Smoke Control Systems

[Willo]

I haven't had the opportunity to look at Stu's spreadsheet. I shall find the time to this week and I will see how I get on.

Stu, I get this impression that you feel that there is a lack of a definitive guide to smoke and its control. If this is so why do you thin sok? Is it the complexity of the subject, a lack of research or something else? To me there seems to be a large number of sources of information, each one slightly different. When I was studying mechanical engineering we were recommended one or maybe two course books which would tell us what we needed know. They may not have been particularly good reads but they did the job. It doesn't seem to be the case here.

[Phoenix]

Hi Willo,

There is plenty of guidance but as you are starting to see it isn't consistent.  

Remember this, the subject is relatively new, lacks historical data and is still subject to people's evolving and differing ideas.
  
The tests that are required to evaluate new or improved models are very expensive and can only be undertaken at a few sites.  

The models that are out there are probably a little conservative (i.e. erring on the side of safety).  

There is diminishing motivation to go to the expense of developing more accurate models as the current range of models seem to be put to cost-effective use time and again and seem adequate for current building practices.
  
The buildings in which the current models are used for design of systems are all quite new, they are generally relatively expensive buildings and therefore relatively well managed.  This means that they tend not to have fires in them anyway.  

No one is quite sure if we are getting away with it for the time being whilst the systems are not up to the job or, as should be hoped (and I generally believe) , the systems are perfectly suitable to meet their objectives.  

The question is, or will be, can these buildings, built to what sometimes appear to be ad hoc standards, stand the test of time.  

All current models are guesses but they are the best guesses we have if we want to build the way we do.  The science, and I use that term reservedly, is young and for the time being we have little choice but to put up with the vagaries of youth.

And it's not just the models that can be controversial but also the acceptance criteria - I know of at least one building where it is anticipated that, during a fire, people will have to make their way some distance through smoky conditions with visibility down to 10m - but not just a few people, as might be intuitively acceptable in a block of flats say, but thousands upon thousands of people - I can't see it working personally.  All I can see is a disaster waiting to happen.

We will know in about 40 years time when the buildings are old and mis-managed and when they've had a few fires.

Ask me again then.

Stu

ps  Having said all that, with judicious application of the models and care taken over all other fire safety matters, we can provide buildings that are safe.

[John Webb]

As someone who used to work with the leading experimenters on smoke control systems at the Fire Research Station, I can echo Stu's comments about the problems of the expenses involved in developing or evaluating new models.

It is particularly difficult to check out a new building designed on the basis of a computer model of smoke control; even with well-designed 'hot smoke' tests they often can't be run to the limits of a system because damage might be caused - which the owners of a new building would not appreciate, of course! So one has to run a test which only reaches perhaps half the maximum capacity of a system and hope that gives a reasonable approximation to confirm the system design.

There is an inherent variability in the growth of fires; I used to reckon that if an experimental fire (using actual fuels, not controlled gas burns) was repeated and the data from the tests agreed to within 10% we were getting a repeatable fire!

[Phoenix]

There is an inherent variability in the growth of fires; I used to reckon that if an experimental fire (using actual fuels, not controlled gas burns) was repeated and the data from the tests agreed to within 10% we were getting a repeatable fire!

That's a point John.  One of the things that I dislike about conventional fire engineering is that deterministic (i.e. pin point and precise, purportedly accurate) results are produced when, in reality, a true reflection of reality would be if the results were given as a distribution - that is to say, as a range of possible outcomes with associated probabilities. 

I appreciate that the data does not exist and the variables are not always foreseeable enough to generate accurate outcome distributions.  What is important is that people recognise that the deterministic results they produce are only indicative of approximately what might happen in reality.  Results should not be taken, as they unfortunately often are, as precisely what will happen in reality.

Think of weather forecasts.  They do their best, poor souls, but are sometimes right and sometimes wrong.  And because they are sometimes wrong, the value of their correct predictions is diminished because people lose confidence in the forecasts generally.  How much better they would be to give probabilities of possible weather instead of deterministic predictions.  I would be much happier knowing that there was a 40% chance of a rain shower lasting between one and two hours tomorrow rather than them either telling me that it will or will not rain.  I could make my own decisions then on how to tackle the potentialities of the weather and would not get annoyed when the deterministic forecast was wrong.  Forecasts could be correct almost 100% of the time if they all adopted this probabilistic practice.  Some do already, I know, but most of the popular ones I see do not and are wrong half the time.   

Where was I...?

Never mind.

Stu

[Willo]

You two have  brought up some very interesting points. Stu's post on the 19th has prompted me to ask a question which I have been asking myself recently. Are there any records of these systems operating during a real building fire?

I understand what is being said about the difficulties of producing fire engineered solutions designed around a design fire size. It isn't that suprising when comparing the variables that other engineering disciplines have to deal with. When modeling other systems for engineering purposes the variables are comparatively fewer and easier to control than those in a fire. Where even slight differences in the composition of the fuel load can make big changes in the outputs that are trying to be analysed and modeled.  Maybe fire engineers should take a leaf out of the books of other engineers like structural engineers. I had a chat with one of them in the office today to get some idea of the scale of the factor of safety (the technical term is a fudge factor). It turns out it can be up to 35% over the design loads, sometimes more. Perhaps designing systems which will cope with a similar variation of design fire size would help provide a more effective solution. I suppose the next question to ask, especially for the case in point, would the clients be willing to pay for equipment with that much more capacity.

[John Webb]

Generally speaking, the suggestions made for smoke control systems by my colleagues at FRS were conservative and include some margins for safety.

For example, later work by myself on sprinklered fires before I retired indicated that the reduction of the '5MW' design fire to 2.5MW for sprinklered scenarios had a considerable safety margin - most sprinklered fires rarely got above 250kW in size and were rapidly reduced to a tenth of that after sprinkler operation. (Whereas similar unsprinklered fires reached anything up to 7MW!) These and other test fires were looked at about ten years ago in the Fire Engineer's Journal, in an article by Penny Morgan I recall. (Sorry, can't find my copy!)

[Benzerari]

You two have  brought up some very interesting points. Stu's post on the 19th has prompted me to ask a question which I have been asking myself recently. Are there any records of these systems operating during a real building fire?

I understand what is being said about the difficulties of producing fire engineered solutions designed around a design fire size. It isn't that suprising when comparing the variables that other engineering disciplines have to deal with. When modeling other systems for engineering purposes the variables are comparatively fewer and easier to control than those in a fire. Where even slight differences in the composition of the fuel load can make big changes in the outputs that are trying to be analysed and modeled.  Maybe fire engineers should take a leaf out of the books of other engineers like structural engineers. ...

In addition to Stu, Civvy, Kurnal and John's Adds..., I would say;

To have real-world genuine records saved in a database style of various case studies..., the existing fire safety systems have to integrate electronics data acquisition facilities to record loads fire safety aspects during real emergencies, and then Failure Analysis have to be carried out to assess the real scenario against the initial FSE design... this assessment and comparison may improve the probabilistic concept; which is sensibly much reasonable then the deterministic solution (as stated by Stu). particularly when enhancing the Probability of success Ps  this will automatically decrease the probability of failure Pf , since their summation, Ps + Pf = 1

Also you may consider the university of Ulster FireSERT institute ; http://prospectus.ulster.ac.uk/course/?id=7759