1. Delivering the Australian Rail Risk Model (ARRM)
Jesse Baker RISSB General Manager Safety and Innovation
25 October 2017
IRSC

2. A brief introduction to RISSB
A non-profit company established in 2007 by the ARA.
SDAC (formerly ABSDO) accredited to produce Australian Standards.
100% owned by the Rail Industry.
Government / industry funded.
Responsible for the Australian Codes of Practice (ACOP).
Standards, Codes of Practice, Safeworking Rules (ANRP) and Guidelines – developed by industry, for industry.
RISSB Standards are free to members.
New Board.
New CEO.
New delivery model.
That, coupled with our flat management structure means we’re more nimble, more responsive than ever. We’re now producing a standard in an average of 18 months.

3. Why do we need decision support tools?
Our brains are HOPELESS at decision making !!!
We’re vulnerable to:
Several types of bias:
Decision making, belief, and behavioural
(eg confirmation bias, illusory superiority, availability heuristic)
Social
(eg Dunning–Kruger effect)
Memory
(eg Hindsight bias)
Logical fallacies
(eg Argument from personal incredulity)
And now that I’ve challenged you perception of the world, let me insult your intelligence!
Many of these biases affect belief formation, business and economic decisions, and human behaviour in general.
(illusory superiority – a 2014 yougov poll found that only 2% of British people described themselves as being below average intelligence)
Refers to the systematic errors made when people evaluate and/or try to find reasons for their own and others' behaviours.
Either an enhancement or impairment to the recall of a memory (either the chances that the memory will be recalled at all, or the amount of time it takes for it to be recalled, or both), or that alters the content of a reported memory.
A logical fallacy is a flaw in reasoning – either intended or accidental.
WE ARE RIDDLED WITH THEM!!!!
Naïve realism
The belief that we see reality as it really is – objectively and without bias; that the facts are plain for all to see; that rational people will agree with us; and that those who don't are either uninformed, lazy, irrational, or biased.

4. An experiment Sound 1 & 3 Sound 2
Why might we be able to hear it better the second time? The sound hasn’t changed….
It’s because neurons have fired in your brain in a particular pattern corresponding to the sentence you heard, those patterns have reinforced themselves through usage giving you a different interpretation of THE SAME environmental factor the second time around.
Nothing out there in the world has changed.
I’ve effectively rewired your brains to a certain degree, so, what to do now that I’m in control of you brains…. Haha, first I better get on with my speech!

5. RISSB’s ARRM
An objective, quantitative tool that will soon provide industry with a deep, robust and comprehensive picture of safety risk.
It will be primarily used by Rail Transport Operators to:
help understand their current safety risk, and
to benchmark their safety risk performance against other RTOs.
ARRM consists of:
a risk model in the form of Hazardous Events, Fault Trees and Event Trees, and
estimates of frequency, probability and harm for events and consequences.
Fault tree
Event tree
Hazardous event / top event
The railways are inherently safe, but the law requires us to manage our safety risk. Given how well we’ve all just done displaying our cognitive abilities, RISSB is progressing a project to help….

6. RISSB’s ARRM A long time in the making c12 months in production
Non-mandated
> 20 organisations voluntarily subscribed
Australia has nearly 200 rail companies, but this 20 represents nearly 95% of the railway!
Consultative (more than 100 industry experts engaged)
Very positive feedback at UAT
Launch at AusRAIL (Australasia’s largest rail conference / exhibition) next month

7. ARRM information
ARRM risk estimates are expressed as FWI/yr, with information on uncertainty.
ARRM risk estimates are based on actual occurrence data supplied by RTOs, including the occurrence data RTOs supply to ONRSR. As well as expert judgement where required.

8. ARRM information

9. ARRM information (harm estimates – also frequency and probability)

10. ARRM Architecture

11. ARRM Architecture

12. ARRM Architecture
User Interfaces

13. ARRM Architecture
Application data

14. ARRM Architecture
Back end processing

15. ARRM Architecture
Publication

16. ARRM website

17. ARRM interface

18. ARRM outputs

19. ARRM outputs

20. ARRM next steps The occurrence association task involves
– Associating each occurrence with events in the fault trees / event trees
– Identifying the harm to each exposed group that resulted for each occurrence
The occurrence association is difficult and expensive
– ~1200 hours of work to associate ~22,000 occurrences
• we are only associating 40% sample of occurrences
– Difficult to find appropriate resources to do the work
• Significant part of the work are tedious and repetitive
• Other parts of the work are intellectually challenging and require good understanding of railways
– With such a huge amount of work, errors are inevitable – QA is essential

21. ARRM next steps

22. Thank you – questions? jbaker@rissb.com.au