DRAFT Brief Overview of New ALCAM History Background

DRAFT Brief Overview of New ALCAM

History & Background DRAFT • Australian Level Crossing Assessment Model (ALCAM) is an assessment tool used to identify key potential risks at level crossings and to assist in the prioritisation of crossings for upgrades. It provides a rigorous defensible process for decision making for level crossings safety improvements; • ALCAM, originally known as ‘Risk Scoring Matrix’, was developed in 1999. In 2002, the matrix was re-named as ALCAM, and a national committee was established to ensure its consistency of development and implementation; • ALCAM was endorsed at the May 2003 Australian Transport Council Meeting. All State & Territory Ministers agreed to adopt this innovative method of level crossing risk assessment; • In 2004, a Microsoft Access database was developed (Level Crossing Management System – LXM) to maintain data and run assessments. This LXM system was adopted formally by the ALCAM committee. The LXM is currently being developed into a SQL database with internet interface; • Significant redevelopment of the ALCAM has taken place over last 3 years. A new ALCAM comprising new Traffic Exposure Model, Infrastructure Model and Consequence Model has been developed and accepted by the national committee; • Both the new ALCAM and the new LXM (i. e. online database) are expected to be released in late 2013. 2

CURRENT ALCAM Structure of Current ALCAM Consequence Factor Exposure Factor X X Trains X ALCAM Risk Score (ARS) = Vehicles Speed Environment Likelihood Factor DRAFT 3

CURRENT ALCAM DRAFT Likelihood Factor • In the current ALCAM, a likelihood factor of a level crossing is determined by a Characteristics Matrix and a Controls Matrix: – The Characteristics Matrix is used to determine the effect that each characteristic would have on each accident mechanism; – The Controls Matrix is used to determine the effect that controls will have on reducing the likelihood of an accident mechanism occurring; – A likelihood factor is a number between 0 and 800. This number is useful for comparisons between level crossings, but is not directly expressed as an accident likelihood/probability. Any feature of a roadway or railway which may have an influence on pedestrian or driver behaviour, e. g. sighting distance, speed of train Any significant pedestrian or driver behaviour that increases the potential for a collision with a train to occur Devices that reduce the risk of an accident by changing pedestrian or driver behaviour, e. g. flashing lights, boom gates, signage 4

NEW ALCAM DRAFT Infrastructure Factor • • • In the new ALCAM, both Characteristics Matrix and Controls Matrix have been refined and the output from the matrix has been renamed as Raw Infrastructure Factor; An Infrastructure Modifier has been introduced in the new ALCAM to turn the Raw Infrastructure Factor into a real accident probability. The Infrastructure Modifier is a liner equation that was determined by correlating 10 years of Australian and New Zealand level crossing crash data against the Raw Infrastructure Factors for all jurisdictions (normalised by vehicle and train volumes); Multiplying the Raw Infrastructure Factor by the Infrastructure Modifier will produce an Infrastructure Factor. It is expressed as a scalar e. g. 1. 08, which represents the expected effect that the crossing condition will have on the accident probability. 5

CURRENT ALCAM Exposure Factor DRAFT • In the current ALCAM, the exposure factor is produced by multiplying the road traffic volume (V) and the rail traffic volume (T) of a level crossing; • This approach gives the traffic volumes the most influential parameters on the ranking of crossings with the highest risk predicted at the extremes of vehicles per day and trains per day. 6

NEW ALCAM Exposure Factor DRAFT • In 2011, the National ALCAM Committee commissioned a study to assess the relationship between vehicle (V) and train (T) volumes in respect to the risk of an accident; • The study investigated different exposure modelling approaches from Australia, UK, and US, and compared their predictions with 10 years of Australian and New Zealand level crossing crash data; • The study found that conventional ‘linear’ approach (V x T) used in the current ALCAM did not best replicate the observed collision record; • The study recommended that ALCAM adopt the Peabody-Dimmick Formula, which is an accident predication model widely used in the US; • The study further recommended to apply an adjustment factor to the result in order to produce more contemporary crash rate predictions. This has been done by many users of the formula in the US; • The adjustment factor was calculated using 10 years of Australian and New Zealand crash data. This ensures the formula is applicable to Australian and New Zealand conditions; • The Exposure Factor in the new ALCAM is expressed as an accident probability per year, e. g. 0. 12. It represents the baseline likelihood of an accident at a level crossing, excluding site-specific conditions that are captured in the Infrastructure Factor. 7

CURRENT ALCAM Consequence Factor • • DRAFT In the current ALCAM, the consequence factor is determined as a relationship between an environmental factor and a train speed factor; It works as a modification factor to inflate or deflate the exposure factor (V x T) for the level crossing by up to a factor of 10. 8

NEW ALCAM Consequence Factor DRAFT • In the new ALCAM, the consequence factor is the expected outcome in the event of a collision. This includes deaths and injuries on both the train and vehicle; • The core component of the new consequence model is an event tree that estimates the likelihood that a given level crossing collision will escalate into more serious consequences, e. g. derailment, overturn, and secondary collision. This approach involves assigning probabilities to a sequence of events occurring, and hence the model produces a number of possible outcomes, each with an associated probability of occurrence; • Each outcome at the end of the event tree has an associated number of fatalities, serious injuries, and minor injuries. These are combined to produce a single consequence factor, which is expressed in terms of equivalent fatalities per collision, e. g. 2. 3; • The probability of occurrence and possible outcomes used in the event tree are based on 10 years of Australian and New Zealand level crossing crash data and assumptions from UK data. 9

NEW ALCAM Consequence Factor • DRAFT Structure of event tree 10

NEW ALCAM Consequence Factor DRAFT • Further development to the new consequence model is currently underway. This will extend and refine the consequence model by: o adding a road vehicle speed element to the existing model; o reviewing Australian and New Zealand level crossing crash data, in comparison with UK data; o recommending a “best available” combined dataset for use in the model. • This development is being managed by the Tf. NSW on behalf of the National ALCAM Committee, and is expected to be completed by the end of this financial year. 11

NEW ALCAM Structure of New ALCAM DRAFT 12

DRAFT Questions? 13