Correlative small scale roof fire testing.

There is only one known roof fire test that is required in the United States. This test is the ASTM E 108 (UL 790). There are no more than five laboratories in the United States that offer this test and the complete test can run anywhere from one to several thousand dollars per roofing system. The sample size is 3 ft. 4 in. by 4 ft. 4 in. and usually at least two samples are required for the first two parts of the test. The last part of the test requires a sample that is a test deck that can be anywhere from 8 ft. to 13 ft. long. The tests is comprised of three parts:

* Inter-mittent flame exposure - requires a minimum of two test decks. The test deck is exposed to a flame that is 36 in. long and has a 12 mph wind fanning the flame onto the deck. The exposure period and frequency depends upon the rating required.

* Burning brand - this part of the test can require either two or four deck samples for testing. A wooden brand composed of thin pieces of wood is placed on the deck and ignited. A 12 mph wind fans the flames of the brand while intensifying them.

* Spread of flame test - this test section requires two test decks. The same burner from the intermittent flame test is used for this part. Except the flame is either kept on for ten minute or four minutes, depending upon the classification.

There are three classifications that can be given for a roofing system. They are based on the level of fire protection that a roofing system can provide in a fire. The maximum fire protection is provided by an "A" classification, "B" is the middle classification and "C" classification is the lowest level. An independent survey performed by a roofing manufacturer indicated that the most often required classification is the "A". The next most required is the "C" classification and the last requirement are roofing systems with no classification. Therefore, the "B" classification is very rarely requested in most roofing systems. In some areas of the United States only "A" classification roofing systems are commercially available. This is driven by the building codes in the area.

The fire exposure severity increases for each test part, so that the A" classification has the most time exposure with the greatest fire intensity of the three classifications. To perform the complete series of tests is expensive and time consuming. The frustration enters when the sample fails on one of the test sections and passes the others. That one failure indicates that the roofing system needs to be reworked to pass that one test. To get to that point can takes weeks from the start of the test series.

A small scale test method has been developed to evaluate a roofing system for the full range of fire tests in a period of one day for a fraction of the cost of the complete test series. A survey of roofing manufacturers has shown that most often the test that the samples fail is the intermittent flame and burning brand test. The small scale test procedure addresses just those two parts of the test with excellent correlative 

Experimental

To establish a small scale test that readily duplicates that of the full size requires the investigation of the unique properties of the test requirements of the full size test. As in most fire tests the intensity of the fire relative to heat output and rate of heat with duration are the most common factors for fire test exposure. Each part of the ASTM E 108 that was under consideration would need to be scrutinized for these physical parameters. The test itself does not provide very accurate descriptions of the heat flux and depends primarily on the airflow and fire load for the exposure. Without going into extensive research at the cost of heading in the wrong direction, the primary approach was to scale down the test while keeping all the parameters the same. One of the major concerns was with the possibility of edge effects. Flame patterns were studied with the same type burner as in the full test but smaller. Temperature measurements where made with correlation to that of the requirements. It appeared initially that the small scale test could be made to work based on geometrical scaling down.

Other scenarios under consideration included using the cone calorimeter to duplicate the heat flux for this purpose. However, work by other researchers proved this to be a lengthy process with correlations not becoming evident. Many other ASTM tests were reviewed to see if they could be more easily modified than to down scale the ASTM E 108. There were too many inconsistencies for correlation between the other ASTM tests and the ASTM E 108. The economics were not evident to design a completely new test procedure from basic information. One of the goals was to keep the test simple and inexpensive. Down scaling was the most viable option for these reasons.

A size was needed to be selected that would be representative of the full size for purposes of obtaining the same results. There were several different ways to change the size. One was by area ratio reduction for all the sizes, the other was to reduce the dimensions or look at the fire exposure and air flow components. Other considerations were the size restrictions for a new piece of test equipment and could it be made in a modular configuration for easy storage when not in use. The final configuration was a combination of the different elements discussed. The sample size is 36 in. long by 20 in. wide. See figure I for construction details of the sample size. The fire temperature exposure of the full size was the same as the small scale test set-up. There was no change in hot gas exposure or air velocity. The final configuration of the set-up is detailed in figure 2.

It was decided to keep the air flow consistent with the full size test fixture and reduce the length of the pipe burner to accommodate the smaller size test sample. The most difficult part of the test set-up was to develop good air flow patterns at the desired speed of 12 mph. This was critical to keep the flames from the pipe burner traveling in a straight uniform direction and for the burning brand to develop the same intensity of heat as the full size. The burning brands for each different classification were also kept the same as the full size.

The small scale test set-up has the flexibility to be adjusted so that it could also perform the flame spread test on roof decks. It was also demonstrated that multiple sample runs were not necessary to obtain the data required. This was especially true when optical pyrometry was used to scan the different surfaces of the roof sample and trends where established for heat transfer models. The information obtained was sufficient for the manufacturers to establish the competency of the formulation for that particular roofing system. The addition of temperature and flux monitoring has been a valuable tool which more than compensated for the difference in size and strict visual subjective evaluations.

Results and discussions

Correlation was performed on "A" rated roofing systems. An "A" system achieved an "A" rating when tested in accordance with the construction details. However, when elements of the construction were altered to compromise the rating then the "A" rating was not achieved. These results were further verified by the manufacturers testing of roofing systems and obtaining different ratings with correlation to full scale. All types of roofing systems were tested on the small scale apparatus. This would include conventional type materials to the unique recycled plastic shingles.

The smaller size has made a tremendous difference in reducing the cost and time which takes into account the following:

* UPS shipping instead of common carrier;

* heat transfer information about the sample to reduce the number of samples that need to be tested;

* cost to manufacture sample;

* test cost.

Conclusion

The main concept that has been demonstrated: is full size multiple sample test can be downsized to single a sample with correlation. This is partly due to the use of sophisticated monitoring methods that were not available when the test was first originated. The new concept in business is to reduce the time and cost elements. Large fire tests or other types of tests are going to be performed less frequently. They will be over run by sophisticated monitoring transducers that have data acquisition. The resultant data will be manipulated and in many cases reviewed within a computer model. The ASTM E 108 is one of the first fire tests that has gone through this scrutiny and passed. More of the large size fire and flammability testing will succumb to the same scenario. The driving force will be private industry that needs to operate in the most efficient manner on an international basis. Other down sized tests will be developed for the rubber and roofing industry that will facilitate the evaluation of multiple compounds in a cost effective, short period of time.