«Is One Tablet Enough?» Development, Norm-Setting, and Design of Fire Safety Systems: Who's Deceiving Whom?.
I visit the discussion section of the site on fire safety topics quite often, although mostly in silence — I look, listen, take note, and go back home to do my own thing. And recently I met a doctor there (no, not Pilyulkin, the doctor's last name was Andorra, but he behaved just as rudely). And I was so offended by his manner of pointing a finger at the naked king that I couldn't resist and put this remark here, on the pages of the TZ. To begin with, I'll quote:
1. How would you feel about the manufacturer of the tablets and the tablets that do not have instructions on how many to take? ONE, TWO or THREE? It is not specified when it is possible, when it is not possible, there is no restriction on the presence of diseases, compatibility with other drugs? What side effects may occur, damage to health, decreased activity (acceptable material losses).
2. To the professor-DIETITIAN who wrote the dissertation, legalized, published a work on diet. He described the treatment procedure, the use of the drug so well, he obscured everything, left so many contradictions and ambiguities that patients argue about how many tablets to take — ONE, TWO or all at once. They argue, and what will come of this — will there be damage to health, or will there be unacceptable material losses… And in general, is this medicine for the patient's case?
3. How would you feel about a pharmacist who gives you medicine without telling you how many tablets to take and under what conditions, without saying that ONE is for a child, THREE for an adult? The manufacturer took a certificate from the pharmacist that the miracle pill can be taken ONE at a time. It declares, advertises on the website, that the savings are here, everywhere and always, it will help you. You just need to follow the diet, namely p. 13.3.3. The pharmacist gives advice that is vague, like predictions, repeated like spells and a stuck record… Follow the work of a nutritionist SP-5 p. 13.3.3… Read SP-5… Read SP-5, p. 13.3.3, p. 13.3.3, and you will be happy. And most importantly, the whole company unanimously keeps silent about the fact that there are restrictions on intake and dosage. And there is harm, and damage to health, and hopes for savings are in vain…
I was very indignant.
I am the manufacturer of those very tablets of the «smoke alarm» type. I love them dearly, every detail is suffered and substantiated, I am sure that my brainchildren, if not the best in the world (I will not make unfounded statements, I have not compared with everyone), then, at least, the best in the country in a similar price range. Well, of course, it was said about them in «The Diamond Arm»: «One tablet is enough …»
And anyway, why is this doctor getting into the high area of fire protection with his pharmaceutical comparisons? We've been chewing over the question of «how many detectors should be installed in one room» for many years now. We've eaten so many dogs here, and we…we've already forgotten where the tradition of installing several of them came from.
Then, however, I remembered that NPB-88 always gave me a dual feeling — simple and clear sections in it always coexisted with very vague ones. And when in a new reincarnation the same text was included in SP-5 and there appeared a charming appendix «R», with its vagueness capable of making a gypsy fortune teller blush, I stopped looking there and in new developments I mainly focus on European standards (EN-54).
But the designers and installers, whose voice was conveyed to me by the Internet doctor, live in the real world (i.e. they are not so much concerned with the quality of fire protection, as with the project's passability through all approvals).
I try to improve reliability, stability, sensitivity, reproducibility, and then suddenly it turns out that no one needs it. I had to come back down to earth. I looked through all the available documentation again (it took only a couple of weeks — fortunately, there is only 2-3 times more regulatory documentation lately than five years ago) and realized that this section of the NPB (and, accordingly, SP) has practically not been rewritten since the early 90s, when addressable-analog systems seemed to our people as infinitely distant as yachts in the Bahamas.
Some other sections are clearly being updated, fire extinguishing describes quite modern technologies, and in the fire alarm section the text is almost identical to the original from 20 years ago (which originates from even older recommendations), with a few new amendments that only confuse the issue. Having emerged from the sea of paper, I was able to confidently say two things:
yes, I am absolutely right when I say that our wonderful addressable analogue fire alarms can be installed one per room;
yes, those experienced people are absolutely right who say: if you don't want problems, always put 3, it doesn't matter — the cheapest, I won't say which ones, or the most modern multi-criteria — put 3 anyway
In terms of the original line — «the pharmacist says that this pill will cure one a month, but since the inspectors appointed by the nutritionist will check how you take your pills three times a day, then you'd better take not one modern pill, but 21 old ones and take them for a week, 3 times a day. Of course, you'll be cured much later and you'll be sick longer, but the inspector won't have any questions. And if you want to be cured and not go to jail for being cured — buy 21 modern expensive pills, eat one, and show the rest to the inspector, and he'll leave you alone.»
In real life, this means that only very rich organizations can afford good equipment. Since it is forbidden to install 100 good ones, there is no money to install 300 good ones, so we will install 300 bad ones, but cheap ones. The result is predictable: after a couple of months of unsuccessful attempts to combat false alarms, the system is switched off. By the way, I was recently told about a high-tech way to create the appearance of a working system: a non-addressable loop with fire detectors is switched to security mode (normally closed). As a result, it ignores signals from the detectors, and the indicators on them blink as if they are working. Isn't it a good way? Excellent protection from the fire inspector. True, it does not protect against fire, but no one cares anymore.
A glimmer of hope arose, however, when I looked through the methods for calculating fire risk. Well, here it will certainly be clear that good systems are better than bad ones. I look carefully, and what do I see? I quote:
The probability of effective operation of the fire alarm system is determined by the technical reliability of the elements of the fire alarm system, given in the technical documentation. In the absence of information on the parameters of technical reliability, it is permissible to accept Rобн = 0.8;
That's it. We've arrived. No matter what — 0.8. Well, okay, I thought, I'll calculate it for my system now. How can I calculate it? It's easy to calculate the probability of one alarm failure. It needs to be calculated in every way. Multiply the number of components by the reference data for them and add the soldering — in general, for ours it comes out to about 100,000 hours. I understand very well that the accuracy of such a calculation based on tabular data is zero. But I don't have reliable statistics on failures yet. So if a faulty alarm is replaced within 4 days after a malfunction signal (optimistic, but possible), then the probability of system failure in the event of a fire will be about 0.1%. That is, the probability of operation is 99.9%. Hmm. Not that much. I expected more nines. By the way, the formula in the method is interesting. If the probability of the alarm being triggered exceeds 6 nines, then the probability of a successful evacuation can be even negative — the result of the formula will be satisfactory. Probably, the formula is considerably simplified and works only in the range of about 0.8. Well, the developers did not count on modern systems. Okay, let's not pick on the first pancake, it is clear that the formula was written, alas, not by Kolmogorov.
So, high-quality addressable detectors with exhaustive (as far as I could think of) self-diagnostics provide a 20-fold reduction in risk compared to conventional ones. By the way, do conventional ones really provide an average of 80%? Well, let's check. I confess, I took the idea for this check from the article by A. Yarygin and D. Katkin.
So, the mean time between failures of cheap detectors (even if we believe that they comply with GOST in this parameter) is 60,000 hours. The service life of the system before replacement is usually 10 years. The probability that after 10 years at least two of these 3 (non-addressable, without diagnostics) detectors work is about 25%. Not 80%. And if we take into account that the control panel (CP) also has the same (or less) mean time between failures, and its operability is also not actually monitored, then it would be good if 10% remained. The developers of the method clearly counted on higher quality products on average. Probably, they did not read the GOST requirements. Or they seriously believed that all 100% of detectors at the facility are checked at least once a year with an aerosol smoke simulator. Okay, let's not fantasize, at least they are checked by pressing a button.
So, what should I answer the doctor who upset me so much with his remark?
Study the works of another nutritionist. He recommends that the patient not read books by alternative theorists, but calculate the required number of pills himself. Moreover, the calculation method in the work of that nutritionist will allow, if necessary, to easily prove that one pill is enough. Even, perhaps, one pill for several rooms.
What? You still don't understand? Well, my friend, then contact Zhvanetsky. In your case, only he can explain what troubles there are in Russia and who cares about preserving your property from fire.