Explosion venting data: A single maximum overpressure

Transcription

1 Energy Research Institute 53 rd UKELG meeting, 23rd April 2015 Explosion venting data: A single maximum overpressure is not sufficient Phylaktou H.N., Fakandu B.M., Andrews G.E. and Tomlin G. Energy Research Institute University of Leeds, UK

2 Venting Terms and Venting Standards Vent Area, A v, non-dimesionalised as o Vent coefficient K v = V 2/3 /A v (V is the vessel volume) - BS EN (2007) Also K A = A x /A v (A x is the cross-sectional area of the vessel in the plane of the vent) British Gas Research and successors And A v /A s ( A s is the internal surface area of the vessel) NFPA 68 (2013) A x and A s can be related to V 2/3 therefore the different definitions are similar 2

3 Mixture reactivity Deflagration index, K G K G = (dp/dt) max / V 1/3 Laminar burning Velocity, S u It can be shown that the two parameters are directly related. 3

4 Venting Standards standards and guidance focus on providing the correct vent area o as function of the mixture reactivity, vessel volume and shape, and some vent properties e.g. for compact vessels BS EN (2007)

5 NFPA 68 (2013) 5

6 Issues with the Standards For a given vent area they give a maximum overpressure effectively based on correlations of experimental data. No insight /understanding of mechanism of pressure generation Effect of positioning, number and shape of vents not included Effect of ignition location not included Explicitly or implicitly the above are taken to have no effect 6

7 What is needed for model development For correct structural design the full pressure profile is needed Experimental data sets of vented explosion overpressures can t be provided for every potential practical scenario and we need to develop reliable models. Ultimately for correct modelling and for validation of such models we need detailed quantitative data that elucidate the mechanisms and processes involved, and give dependencies on the important parameters. 7

8 Physical Causes of venting overpressures 8

9 Peak pressure events Peak due to vent opening at pressure This work Fakandu et al. [2011,2012] Kasmani et al. [2010b] Cooper et al [1986] Central ignition P burst P 1 P 1 Harrison and Eyre [1987] End ignition Cates and Samuels [1991] Bauwens et al. [2010] Central ignition P stat Peak due unburned gas flow through the vent (onset of burnt gas venting) Peak due the external explosion Peak due to maximum flame area inside the vessel Peak due to the reverse flow into the vented vessel Peak due pressure oscillations. P fv P 2 P emerg P P ext P 3 P 2 P ext Dominant P 1 P mfa P 4 P 3 P max Max. burning rate P rev P 5 P ac P 6 P 4 P 2 P 3 9

10 10

11 Not all pressure peaks are always present. Which one dominates depends on the test conditions 11

12 Test vessel m 3 a Test vessel m 3 b Experimental set-up for test vessels (a) Schematic diagram, (b) Photograph. 12

13 Gas mixtures and vent designs Different gas mixtures: Methane-air(10%), Propane-air (4%, 4.5%), Ethylene-air (6.5%, 7.5%) Hydrogen-air (30%, 40%). A range of investigated. Kv was 16 Joule spark ignition. Central and end ignition compared Vent orifice Grid-Plates Influence. of the number of vents, the shape of vent and the position of the vent were Repeatability Each test conducted at investigated, as these are stated in the US least three times with and EU standards as having no effect. each individual result The results show a significant effect for all. plotted in the graphs. 13

14 Vent Static Burst Pressure effects Pred(bar) P burst 10% Methane-air (K v =7.2) P fv P ext T 4 P stat = 57mb P stat = Time (ms) Comparison of the pressure time records for 10% methane-air for K v = 7.2 for free venting and for P stat = 57mb 14

15 Effect of vent area Overpressure-P red (bar) a 10% Methane-air P fv P ext Overpressure-P red (bar) b 4.5% Propane-air P fv P ext K v K v Overpressure against K v with end ignition for (a)10% methane-air(b) 4.5% Propane-air 15

16 End Vs Central ignition 16

17 17

18 Effect on vent area distribution EN : NFPA 68, 2013: 18

19 Effect on vent area distribution Overpressure(bar) Overpressure(bar) a % Ethylene-air(K v =10.9) b 7.5% Ethylene-air(K v =5.4) P fv P fv Vent Vent P ext P ext Time(ms) Single vent 16 vents Time(ms) Single vent 16 vents Fig. 13 Pressure-time records of single and multiple vents for 10% Methane-air and 7.5% ethylene-air (Kv=10.9 and 5.4) Flame Speed-S f (m/s) % Methane-air(Kv =10.9) Single vent 4 vents 16 vents Vent Distance from spark(mm) 19

20 What influences the external explosion In all tests except for one (methane, small vent area, 4 hole) the external explosion was the highest pressure peak. The external explosion is a turbulent combustion fire-ball, dependent on the turbulent burning velocity, often expressed in terms of the turbulent Reynolds number R u depends on flow through the vent and the pressure loss coefficient (upstream flame speed and BR). is of the order of the width of the solid material between the holes (vents) 20

21 External Pressure Vs length scale 21

22 Vessel Shape : Square vs Circular Vents 22

23 Overall results Methane P red (bar) NFPA Vent C d = 0.7 Present results Methane-air NFPA 68 (2013), L.Flame theory(c d =0.61) Bartknecht(1993)(P stat =0.1bar), Bartknecht(1993)(-0.1bar) L. Flame theory(c d =0.7), Bartknecht(1993)(P stat =0) 2.7m 3 (Cooper et al,1986), 0.2m 3 (Kasmani et al,2010) 30m 3 (Bartknecht,1993), This work 35m 3 (Solberg 1979), 0.68m 3 (Cooper et al,1986) 49m 3 (Hochst and Leuckel,1998), 1m 3 (Bartknecht,1993) 1/K v Sonic Bartknecht s results and vent correlation as used in the EU venting standard Laminar flame Theory vent C d = 0.61 C d = A Av/As s = Surface area of vessel Comparison of vent design equations with experiments for Methane-air as P red v. 1/K v = A v /V 2/3 23

24 Overall results Propane 1 Sonic P red (bar) Propane-air NFPA 68(2013), Bartknecht(1993)(-0.1bar) L.Flame theory(c d =0.61), Bartknecht(1993)(P stat =0) Bartknecht(1993)(P stat =0.1bar), L.Flame theory(c d =0.7) 25m 3 (Bromma,1957), 35m 3 (Solberg,1979) 70m 3 (Bromma,1957), 81m 3 (Howard,1980) 200m 3 (Bromma,1957), 35m 3 (Papas and Foyn,1983) 64m 3 (Bauwens et al,2010), 500m 3 (Bimson et al,1993) 0.2m 3 (Kasmani et al,2010), m 3 (Sato et al,2010) 2m 3 (Bartknecht,1993), 10m 3 (Bartknecht,1993) 30m 3 (Bartknecht,1993), 60m 3 (Bartknecht,1993) 1m 3 (Bartknecht,1993), Present work 0.01 A v /A s /K v 1 Comparison of vent design equations with experiments for propane-air as P red v. 1/K v = A v /V 2/3 24

25 Conclusions Present results from small vessel tests comparable to results from very large scale tests. Able to identify and study the various mechanisms of pressure generation For low K v <~7 the external explosion dominates P max and for K v > ~7 the flow through the vent dominates P max. Vent number, shape, position and ignition position are all important but not recognised as such in the standards. Bartknecht s results (on which the european standards are based) are higher than anybody else s. o Possibly because of the coanda effect on the discharge jet when the vessel is flush with the ground. Similar effects observed in some of our tests.. 25