View the PDF here

Fire Development on Sloped Roofs with PV Modules: An Experimental Study

By: Reidar Stølen, Ragni Fjellgaard Mikalsen, Janne Siren Fjærestad, Anne Steen-Hansen

(RISE Fire Research and NTNU Norwegian University of Science and Technology, Trondheim Norway)

The number of photovoltaic (PV) installations on rooftops has increased rapidly over the last years. The changes in fire dynamics caused by the addition of PV modules have been studied previously and it is clear that the modules influence the risk of fires spreading on the roof [1,2].

RISE Fire Research has recently investigated these effects on a sloped roof based on the test method CEN/TS 1187, test 2 [3] that is used for classification of roofing surfaces in Scandinavia. In this test, a burning wood crib, weighing approximately 40 g, is placed on the roofing surface where a wind velocity of 2 or 4 m/s is maintained. After the test, the length of the damaged roofing material is measured, and should be less than 55 cm for the roofing to qualify for the class BROOF(t2) according to EN 13501-5 [4]. To simulate the installation of a PV module, a steel panel was installed above the roofing surface, and the effects on fire propagation were studied. In addition to the experiments in small-scale based on the CEN/TS 1187 test method, the setup was scaled up to medium- and full- scale to investigate how the small-scale data could relate to a more realistic building construction. The same materials, and a roof inclination of 30 ° were used in all the experiments.

A steel plate was used to represent an inert PV module without any contribution of combustible materials based on experiments from Kristensen et al. [1]. The small-scale experiments showed that a simulated PV module installed with a gap of 6 cm above the roofing surface clearly increased the damaged length of the roofing surface compared with no module. When the simulated module was installed with a gap of 9 cm or more, no significant effect at this scale was found for module versus no module. The critical gap height between 6 and 9 cm, as observed in these experiments is relatively low compared with other experiments, on flat roofs with different materials and ignition sources, where large changes in fire spread occurred in the range 17 – 20 cm [5] and 11 – 12 cm [1]. A photo from a small-scale experiment can be seen in Figure 1.

Figure 1: Small-scale experiment based on test 2 in CEN/TS 1187 with a steel plate representing a PV module installed over the roof. The steel plate is marked with a blue dashed line and the roofing surface is marked with a green dashed line Photo: RISE Fire Research.

In a medium-scale experimental setup, the chosen distance between roof surface and PV module was 12 cm, and different sizes of ignition sources were studied. The largest ignition source (corresponding to the B-brand from the UL 790 test method, weighing approximately 500 g [6]) caused the fire to spread all the way up along the roof in the cavity below the module. A high velocity buoyancy-driven flame was observed in the cavity when the gases in the cavity heated up and started rising as seen in Figure 2.

Figure 2: Medium-scale experiment with a fire propagating in the cavity between the roofing surface and the steel PV module. Photo: RISE Fire Research.

Two full-scale experiments with dimensions representative of a small house, with a roof length of 5.4 m, were performed using the same gap height and ignition source as in the medium-scale experiments. Like the experiment in medium scale, the fire propagated and spread up the roof surface. Melted and burning debris from the roofing membrane also caused the fire to spread down the sloped roof. This can be seen in Figure 3.


Figure 3: Full-scale experiment with a fire propagating up and down a sloped roof with 6 steel PV modules. Note the flames extending from the top of the simulated PV modules to the top of the roof and beyond. The burning melted material can be seen between the lower edge of the modules and the ground.  Photo: RISE Fire Research.

The overall conclusion of the study is that installed PV modules can cause increased fire propagation on a sloped roof surface. The extent of this effect was found to depend on parameters such as the gap distance and the size of the ignition source. Although not studied in these experiments, the angle of the PV modules may also influence the fire spread, as found in previous studies [7,8].

The study also included a survey among firefighters in Norway. The respondents emphasized the need for training, knowledge, and necessary tools for firefighting in buildings with PV installation. Two reports from this study are available for download (see link below) [9,10]. They are in Norwegian, with an English summary. The results from the study will also be published in English in an international scientific journal.

Having PV modules installed on roofs is well established, although the effect on fire safety is not fully understood. Currently, the committee working with the standardization of fire testing of external fire exposure to roofs, CEN TC 127 WG5 has initiated work on a future standardized test method to include these effects. 

The full reports by RISE Fire Research with details on the study including the experimental work (RISE Report 2022:82 and RISE Report 2022:83) can be downloaded here:

[1]          J. S. Kristensen, B. Jacobs, and G. Jomaas, ‘Experimental Study of the Fire Dynamics in a Semi-enclosure Formed by Photovoltaic (PV) Installations on Flat Roof Constructions’, Fire Technol., Mar. 2022.

[2]          B. Backstrom, M. Tabaddor, and P. Gandhi, ‘Effect of Rack Mounted Photovoltaic Modules on the Fire Classification Rating of Roofing Assemblies Phase 1’, Underwriters Laboratories Inc., 2009.

[3]          ‘CEN/TS 1187_2012. Test methods for external fire exposure to roofs’. CEN, European Committee for Standardization, 2012.

[4]          ‘EN 13501-5:2016 Fire classification of construction products and building elements – Part 5: Classification using data from external fire exposure to roofs tests’. CEN, European Committee for Standardization, Jun. 2016.

[5]          J. S. Kristensen, F. B. M. Faudzi, and G. Jomaas, ‘Experimental study of flame spread underneath photovoltaic (PV) modules’, Fire Saf. J., p. 103027, May. 2020.

[6]          ‘UL790, Standard Test Methods for Fire Tests of Roof Coverings’. Underwriters Laboratories, 22 Apr. 2004.

[7]          X. Ju et al., ‘Impact of flat roof–integrated solar photovoltaic installation mode on building fire safety’, Fire Mater., 2019.

[8]          F. Tang et al., ‘Experimental study and analysis of radiation heat fluxes received by a floor beneath an inclined ceiling’, Fire Mater., vol. 45, no. 2, pp. 205–214, Mar. 2021.

[9]          R. F. Mikalsen, J. S. Fjærestad, R. Stølen, and O. A. Holmvaag, ‘EBOB – Solcelleinstallasjoner på bygg. Brannspredning og sikkerhet for brannvesen. Del 1: Hovedrapport.’, RISE Rapport 2022:82.

[10]        R. Stølen, J. S. Fjærestad, and R. F. Mikalsen, ‘EBOB – Solcelleinstallasjonar på bygg. Eksperimentell studie av brannspreiing i holrom bak solcellemodular på skrå takflater. Del 2: Teknisk rapport.’, RISE Fire Research, RISE Rapport 2022:83, 2022.