This project is funded by the European Union Projekat finansira Evropska Unija TOP EVENTS CONSEQUENCE ANALYSIS MODELS Antony Thanos Ph.D. Chem. Eng. antony.thanos@gmail.com This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Consequence analysis framework Hazard Identification Release scenarios Event trees Dispersion models Consequence results Accident type Release models Release quantification Fire, Explosion Models Domino effects Limits of consequence analysis This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire Ignition of flammable liquid phase Main consequence Thermal radiation Liquid fuel tank fire This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire characteristics Pool dimensions (diameter, depth) o Confined pool (liquid fuels tank/bund fire) : Tank fire pool : diameter equal to tank diameter dimension bunds : pool diameter estimated by equivalent diameter of bund Dp This Project is funded by the European Union 4 Abund Project implemented by Human Dynamics Consortium • Pool fire characteristics (cont.) Pool dimensions (diameter, depth) (cont.) o Unconfined pool (LPG pool from LPG tank failure –no dike present) : Theoretically maximum pool diameter is set by balance of release feeding the pool and combustion rate from pool Release to pool This Project is funded by the European Union Combustion rate Project implemented by Human Dynamics Consortium • Pool fire characteristics (cont.) Pool dimensions (diameter, depth) (cont.) o Unconfined pool : M in M comb mcomb A Apool V pool / Depth Min : release rate (kg/sec) Mcomb : combustion rate (kg/sec) mcomb : specific combustion rate (kg/m2.sec) In real life, pool is restricted by ground characteristics. Typical values for assumed depth: 0.5-2 cm (depending on ground type, higher values reported for sandy soils) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire characteristics (cont.) Flame height, inclination (angle of flame from vertical due to wind) Long duration (hours to days) Combustion rate This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models Combustion rate per pool surface on empirical equations (Burges, Mudan etc.) o Example : Tb Ta Tb Ta m 0.001 H C H v C p (Tb Ta ) (liquefied gases ) m 0.001 H C H v m = specific comb.rate (kg/m2.sec) Tb = Boiling point βρασμού (Κ) Ta = ambient temperature (K) ΔHc = Combustion heat (J/kg) ΔHv = latent heat (J/kg) Cp = liquid heat capacity (J/kg.K) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Combustion rate for liquids not exceeding 0.1 kg/m2.sec. Upper range for low boiling point hydrocarbons Flame dimension from empirical equations (Thomas, Pritchard etc.) o Example, Thomas correlation : Hf m 42 a g Dp Dp 0.61 Hf= flame height Dp= pool diameter m= specific comb.rate ρa= air density g=gravity constant o Big pools : Hf/Dp in the range of 1-2 This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Point source model (cont.) o No flame shape taken into account o A fraction of combustion energy is considered to be transmitted by ideal point in pool center Thermal radiation transmitted semi spherically This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Point source model (cont.) f M H C t a q 4 x2 q = thermal radiation flux at “receptor” (kW/m2) f = thermal radiation fraction (0.1-0.4, depending on substance and pool size. Big pools, low values) Μ = combustion rate (kg/s) ΔHc= combustion heat (kj/kg) ta = transmissivity coeff., x = distance of pool center from “receptor” o Increased inaccuracies near pool end (important for Domino effects) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model, radiation emitted via flame surface Pool diameter Flame height Pool depth This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o Calculation based on : flame shape (usually considered cylinder -tilted or not-), distance from flame (View Factor), emissive power (thermal radiation flux at flame surface) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o Calculation equation : q VF E ta q = thermal radiation flux at “receptor” (kW/m2) VF = view factor for flame shape at receptor Ε = emissive power (kW/m2) ta = transmissivity coefficient This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o View Factor : function of distance of receptor from flame and flame dimensions. Different equation for different flame shapes o Transmissivity coefficient : Absorbance of thermal radiation by atmosphere components - e.g. humidity, CO2 – Correlation with relative humidity (R.H.) level and distance to “receptor”) High R.H, low transmissivity coefficient More important for far-field effects (due to increased distance) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o Emissive power : Depending on pool size, substance For big pools, soot formation (20 kW/m2), masking of flame, significant reduction of average flame emissive power This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o Emissive power : (cont.) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o Emissive power : (cont.) Experimental Gasoline pool examples : Dp=1 m, E=120 kW/m2 Dp=50 m, E= 20 kW/m2 Medium to low emissive power for big pools (thermal radiation flux, up to 60 kW/m2 for liquid fuels) LPGs, LNG, provide higher emissive power (up to 150-270 kW/m2 for LPG, 250 kW/m2 for LNG) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o Emissive power : (cont.) One example of correlations available for max emissive power : m Hc f Emax H 1 4 f Dp Ε= emissive power (kW/m2) m= specific combustion rate (kg/m2.sec) ΔHc= combustion heat (kJ/kg) f = thermal radiation ratio Hf = flame length Dp = pool diameter This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o Emissive power : (cont.) Final emissive power must take into account smoke production. Example correlations : E Emax (1 s) Esmoke s E 140 e 0.12 D p 20 (1 e 0.12 D p ) s, smoke coverage of surface Dp, pool diameter (m) Esmoke, emissive power of smoke (kW/m2) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Solid flame radiation model (cont.) o Emissive power : (cont.) Please be careful !!!! Make sure radiation fraction used is in-line with experimental data if available Evaluate calculation results for emissive power with experimental results, if available This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) UK HSE suggestions for LPGs : o Emissive power : 200 kW/m2 over half flame height Some conservative assumptions o For unconfined LPG cases, for theoretical pool calculation : butane fire instead of similar propane release (lower boiling rate, higher pool diameter low ambient temperature examined (as above) o Low relative humidity examined (high transmissivity coefficient) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Example results for propane pool fire Dp=10 m, wind speed 5 m/sec T=25 °C (confined fire, Aloha), o flame height Hf : 21 m o combustion rate M : 400 kg/min This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pool fire models (cont.) Example results for Methanol tank, Dtank=20 m, H tank=20 m, T= 25 C°, atmospheric conditions D5, 2 in hole on tank shell at ground level (burning unconfined pool, Aloha) o pool diameter Dp = 27 m o flame length : 11 m This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball, BLEVE (Boiling Liquid Expanding Vapour Explosion) Rapid release and ignition of a flammable under pressure at temperature higher than its normal boiling point Main consequence Thermal radiation Secondary consequences: oFragments (missiles) oOverpressure LPG BLEVE (Crescent City) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE characteristics Very rapid phenomenon (expanding velocity 10 m/sec) Limited duration (up to appr. 30 sec, even for very large tanks) Significant extent of fireball radius (in the order of 300 m for very big tanks, ≈ 4000 m 3) Very high emissive power (in the order or 200350 kW/m2) No precise capability for prediction of when it will happen (usual initial step for tanks exposed to heat -pool fire, jet flame-, opening of PSVs) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE characteristics and models (cont.) Radius and duration from correlations with tank content, example (AIChE CCPS) : D 5.8 m1 / 3 t 0.45 m1 / 3 m 30 tn t 2.6 m1 / 6 m 30 tn o t, duration (sec) o m, mass (tn) o No significant deviations for various correlations available, example results for full propane tank BLEVE (100 m3) Radius (max), m Duration, sec This Project is funded by the European Union ADL TNO AIChE Aloha 122 112 110 106 16 14 16 13 Project implemented by Human Dynamics Consortium • Fireball/BLEVE characteristics and models (cont.) Radius and duration from correlations with tank content, example (AIChE CCPS) : D 5.8 m1 / 3 t 0.45 m1 / 3 m 30 tn t 2.6 m1 / 6 m 30 tn o t, duration (sec) o m, mass (kg) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE characteristics and models (cont.) Mass in fireball calculations : o Typically whole tank content (worst case approach.) o Netherlands (BEVI method) : gas phase + 3 x flash fraction of liquid phase at failure pressure. For typical failure pressure in LPGs with hot BLEVEs, results to whole tank content For propane at usual atmospheric conditions, results to whole tank content This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE characteristics and models (cont.) Solid flame model o radiation emitted via fireball surface, o Usually fireball considered as sphere touching ground (conservative approach, adopted by UK HSE) Evolution of fireball/BLEVE This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE characteristics and models (cont.) Solid flame radiation model (cont.) o Calculation based on : sphere shape at contact with ground, distance from fireball (sphere View Factor), fireball emissive power (thermal radiation flux at fireball surface) q VF E ta q = thermal radiation flux at “receptor” (kW/m2) VF = view factor ar receptor for sphere shape fireball Ε = emissive power (kW/m2) ta = transmissivity coefficient This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE characteristics and models (cont.) Solid flame radiation model (cont.) o View Factor : function of distance of receptor from flame and fireball radius o Transmissivity coefficient : as in pool fire case This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE characteristics and models (cont.) Emissive power in fireball calculations : o Correlations are available for emissive power calculation based on : o vapour pressure at failure conditions (AIChE CCPS) E 235 Pv0.39 Pv, vapour pressure at failure (MPa) o and/or mass involved, duration, size of fireball o Experimental data provide values up to 350 kW/m2 o UK, HSE suggestion >270 kW/m2 This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE models (cont.) Example results (full 100 m3 propane tank BLEVE, Aloha) o But, duration is only 13 sec. For limit values set in TDU (not in kW/m2), the relevant thermal radiation flux limit must be calculated This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Fireball/BLEVE models (cont.) Example results (full 100 m3 propane tank BLEVE, Aloha) (cont.) o For t=13 sec, 1500 TDU corr. to 35 kW/m2 450 TDU corr. to 14 kW/m2 170 TDU corr. to 6.9 kW/m2 This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame Ignition of gas or two-phase release from pressure vessel Main consequence Thermal radiation Propane jet flame test This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame characteristics Results as outcome of gas or two phase releases of flammable substances Cone shape Long duration (minutes to hours, depends on source isolation) Very high emissive power (in the order or 200 kW/m2) Soot expected, but not affecting radiation levels This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame models Combustion rate determined by release rate Dimensions from empirical equations. Example of simplified Mudan-Cross equation L 15 d Ct MWa MW f L= jet flame length d= release point diameter Ct= fuel content per mole in stoichiometric mix of fuel/air ΜWa= air molecular weight MWf= fuel molecular weight This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame models (cont.) Dimensions from empirical equations. Example of simplified Considine-Grint equation for LPGs L 9.1 M W 0.25 L L= jet flame length M= release rate (kg/sec) W= jet radius at flame tip (m) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame models (cont.) Point source models o Single point : all energy is released from flame “center”. Similar to relevant point source model for pool fires o Multipoint source : several point along jet trajectory taken into account This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame models (cont.) Solid flame radiation model o Radiation emitted via flame surface o Calculation based on : shape (cylinder, tilted or not) distance (View Factor) emissive power This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame models (cont.) (cont.) Solid flame radiation model (cont.) o Calculation equation : q VF E ta q = thermal radiation flux at “receptor” (kW/m2) VF = view factor for flame shape at receptor Ε = emissive power (kW/m2) ta = transmissivity coefficient This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame models (cont.) Solid flame radiation model (cont.) o View Factor : function of distance of receptor from flame and flame dimensions for shape assumed o Transmissivity coefficient : as in pool fires, fireball/BLEVE o Emissive power : Estimated by flame dimension (surface) and energy released E= Emissive power (kW/m2) M= release rate (kg/s) ΔΗc= combustion energy (kJ/s) A= jet surface area, m2 M c E Fs A jet Fs 0.21 e 0.00323u j 0.11 Fs= fraction of combustion energy radiated uj= expanding jet velocity (m/sec) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame models conservative approaches Examination of horizontal jet o Produce more extended thermal radiation zones o Have direct effect via impingement in near by equipment Wind speed (for models taking into account flame distortion due to wind) : Vertical jets : High wind speed (UK HSE suggestion 15 m/sec) Horizontal jets : Low wind speed (UK HSE suggestion 2 m/sec) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Jet flame models example results (cont.) Example results, 2 in hole in top of propane tank/gas phase, vertical jet (Aloha) This Project is funded by the European Union Project implemented by Human Dynamics Consortium END OF PART A This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud dispersion (cont.) Extent of cloud : dimensions, downwind/crosswind till specific endpoints (concentration) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud dispersion (cont.) Endpoints : o Toxics : several toxicity endpoints (e.g. IDLH, LC50) o Flammables : LFL, ½ LFL Deaths expected within cloud limits where ignition is possible (Flash fire) due to thermal radiation and clothes ignition Reporting of LFL, ½ LFL is for theoretical extend of cloud, as no ignition is assumed on cloud path Very extended clouds expected for LPGs, especially in catastrophic failure cases (in the order of 500-1500 m) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud dispersion (cont.) Endpoints : (cont.) o Flammables : (cont.) Usually ignition sources outside establishment premises limit actual cloud Protection zones not justified to take into account flammable dispersion till LFL, ½ LFL This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud dispersion (cont.) Example results for LPG dispersion (SLAB) at ground level centerline 100% Centerline "ground" concentration (v/v) 0 10% 50 100 150 200 250 300 350 400 UFL LFL 1% 1/2 LFL 0% Downwind distance, x (m) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud dispersion (cont.) Example results for LPG dispersion (SLAB) (cont.) at ground level 100 1/2 LFL Crosswinf distance (CW) 50 44 DW (UFL) LFL 20 CW (UFL) UFL 128 DW (LFL) 0 0 50 100 150 200 250 67 CW (LFL) 204 DW (1/2LFL) 94 CW (1/2LFL) -50 -100 Downwind distance (DW), x(m) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Release conditions affecting dispersion : o substance properties (Boiling Point etc.) o pressure, temperature at containment o release rate and area o release point height o release direction (upwards –PSV-, horizontal) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Meteorological conditions affecting dispersion : o atmospheric stability class (A-F), o wind speed, o air temperature, o humidity (for some substances reacting with water as for example HF or other polar substances : SO2, NH3 etc.) o Type of area : rural/industrial/urban, roughness factor This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Atmospheric stability : o Expression of turbulent mixing in atmosphere. Related with atmospheric vertical temperature gradient (dT/dz) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Atmospheric stability : (cont.) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Atmospheric stability : (cont.) o Usually attributed to standardized class A-F (Pasquill) A : unstable, in combination with high winds favors dispersion D : neutral F : stable, minimum mixing in atmosphere o Other parameter to attribute atmospheric stability, Monin-Obukhov length (positive for stable conditions, negative for unstable conditions) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Atmospheric stability : (cont.) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Atmospheric stability : (cont.) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Wind speed : o Wind speed referred in meteorological data usually refer to measurement at 9-10 m height o Boundary layer effect (variation of speed with height) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Wind speed : (cont.) o Simplified function : u z uref p, function of : z z ref p stability class surface roughness This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Wind speed : (cont.) o Variation with stability class for rural environment : 32 Wind velocity, m/sec 28 C A 24 D F E 20 16 12 8 4 0 0 5 10 15 20 25 30 35 Height, m This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Meteorological conditions : o Typical set under interest in Safety Reports : D5 : stability class D, uref=5 m/sec (unstable conditions) F2 : stability class D, uref=2 m/sec (stable conditions). Worst case for extent of vapour cloud, especially in heavy gas dispersion This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Type of surroundings : rural/industrial/urban o Refers to variation of height in elements of surrounding o Usually attributed via “roughness factor” This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Type of surroundings (cont.) o Conservative approach : open country (rural) Averaging time : o Variation in time, due to turbulence, of wind characteristics : speed direction This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Averaging time : (cont.) y o For continuous releases, concentration at constant location (x, y, z) is not constant x not exposed T=0 average wind direction y exposed x T=t This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Averaging time : (cont.) o increase of averaging time : plume boundaries widen concentration distribution flattens This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Averaging time : (cont.) o Very important to use averaging time in models, suitable to exposure time under interest o Models may use parameters for certain averaging time, which might not be suitable for application in Safety Report. PLEASE ALWAYS CHECK !!! o Gaussian models use implicit 10-min averaging time but… This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Averaging time : (cont.) o Toxics exposure usually under interest for period of 30 min (due to LC50 30 min endpoints etc.) o Aloha-DEGADIS (Heavy Gas Dispersion) uses 5 min for toxics o Ignition of flammable cloud is related with very low exposure time (time just for ignition to happen). o Aloha-DEGADIS uses 10 sec for flammables (e.g. LPGs, no matter if toxic effect is examined) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Averaging time : (cont.) o Example results for propane release from liquid phase piping (SLAB) 100 100 Averaging time = 1 sec 1/2 LFL 50 50 44 DW (UFL) LFL 20 CW (UFL) UFL 128 DW (LFL) 0 0 50 100 150 200 250 67 CW (LFL) 204 DW (1/2LFL) 94 CW (1/2LFL) -50 Crosswind distance (CW) Crosswinf distance (CW) Averaging time = 30 min 1/2 LFL 28 DW (UFL) 10 CW (UFL) LFL 86 DW (LFL) 0 0 UFL 50 100 150 200 250 47 CW (LFL) 136 DW (1/2LFL) 72 CW (1/2LFL) -50 -100 -100 Downwind distance (DW), x(m) This Project is funded by the European Union Downwind distance (DW), x(m) Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Averaging time : (cont.) o Reason for reporting both LFL, ½ LFL in flammable dispersion o ½ LFL reporting contributes to uncertainty of averaging time (conservative approach) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Passive (neutral) dispersion (Gauss) : o Release of gas with density equal or higher than air This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Passive (neutral) dispersion (Gauss) : o Basic characteristics: Maximum concentration at centreline Concentration reducing with increasing distance from source If release at ground level, maximum concentration at ground level This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Passive (neutral) dispersion (Gauss) : o Basic equation for point source continuous release C ( x, y , z ) M 2 y z u e y2 2 2 y ( z H2e ) 2 ( z H2e ) 2 2 z 2 z e e u, wind speed at z (m/sec) M, release rate flow (kg/sec) Hd, active release height (m) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Passive (neutral) dispersion (Gauss) : (cont.) o σy, σz : functions of stability class with x and roughness factor usually given for 10-min averaging time proper correction of σy based on necessary averaging time is required t a ver y 10 min y This Project is funded by the European Union t aver 10 min 0. 2 Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Passive (neutral) dispersion (Gauss) : (cont.) o Example results for dispersion for NH3 release by 2 in hole in 6 bar gas vessel, D5 (Aloha) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Passive (neutral) dispersion (Gauss) : (cont.) o Equations provided for point stationary source (no momentum) o But jets of releases have significant momentum due to high velocity… modifications needed in model to take into account release momentum This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Passive (neutral) dispersion (Gauss) : (cont.) o For jets of releases modifications are needed, typical example : plume rise parameter to modify the release source point at downstream location Modified Original This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Flue gases dispersion o Example : pool fire combustion products, e.g., SO2 o Special characteristics : Large area of source (e.g. tank area, bund area), (not point source). Modifications are available to models or sources are treated as point ones High temperature of flue gases Relevant plume rise equations provided for stacks (Briggs, Holland equations), provide unrealistic plume rise height o Conservative approach, no plume rise, dispersion begins from flame end This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Flue gases dispersion (cont.) o Special atmospheric condition to be considered (temperature inversion conditions, trapped plume) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Plume rise effects o Concentration at centrelines not continuously decreasing with distance o Max concentration at centreline appears at distance from source This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Plume rise effects (cont.) o Example results from calculation of SO2 dispersion from dike fire of heating oil tank (D5, release rate 0.25 kg/sec SO2, dike equivalent diameter 66 m) Ground level centreline concentration (mgr/m3) 0,8 0,6 0,4 0,2 0,0 0 1000 2000 3000 4000 5000 6000 Downwind distance (m) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Heavy gas dispersion o Special complex models CFD Box models (instant releases) Grounded plume models (continuous releases) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Heavy gas dispersion (cont.) o Maximum concentration expected at centerline o Concentration decreases with increasing distance o More extended plume compared to neutral dispersion This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion (cont.) Heavy gas dispersion (cont.) o Meteorological data F2 produce more extended cloud o Propane/butane cases same release source (e.g. same hole size) will produce more extended cloud for propane due to higher release rate This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion models Heavy gas dispersion (cont.) o Example results for propane release from 2 in hole in liquid phase of tank (D5) (Aloha) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour cloud (gas) dispersion models Heavy gas dispersion (cont.) o Example results for propane release from 2 in hole in liquid phase of tank (F2) (Aloha) This Project is funded by the European Union Project implemented by Human Dynamics Consortium END OF PART B This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) Delayed ignition of flammable vapour cloud under partial confinement (obstacles within cloud) producing overpressure during flame front propagation Main consequence Overpressure VCE results (Flixborough) This Project is funded by the European Union Secondary consequences: oFragments (e.g. broken glasses) Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) characteristics Very short duration (<1 sec) Models, high uncertainty due to several assumptions used in every model Type of models: o CFD (FLACS, PHOENIX etc.) o TNT blast charge (TNT equivalency) o Air-fuel charge blast (Multi-Energy, Baker Strehlow -Tang etc.) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) models TNT equivalency model o Simple, based on analogy with explosives effects o A fraction of combustion energy released in cloud is attributed to produce overpressure o The former energy fraction is recalculated as equivalent (on energy basis) mass of TNT o The effects are defined based on known correlation of overpressure with TNT mass This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) models (cont.) TNT equivalency model (cont.) WTNT e W f Hc f HcTNT Wf= flammable in cloud WTNT= TNT equivalent mass (combustion energy basis) αe = TNT equivalency coefficient (energy basis) Hcf = combustion energy of flammables (kJ/kg) o αe, refers to part of combustion energy released producing overpressure (1-10%) o High uncertainty in both αe value and quantity of flammables (released mass –till what time ???-, mass within LFL-UFL) to be used o Review on topic by TNO Yellow Book and AICheJ CCPS Guideline This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) models (cont.) TNT equivalency model (cont.) o Some comments/examples on selection of mass and αe : αe must be selected along with suitable flammable mass for αe 1-5%, mass must not contain only the part of cloud in LFL-UFL section flammable mass must take into account not only gas but also liquid droplets (aerosol) in 2-phase releases Dow approach : mass defined by release rate and time to maximise LFL distance This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) models (cont.) TNT equivalency model (cont.) o Some comments/examples on selection of mass and αe : (cont.) mass defined by time to reach ignition source or time to stop release (time for energizing isolation valves) HSE suggests TNT mass double the gas mass in confined areas o Explosives blast and VCE present differences, as explosives have short duration higher shock wave peak values. o TNT equivalency model is approximation of phenomenon based on statistical analogies This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) models (cont.) TNT equivalency model (cont.) R R 3 WTNT HcTNT= combustion energy for TNT (4,680 kJ/kg) R = Hopkinson distance (m/kg0.33) R= distance from explosion centre o Overpressure calculated by diagram for distances required o Uncertainty on centre of explosion to be considered o Similar diagrams for positive phase duration, impulse This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) (cont.) TNO Multi-Energy model : o Only confined areas of cloud are considered o Partial explosions from confined areas expected o Energy released assuming stoichiometric combustion, based on air contained in areas taken into account (average 3.5 MJ/m3 of air for most hydrocarbons) uniform concentration of flammable in confined areas assumed This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) (cont.) TNO Multi-Energy model : (cont.) o Overpressure from Berg graph using Sachs distance R R E 3 P0 E= combustion energy R = Sachs distance R= distance from explosion centre This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) (cont.) TNO Multi-Energy Model : (cont.) o Similar graphs for positive phase duration, dynamic pressure o Blast strength 10 : detonation, explosives case, not valid for VCEs as propagation of blast via detonation requires high homogeneity in cloud This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) (cont.) TNO Multi-Energy Model : (cont.) o Disadvantage : complex empirical rules for (TNO Yellow Book, Assael) : definition of confined areas definition of successive or simultaneous blast in confided areas selection of blast strength (confinement increase, increases blast strength o HSE suggests blast strengths 2 and 7 This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) (cont.) Baker-Strehlow-Tang model o Similar principles as TNO Multi-Energy model confined areas only taken into account stoichiometry of air with fuel in confined areas o Gas type “reactivity” (susceptibility to flame front acceleration) taken also into account along with obstacle density o methane, CO : low reactivity o H2, acetylene, ethylene/propylene oxide : high reactivity o other substances : medium This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) (cont.) Baker-Strehlow-Tang model (cont.) o Flame speed defined by table on gas reactivity and confinement type This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) (cont.) Baker-Strehlow-Tang model (cont.) o Overpressure from graph using Sachs distance and flame speed Energy to be used double to actual as graph presents free air blast (not surface blast) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Vapour Cloud Explosion (VCE) (cont.) Example results for propane release from 2 in hole in liquid phase of tank (D5) (Aloha, Baker-Strehlow-Tang method) Ignition time 2 min This Project is funded by the European Union Composite for unknown ignition time Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion Variation of stored substances quantities within year due to seasonal production of some products. o Evaluation of stored quantities distribution could be required to evaluate quantities to be taken into account in calculations. Some times, active substance stored in powder form Specific combustion rate rather low (TNO Green Book) : in the range of 0.02 kg/m2.sec This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion (cont.) Special characteristic of pesticides : when burnt, not all substance is consumed, flue gases contain unburned pesticide substances (“survivor” fraction) In case of fire, dispersion of flue gases must examine : o Combustion products (e.g. SO2, HCl, NO2 etc.) o Unburned pesticide active substance This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion (cont.) UK HSE suggests stoichiometric conversion of S, Cl to SO2 and HCl. Conversion ratios : o C to CO : 5% o N to HCN and NO2 : 5% According to TNO Green Book, formation rate of NO2, HCN and NO decreases with this order, Taking into account the similar toxicity of the former, conversion of N to NO2 only is conservative This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion (cont.) Survivor fraction in flue gases : 0.5-10% of combustion rate of substance at source Lower survivor fraction for high boiling point substances UK HSE suggests survivor fraction 10% otherwise justification must be provided This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion (cont.) Especially in closed warehouse cases : o what is plume rise for flue gases ?? o which is the combustion rate ?? Plume rise in warehouse flue gases o For fire in full development plume rise might be high, but potentially low in initial phase of fire o Typical equations fail, as producing unrealistic plume rise This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion (cont.) Combustion rate, affected by type of fire o Roof collapse As in open area, fuel controlling High flue gas temperature o Ventilation controlled roof intact, some window breakage (limited release area) fire rate controlled by availability of oxygen in warehouse low temperature of flue gases, low plume rise This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion (cont.) UK HSE suggests : o plume rise set at max 50 m o calculations for source via a few m2 area of window (nevertheless, recognized as pessimistic), (NTUA methodology refers to 3 m2) o special models UK HSE suggests meteorological condition to be examined as worst case ones : o F2 o D5, D15 with low inversion height (400 m) This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion (cont.) NTUA methodology suggests the following cases : o Roof collapse : flue gas rate 8 kg/m2.sec (per warehouse area), T=500 °C o Ventilation controlled (roof intact) : flue gas rate 5 kg/m2.sec (per opening area, assumed 3 m2), T=140 °C This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fires and dispersion (cont.) NTUA methodology classification of fires (as per HSE FIRE Pest II computer program) Combustion rate Duration High intensity fire Flammable liquids (product solutions in solvents) purring in floor and pool fire. Inert “technical” substances or products containing significant percentage of flammable solvents High 6500 sec Medium intensity fire Inert “technical” substances or products containing significant percentage of flammable solvents Medium 7000 sec Low intensity fire Inert “technical” substances or products in flammable packaging Low 13000 sec Combustibles This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Pesticides fire and dispersion (cont.) Survivor fraction according to NTUA methodology (as per HSE, Risk Assessment Method for Warehouses 1995) Solid substance Liquid substance Liquid substance (2) particles <2 mm, high storage height (1) cans < 10kg, large storage height (1) cans < 10kg, small storage height (1) metal drums High intensity fire 10 10 0.5 10 Medium intensity fire 5 5 0.5 4 Low intensity fire 2 2 0.5 1 Liquid substance (1) Medium and large storage height : > 2 m, small storage height: < 2 m (2) Same for particles <2 mm and small storage height This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Accidents with effects to environment No mature and wide-used quantitative models for estimation of effects to environment Qualitative models (applied some times, examples : o Energy Institute (ex. IP) Screening Tool o Belgium (Flanders) Richtlijn Milieurisicoanalyse o IPC Guidance Note on Storage and Transfer of Materials for Scheduled Activities, Irish EPA No unique approach in EU members (in many countries no specific approach) in relevant requirements This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Literature for Top Events Consequence Analysis Models Lees’ Loss Prevention in the Process Industries, Elsevier Butterworth Heinemann, 3nd Edition, 2005 Methods for the Calculation of Physical Effects due to Releases of Hazardous Materials (Liquids and Gases), Yellow Book, CPR 14E, VROM, 2005 Methods for the Determination of Possible Damage to People and Objects Resulting from Releases of Hazardous Materials , Green Book, CPR 16E, TNO, 1992 Guidelines for Chemical Process Quantitative Risk Analysis, CCPSAICHE, 2000 Guidelines for Consequence Analysis of Chemical Releases, CCPSAICHE, 1999 Guidelines for Evaluating the Characteristics of Vapour Cloud Explosions, Flash Fires and BLEVEs, CCPS-AICHE, 1994 This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Literature for Top Events Consequence Analysis Models (cont.) Safety Report Assessment Guides (SRAGs), Health and Safety Executive, UK Risk Assessment Methods for Warehouses - Computer Program FIREPEST II, Health and Safety Executive, 1997 Assael M., Kakosimos K., Fires, Explosions, and Toxic Gas Dispersions, CRC Press, 2010 Benchmark Exercise in Major Accident Hazard Analysis, JRC Ispra, 1991 Rew P., Humbert W., Development of Pool Fire Thermal Radiation Model, HSE Contract Research Report 96, 1996 McGrattan K., Baum H., Hamins A. Thermal Radiation from Large Pool Fires, National Institute of Standards and Technology, NISTIR 6546, Nov 2000 Taylor J., Risk Analysis for Process Plant, Pipelines and Transport, E&FN SPON, 1994 This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Literature for Top Events Consequence Analysis Models (cont.) Drysdale D., Fire Dynamics, J. Wiley and Sons, 2nd Edition, 1999 Beychok M., Fundamentals of Stack Gas Dispersion, 3rd Edition, 1994 C. Delvosalle, F. Benjelloun, C. Fiévez,, A Methodology for Studying Domino Effects, Faculté Polytechnique de Mons, Ministere Federal de l’;Emploi et du Travail, July 1998 RIVM, Reference Manual Bevi Risk Assessments, 2009 ALOHA, Users Manual, US EPA, 2007 ALOHA Two Day Training Course Instructor's Manual Environmental risk assessment of bulk storage facilities: A screening tool, EI, 2009 Richtlijn Milieurisicoanalyse, 2006 This Project is funded by the European Union Project implemented by Human Dynamics Consortium • Literature for Top Events Consequence Analysis Models (cont.) IPC Guidance Note on Storage and Transfer of Materials for Scheduled Activities, Irish EPA, 2004 N. Markatos, NTUA, Chemical Engineering Department, Methodology of Assessment of Consequence from fire in Pesticide installations, 2001 (in Greek) This Project is funded by the European Union Project implemented by Human Dynamics Consortium