Re: [Fwd: Re: [sc] Re: Fault Tree Analysis (was: Legal and Ethical Issues forSafety Practitioners)]



Re: [Fwd: Re: [sc] Re: Fault Tree Analysis (was: Legal and Ethical Issues forSafety Practitioners)]

From: Daniel Johnson <drdpj_at_xxxxxx>
Date: Sat, 9 Apr 2011 15:39:32 -0500
Message-ID: <0AEABA3AE14542B997DB4E5E63B86A0B@xxxxxx>
As discussed in my previous posting (the example with the light bulbs), the
10^-23 would represent an instantaneous failure rate, not an actual 
lifetime.

--Dr. Daniel P. Johnson

-----Original Message----- 
From: Peter Bishop
Sent: Saturday, April 09, 2011 12:37 PM
To: safety-critical@xxxxxx
Subject: [Fwd: Re: [sc] Re: Fault Tree Analysis (was: Legal and Ethical 
Issues forSafety Practitioners)]

To remove confusion, this should have been posted to the list
Rather than Daniel Johnson

Peter Bishop

-------- Original Message --------
Subject: Re: [sc] Re: Fault Tree Analysis (was: Legal and Ethical Issues
forSafety Practitioners)
Date: Fri, 08 Apr 2011 17:30:26 +0100
From: Peter Bishop <pgb@xxxxxx>
Organisation: Adelard
To: drdpj@xxxxxx
References: <380-22011458143327861@xxxxxx>

drdpj@xxxxxx wrote:
...
> Let us dig deeper into the Byzantine Fault problem. SAFECOMP 2003:
> http://www.mail2web.com/cgi-bin/redir.asp?lid=0&newsite=http%3A%2F%2Fwww%2Ec
> s%2Eindiana%2Eedu%2Fclasses%2Fb649%2Dsjoh%2Freading%2FDriscoll%2DHall%2DSive
> ncrona%2DXumsteg%2D03%2Epdf
>
...
>
> The cited example of a 10^-23 calculated failure rate for a failure that
> was then observed during testing is a real-life example of this- there was
> a common failure mode due to a byzantine fault in an integrated system 
> that
> was not recognized until late in the testing cycle.

> But the issue was not the probability calculations, it was the lack of
> recognition that a single failure mode was a common point of failure for
> redundant subsystems. The FTA then faithfully reported the results of that
> gap.

You might however question the number you get from the FTA when you
consider that the current age of the Universe (not our relatively
youthful solar system) is a mere 10^14 hours.

Peter Bishop








[The content of this part has been removed by the mailing list software] 
Received on Sat 09 Apr 2011 - 21:39:39 BST