[Prev][Next][Index][Thread]

Re: Reliability with redundancy



Jim Fitzpatrick <jim@jfitz.demon.co.uk> wrote:
>nick@vu-vlsi.ee.vill.edu (Nick Pine) wrote:

>>Sometimes I leave errors in postings to see if people read them closely.
 
>Do I win a prize?  :-)

You have my gratitude for finding and fixing the mttf/bf error,
but not for finding the minus sign...

>>I like to think about the marginal return on an investment in reliability:
>>where would you spend the first dollar, in making a system more reliable?
>>Where would you spend the next one? How hard is the switchover? How
>>fine-grained should the redundancy be? It's an interesting game to play
>>with a block diagram... And how does the cost of making a system more
>>reliable compare to the cost of a failure? I like to use numbers for this.
 
>You seem to have identified the need for a cost-benefit analysis.

I'm thinking more about design, ie synthesis, than analysis. "Marginal
economic availability synthesis"? Seems like it could be done: take each
piece of a block diagram and ask questions like these:

"Suppose this block were doubly-redundant, or (N-1)/N redundant, or...?"
"What would be the new cost of the components that do that job?"
"Would they each do the whole job or share the job or be hot standbys?"
"How does the system performance degrade if one fails?"
"How would the alarm (if any) and switchover mechanisms work?"
"What would they cost, and how reliable would they be?"
"What would the unavailability improvement of the overall system be
 with this particular change?"

One might systematically go over a whole block diagram this way, with
some computer assistance, and explode each block down to the lowest level
and do it again and again. Not easy, and there are local optima. It would
probably be more complicated than a FMECA.

>I rather suspect that, in some instances at least, the benefit would be
>found to outweigh the cost. However, since (going back to the original
>scenario) it is the customer who is inconvenienced by a power outage,
>and 'captive' customers don't count for a bean in the eyes of a
>virtual monopoly, the money isn't spent.

OTOH, regulated utility monopolies sometimes have an public arrangement
where they earn a percentage of what they manage to spend on equipment, etc.
So if they decided to go bananas about reliability, we might have gold-
plated wires everywhere, at the public's expense. But in the US, we
seem to be going in the other direction.

>>if things can be arranged so that there are very few common elements 
>>compared to the more complicated subsystems, with a much lower failure 
>>rate for the common elements, then one is ahead of the game
>
>why not just build your original system with components that have a
>'much lower failure rate'?

That sounds like a good idea. What do they cost? What is the most
cost-effective arrangement to achieve a certain level of unavailability?
 
>>Interesting typical figures.
>
>Look up the 'Beta Factor' model.

Thanks. I may do that, if somebody pays me to do reliability work again :-) 

>>I'm thinking more about full diversity, with an attempt at clever design,
>>and picking the right things to diversify. 
 
>You can never get full diversity. No matter how different your systems
>may be, you'll still have wiring, switches, fuses...etc., in both. You
>can almost get there - for example, buy different types of switches
>from different manufacturers (in different countries, or on a
>different planet, if you like). But in the end, they're still
>switches, and so not fully diverse.

I have a nutty friend who lives in New York City. He just bought about 5 kW
of solar PVs and an inverter. He's planning to fill up his house with
batteries and a whole other set of 110V outlets, wiring, fuse boxes, etc.
(I think he still heats his house with oil :-) Now assuming the outlets
and wiring, etc. are in different places, that seems pretty diverse to me.

Of course, the battery explosions may not be good for the his grid
system wiring, and if his house catches on fire, one or both electrical
systems may stop working, and if a black cloud covers New York for a month,
everyone in the city will use lights more during the day, and that will put
some sort of strain on the grid system, as well as his solar system, and
if a tornado tears his house into little bits, or someone nukes New York,
both electrical systems may stop working, but will he care? Where do you
draw the line? How do you define the universe, in the probabilistic sense,
for common causes?

Nick


References: