin Parkway.

Again, applying the initial first pass approximation (before the event) was can see the ENTIRE parkway is 382,000 square metres (incluidng buildings/roads/trees). If this were COMPLETELY full (typically Winter clothing packing density 2.5 to 3 people per square metre) then this area could hold a maximum of approximately 1,000,000 people. But we stress, this is side to side, front to back packing.

A priori commentary we anticipating 2,000,000 - 3,000,000 people. So, 2 to 3 times the area indicated below, completely full, side to side, front to back.

We have three upper limits for the available space. We can assess percentages of the occupied spaces against the potential available space against these upper limits. 

For example, Franklin Parkway (above) is the available, below the occupied space.

Taking all three parts into consideration, we estimated the following total (in real-time). We typically round this number up (in this case to approximately 700,000 people) to allow for buildings, lines of sight, crowd migration etc.

Footnote: To the claims of being “out by a few million” then we challenge you to produce your evidence of the missing millions.

© CSMMU 2018. All rights reserved. All images, videos, audio files and text are protected under international copyright law  You may not copy, modify or use any of the website content, in whole or in part, for any commercial or non-commercial purpose without permission.