Experimental and numerical study for the principal fire dynamics within thermally-thin bounded compartments (case study: informal settlements in Cape Town, South Africa)
Item statusRestricted Access
Embargo end date24/06/2023
Beshir, Mohamed Mohyeldin Gaber
It is estimated that there are one billion people living in Informal Settlements (ISs) around the globe and that this number is increasing. Fire risks in ISs arise not just from poor infrastructure (e.g., lack of water or affordable safe energy) but also, dwelling proximity and flammable construction materials. Globally, over 95% of fire deaths and burn injuries are in low – and middle – income countries. Within ISs, the risk of a fire resulting in injury or death is particularly high. Informal settlement dwellings (ISD) are often built using cheap materials (e.g., steel sheets, timber) and poor construction designs (e.g., gaps between walls and walls collapsing during fires). This impacts the compartment fire dynamics, making them different compared to the "ordinary/normal" residential compartments fires. This thesis aims to provide a robust understanding of the fire dynamics within ISD through experimentally and numerically investigating the different physical aspects of IS compartment fires dynamics; with dwellings from the Western Cape, South Africa taken as a case study. IS compartments in Cape Town are commonly made out of a lightweight timber frame, with thin corrugated steel walls (thermally-thin), and often lined with cardboard for thermal insulation. The literature review revealed that, whilst under-ventilated compartment fires are well understood, there is still a lack of knowledge related to; the conditions for flashover in thermally-thin bounded compartments; the effect of vertical ventilation location or size on the time and Heat Release Rate (HRR) required for flashover (𝑞�̇𝑓�𝑜�); ventilation change post-flashover (which is one of the aspects seen in IS fires); robust validated CFD models for under-ventilated thermally-thin bounded compartment fires; or material testing databases to feed in these models. Additionally, little attention has been given to fire dynamics and spread in the presence of wind and most studies conducted reduced-scale models with wind tunnels which do not replicate reality. To fill the gap of knowledge and establish an understanding of fire behaviour in the thermally-thin bounded compartments of interest, quarter scale and full-scale under-ventilated lab based compartment fires were conducted. The compartments design was based on the ISO-9705 standard room dimensions with 0.51 mm thick steel cladded walls. These experiments were then used to validate the CFD model, namely Fire Dynamics Simulator (FDS), to further understand the physics at work in these compartments via parametric studies. Eight reduced-scale compartment fire experiments were conducted to study the 𝑞�̇𝑓�𝑜� and the heat fluxes to the surroundings from thermally-thin bounded compartments, and to provide an a-priori understanding of these novel compartments. It was found that heat transfer to and from thermally-thin compartment walls is dominated by radiation. The radiative heat transfer coefficient was theoretically resolved and correlated with the wall temperatures, gas layer temperature and the 𝑞�̇𝑓�𝑜�, to create a semi-empirical correlation for estimating the 𝑞�̇𝑓�𝑜�. Based on the a-priori reduced-scale experiments and modelling, thirteen full-scale under-ventilated compartment fires were conducted. It was demonstrated, both experimentally and numerically, that fuel location had a significant effect on the time to flashover and the magnitude of 𝑞�̇𝑓�𝑜�, while external heat fluxes and wall temperatures were only slightly affected. FDS modelling results showed that fuel location highly affects air entrainment and mixing within the wood crib, and thus affects the time to flashover and the magnitude of 𝑞�̇𝑓�𝑜�. Five different wall boundary conditions (BC) were studied, namely: sealed; non-sealed (leaky walls); highly insulated, thermally-thick bounded; and cardboard lined compartment walls. It was found that these BC had a significant effect on the internal and external fire dynamics, and that these dynamics are not accurately captured by the analytical and empirical equations available in literature. FDS was able to replicate the experimental results well and captured the main fire dynamics both qualitatively and quantitatively. A parametric study was conducted investigating the conditions for flashover in thermally-thin bounded compartments and an empirical equation for estimating the 𝑞�̇𝑓�𝑜� for thermally thin bounded compartments was conducted. Six different compartment ventilation conditions fire experiments and FDS models found, that the size and location of the window highly affects the conditions for flashover (i.e., time and 𝑞�̇𝑓�𝑜�) due to changes in the flow fields (i.e., cross vs. perpendicular flows). While the increase of the ventilation factor only slightly affected the time to flashover and the heat fluxes from the openings, it significantly increased the 𝑞�̇𝑓�𝑜�. The window’s aspect ratio was found to have no effect on the 𝑞�̇𝑓�𝑜� or the heat fluxes from the openings, however, taller narrower windows did lead to a slightly delayed flashover time. Two ISD were placed with opening-to-opening orientation at 1.0 m apart to investigate the fire development and spread mechanisms within and between the two dwellings. It was found that the fire spread in these conditions is mainly related to the wall collapse or large external plumes from the dwelling of origin and that the FDS model qualitatively captured the experimental results. An experimental and numerical study was also conducted to quantify the effect of adding horizontal roof openings, as a fire spread mitigation adaptation, on reducing the flame length and radiation out of vertical openings. In total, 19 reduced-scale compartment fires were conducted and used to validate an FDS model. It was demonstrated that adding horizontal openings significantly reduced the average heat flux measured at the door. Furthermore, a new empirical ventilation factor was generated to describe the flow field through both opening configurations showing a strong coupling with the inlet mass of fresh air to the compartment. The effect of wind on IS fire dynamics and spread was studied at three levels. First, the validated reduced-scale FDS model studied the influence of different wind speeds and directions on the 𝑞�̇𝑓�𝑜� for both thermally-thin and thermally-thick under-ventilated compartment fires (using a burner as a fuel source). The second study used the full-scale validated model to investigate the ability to scale-up the reduced-scale wind study’s conclusions using wood cribs instead of a burner (i.e., more realistic conditions). It was found that regardless of the wind direction, increasing the wind speed considerably reduced the time to flashover by enhancing the burning rate of the wood cribs, while wind direction had a relatively minor effect. Additionally, thermally-thin ISD presented a higher fire severity, with higher HRR, gas temperatures and radiation from the openings, compared to thermally-thick walled dwellings. Both the full-scale and reduced-scale studies show that both wind speed and wall thermal characteristics play important roles in the fire severity of IS fires and can highly affect the risk of fire spread. Lastly, the two dwellings fire spread model was updated to include wind effects, and provided further in-depth understanding of the effect of wind on the fire spread in more realistic conditions and highlighted the complexity of these systems (i.e., ISs). The work presented in this thesis has provided knowledge to fill gaps in the understanding of the fire dynamics within thermally-thin bounded compartments (with ISs in Cape Town as a case study). The thesis has also provided some best-practice guidance and material input parameters for numerical modelling of under ventilated thermally-thin compartment fires and has proposed empirical equations, in addition to engineering based solutions to reduce the fire spread risk in ISs. This thesis also showed that the fire spread in ISs can be highly affected by many factors and provides data that can feed into risk mapping or fire spread risk models for IS fires.