Quantitative infrared imaging of impinging buoyant diffusion flames

Paper # 070FR-0154 8 th US National Combustion Meeting Organized by the Western States Section of the Combustion Institute and hosted by the University of Utah May 19-22, 2013. Topic: Fire Quantitative infrared imaging of impinging buoyant diffusion flames Ashish S. Newale 1 Brent A. Rankin 1,2 Jay P. Gore 1 1 School of Mechanical Engineering, Purdue University, West Lafayette, IN-47907, USA 2 Innovative Scientific Solutions Inc., Dayton, OH-45440, USA Studies of infrared radiation from buoyant diffusion flames impinging on structural elements have applications to the development of fire models. An experimental study of radiation from buoyant diffusion flames with and without impingement on a flat plate is reported. Quantitative images of the radiation intensity from the flames are acquired using a high speed infrared camera. The measured radiation intensities reveal necking and bulging of the flame with a characteristic frequency of 7±1 Hz which is in agreement with previous empirical correlations. The results demonstrate the effects of the stagnation point boundary layer on the upstream buoyant shear layer. The coupling between those two shear flows presents a model problem of practical significance for sub-grid scale modeling necessary for future large eddy simulations. 1 Introduction Fire structure interaction studies explore the complex interaction between gas phase heat release rate, structural and thermal properties of load bearing elements. Heat transfer from a fire to a structure has been the subject of many investigations [1 3]. Radiation heat transfer has been established as an important link between the gas phase and the solid [4]. The current work reports the effect of impingement on flame radiation for the model problem of buoyant diffusion methane flame impinging on a flat plate. Improved estimations of flame structure and flow field will aid in the further improvements in the estimations of heat transfer to structural elements. One of the most challenging aspects of a coupled fire structure interaction simulation is the significant variation in the length and time scales. The different numerical methods used for fire, thermal, and structure simulations lead to difficulties in determining an efficient method of transferring information between these simulations [4 6]. Wickstrom et al. [5] demonstrated the use of adiabatic surface temperature as an effective technique of transferring heat flux boundary conditions between the fire model and the thermal, structural models. Prasad and Baum [6] developed a radiative transport model adapted from zone models for compartment fires to estimate the radiative heat flux incident on complex structural elements. The authors used simulations to determine the thermal response of typical floor trusses and columns impacted by the World Trade Center 1

fires resulting from September 11, 2001 terrorist attacks and rendered their simulations as movies for comparison with video clips from the event. Hasemi and Weng [1] developed an analytical model to determine the thermal response of an unconfined non-burning ceiling to an impinging buoyant diffusion flame. The model showed good agreement with previous experimental heat flux data away from the stagnation point. For locations near the stagnation point, the model estimated higher heat fluxes. You and Faeth [3] studied impinging turbulent fires and fire plumes during initial stages of ceiling heating. The authors reported correlations for free flame height, radial extent of the impinging flames, stagnation point and radial profiles in the wall jet region of heat flux. Confinement led to higher heat fluxes in both the stagnation point heat flux and the wall jet regions. You [7] developed an integral model for buoyant, axisymmetric, turbulent diffusion flames to determine properties in the plume region. The model provided good agreement with experimental data from previous work for vertical velocities; however, characteristic flow radius was under-estimated and the temperature was over-estimated. You [2] extended his integral model for the plume region to the ceiling jet region. The model provided reasonably good agreement with measurements for flow properties and ceiling heat fluxes. Ranga Dinesh et al. [8] performed three-dimensional direct numerical simulations for transitional hydrogen-air non-premixed impinging jet flames. The authors suggested the near wall vortical structures are important in the flame to wall heat transfer. Chung et al. [9] showed the unsteady nature of heat transfer in non-reacting jets due to primary vortices emanating from a jet nozzle. Tuttle et al. [10, 11] reported time-averaged and time-resolved measurements of axial and radial profiles of local heat fluxes of methane-air jets impinging on a cooled plate. A range of flames were explored by varying the equivalence ratio, Reynolds number and nozzle-to-plate spacing. The coupling between flame structure and resultant heat transfer was demonstrated in this work by correlating visible photographs of the flame to measured heat flux profiles. Premixed flames typically produce the highest heat flux and are used in industrial applications to provide localized heat transfer. Baukal and Gebhart [12] determined that the contribution of thermal radiation and thermochemical heat release to total heat flux are low compared with convective heat transfer for premixed natural gas flames they studied. Baukal and Gebhart [13 16], Viskanta [17] and Chander and Ray [18] have reported comprehensive reviews on flame impingement heat transfer for manufacturing applications. The quantitative comparison of measured and computed images of the radiation intensity is an effective method for studying reacting flows and prompting improvements in combustion and radiation models [19]. Rankin [20] measured and computed quantitative images of infrared radiation intensity from a representative turbulent non-premixed jet flame (flame A) from the International Workshop on Measurement and Computation of Turbulent Nonpremixed Flames [21]. The computed images utilized scalar results from LES of flame A performed by Ihme and Pitsch [22, 23]. The measured and computed radiation intensities agreed to within approximately 20 % from the burner exit to 40 diameters downstream. Differences between the measured and computed radiation intensities near and downstream of the stoichiometric flame tip suggested that including 2

radiation heat loss effects is important even for weakly radiating flames. The current study involves measurements of buoyant diffusion flames with and without impingement using Rankin s infrared imaging technique. Motivated by a review of the existing literature on fire-structure interactions and the availability of the computational and experimental quantitative infrared imaging technique, the objectives of the current study are as follows : 1. Measure images of the infrared radiation intensity from a representative buoyant diffusion flame with and without impingement on a steel plate; 2. Study the effects of flame impingement on the measured infrared radiation intensities. 2 Experimental Methods 2.1 Buoyant Diffusion Flame A buoyant diffusion flame is established on a diffuser burner with an exit diameter of 7.1cm. The diverging angle of the burner is 7 such that the gaseous fuel(methane) is decelerated and forms a uniform velocity distribution at the burner exit [24]. The methane (CH 4 ) mass flow rate (84.3 mg/s) is calibrated using a dry test turbine meter and controlled by setting the pressure upstream of a choked orifice plate. The buoyant diffusion flame burns in quiescent atmospheric pressure environment. The Froude number is 0.109 and matches that of a 7.1 cm diameter liquid toluene pool fire [24]. The total heat release rate of the methane flame is 4.2 kw under the assumption of complete combustion, and the visible flame height is approximately 36 cm [24]. Measured and computed vertical and horizontal velocity, mixture fraction, and temperature values for this flame have been reported by Zhou et al. [25, 26] and Xin et al [24, 27]. A schematic of the experimental arrangement and coordinate system definitions for the radiation measurements are illustrated in figure 1. The flame (observed) coordinate system (X,Y,Z) are defined with an origin located at the center of the burner exit. The infrared camera (observer) coordinate system (x,y,z) is defined with the origin located at (x,y,z) = (0,d,0) where d is the distance between the flame axis and the camera lens. A square steel plate (0.254 m 0.254 m 0.0254 m) was oriented perpendicular to the flame centerline and suspended at an axial location of 0.142 m (Z/D = 2) downstream of the burner exit. The plate was supported at each of the four corners on the bottom surface using thin brackets to minimize flame disturbance. 3

Figure 1: A schematic of the experimental arrangement for radiation intensity measurements of the buoyant diffusion flames without (left) and with (right) impingement on a steel plate. Note: (X, Y, Z) are flame-based (observed) coordinates and (x, y, z) are camera-based (observer) coordinates. 2.2 Radiation Intensity Measurements Time dependent images of the radiation intensity from the flame and the plate were measured using an infrared camera (FLIR Phoenix) with a 25 mm lens (f/2.3) and an indium antimonide (InSb) detector. A narrowband filter (2.77 ± 0.12 µm) was used to measure the radiation intensity originating from water vapor, carbon dioxide, soot and steel surface emissions. The spatial resolution of the radiation intensity measurements was approximately 0.97 mm for each pixel at the center of the flame. The infrared camera sampling frequency was 330 Hz and the exposure time was 10 µ s. The experimental uncertainty in the radiation intensity is estimated to be ±15% (95% confidence) based on repeated measurements of a turbulent jet diffusion flame using the same infrared camera [28]. 3 Results and Discussion Figure 3 displays the measured time dependent images of the radiation intensity from the nonimpinging flame. The time between successive images is 30 ms. The measured images display the necking and bulging of the flame. The series of images shows the intermediate phases of an entire pulsation cycle. The measured pulsation frequency is 7 Hz. Large vortical structures are observed in the measured images of the radiation intensity. Figure 4 shows the measured time dependent radiation intensity images for the impinging flames. The time between successive images is 30ms. Both the measured pulsation frequencies of the buoyant diffusion flames with and without impingement are not affected significantly by the impingement. Figure 5 displays the measured time-averaged radiation intensity images for the non-impinging flame. The measured image shows a low intensity region near the fuel rich central portion close 4

8th US Combustion Meeting Paper # 070FR-0154 Topic: Fire Figure 2: Measured time dependent images of the radiation intensity from the non-impinging buoyant diffusion flame. The time between consecutive images is 30 ms and the exposure time is 30 µs. Figure 3: Measured time dependent images of the radiation intensity from the buoyant diffusion flame impinging on the plate suspended at Z = 14.2 cm. The time between consecutive images is 30 ms and the exposure time is 30 µs. 5

to the burner exit. Larger intensities are observed in the flame stabilization region near the burner rim. The highest intensities are observed at approximately 12 cm downstream of the burner exit where mixing between the fuel and entrained air results in the highest path-integrated temperatures, radiating gas species concentrations, and soot volume fractions. Figure 4: Measured image of time-averaged radiation intensity from the non-impinging buoyant diffusion flame. Figure 6 depicts the measured time-averaged radiation intensity image for the impinging buoyant diffusion flame. A region of high radiation intensity is observed at small distances upstream of the plate. Significant time-averaged radiation intensity was not observed in the wake region due to the general absence of hot combustion gases for the conditions utilized in the current study. A small number of instanteneous realizations showed reactions in the wake region as expected based on the vast literature of bluff body stabilizers. The magnitude of the peak time-averaged radiation intensities observed in the measurements is 750 W/m 2 -sr which is three times larger than the non-impinging case. The peak radiation intensity occurs near the image centerline in the measurements. Average radiation intensities from diametric paths are plotted as a function of the distance downstream of the burner exit in figure 7 for both the non-impinging and impinging flames. The impinging flame results are reported from the burner exit to the stagnation point. The location of the peak radiation intensity occurs upstream of the stagnation point and relatively short duration of the experiment and relative low initial temperature of the plate. The normalized root mean square (RMS) of the radiation intensity as a function of the distance downstream of the burner are plotted in figure 9. For the impinging flame, a peak is observed at a location upstream of the stagnation point in the measured RMS values. The decrease in the 6

8th US Combustion Meeting Paper # 070FR-0154 Topic: Fire Figure 5: Measured image of the time-averaged radiation intensity emitted from the buoyant diffusion flame impinging on the plate suspended at Z = 14.2 cm. Figure 6: Measured mean radiation intensity emitted from diametric paths (Y = 0cm) as a function of distance downstream of the burner exit 7

8th US Combustion Meeting Paper # 070FR-0154 Topic: Fire normalized RMS upstream of the plate (Z=14.2 cm) indicates that the plate has a stabilizing effect on the flame radiation. Figure 7: Measured root mean square (left) and normalized root mean square (right) of the radiation intensity emitted from diametric paths (Y = 0 cm) as a function of distance downstream of the burner exit Figure 10 reports the measured probability density functions (PDF) of the radiation intensity from the flames with and without impingement at a representative location (Y= 0 cm, Z= 13.7 cm). The location has been selected to correspond to the peak mean intensity observed for the impinging flame (Fig. 8). The PDF for the non-impinging flame are skewed indicating that the mean intensity is dominated by occasional periods of high intensity. The PDF for the impinging case is more symmetric about the mean intensity. Figure 11 reports the temporal autocorrelation correlations and power spectral density functions of the radiation intensity from the non-impininging and impinging flames at a select location (Y = 0 and Z = 13.7 cm). The magnitude of the oscillations is damped for the impinging flame relative to the non-impinging flame. This observation suggests that the flat plate has a local stabilizing effect on the flame. The magnitude of the PSD increases by approximately an order of magnitude at the low frequencies for the impinging flame relative to the flame without impingement. The low frequency fluctuations are increased significantly and the peak at the characteristic frequency is reduced significantly by the presence of the obstructing plate. 4 Summary and Conclusions A quantitative comparative study of measured images of the radiation intensity from non-impinging and impinging buoyant diffusion flames is reported. 8

8th US Combustion Meeting Paper # 070FR-0154 Topic: Fire Figure 8: Measured probability density functions of the radiation intensity at a select location for the non-impinging and impinging buoyant diffusion flames. Figure 9: Temporal correlations(left) and power spectral density functions (right) of the radiation intensity from buoyant diffusion flames with and without impingement on the plate. 9

Specific results and conclusions of the present study are described below: 1. The plate is shown to have an upstream propagation effect on both the mean and the normalized RMS of the impinging flame radiation. 2. The pulsation frequency is quantitatively similar for both the impinging and non-impinging flames. 3. The plate acts to locally stabilize the impinging flame near the stagnation point as demonstrated by a decrease in the magnitude of the oscillations observed in the temporal correlations. Acknowledgments The National Institute of Standards and Technology and National Science Foundation are gratefully acknowledged for supporting previous fire structure interaction work with Professor Jay P. Gore and Professor Amit H. Varma. References [1] W.G. Weng and Y. Hasemi. Heat and Mass Transfer, 42 (2006) 652 659. [2] H.Z. You. Fire and Materials, 9 (1985) 46 56. [3] H.Z. You and G.M. Faeth. Fire and Materials, 3 (1979) 140 147. [4] H.R. Baum. Mechanics Research Communications, 38 (2011) 1 11. [5] U. Wickstrom, D. Dutinh, and K.B. McGrattan. Adiabatic surface temperature for calculating heat transfer to fire exposed structures. In Proceedings Interflam 2007, Vol. 2, 2007. [6] K. Prasad and H.R. Baum. Proceedings of the Combustion Institute, 30 (2005) 2255 2262. [7] H.Z. You. Fire and Materials, 8 (1984) 28 39. [8] KKJ Ranga Dinesh, X. Jiang, and JA van Oijen. International Journal of Hydrogen Energy, 37 (2012) 4502 4515. [9] Y.M. Chung, K.H. Luo, and N.D. Sandham. International Journal of Heat and Fluid Flow, 23 (2002) 592 600. [10] S.G. Tuttle, B.W. Webb, and M.Q. McQuay. International Journal of Heat and Mass Transfer, 48 (2005) 1236 1251. [11] S.G. Tuttle, B.W. Webb, and M.Q. McQuay. International Journal of Heat and Mass Transfer, 48 (2005) 1252 1266. [12] C.E. Baukal and B. Gebhart. Experimental and Thermal Fluid Science, 15 (1997) 323 335. [13] C.E. Baukal and B. Gebhart. Combustion Science and Technology, 104 (1995) 339 357. [14] C.E. Baukal and B. Gebhart. Combustion Science and Technology, 104 (1995) 359 385. [15] C.E. Baukal and B. Gebhart. International Journal of Heat and Fluid Flow, 17 (1996) 386 396. [16] C.E. Baukal and B. Gebhart. International Journal of Heat and Mass Transfer, 39 (1996) 2989 3002. [17] R. Viskanta. Experimental and Thermal and Fluid Science, 6 (1993) 111 134. 10

[18] S. Chader and A. Ray. Energy Conversion and Management, 46 (2005) 1 3. [19] B.A. Rankin, D.L. Blunck, V.R. Katta, S.D. Stouffer, and J.P. Gore. Combustion and Flame, 159 (2012) 2841 2843. [20] B.A. Rankin. Quantitative experimental and model-based imaging of infrared radiation intensity from turbulent reacting flows. PhD Thesis, Purdue University, West Lafayette, IN, 2012. [21] International Workshop on Measurement and Computation of Turbulent Non-Premixed Flames. Sandia National Laboratories. http://www.sandia.gov/tnf. [22] M. Ihme, H. Pitsch, and D. Bodony. Proceedings of the Combustion Institute, 32 (2009) 1545 1553. [23] M. Ihme and H. Pitsch. International journal of aeroacoustics, 11 (2012) 25 78. [24] Y. Xin, J.P. Gore, K.B. McGrattan, R.G. Rehm, and H.R. Baum. Combustion and Flame, 141 (2005) 329 335. [25] X.C. Zhou and J.P. Gore. Proceedings of the Combustion Institute, 27 (1998) 2767 2773. [26] X.C. Zhou. An experimentanl and theoretical investigation of flows induced by buoyant diffusion flames. PhD Thesis, Purdue University, West Lafayette, IN, 1999. [27] Y. Xin. A theoretical and experimental study of buoyant turbulent flames with emphasis on soot and radiation. PhD Thesis, Purdue University, West Lafayette, IN. [28] B.A. Rankin, D.L. Blunck, and J.P. Gore. Infrared imaging of a turbulent non-premixed jet flame and plume. In 7 th US National Combustion Meeting, 2011. 11