Exploring the complexity of the heroin drug use in Melbourne (Australia)

A. Dray and P. Perez


SimDrug is a multi-agent system model, built with Cormas, as part of the Drug Policy Modeling Project, led by Turning Point, a specialist alcohol and drug organization based in Melbourne. The project aims at addressing a demand for new integrative approaches to support policy makers and practitioners who implement illicit drug policy in Australia.

SimDrug focuses on the representation of dynamic relationships between law enforcement, treatment, harm reduction and prevention. Its design is based upon a real case-study: the heroin 'drought' that struck the illicit drug market in Melbourne in 2000. The supply of heroin in Melbourne suffered a dramatic decline between late 2000 and early 2001, after a strong increase in heroin use and related harms in the late 1990s. This change in heroin supply led to a substantial decrease in overdoses. Field experts argue that the drought was actually shadowed by a dramatic increase in psycho stimulants, resulting in a fairly constant number of injecting drug. On the other hand, the heroin drought is held up by the Australian Government as an example of law enforcement (seizure, syndicates dismantlement) having a significant impact on the supply of drugs.

Model description

Time scale
One time step is equivalent to a 24h-day in reality as injecting behaviors need accurate time scale. Each simulation is run over a 5-year period (1998 - 2002 dataset) to allow the model to be able to consistently reproduce pre-drought, crisis, and post-drought dynamics of the system.

Spatial environment
SimDrug works on an archetypal representation of Melbourne based on a regular 50*50 square mesh, each cell representing a street block. Five suburbs (defined as an aggregation of neighboring cells) are created with different sizes and shapes, regardless of realistic features. Two special cells represent the Police Station and the Treatment Center.

Social entities
SimDrug includes different types of social agents: 3000 users, 150 dealers, 10 wholesalers, 10 constables, and 10 outreach workers. Figures are based on a 1:10 scale of the reality.

Model dynamics
Each time step encompasses the following stages:

  1. The population of agents, based on the changes triggered during the previous time step is updated. All detainees are retrieved from the system and new users, dealers and wholesalers are created accordingly.
  2. Wholesalers and dealers interact towards drug supply. Wholesalers are given the opportunity to refill their supply once a month while dealers can visit their wholesaler as soon as their drug stock is sold out.
  3. Outreach workers interact with users in order to decrease their readiness for treatment and convince them to enter into treatment.
  4. Users interact with their environment and other agents. They start by assessing their need looking at their available cash and drug and decide whether they need to commit a crime. They, then, find their usual dealer (or alternatively a new dealer) and buy some drug. They use it at once and might declare an overdose.
  5. The Treatment Center manages new users entering treatments and on-going treated users reaching the end of their treatment duration.
  6. Street blocks risk and conductivity are updated and the new suburbs' protest values are calculated accordingly.
  7. The police station adapts its strategy by reallocating constables on the grid and eventually performing successful crackdowns.

Model code and documentation

The Cormas model -including the smalltalk code, a set of UML diagrams and a description of the variables- is available here.


Perez, P. and Dray, A. 2005. Monograph No. 11: SimDrug: Exploring the complexity of heroin use in Melbourne. DPMP Monograph Series. Fitzroy: Turning Point Alcohol and Drug Centre.

For more information, contact the corresponding author


Le Cirad Centre de coopération internationale en recherche agronomique pour le développement
Informations légales © Copyright Cirad 2001-2015