The fire tests are quite realistic; a very large piece of ceiling is constructed (appx 20 ft. sq), and is subjected to an enormous, controlled natural gas fire for a specific time. Temperatures are monitored on the "unexposed" face; for it will not do much good if the outside itself gets hot enough to light a new fire! Nor will it help if the structure is weakened to the point of collapse.
That test is based on the standard time temperature curve and actual structure fires are quite a bit hotter. The curve was developed from measurements of actual fires in ordinary occupancies (masonry bearing walls with wood joists and rafters). The testing was done in the days prior to the use of large amounts of synthetic materials in the built environment. The synthetic materials cause the fires to develop faster and they give up their BTUs much more quickly. So even though the synthetics average fewer BTUs per pound by burning so much more quickly they cause a much sharper temperature rise. The standard time temperature curve is still useful in comparing materials and techniques to each other but it no longer reflects the stresses that todays fires subject the materials to.
