[SystemSafety] Meta

[Peter Bernard Ladkin]

On 2014-12-09 03:06 , Les Chambers wrote:

> Hazard meta data seems like an obvious idea. 

Yes, it does.

Two points.

First, harmonisation of vocabulary. Developing a quasi-XML tag set as you suggested depends upon
attaining general agreement upon vocabulary. I can tell you first hand that there will be a lot of
work involved in that, and many of my colleagues with whom I have discussed the matter are sceptical
that it is organisationally possible.

What might have more chance is a plethora of different conceptual definitions, along with
intertranslations.

We have one project, called SmartTerms, running from October 2013 to end September 2015, which
attempts to capture the vocabulary currently being used for "new" electrotechnical domains and
perform Def-Use analysis amongst the terms, and suggest a harmonised standard vocabulary. In
Germany, these "new domains" are currently "Smart Grid", "Smart Cities", "Smart Homes and
Buildings", "EMobility", and "Industrie 4.0". EMobility is electric road vehicles; Industrie 4.0 is
robotic factory automation). We have been working with colleagues in standardisation (on both
national and international levels) to achieve a similar harmonisation wrt Safety and Security, in
particular for critical infrastructure, and we are currently preparing a project proposal. As I
said, many colleagues are sceptical that such an effort can bear much fruit, but it does seem worth
trying systematically.

Second, you don't just need hazards, you need possible consequences and possible sources, because
talk of what a hazard is without having any idea of the possible deleterious consequences is
pointless (how would you even tell it's hazard?), and talk of what a hazard is without suggesting
how it can have come about is not very useful. Call it qualitative abstract risk analysis, or QARA
(nice Australian-sounding word).

Causalis has a means of doing it, called Causal Failure Analysis, and a visual depiction of the
results, called a Causal Fault Graph. CFA and CFG can be as specific or as general as you like - the
level of detail is up to you. Causalis has used CFA and CFG successfully to fulfil the needs of
clients (given the competition, success is all but guaranteed!). The CFG is depicted on
screen/printout using Graphviz, dot in particular, and the specification language is CausalML,
developed a decade ago jointly with our partners in IfEV at TU Braunschweig and at Siemens Rail
Automation, and more recently by my Uni group RVS alone. So CausalML suffices for the task.

PBL

Prof. Peter Bernard Ladkin, Faculty of Technology, University of Bielefeld, 33594 Bielefeld, Germany
Tel+msg +49 (0)521 880 7319  www.rvs.uni-bielefeld.de