Simulations of Heat Transfer through Multilayer Protective Clothing Exposed to Flame


 In this paper, the safety and thermal comfort of protective clothing used by firefighters was analyzed. Three-dimensional geometry and morphology models of real multilayer assemblies used in thermal protective clothing were mapped by selected Computer-Aided Design (CAD) software. In the designed assembly models, different scales of the resolution were used for the particular layers – a homogenization for nonwoven fabrics model and designing the geometry of the individual yarns in the model of woven fabrics. Then, the finite volume method to simulate heat transfer through the assemblies caused by their exposure to the flame was applied. Finally, the simulation results with experimental measurements conducted according to the EN ISO 9151 were compared. Based on both the experimental and simulation results, parameters describing the tested clothing protective features directly affecting the firefighter’s safety were determined. As a result of the experiment and simulations, comparable values of these parameters were determined, which could show that used methods are an efficient tool in studying the thermal properties of multilayer protective clothing.


Introduction
The problem of thermal comfort is inextricably linked to the issue of heat balance between the human body and its environment. The human body continuously exchanges heat with the surrounding environment. This exchange occurs via four different phenomena: conduction, convection, radiation, and sweat evaporation. Each phenomenon is conditioned on the characteristics of the human body including metabolism, temperature, sweat rate, breathing rate, as well as the environment described in terms of air temperature, heat radiation, relative humidity, and velocity of airfl ow.
Clothing is an essential factor playing a signifi cant role in the heat balance because it creates a barrier between the human body and its environment. In hot environments, clothing protects against excessive heat and help thermoregulatory mechanisms by producing an artifi cial microclimate next to the human body. These usages are crucial to fi refi ghters, heavy industry workers (ironworks employees), rescuers, and special forces soldiers. Because during their professional activities they may be exposed to fl ame and the thermal radiation and because of this, their clothing should be designed to protect them against thermal hazards and prevent skin burn.
There are several types of fl ame-retardant treated fabrics (cotton FR and viscose FR) and fabrics made from inherently fl ame resistant fi bers (aramid, PBI). Protective clothing made from such fabrics does not ignite due to contact with the fl ame.
However, during various professional activities, the heat fl ux intensity in the surroundings may even exceed 60-200 kW⋅m −2 [1]. These numbers imply that the thermal protective clothing should be evaluated not only for protection against fl ame but also against heat fl ux that often results in skin burns.
The heat transfer through protective clothing depends on many factors, such as the geometry of the textiles (thickness, porosity, mutual positions of layers, and fi ber number in a yarn cross-section) and the thermal properties of raw materials (specifi c heat, thermal conductivity, and emissivity). The thermal protection features of multilayer protective clothing are tested following EN ISO 9151 [2] (fl ame exposure) and EN ISO 6942 [3] (radiation exposure). These standards provide two metrics that are critical for evaluating the protective properties of clothing against thermal hazards [2,3]. The fi rst metric describes the amount of time needed for the temperature of the inner side of clothing to increase by 12°C. This temperature increase is assumed to be a pain threshold for users. The second metric describes the amount of time needed for the temperature of the inner side of clothing to increase by 24°C. When crossing this temperature threshold indicates that the user experienced second-degree burns.
The test methods presented in EN ISO 6942 and EN ISO 9151 are based on the measurement of the temperature change in the copper calorimeter. The main part of the calorimeter is a copper plate as a sensor placed right behind a clothing specimen, which is exposed to heat fl ux. However, the thermal properties of the copper plate, such as heat conductance or thermal capacity, deviate considerably from the human skin. This means that the results of testing the thermal protective properties of clothing obtained by this method may not refl ect real situations. For this reason, Keltner [4] suggested replacing the calorimeter with a thermal skin simulant sensor.
Many studies have been undertaken to develop new materials for the manufacturing of protective clothing as well as to create heat transfer models for the prediction of their thermal properties [5][6][7][8][9]. One such model was proposed by Torvi [10] for textiles under high heat fl ux conditions. Song et al. [11] presented a numerical model of heat and moisture transfer through multilayer protective clothing during exposure to a fl ash fi re. The model predictions were compared with experimental outcomes from various textiles systems.
Nowadays, numerical simulations are effective and widely applied in tools to study thermal processes occurring in thermal protective clothing. The simulations are performed on real clothing models and apply the numerical analysis of physical phenomena to predict the properties of the clothing. Currently, the numerical simulations are a supplement of typical traditional experiments with actual materials [12][13][14][15][16][17][18]. The goal of the research presented by Onofrei et al. [14] was to create a heat transfer model in the fi refi ghter's garments exposed to heat radiation. Zhu and Zhou [16] developed a model for moisture and heat exchange through a fi refi ghter's protective clothing when exposed to heat radiation.
The presented work is a continuation of research on modeling physical phenomena that occurred in textiles [19][20][21]. However, mentioned studies concerned the modeling of thermal phenomena in solids under steady-state conditions. In the current paper, the CAD software to predict heat transfer in textile assemblies intended for thermal protective clothing when exposed to the fl ame was used. The 3D models of the two actual multilayer assemblies were designed. Different scales of mapping were used, ranging from a homogenization in nonwoven models to mapping of real geometry of the yarns in woven fabrics models. Next, the computational fl uid dynamics to model heat transfer induced by fl ame through multilayer clothing under transient state conditions was used. Finally, the utility of the models through comparisons with experimental outcomes was verifi ed.

Materials
Two multilayer assemblies used in thermal protective clothing were studied ( Figure 1). The assemblies consisted of fi ve layers: A -outer shell (woven fabric), B -moisture barrier (membrane), C -nonwoven fabric, D -nonwoven fabric, and E -lining (woven fabric). In both assemblies, Layer B was laminated with Layer C, while in the Assembly 2 Layer D was laminated with Layer E.
Characteristics of multilayer assemblies were presented in Table 1.
Total porosity P were determined according to Eq (1) [22]. (1) where: M p -mass per unit area in g⋅m -2 , d -thickness in mm, and ρ -density of the raw material in g⋅cm -3 . Total porosity P of woven fabrics (Layers A and E) resulted from porosity of the yarn P yarn and the free spaces between the yarns from which these fabrics are formed. P yarn was determined by analyzing the SEM images of the yarn cross-section using Image J software.
All fi ve layers are built of fl ame-resistant materials but play different roles. Layer A protects against mechanical wear and tear. Layer B (membrane) does not allow water to enter, however, being vapor permeable, it provides air and water vapor (sweat) exchange between the skin and the environment. Layers C and D are the thermal insulations, while Layer E is the lining.

Test apparatus and method
The heat transfer through the multilayer clothing caused by exposure to fl ame in a transient state was investigated. The tested samples were conditioned in an atmosphere with relative humidity (RH) of 65% and a temperature of 20°C before testing for at least 24 h. The studies were carried out using the test apparatus according to EN ISO 9151, under conditions as follows: air temperature of 20°C and RH of 40%. Figure 2 illustrates the experimental setup.
The measuring system consisted of a Meker gas burner, a copper disc calorimeter, a thermometer, and a PC with dedicated software. During the measurement, a horizontally oriented test sample is subjected to an incident heat fl ux induced by the fl ame of a gas burner located under it. The heat falls to the outer surface of the tested sample (Layer A) and passing through other layers of assembly is measured by the calorimeter

Model design
In the designed 3D models of two tested assemblies ( Figure  3) both woven fabrics (Layers A and E) were mapped with the following parameters: (1) layer thickness, (2) distance between yarns, and (3) elliptic cross-section of yarns.
In the designed models, each layer was designed separately, and then the layers were combined into the fi nal assembly.
In the designed models, the layers lie on top of each other in direct contact (no air gap). The model does not take into account sewing threads connecting layers due to its negligible low weight and negligibly low impact on heat transfer inside the assembly.
Both nonwovens (Layers C and D), membrane (Layer B) and yarn in both woven fabrics (Layers A and E) due to their complex internal structure, were designed as the homogenized three-having direct contact with Layer E. The calorimeter was linked to a digital, computer-driven thermometer, which records the growth of Layer E temperature in time. Measurements of the temperature of the inner surface of the sample (Layer E) were recorded in 1 second intervals until the temperature of the Layer E increased to 50°C. The measurements were carried out for two intensity levels of incident heat fl ux: 45 kW⋅m −2 and 58 kW⋅m −2 . The fi nal results of these tests were the average values of three individual tests made for each of the two above mentioned intensity levels of heat fl ux. Based on the experiment, following thermal parameters were determined: -heat transfer index (HTI 12 and HTI 24 ) means time to achieve a temperature rise of 12°C and 24°C, respectively, in the calorimeter, when a specifi ed heat fl ux acts on the sample.

Results and discussion
In Figures 5 and 6, a comparison between the experimental and the simulated results was presented. It shows a pronounced relationship between the temperature of the inner surface of tested assemblies and the intensity level of heat fl ux directed to the outer surface of the sample.
Both experimental and simulated results obtained for tested levels of heat fl ux density show that the temperature rise process of Layer E could be imparted into two stages. The fi rst stage takes about 3-5 s. During the stage, the temperature of Layer E remains almost unchanged. This could be caused by the multilayer, porous sample structure, the thermal capacity dimensional objects with physical features (density, thermal conductivity, and specifi c heat) resulting from corresponding porosity showed in Table 1. Table 2 presents the basic physical parameters of the materials from which the tested assemblies are made [23,24].

Physical basis of heat fl ow simulation
The fi nite volume method was carried out using Solidworks Flow Simulation 2014 software to analyze the heat transfer through the tested assembly. The software allows to predict fl uid fl ow solving energy conservation equations and Navier-Stokes formulas [18] and allows estimating simultaneous heat transfer in solid, liquid, and gaseous states and incorporates energy exchange between these states. The above-mentioned equations are augmented by fl uid state equations and by empirical dependence of fl uid density, viscosity, and thermal conductivity on temperature. The software allows to analyze following physical phenomena: (1) heat transfer in solids (conduction), (2) free, forced, and mixed convection, and (3) radiation both in the steady-state and transient state [25]. A broader description of the physical laws based on which the simulations were carried out was presented in earlier work on modeling of thermal performance of multilayer protective clothing exposed to radiant heat [22].

3.2.2.Conditions of heat fl ow simulations
The main aim of the heat transfer simulations was to determine the time dependence of the temperature of Layer E of the two assembles models for two intensity levels of heat fl ux which were used in the experimental measurements. For this purpose, the assembly model was placed inside a rectangular computational domain fi lled with air (presented in Figure  4)

Conclusions
In this paper, the outcomes of the modeling of the two textile assemblies intended for multilayer protective clothing to calculate the heat transport through the assembly caused by fl ame were obtained using the computational fl uid dynamics (CFD). Two assembly models built of fi ve layers consisting of two woven fabrics, two nonwoven fabrics, and membrane, were designed. The models assumed the simplifi cations regarding both geometry and the internal structure of individual layers. The model has been validated by experiment results performed according to EN ISO 9151. Simulations carried out on both models gave results that correlated with experimental curves describing the relationship between the temperature of lining and outer shell exposure time to heat fl ux generated by gas Meker burner. The two parameters of protective clothing (HTI and HTF) were determined experimentally and by simulation. The experimental results are compatible with simulated of the sample, and the thermal inertia of the calorimeter used. The inertia could infl uence the precision of temperature measurements of the sample. During the second phase of the temperature rise, the temperature of the sample inner surface (Layer E) is approximately in a linear time dependence with the slope proportional to the heating rate of the sample. One can see that for both the tested intensity levels of heat fl ux, the modeled heating rate of the sample model consistently overestimated the heating rate of the real sample during experimental measurements. The slope of the temperature versus time modeled curves is higher than the slope of the curves obtained experimentally. The thermal parameters obtained in the experiment and by simulation for two tested assemblies were presented in Table 3.
Relative disagreements between experimental and modeled parameter estimates depended on the intensity level of heat fl ux, they are: 8-13% for HTI 12 , 0-6% for HTI 24 , and 0-19% for HTF. The comparative analysis of the simulated and experimental values of these parameters revealed the biggest compatibility between the outcomes in the case of HTI 24 (Assembly 2, Q = 58 kW⋅m -2 ), however, the smallest one in the case of HTF (Assembly 1, Q = 58 kW⋅m -2 ). As expected, the HTI 24 value that expresses the time of the clothing exposure to heat fl ux at which second-degree burns may occur, decreases   predictions and the dissimilarity between them depends on the specifi c parameter and the applied incident heat fl ux density. These differences could be related to the main diffi culty of textiles modeling (referring to above all simplifi cation concerning the complicated geometry of real textiles) like: -Homogenization (fi bers with air) instead of mapping in the models the single fi bers and free gaps between them, -In the woven fabrics model, a constant elliptic crosssection of yarn constant in all repeat was adopted. In the real woven fabrics, the shape of the cross-section of yarn changes constantly due to frictional forces between warp and weft, and -Real direct contact of the nearest layers (infl uencing the interlayer thermal conductivity) may be different in the model than in the actual multilayer textile assembly.
Despite the simplifi cations mentioned above, the designed 3D models of multilayer assemblies and performed simulations allow to: -predict parameters characterizing protective clothing with an error of 0-16% (Assembly 1) and with an error of 0-10% (Assembly 2), -understand the phenomenon of heat transfer through multilayer textile structures, and -optimize the thermal protection properties of protective clothing.