Crowd movement using Discrete Element Methods
Movement of people in crowds is highly complex and, as population density grows, increasingly important in design of cities and buildings[1]. It is also sometimes a matter of life and death, when buildings and transport vehicles have to be evacuated. Some of the things that affect the rate at which a crowd might leave a room, for example, include the shape of the room and the exit, the people density per m2 and the age, age distribution and cultural mix of the people making up the crowd.
How could such a complex situation be modelled? In the entirely unrelated field of granular materials, movement of particles in flow has been modelled very successfully using the Discrete Element Method, DEM (see, for example, http://www.dem-solutions.com/ which concerns a commercialised DEM code), in which every particle and its interactions with every other particle are individually modelled. The elements can each have their own individual properties such as size and density. The key elements in such a model are the “housekeeping”, which keeps track of where every element is in time (and much work has been done on making this as efficient as possible) and the laws of motion and of interaction between elements (which in the case of granular flow are usually modelled as Newtonian). What are the corresponding laws of motion and of interaction in the case of people? No-one knows (and you may think that “laws” is stretching the analogy between granular materials and people a bit far!), although there has been a limited amount of work on (largely) 1D people motion
(see, for example, http://www.ped-net.org/index.php?id=12&ID=238 ).
The project will start with an existing working DEM code, set up for modelling granular materials. The objectives are
· to adapt it to model people flow
· to carry out some simple simulations
· …and to compare the results with some available experimental data.
In the time available, it is probably realistic to consider 1D situations only. Some knowledge of at least one programming language is required, preferably FORTRAN.
This work is speculative, but given the wide interest in the topic, a PhD project could definitely result. The field is of sufficient interest to design companies such as Arup that some form of sponsorship is likely.
Jonathan Seville, Engineering
[1] There is a strong current interest in airports, not surprisingly, especially since the advent of the high-capacity Airbus 380.