HYDROGEN COMBUSTION EXPERIMENTS IN A VERTICAL SEMI-CONFINED CHANNEL

1. HYDROGEN COMBUSTION EXPERIMENTS IN A VERTICAL SEMI-CONFINED CHANNEL Friedrich, A.*1, Grune, J.1, Sempert, K.1, Kuznetsov, M.2 and Jordan, T.2 . 1Pro-Science GmbH, Parkstrasse 9, Ettlingen, 76275, Germany,2Karlsruhe Institute of Technology, IKET, 76131 Karlsruhe, Germany. Funded by:

2. Background/Motivation • In hydrogen safety considerations phenomena of effective flame acceleration (FA) and deflagration-to-detonation transition (DDT) are very important, especially for large semi-confined spaces, such as rooms or tunnels, • Criteria for the spontaneous transition processes of FA and DDT were derived empirically, numerical simulations of these processes on full reactor scale currently not possible. • These criteria that are currently used worldwide in accident simulations were derived for homogenous mixtures and complete inclusion.

3. Background/Motivation • In a previous experimental series in a large-scale horizontal combustion channel (9 x 3 x 0.6 m³) with an open ground face were formulated. These criteria were derived by analyzing numerous experiments with various layer thicknesses, obstacle configurations and mixture properties. extended criteria for the onset of FA and DDT in semi-confined geometries • The main objective of this work was to evaluate the critical conditions for FA and DDT in a semi-confined obstructed vertical layer and to compare these conditions with results of the previous campaignin the semi-confined horizontal channel.

4. Experimental Set-Up • All experiments were performed in a vertical channel installed to the safety vessel A3 at the Hydrogen Test Site HYKA at KIT: • dimensions 6 x 0.4 x 0.4 m³ (one open side face), • framework structure covered by wooden plates, • welded steel shell seals and protects wooden inner surface against flames, • 16 grid obstacles (BR 50%, spacing 25 cm) in first 4 m from ignition in all experiments. h = 8 m, di = 2.5 m, V = 33 m³, pStat = 60 bar

5. Test-ProcedureGeneral • Open channel side face sealed by thin plastic Film, • Test mixture generated by mass flow controllers for H2 and air, • Initial channel atmosphere replaced with test mixture (from top) and pushed out of channel through outlet valve at lower end, • When the desired H2-concentration or H2-gradient is reached inlet and outlet valves are closed, • Film is destroyed prior to ignition.

6. Test-ProcedureHomogeneous Mixtures cH2(in) • Open channel side face sealed by thin plastic Film, • Initial channel atmosphere replaced from top with test mixture and pushed out of channel through outlet valve at lower end, • Procedure completed when cH2(in) = cH2(out). Film Procedure homogeneous mixtures H5875 H4875 H3875 • Film is destroyed over complete height by falling knife prior to ignition (cut observed by cameras distributed in safety vessel) • Experiment not further evaluated if cut failed or was incomplete. H2375 H1125 Time [s] H125 cH2(out) homogeneous mixture

7. Test-ProcedurePositive Concentration Gradients cH2(in) • Open channel side face sealed by thin plastic Film, • Channel filled homogeneously with mixture of concentration cH2(0) that corresponds to lowest value in H2-concentration gradient, Film • Then further injection of H2-air mixture with increasing H2-concentration (programmable massflow-controller), • Procedure completed when cH2(in) = cH2(max), H5875 • Film destroyed prior to ignition. H4875 Procedure opsitive cH2-gradients H3875 t = 720 s Ignition H2375 c(max) H1125 dc c(0) H125 cH2(0) cH2(out) Time [s] positive gradient

8. Test-ProcedureNegative Concentration Gradients Luft cH2(in) • Open channel side face sealed by thin plastic Film, • Channel filled homogeneously with mixture of concentration cH2(0) that corresponds to highest value in H2-concentration gradient, Film • Then air is inject from top, • Procedure completed when calculated air-volume is injected, • Film destroyed prior to ignition. H5875 H4875 Procedure negative cH2-gradients H3875 t = 135 s Ignition H2375 H1125 H125 cH2(0) cH2(out) Time [s] negative gradient

9. Test-ProcedureGradients Available • 2 variables in the procedure for positive gradients: • variation of cH2(0) “shifts” gradient along abscissa, • variation of dc (= cH2(max) – cH2(0)) influences slope of gradient, • Gradients are rather stable (several minutes). • In procedure for negative gradients only cH2(0) can be varied,  all gradients have similar slopes, • Gradients not stable  ignition initiated 10 s after end of air-injection in all experiments with negative gradients.

10. Test-ProcedureIgnition & Instrumentation Top Ignition • Ignitionvia glow wire in perforated tube ( almost planar flame front), HT PT NT PB HB NB Negative Gradient Positive Gradient Homogeneous • Mixture ignited either at top or at bottom end of channel •  6 cases (HT/HB, PT/PB, NT/NB). Film Film Film Instrumentation: Bottom Ignition 1375 1875 2875 3375 4375 4875 5375 5875 125 625 1125 2125 2625 3625 4125 5125 5625 1375 1875 2875 3375 4375 4875 5375 5875 125 375 625 1125 1625 2125 2625 3625 4125 5125 5625 Bottom Ignition Top Ignition 0 375 1625 2000 2500 3000 3500 4000 4500 5000 5500 6000 2250 2750 3250 3750 4250 4750 5250 5750 0 2000 2500 3000 3500 4000 4500 5000 5500 6000 2250 2750 3250 3750 4250 4750 5250 5750 Thermocouple Thermocouple Pressure Transducer Pressure Transducer Ionization Probe Ionization Probe Obstacle Obstacle Ignition Device Ignition Device - flame front: 18 modified thermocouples, 9 ionization probes - shock wave: 8 pressure sensors

11. Test-Matrix Homogeneousmixtures Positive gradients Negative gradients • 25 experiments with homogeneous mixtures(13 HT and 12 HB). • 43 experiments with vertical gradients • 25 experiments with positive gradients (12 PT and 13 PB) • 18 experiments with negative gradients (9 NT and 9 NB)

12. Data Evaluation • x-t-diagrams of ionization probe and pressure sensor signals recorded during the experiments: cH2[Vol%] 18.5 cH2[Vol%] cH2[Vol%] 10.6 16.3 24.5 14.9 10.3 Slow deflagration v < cp pmax < 3,5 p0 Fast turbulent deflagration v ≈ cp 5·p0 < pmax < 8·p0 Detonationv ≈ 2·cp 12·p0 < pmax < 17·p0

13. Results and DiscussionFlame Acceleration • Transition between regimes not continuous but stepwise, as results of experiments with homogeneous mixtures indicate: • Flame acceleration: • from graph: cH2(krit) ≈ 14,5 Vol% s* • dimensionless: s* ≈ 4.5 • equation from earlier work: • With: s*0 = 3.75 KHC = 0.66 (horiz. channel) • s/h = 0.25 m/0.4 m = 0.625 •  s* ≈ 4.16 s* = s*0 (1 + K · s/h) • or, due to differences between facilities:KVC = 0.32 (vert. channel) with s* = 4.5 • Criterion derived for horizontal semi-confined channel in earlier work also applicable for homogeneous mixtures in vertical semi-confined channel, Also applicable on mixtures with concentration gradients?

14. Results and DiscussionFlame Acceleration • All experiments were performed in same facility, so representation of σ over s/h not suitable for comparison of different gradient orientations, • so representation of σ over configuration is used: FA no FA FA no FA Horizontal channel PB-G PB-G s* s* HT HB PT PB NT NB • No good agreement of data with FA-criterion when maximum H2-concentration is used for determination of σ (earlier work: Concept of Maximum Concentration), • But very good agreement of data with FA-criterion when average H2-concentration in obstructed region (cH2av#) is used, • Strong deviations only for experiments with very steep concentration gradients (series “PB-G” (triangular symbols)).

15. Results and DiscussionDetonation-Transition • For evaluation of DDT-criterion representation of h/λ over configuration is used, but again cH2max or cH2av# can be used for calculation of λ: DDT no DDT DDT no DDT h/l = 13.5 • Homogeneous mixtures show good agreement with DDT-criterion determined for the horizontal channel, • In general better agreement for λ calculated with cH2max, but overestimation for case PT (and very steep concentration gradients (series “PB-G”)), • Use of cH2av# gives better agreement for case PT (and “PB-G”), but underestimates most other cases.

16. Summary & Conclusion • Slow (subsonic) and fast (sonic) deflagrations as well as detonations were observed and the conditions for an FA and DDT were determined and compared with the results of a previous campaign with semi-confined horizontal layers, • σ-criterion for FA in semi-confined horizontal layers was • successfully adapted to vertical layers with homogeneous H2-concentrations, • also applicable to vertical layers with vertical H2-concentration gradients (better results when mean concentration in gradient is used, with maximum concentration in gradient estimation is very conservative) • λ-criterion for DDT in semi-confined horizontal layers was • successfully applied to vertical layers with homogeneous H2-air mixtures, • also applicable to vertical layers with vertical H2-concentration gradients (better results when maximum concentration in gradient is used, with mean concentration in gradient underestimation of many cases). • The previously formulated criteria for FA and DDT in semi-confined horizontal H2-air layers were applied successfully to vertical H2-air layers with and without concentration gradients, which is a proof for the capacity of the methods used. Thank you very much for your attention!