4th International Conference on Hydrogen Safety, San Francisco, USA, September, J. Yanez, A. Kotchourko, M. Kuznetsov, A. Lelyakin, T. Jordan Karlsruhe Institute of Technology, Germany

Introduction 4th International Conference on Hydrogen Safety, San Francisco, USA, September,  Evaluation of explosion potential for various combustible mixtures appearing during possible accident is of primary importance for safety analysis  Many of the topical studies on FA and DDT in obstructed configurations were performed resulting in criteria able to predict flame propagating regime, as e.g., σ- and 7λ –criteria  Recent experimental studies extended these criteria for partially vented configurations and for semi-confined (flat layer) configurations  The available data from the performed experimental works can not provide all information which potentially can influence the conditions of the initiation of the high speed combustion regimes or DDT event

Objective 4th International Conference on Hydrogen Safety, San Francisco, USA, September, The main goal of the current work was, on the basis of the parametric numerical study, to refine the requirements to geometrical configurations necessary to induce fast flame propagation regime. As a result of the numerical experiments, a criterion accounting the geometrical configuration of the facility and obstacles required for flame acceleration has to be elaborated

Experiment facility 4th International Conference on Hydrogen Safety, San Francisco, USA, September, I – ignition point P, I – pressure and light gauges L = 12 m – A1 length D = 3.5 m – A1 diameter V = 100 m 3 (+30 m 3 )– total volume BR = variable δ – layer thickness A1 Facility at KIT

Experimental details 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Explosion box (9 x 3 x 0.6 m):  mixture volume / safety vessel = 0.16  the distance from the underside of the box to the floor was a minimum of 1.7 m  robust construction  windows for optical measurements / visualisations  effective venting ratio α = 0.46 (layer thickness d = 0.15 m)

Numerical representation 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Explosion box Obstacle set Mixture layer Simplified experimental conditions  No protective vessel  Only geometry variations  2D geometry Type of gridGrid size, cellsControl cellsCell size, m Cartesian structured400 × ,

Numerical models 4th International Conference on Hydrogen Safety, San Francisco, USA, September, The turbulent reactive flow code COM3D code (KIT) is developed wih a view to model large-scale combustion events in geometrically complex environment with multiple compartments and internal structures in a multi-block computational domain. COM3D is a finite-differences code based on solving compressible Navier-Stokes equations in three-dimensional Cartesian space. KYLCOM using Schmidt turbulent flame speed correlation. The current simulation utilizes standard κ−ε model together with burning velocity model KYLCOM using Schmidt turbulent flame speed correlation. The mesh used in COM3D code is rectangular Cartesian equidistant uniform multi-block simply connected domain with cell-centered variables. The transport of the basic variables is provided using HLLC algorithm. The time integration accuracy is of the 2nd order. OpenMPI version of the code using Linux cluster with 64 CPUs was used.

Simulation matrix 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Mixture layer thickness H Distance between obstacles D Height of obstacles h Gap between obstacles I First obstacle gap I 0

Simulation matrix 4th International Conference on Hydrogen Safety, San Francisco, USA, September, CaseD, [m]H, [m]BRh, [m]l 0, [m]l, [m] 40,60 0,500,06 50,600,200,500,06 60,600,300,500,06 70,600,150,500,06 80,300,200,500,06 90,300,150,500,06 140,30 0,500,06 150,150,300,500,06 160,150,100,500,06 170,150,100,500,06 180,900,300,500,06 190,900,200,500,06 200,900,150,500,06 210,900,100,500,06 220,600,30 0,040,08 230,600,200,300,040,08 240,600,150,300,040,08 250,600,300,700,090,04 260,600,200,700,090,04 270,600,150,700,090,04 280,600,300,900,110,01 290,600,200,900,110,01 300,600,150,900,110,01 310,600,300,500,100,11 320,600,200,500,100,11 330,600,150,500,100,11 340,600,300,500,30 350,600,300,500,15 360,600,200,500,30 370,600,300,700,090,000,04 380,300,150,500,060,000,06 390,600,300,900,110,000,01

Density field after ignition 4th International Conference on Hydrogen Safety, San Francisco, USA, September, D = 0.60 m H = 0.30 m BR = 0.30 h = 0.04 m l 0 = 0.08 m l = 0.08 m t = st = s t = s t = s Depending on obstacle configuration, burnable mixture thickness and concentration different combustion regimes are realized H 2 -air mixture

Layer thickness dependence 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Layer thickness 15 cm Obstacle blockage ratio 0.5 Obstacle distance 60 cm Obstacle height 60 cm Density filed: 0 – 3.5 kg/m 3 Terminal speed: < 100 m/s

Layer thickness dependence 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Layer thickness 30 cm Obstacle blockage ratio 0.5 Obstacle distance 60 cm Obstacle height 60 cm Density filed: 0 – 3.5 kg/m 3 Terminal speed: ~ 500 m/s

Layer thickness dependence 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Layer thickness 60 cm Obstacle blockage ratio 0.5 Obstacle distance 60 cm Obstacle height 60 cm Density filed: 0 – 3.5 kg/m 3 Terminal speed: ~ 600 m/s

Pressure evolution: fast flame 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Layer thickness 60 cm Obstacle blockage ratio 0.5 Obstacle distance 60 cm Obstacle height 60 cm Pressure filed: 0 – 10 bar Terminal speed: ~ 600 m/s

Combustion regimes 4th International Conference on Hydrogen Safety, San Francisco, USA, September, CaseD, [m]H, [m]BRh, [m]l 0, [m]l, [m] 150,150,300,500,06 240,600,150,300,040,08 260,600,200,700,090,04 Case 15: fast flame (choked) Case 26: fast flame (choked) Case 24: slow flame (decaying)

Governing processes 4th International Conference on Hydrogen Safety, San Francisco, USA, September, CaseD, [m]H, [m]BRh, [m]l 0, [m]l, [m] 90,300,150,500,06 250,600,300,700,090,04 280,600,300,900,110,01 370,600,300,700,090,000,04 380,300,150,500,060,000,06 390,600,300,900,110,000,01 Fast flame For high BR existence of top gap can be important Larger openings  lower turbulence Smaller openings  higher turbulence D= 0.6 H = 0.3 BR = 0.3 D= 0.6 H = 0.3 BR = 0.9 t = s

Blockage Ratio & Co 4th International Conference on Hydrogen Safety, San Francisco, USA, September, BR as a single explanatory variable Explanatory variable candidates:  BR  Layer thickness  Distance between obstacles  Obstacle height  …  Mixture composition  …

Combustion regime analysis 4th International Conference on Hydrogen Safety, San Francisco, USA, September, CaseE(v)v max v min σ(v)E(v)/C u E(v)/C p v max /C p v min /C u E(v)>C u v max >C u 4640,4899,3303,8193,11,560,800,880,74 ## ,2766,0310,6151,21,340,680,750,76 ## 7266,2360,3177,958,60,650,320,350, ,1682,5334,7132,81,240,610,670,82 ## 9483,7658,2322,5123,21,180,590,650,79 ## 14589,4748,0439,9108,51,440,670,731,07 ## 15594,2603,1588,73,01,450,540,591,44 ## 16296,8407,5247,448,80,720,360,400, ,3288,293,651,20,460,260,280, ,4147,045,430,90,210,130,140, ,3361,3200,940,20,670,320,350, ,3105,561,912,20,200,090,100, ,577,647,49,10,160,070,080, ,4806,3220,9210,01,300,720,790,54 ## 26488,4782,4194,9188,01,190,700,770,48 ## 27360,1548,2168,799,50,880,490,540,41 # 28501,1737,7134,3196,51,220,660,720,33 ## 29445,5667,3127,7169,91,090,600,650,31 ## ,2799,6265,3183,81,330,710,780,65 ## 32120,2179,283,530,80,290,160,180, ,3150,558,231,20,270,130,150, ,3282,2108,343,70,450,250,280, ,2482,2475,91,81,170,430,471,16 ##

Grid similarity 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Turbulent kinetic energy and the dissipation of turbulence generated by grid decay as power laws (n = 1.3) Schmidt correlation for turbulent flame speed and Roach approach

Dimensionless dependencies 4th International Conference on Hydrogen Safety, San Francisco, USA, September,  Flame speed St ~ 1/D  Flame speed St ~ h  Blockage ratio is an important factor characterizing the problems  If BR is high, the dependence on l 0 looks to be also significant  The total energy ~ H can be nondimensionalized using l or h variables. The trial and error approach provides better separation for h variable  Set of variables H/h, l 0 /l, D/h, BR can be combined to obtain a criteria for the separation of fast and slow flames  dimensionless variable D/h can be utilized to characterize the system  dimensionless variable BR  dimensionless variable l 0 /l  dimensionless variable H/h

Flame acceleration criterion 4th International Conference on Hydrogen Safety, San Francisco, USA, September, Expressing in conventional form against expansion ratio of the mixture K = 0.17

Summary   Numerical study of FA in obstructed semi-confined flat layer of uniform stoichiometric H2-air mixture in large scale was performed   The effects of obstruction arrangement described by the distance between obstacles, the thickness of the hydrogen layer, the blockage ratio, the sizes of the obstacles and the orifices, on the flame propagation regime was analyzed   It was shown that the standard deviation of the velocity characterizing the effect of obstacle structure on the flame behavior can be described using the variable D 1.65 l   The set of the dimensionless parameters and the dependence linking them permitting credible separation of the fast and slow propagation regimes were proposed 4th International Conference on Hydrogen Safety, San Francisco, USA, September,