Comparative Study on Validation of Multi-Step Chemical Reaction Model in Fire Simulation

Daehwan Yu; Joonho Jeon

doi:10.7731/KIFSE.42ad2fc1

Abstract

Fire involves a number of chemical reactions, but most fire simulations do not consider multi-step chemical processes. Using multi-step chemical reactions enalbles more realistic fire simulation. In this study, ANSYS FLUENT^® used to simulate pool fire with multi-step chemical reactions. It is possible to predict various species concentrations involved in the fire process, not only thermal properties. In this study, the fire is modeled using two different computational fluid dynamics software ANSYS FLUENT^® and Fire Dynamic Simulator (FDS). The validity of the ANSYS FLUENT^® fire simulation is assessed by comparing temperature and species concentration results with experimental and FDS results. Unlike FDS fire simulation, ANSYS FLUENT^® fire simulation with multi-step chemical reactions is able to analyze intermediate species concentration compared with the experimental results. Furthermore, ANSYS FLUENT^® fire simulation can measure various species involved in fire, so it is expected to predict various aspects of fire.

Keywords: Computational fluid dynamics; Fluid dynamics; Fire chemistry; ANSYS FLUENT^®; FDS; Chemical reaction

1. Introduction

Fire is a thermal and chemical chain reaction influenced by the composition of combustibles, and combustion conditions. Predicting these multiple features of the fire accurately, experiment data is essential. Acquiring such data, though, can be constrained by difficulties in securing experimental facilities, high costs, and the danger of fire itself. With the improvement of computational performance, computational fluid dynamics (CFD) software have been developed. Using CFD software can be able to retain data and predict fire without experiment, then it is possible to release difficulties of fire experiment partially. However, simulating fire was challenging due to the intricate nature of fire, which involves complex chemical chain reactions. To resolve these problems, the simplification of chemical reactions is mainly proposed in CFD software.

Using CFD software with these simplifications, the fire phenomenon has been simulated continuously. One of the CFD software widely used to simulate fire is the Fire Dynamic Simulator (FDS) developed by National Institute of Standards and Technology (NIST). FDS has the advantage of fast calculations and numerical stability using simplification. Another CFD software simulated fire phenomenon is ANSYS FLUENT^®. ANSYS FLUENT^® developed by ANSYS is a software used to simulate various industries because it has a lot of reliable models to utilize chemical processes, combustion, energy, medical and so on[1].

There have been processed numerous studies simulating the fire. Initially, basic forms of fire, such as fire plumes and pool fires, were simulated[2-4]. As time goes on, computational performance has improved more increases, it can be possible to simulate large-scale fires occurring in compartments, buildings, apartments, tunnels, and even wildfires. Comparative studies have been conducted between ANSYS FLUENT^® and FDS. P. Lin et al.[5,6] investigated a semi-transversal smoke control system in tunnel fires and compared both simulations’ results. W. Węgrzyński et al.[7] evaluated the visibility according to soot yield results with experiment, ANSYS FLUENT^® and FDS. M. Król et al.[8] analyzed the phase of fire spreading in the furnished room by validating experiments and simulations with ANSYS FLUENT^® and FDS. These studies have processed by simplifying chemical reactions in fire simulations.

However, fire does not fundamentally occur simplified processes. As fire scenarios are diversified and advanced, it is crucial to analyze species that participated in the fire. Despite fire simulation with multi-step chemical reactions also needed, few studies on multi-step chemical reactions have been found in the literature[9,10].

For this reason, the present study performed the fire simulation with multi-step chemical reactions using ANSYS FLUENT^®. The purpose of this study is to verify whether complex calculated fire model shows a reasonable consequence compared to experiment and simulation.

2. Numerical Methods

2.1 Fire modeling

In this study, ANSYS FLUENT^® and FDS were used to simulate fire. Pool fire, one of the simplest forms of fire, was selected because it is almost a steady-state fire and experiments have been performed a lot. Referring to the methanol pool fire experiment[11], the schematic of analysis space is 1 m × 1 m × 1 m and the pool fire diameter is 0.3 m shown in Figure 1. The ambient air temperature is taken as 298 K and methanol is selected as fuel to advantage simple combustion mechanism and widely used in pool fire experiment. Methanol properties are shown in Table 1 and the mass burning flux of methanol was measured at 12.4 g/m2⋅s through the experiments[12,13]. The total simulation time is 60 s and results are extracted in the period from 25 s to 60 s because referred experiment results are also set to be obtained as steady-state fire[11].

2.2 ANSYS FLUENT^® modeling

The models used in ANSYS FLUENT are shown in Table 2. The turbulence was resolved with the SST k-ω model which is widely used for modeling turbulence flow. The species transport reacting model in ANSYS FLUENT^® has been selected to model the fire simulation with multi-step chemical reactions. This model is a chemical solver that can calculate temperature as well as lots of species concentration by importing detailed combustion mechanism. This study imported the GRI-3.0 mechanism that consists of 53 species and 325 reactions and it is possible to predict combustion of methane and other hydrocarbon mixtures[14,15]. The finite-rate/eddy-dissipation model has been selected to consider interaction of finite rate chemistry and turbulence of fire. In this model, the slower rate is selected between finite rate and eddy dissipation rate. When fire produces various species such as carbon monoxide, water vapor, and smoke then, the surrounding space is raised in optical density and it has a significant effect on absorption, emission, and scattering. The radiation discrete ordinates model is able to calculate these absorbing, emitting, and scattering effectively and also can deal with a wide spectrum of optical density in each medium respectively.

2.3 FDS modeling

In this study, FDS ver 6.7.5 developed by the NIST was used to implement methanol pool fire. The FDS calculated turbulence using very large eddy simulation (VLES) model by default. The chemical reaction of methanol pool fire is simple chemical reaction consist of reactants and products. The CO and soot yield is set to 0 and fire growth of methanol is set to ultrafast[12]. To implement pool fire, the burner schematic uses the circular vents model[16].

2.4 Mesh sensitivity analysis

CFD simulations always require an appropriate and well-defined mesh size to calculate accurate analysis. Through Eq. (1), both simulations make an appropriate mesh size with the value of characteristic fire diameter, [16,17].

(1)

D*=(Q˙ρ∞cpT∞g)25

where, Q˙ is the maximum heat release rate [k,W], ρ_∞ is the ambient air density [kg/m³], c_p is the specific heat of air at constant pressure [kJ/kg⋅K], T_∞ is the ambient temperature [K], g is the gravitational acceleration [m/s²]. The maximum heat release rate is calculated by Eq. (2).

(2)

Q˙=m˙″AΔHc

where, m˙″ is the mass burning flux [g/m²⋅s],  is the area of fire source [m²], △H_c is the heat of combustion [kJ/g]. Using this equation, the appropriate mesh size was selected for two simulation models, as shown in Table 3. Using Eqs. (1) and (2), the maximum heat release rate was determined to be 17.38 kW and the characteristic fire diameter was 0.1889 m. According to NUREG-1824, the recommended fire simulation mesh size select D*/dx between 4 to 16[17]. In this study, the appropriate mesh size range was calculated as 0.012 m to 0.047 m. Within this range, mesh sensitivity analysis was conducted for ANSYS FLUENT^® and FDS using different mesh sizes. The temperature results using mesh sensitivity analysis were measured at 0.1 m height away from the center of the fire source. As shown in Figure 2, both simulations measured lower temperature at the mesh size of 0.05 m. The FDS temperature results of 0.02 m and 0.04 m mesh sizes were similar temperature results. And ANSYS FLUENT^® temperature results of 0.02 m and 0.04 m mesh sizes were closed to experimental results as steady-state fire. But, the measurement points spacing is less than 0.40 m, so both simulations selected 0.20 m mesh size. The detailed mesh properties of FDS and ANSYS FLUENT^® are shown in Table 3.

3. Results and Discussion

3.1 Mean temperature measurement

The mean temperature measurement positions in this study matched used in the experiment[11], along the centerline of the fire source, as shown in Figure 1(a). Figure 3(a) shows temperature results of ANSYS FLUENT^® and FDS. The temperature results from ANSYS FLUENT^® and FDS were nearly identical both result and tendency. In the range of 0.1 m to 0.2 m, the temperature predicted by ANSYS FLUENT^® was overestimated compared to experimental results. Figure 3(b) shows the temperature error analysis with the error margin of 30%, based on experimental temperature results. All temperature results show within the error margin, indicating that ANSYS FLUENT^® and FDS accurately predict pool fire temperatures. These results confirm that fire simulation with multi-step chemical reactions is possible to predict reliable pool fire temperature.

3.2 Species concentration measurement

The species concentration results were compared with the experimental and two fire models. Figure 4 presents volume fraction results for five species along the centerline, classified into reactants (CH₃OH and O₂), products (H₂O and CO₂), and inert material (N₂). The concentration results of CH₃OH indicated the highest volume fraction, while O₂ indicated the lowest volume fraction near the fire source, as shown in Figure 4(a) and 4(b). In Figure 4(a), FDS calculated more accurate results than ANSYS FLUENT^® at the lower measurement positions. The accuracy of ANSYS FLUENT^® improved with increasing height. As shown in Figure 5(b), the volume fraction of O₂ was nearly identical in both ANSYS FLUENT^® and FDS. In FDS, the lower oxygen consumption near the fire source leads to the lower temperature than the experimental results, as shown in Figure 3. And volume fractions of CO₂, H₂O and N₂ also show similar tendencies in ANSYS FLUENT^® and FDS. Based on species concentration analysis, ANSYS FLUENT^® fire model with multi-step chemical reactions successfully calculated pool fire behavior as validated by temperature and species concentration measurements. Further analysis of ANSYS FLUENT^® methanol concentration results means that the low methanol concentrations at lower measurement are caused by rapid methanol decomposition as shown in Figure 4(a). Rapid methanol decomposition should lead to the increase in product concentraions. However, the measure product concentrations did not reflect this expected increase. Therefore, ANSYS FLUENT^® with multi-step chemical reactions was used to analysis intermediate species concentrations.

3.3 Intermediate and other species concentration measurement

Figure 5 shows the species concentration of intermediates (CO and H₂O) calculated from ANSYS FLUENT^® fire model. The volume fraction of intermediates were similar to experimental results at higher positions but were overestimated at lower positions. This means the rapid decomposition of methanol led to lower CH₃OH concentrations and the higher intermediate concentrations, as shown in Figures 4(a) and 5. Also, both intermediates are combustible gases with high reactivity and exothermic properties. As a results, the higher intermediate species concentration of ANSYS FLUENT^® leads to the higher temperatures than the experimental results, as shown in Figure 3.

The ANSYS FLUENT^® fire model was able to measure additional species concentrations using the detailed chemical mechanism. This is one of the adventages of using this simulation. Figure 6 shows the volume fraction of radical (OH) and toxic gases (CH₄, HCN, NO), which are difficult to calculate using FDS fire model. Regardless of production, species involved in detailed chemical mechanism could be measured.

4. Conclusions

The existing fire simulations rely on simplified chemical reactions. However, fire involves complex chemical processes, such simplification may not provied an accurate representation of fire behavior. Therefore, this study performed fire simulations using the CFD software ANSYS FLUENT^® which can simulate complex fluid phenomena and calculate detailed species involved in fires. The 0.3 m methanol pool fire simulations were performed and results were compared with experiment and FDS.

The ANSYS FLUENT^® fire model produced accurate predictions of both temperature and species concentrations compared to the experiment and FDS, is widely validated fire simulation. It means that pool fire simulation considering multi-step chemical reactions verified its validity. In addition, ANSYS FLUENT^® simulation calculated with multi-step chemical reactions was able to predict intermediates and other species concentrations involved in the detailed mechanism.

Further studies are needed to improve the accuracy of intermediat species predictions and to verify whether fire simulation using multi-step chemical reactions can accurately represent other indicators, such as radical, and toxic gases. Additionally, different types of fire, such as large-scale compartment fires and composite material fires, should be simulated whether fire model with multi-step chemical reactions can be applied. Through these verifications, fire simulation is expected to predict various aspects of fire accurately.

Notes

Author Contributions

Methodology, investigation, writing-original draft preparation, Daehwan Yu; conceptualization, writing-review and editing, supervision, Joonho Jeon. All authors have read and agreed to the published version of the manuscript.

Conflicts of Interest

The authors declare no conflict of interest.

Acknowledgments

This research was supported by the Basic Science Research Program through the National Research Foundation of Korea (NRF) funded by the Ministry of Education (2021R1I1A305197611). The authors would also like to acknowledge Jung-hun Noh, Institute for Advanced Engineering for research support ANSYS FLUENT^®.

Figure 1.

(a) Schematic of reference methanol pool fire experiment[11] and (b) numerical methanol pool fire modeling.

Figure 2.

Mesh sensitivity analysis of (a) ANSYS FLUENT^® and (b) FDS.

Figure 3.

Mean temperature (a) results and (b) error analysis.

Figure 4.

Species volume fraction of (a) CH₃OH, (b) O₂, (c) CO₂, (d) H₂O, (e) N₂.

Figure 5.

Species volume fraction of (a) CO, (b) H₂.

Figure 6.

Species concentration contour of (a) OH, (b) CH₄, (c) HCN, (d) NO.

Table 1

Physical and Chemical Properties of Methanol[12]

Fuel	Chemical Formula	Density	Molecular Weight	Boiling Temperature
Methanol	CH₃OH	796 kg/m³	32.04 g/mol	64.8 °C

Table 2

Summary of ANSYS FLUENT^® Solver Models

ANSYS FLUENT^® Solver Models
Turbulence model	SST k - ω model
Reacting model	Species transport model
Combustion model	Finite-rate/eddy-dissipation model
Radiation model	Discrete ordinates model
Fire source	Methanol (CH₃OH)

Table 3

Mesh Properties

	ANSYS FLUENT^®	FDS
Mesh structure	Hexahedron	Regular Hexahedron
Mesh size	.02 m	.02 m
No. of meshes	117,502	125,000

References

1. ANSYS Co, “ANSYS FLUENT User’s Guide”, Canonsburg, PA, USA (2015).

2. K. McGrattan, S. Hostikka, R. Mcdermott, J. Floyd, C. Weinschenk and K. Overholt, “Fire Dynamics Simulator Technical Reference Guide Volume 3: Validation”, 6th ed., NIST Special Publication, pp. 1018-3 Gaithersburg, MD, USA (2013).

3. Y. Xin, JP. Gore, K. McGrattan, R. Rehm and H. Baum, “Large Eddy Simulation of Buoyant Turbulent Pool Fires”, Proceedings of the Combustion Institute, Vol. 29, No. 1, pp. 259-266 (2002), https://doi.org/10.1016/S1540-7489(02)80035-9.

4. S. Hostikka, K. McGrattan and A. Hamins, “Numerical Modeling of Pool Fires Using LES and Finite Volume Method for Radiation”, Fire Safety Science, No. 7, pp. 383-394 (2003), https://doi.org/10.3801/IAFSS.FSS.7-383.

5. P. Lin, S. Lo and T. Li, “Numerical Study on the Impact of Gradient on Semi-Transverse Smoke Control System in Tunnel”, Tunnelling and Underground Space Technology, Vol. 44, pp. 68-79 (2014), https://doi.org/10.1016/j.tust.2014.07.011.

6. P. Lin, Y. Zhang, T. Li and Y. Si, “A Numerical Study on the Impact of Vehicles’ Blockage on the Performance of Semi-Transversal Smoke Control System in Tunnel Fire”, Procedia Engineering, Vol. 135, pp. 248-260 (2016), https://doi.org/10.1016/j.proeng.2016.01.120.

7. W. Węgrzyński and G. Vigne, “Experimental and Numerical Evaluation of the Influence of the Soot Yield on the Visibility in Smoke in CFD Analysis”, Fire Safety Journal, Vol. 91, pp. 389-398 (2017), https://doi.org/10.1016/j.firesaf.2017.03.053.

8. M. Król and A. Król, “An Experimental and Numerical Study on Fire Spread in a Furnished Room”, Buildings, Vol. 12, No. 12, (2022), https://doi.org/10.3390/buildings12122189.

9. A. Fedorov, N. Fedorova, O. Vankova and D. Tropin, “Verification of Kinetic Schemes of Hydrogen Ignition and Combustion in Air”, AIP Conference Proceedings, Vol. 1939, (2018), https://doi.org/10.1063/1.5027331.

10. H. Lu, F. Liu, K. Wang, G. Xu and H. Curran, “Numerical Study on the Minimum Ignition Energy of a Methane-Air Mixture”, Fuel, Vol. 285, (2021), https://doi.org/10.1016/j.fuel.2020.119230.

11. A. Hamins and A. Lock, “The Structure of a Moderate-Scale Methanol Pool Fire”, “NIST Publication 1928”, National Institute of Standards and Technology, Gaithersburg, MD, USA (2016).

12. M. Hurley, D. Gottuk, J. Hall, K. Harada, E. Kuligowski, M. Puchovsky, J. Torero, J. Watts and C. Wieczorek, “SFPE Handbook of Fire Protection Engineering”, Springer, New York, USA (2016).

13. R. Falkenstein-Smith, K. Sung, J. Chen, K. Harris and A. Hamins, “The Structure of Medium-Scale Pool Fires”, NIST Technical Note 2082, NIST, Gaithersburg, MD, USA (2022).

14. G. Smith, D. Golden, F. Michael, N. Moriarty, B. Eiteneer, M. Goldenberg, C. Bowman, R. Hanson, S. Song, W. Gardiner, V. Lissianski and Z. Qin, “GRI 3.0 Mechanism”, (1999), http://www.me.berkeley.edu/gri_mech/.

15. A. Vargas, A. Arrieta and C. Arrieta, “Combustion Characteristics of Several Typical Shale Has Mixtures”, Journal of Natural Gas Science and Engineering, Vol. 33, pp. 296-304 (2016), https://doi.org/10.1016/j.jngse.2016.03.039.

16. K. McGrattan, S. Hostikka, R. Mcdermott, J. Floyd, C. Weinschenk and K. Overholt, “Fire Dynamics Simulator User’s Guide”, 6th ed., “NIST Special Publication 1019”, Gaithersburg, MD, USA (2013).

17. K. McGrattan, R. Peacock and K. Overholt, “Verification and Validation of Selected Fire Models for Nuclear Power Plant Applications Supplement 1”, U.S. Nuclear Regulatory Commission NUREG-824, (2014).