Section 2.6.1 of the REGULATION (EC) No 1272/2008 (CLP) defines a flammable liquid as:

Flammable liquid means a liquid having a flash point of not more than $60 ^{\circ}$C.

So data on flash point are needed for the classification of a liquid. The CLP goes on to state that data can be determined by testing, found in literature or calculated.

In the case of mixtures the flash point need not be determined experimentally if the calculated flash point of the mixture, using the method described in Gmehling and Rasmussen (Ind. Eng. Fundament, 21, 186, (1982))], is at least 5°C greater than the legally defined threshold/s. This assumes that the liquid is homogenous and we know certain details about the components.

I have had a look at this method which appears to be called the UNIFAC method. As a non-scientist, it looks overly complicated for situations where you just intuitively know that a mixture is not going to be flammable.

I have a hunch that if I pour 1 ml of food flavouring into a 5 litre bottle of propylene glycol (PG) the resulting mixture is not going be classed as 'flammable' (where the flash point of PG is deemed to be 104°C ).

Is this hunch one of those areas where science and intuition conflict?

Does it make sense to perform the whole UNIFAC process to confirm this conclusion or is there a simpler (but still scientific) 'rule of thumb' that we can use to discount having to perform a UNIFAC method?

For example, if I have a mixture containing:

  • 1 ml of pure water, which as far as I can tell has no flash point, and
  • 2 litres of PG (flashpoint 104°C)

Is the UNIFAC process the simplest way to determine the flammability of this mixture or is there a simpler, yet reliable, method to determine the flammability of this mixture?

  • $\begingroup$ A hunch is not a scientific observation, nor scientific evidence. The reason for needing scientific standards is to keep some moron with a hunch from doing something that is scientifically known as being stupid and injuring or killing someone. Hunches are the fodder for Darwin Awards. $\endgroup$
    – MaxW
    Commented Feb 19, 2019 at 17:39
  • $\begingroup$ @JohnWalker it basically depends how stringently you need to confirm this. If its just something you are doing at home, you can presumably just test it your self and be as confident as you feel you need. If you are trying to sell a product however then you likely have to do things by the book and perform tests to ensure it isn't flammable, regardless of your initial hunches. $\endgroup$
    – Tyberius
    Commented Feb 19, 2019 at 17:43
  • $\begingroup$ Thanks for your responses. I agree with using scientific method and logic in place of 'hunches'. $\endgroup$ Commented Feb 19, 2019 at 18:08
  • $\begingroup$ I am afraid that you took it too literally. If propylene glycol would be flammable, then you will have to demonstrate that adding something to it makes it inflammable. Not that if you mix two inflammable solvents/compounds you start to scratch. At least is what I think, just check the safety regulations. For instance "mixtures can be classified based on components" means that salt (not flammable) plus water (not flammable) means salted water is not flammable. Chem. expertise required if one can envision reactions able to drastically change the nature of the mixture, but it is just an extreme. $\endgroup$
    – Alchimista
    Commented Feb 20, 2019 at 9:09

1 Answer 1


UNIFAC is not a procedure, unless by chance in a similar field two things got the same name, which is unlikely.

UNIFAC is used by activity coefficient models for predicting liquid properties. It is often coupled with a vapor phase equation of state, and quite often simply the ideal gas law...

A flash point calculation is a procedure, and must be done in whole. Note: you can do either an experimental flash calculation, or a computational one!. UNIFAC is used in the latter. It combines the fact that at equilibrium a specie in both liquid and vapor must have the same chemical potential (gotten through via activity coefficients in the liquid phase and fugacity coefficients in the vapor typically) coupled with mass balance. You cannot do half of it and have an answer... like all calculations it must be done fully and correctly. It is not that difficult, but it should be done by someone who is a scientist or engineer... This is the type of thing that would be expected of a chemical engineer.

Most chemical engineering Thermodynamics textbooks have a decent section covering flash and bubble/dew point calculations.

There are many programs that can do flash calculations for you. VMGSim, Aspen/HYSIS to name just two of the many commercial/industrial simulators plus a plethora of small programs online made by people like me (mine isn't online ;) ).

I just did a quick google and this looks like it would work https://checalc.com/solved/multi_flash.html

Note, you do not need to use UNIQUAC in the liquid, but, it usually by default will use UNIFAC to fill in unknown interaction parameters, so it is basically a safe activity coefficient model. Depending on what is in the vapor, the peng-robinson or even ideal gas equations of state are probably good enough. If there is association or the molecules are polar, then I recommend the Haydon O Connell.

The link I sent looks like it uses the Peng-Robinson equation of state for liquid and vapor. It is probably okay for what you are doing. Actually the link I sent doesn't look like it has PG in its database... this one may... https://www.daoftx.com/usingflsc.htm

But you know what to look for now... These premade programs just need you to tell them the species, initial composition, T, P and then you just hit calculate. You just need to find one that has your molecules in its database.

  • $\begingroup$ Thanks. This is very informative. I would give you a +1 but I don't have enough ranking. :) $\endgroup$ Commented Feb 21, 2019 at 11:05
  • 1
    $\begingroup$ If it answered your question you can accept it, if it did not, let me know what my answer is missing. $\endgroup$
    – B. Kelly
    Commented Feb 23, 2019 at 20:03

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service and acknowledge you have read our privacy policy.

Not the answer you're looking for? Browse other questions tagged or ask your own question.