To be able to decide on the correct thickness of fire protection material required to protect steel sections of various dimensions for various periods of fire resistance, it is necessary to test a range of small-sized specimens with different protection thicknesses as well as some larger, loaded specimens. This test package has to be carefully designed to suit the type of fire protection material in order to cover the widest possible range of applications.
Having obtained all this data, (from 15 - 20 test samples), the fun begins!. Sometimes a graphical assessment can be made but in the majority of cases mathematical solutions using a Multiple Linear Regression (MLR) formula are required.
The MLR favoured by the British ASFP is not too difficult :-
Fire resistance = k₀ + k₁A/H_p + k₂t,
where the k's are constants derived from the data,  A/H_p  is derived from the section factor of the steel sample and t is the thickness of protection applied. Once you've solved the MLR, a table of data can be produced and published in the "Yellow Book".

However, the European version of the MLR solution is rather more complicated........!!!