1. This Project is funded by the European Union Project implemented by Human Dynamics Consortium This project is funded by the European Union Projekat finansira Evropska Unija Project implemented by Human Dynamics Consortium Projekat realizuje Human Dynamics Konzorcijum ACCIDENT SCENARIOS AND CONSEQUENCE ANALYSIS Antony Thanos Ph.D. Chem. Eng. antony.thanos@gmail.com

2. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Risk Analysis Framework Hazard Identification Hazard Identification Consequence Analysis Consequence Analysis Accepted Risk NO Risk reduction measures END Accident Scenarios Accident Scenarios Accident Probability Accident Probability Risk Assessment Risk Assessment YES

3. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Hazard identification usually specify release expected and not final accident (top event) Typical release scenarios per equipment type failure :  Pipes oCatastrophic failure (Full Bore Rupture – FBR- or guillotine break) oPartial failure (hole diameter equivalent to a fraction of pipe diameter, e.g. 20%)

4. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Typical release scenarios per equipment type failure (cont.) :  Pressure vessel (process vessel, tank, tanker) oCatastrophic failure: “instantaneous” rupture (complete release of content within short time e.g. 3-5 min) oMechanical failure : equivalent hole set to e.g. 50 mm oSmall leakage (e.g. corrosion), smaller hole with equivalent diameter of e.g. 20 mm

5. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Typical release scenarios per equipment type failure (cont.) :  Pressure vessel connected equipment oRelease from PSV oFailure of connecting pipes (as for pipes above)  Pumps/compressors oRelease from PSV oLeakage from seal (equivalent small hole diameter set, e.g. 20 mm)

6. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Typical release scenarios per equipment type failure (cont.) :  Atmospheric liquid fuel tanks oIgnition in floating roof tank (tank fire) oIgnition of constant roof tank (tank fire) oFailure of tank with release to dike (bund) of tank and subsequent fire in dike (dike fire)

7. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Worst case scenarios  Although low probability expected, indispensible for Land Use Planning and Emergency Planning  Worst case releases/scenarios to be provided for the different sections of Plant (type of activities) : oEach Production Unit oTank-farm oMovement facilities (road/rail tanker stations, ports)

8. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Worst case scenarios (cont.)  Worst case releases/scenarios within sections : oCatastrophic failure of vessel (process vessel, tank, tanker) with maximum inventory size oCatastrophic failure of pipe : Full Bore Rupture (FBR)/Guillotine Break) for pipes, especially for movement facilities (import/export pipelines, hoses/loading arms)

9. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Worst case scenarios (cont.)  Worst case releases/scenarios within sections : oFor liquid fuels tanks, fire in : Largest diameter tank Dike with largest equivalent diameter

10. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Worst case scenarios (cont.)  Worst case releases/scenarios must take into account : oDifferent operating conditions (P/T/phase) e.g. : For liquefied gases piping, worst case is usually expected from liquid phase pipe failure For LPGs, worst case is usually expected from pure propane compared to butane (due to higher pressure)

11. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Worst case scenarios (cont.)  Worst case scenarios selection criteria (cont.) : oDifferent operating conditions (P/T/phase) e.g. (cont.) : Smaller tank of pressurized ammonia can produce more extended consequences than larger refrigerated ammonia tank

12. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Worst case scenarios (cont.)  Worst case releases/scenarios must take into account : oDifferent substances, e.g. smaller tank of a very toxic substance can produce more extended consequence than a larger tank of a toxic substance oProximity to site boundaries, especially if vulnerable objects are close

13. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Worst case scenarios (cont.)  Worst case scenarios usual convention : Only one failure can happen at a certain time oNo simultaneous accidents expression, e.g. only single tank BLEVE in LPG tank farm at a time oNo double containment failure, e.g. in refrigerated tanks with secondary containment only primary containment failure is taken into account, if no special reasons are present

14. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Release rates models from vessels  Release of liquids (Bernoulli equation)  Release of gases (adiabatic expansion at hole)  Release of liquefied gases : oGas phase release, as for usual gases oLiquid phase release, special two-phase release models to be used, taking into account equilibrium (or not) at release point  Evaporation from pools : complex models, taking into account : substrate type, substance properties, atmospheric conditions etc.

15. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Hazard identification usually specify release expected and not top event (final accident)  Example : Release of LPG (gas phase) from tank identified in a HAZOP. Various types of top events can be evolved (Jet flame, flash fire, UVCE) Consequence analysis requires top events to be specified Gap closed by techniques such as “Event tree”

16. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Event tree  Logic evolution of initial release identified, as far as its outcome type (top event)  Top events identified per initial release event (e.g. jet flame after failure of pipeline due to corrosion)  Technique in the borderline of hazard identification and consequence analysis

17. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Event tree (cont.)  Example: Gas phase release from LPG tank PHASEIGNITIONCONFINEMENTTOP EVENT DIRECT JET FLAME GASDELAYED NO CONFINEMENTFLASH FIRE CONFINEMENTUVCE NO IGNITIONSAFE DISPERSION

18. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Consequence analysis framework Release scenarios Release scenarios Accident type Accident type Event trees Release quantification Release quantification Hazard Identification Release models Consequence results Consequence results Domino effects Limits of consequence analysis Dispersion models Fire, Explosion Models

19. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Main top event categories Toxic dispersion Toxic dispersion Explosion Fire Hazardous substance release Initial eventTop event Toxic effects Overpressure Thermal Radiation Thermal Radiation Consequences Fire Thermal Radiation Thermal Radiation Toxic dispersion Toxic dispersion Toxic effects Fire Thermal Radiation Thermal Radiation

20. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Pool fire  Ignition of flammable liquid phase Liquid fuel tank fire Main consequence Thermal radiation Main consequence Thermal radiation

21. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Pool fire characteristics  Confined (liquid fuels tank/dike fire) / Unconfined (LPG pool from LPG tank failure –no dike present)  Pool dimensions (diameter, depth)  Flame height, inclination  Medium to low emissive power (thermal radiation flux, up to 60 kW/m 2 for liquid fuels)  Long duration (hours to days)  Combustion rate

22. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Pool fire models  Combustion rate per pool surface based on empirical equations (Burges, Mudan etc.)  Flame dimension from empirical equations (Thomas, Pritchard etc.)  Radiation models : oPoint source No flame shape taken into account Fraction of combustion energy considered to be transmitted by point in pool center

23. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Pool fire models  Solid flame, radiation emitted via flame surface, calculation based on : flame shape, distance (View Factor), emissive power Pool diameter Flame height Pool depth

24. 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 LPG BLEVE (Crescent City) Main consequence Thermal radiation Main consequence Thermal radiation Secondary consequences: oFragments (missiles) oOverpressure

25. This Project is funded by the European Union Project implemented by Human Dynamics Consortium BLEVE characteristics and models  Fireball radius  Duration (up to appr. 30 sec, even for very large tanks)  Very high emissive power (in the order or 200- 350 kW/m 2 )  Radius and duration from correlations with tank content

26. This Project is funded by the European Union Project implemented by Human Dynamics Consortium BLEVE characteristics and models (cont.)  Solid flame radiation model, radiation emitted via fireball surface, calculation based on : sphere shape at contact with ground, distance (View Factor), fireball emissive power Evolution of BLEVE

27. 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 Propane jet flame test Main consequence Thermal radiation Main consequence Thermal radiation

28. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Jet flame characteristics and models  Cone shape, dimensions from empirical equations  Long duration (minutes to hours, depends on source isolation)  Very high emissive power (in the order or 200 kW/m 2 )  Combustion rate determined by release rate  Solid flame model, radiation emitted via flame surface, calculation based on : shape (cylinder), distance (View Factor), emissive power

29. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Vapour cloud (gas) dispersion  Neutral dispersion (stack type)  Heavy gas dispersion, e.g. liquefied under pressure gas releases as for LPG. Vapour cloud remains for long distance at ground level Propane cloudHeavy gas behaviour

30. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Vapour cloud (gas) dispersion (cont.)  Extent : dimensions, downwind/crosswind till specific endpoints (concentration)  Endpoints: oFlammables : LFL, ½ LFL Deaths expected within cloud limits where ignition is possible (Flash fire) due to thermal radiation and clothes ignition oToxics : several toxicity endpoints (e.g. IDLH, LC50)

31. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Vapour cloud (gas) dispersion (cont.)  Affecting parameters: oRelease conditions : substance properties, flowrate, hole diameter, pressure, temperature, release point height, release direction (upwards –PSV-, horizontal) oMeteorological conditions : atmospheric stability class (A-F), wind speed, temperature, humidity oType of area : rural/industrial/urban, roughness factor

32. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Vapour cloud (gas) dispersion models  Passive (neutral) dispersion : Gauss model  Heavy gas dispersion : special complex models  Flue gases : Gauss model modified for plume rise effects

33. 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 Main consequence Overpressure VCE results (Flixborough) Secondary consequences: oFragments (e.g. broken glasses)

34. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Vapour Cloud Explosion (VCE)  Very short duration (sec)  Models (several assumptions used in every model) oTNT equivalency : Simple, based on explosives effects Fraction of combustion energy attributed to overpressure development High uncertainty in both fraction value and assumed quantity of flammables to be used

35. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Vapour Cloud Explosion (VCE) (cont.)  Models (cont.) oTNO Multi-energy : Only confined areas of cloud considered Complex empirical rules for definition of confined areas and blast strength Overpressure from Berg graph using Sachs distance

36. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Vapour Cloud Explosion (VCE) (cont.)  Models (cont.) oBaker-Strehlow-Tang Similar principles as TNO Multi-Energy model Gas type reactivity taken also into account along with obstacle density Overpressure from graph using Sachs distance

37. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Impacts  Probit functions oRelation of probability for a certain damage level (e.g. 2 rd degree burn, death) and cause value (e.g. thermal dose value) P =  (Pr), Pr = A + B ln(D), P : probability value Pr : probit value  : standard function of probability with probit value A, B : probit constants for a specific harm D : cause value

38. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Impacts (cont.)  Thermal radiation oImpacts depend on both thermal radiation flux and exposure duration, e.g. oThermal radiation flux 37,5 kW/m 2 : damage to equipment after 20 minutes 100% lethality in 1 minute 1% lethality for 10 seconds

39. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Impacts (cont.)  Thermal radiation (cont.) oBest practice the use of Thermal Dose : TDU = Q 4/3 t Q (W/m 2 ), emissive power (thermal radiation flux) at flame/fireball surface t (sec), exposure time : BLEVE event : BLEVE duration other events : escape time, usually 0,5-1 minutes

40. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Impacts (cont.)  Thermal radiation (cont.) oProbit constants available in literature for several levels of harm from thermal radiation oEndpoints for thermal radiation defined usually for effects (e.g. lethal effects, irreversible damage) to humans oEffects to structures usually useful only for Domino effects

41. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Impacts (cont.)  Toxic effects oDose concept : Dose = C n t C, concentration t, exposure time (in the order of 30-60 minutes) n, exponent depending on substance (available on literature for several toxics, usually 1-2)

42. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Effects (cont.)  Toxic effects (cont.) oProbit constants available in literature for several toxics oToxic endpoints definitions must include exposure time, e.g. LC50 (30 min) oLiterature toxicity data must be adjusted to humans and for the required exposure time, e.g. literature data for LC1 (2 hours) on rats must be adjusted to LC50 (30 min) for humans

43. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Impacts (cont.)  Overpressure oUsual endpoints defined on constant values for expected effects to structures (light damage, severe damage etc.) oEffects to humans are present at similar or higher overpressures than for effects to structures

44. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Effects (cont.)  Environment oNo mature and wide-used quantitative models for estimation of effects to environment oQualitative models applied some times oNo unique approach in EU members in relevant requirements

45. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Risk : The probability of cause of harm from accident  The probability of dead from fall of lightning is 10 -7 per year (1 person per 10.000.000 persons will die from lightning per year) Individual Risk : Risk of harm from accident, at specific location, independent of affected subjects  Example : The risk of lethal effects from thermal radiation at distance of 100 m from a specific gasoline tank is 10 -6 per year from fire in the gasoline tank

46. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Societal Risk :  Relationship between frequency and the number of people suffering from a specified level of harm in a given population from the realisation of specified accidents  Concerns estimation of the chances of more than one individual being harmed simultaneously by an incident

47. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Consequence/Risk acceptance in EU  Probabilistic approach oLimits usually set for individual risk oStrong dependency on quality of data oDifferences in data from different sources (e.g. failure rates in UK and Netherlands, or for probit function of toxics) oUsually requires large set of scenarios oSpecialized software required for efficient implementation

48. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Consequence/Risk acceptance in EU (cont.)  Deterministic approach oSimpler to implementation oNo probabilities of accidents used oSmaller set of scenarios required oMore conservative oWorst case scenarios included oSafety Zones usually set in-line with Zones for emergency planning

49. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Consequence/Risk acceptance in EU (cont.)  Hybrid approach oProbability band use oResults not so strongly related to probability value quality oAcceptance criteria defined by Risk Matrix oCloser to Rulebook approach

50. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Consequence/Risk acceptance in EU (cont.)  No unique methodology in determination of risk values  No unique approach in perception of risk (only vulnerable objects taken into account in Netherlands)  Diversity in limit values for same approach  Not always unique approaches for permitting, Land Use Planning and Emergency Planning

51. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Literature for Accident Scenarios and Consequence Analysis  Lees’ Loss Prevention in the Process Industries, Elsevier Butterworth Heinemann, 3 nd Edition, 2005  Methods for the Determination of Possible Damage to People and Objects Resulting from Releases of Hazardous Materials, Green Book, CPR 16E, TNO, 1992  Methods for the Calculation of Physical Effects due to Releases of Hazardous Materials (Liquids and Gases), Yellow Book, CPR 14E, VROM, 2005  Guidelines for Quantitative Risk Assessment, Purple Book, CPR 18E, VROM, 2005  Methods for Determining and Processing Probabilities, Red Book, CPR12E, VROM, 2005  RIVM, Reference Manual Bevi Risk Assessments, 2009

52. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Literature for Accident Scenarios and Consequence Analysis (cont.)  Guidelines for Chemical Process Quantitative Risk Analysis, CCPS- AICHE, 2000  Guidelines for Consequence Analysis of Chemical Releases, CCPS- AICHE, 1999  Guidelines for Evaluating the Characteristics of Vapour Cloud Explosions, Flash Fires and BLEVEs, CCPS-AICHE, 1994  Guidelines for Process Equipment Reliability Data with Data Tables, CCPS-AICHE, 1989  Assael M., Kakosimos K., Fires, Explosions, and Toxic Gas Dispersions, CRC Press, 2010  Crowl D., Louvar J., Chemical Process Safety Fundamentals with Applications, Prentice Hall, 2nd Edition, 2002

53. This Project is funded by the European Union Project implemented by Human Dynamics Consortium Literature for Accident Scenarios and Consequence Analysis (cont.)  Taylor J., Risk Analysis for Process Plant, Pipelines and Transport, E&FN SPON, 1994  Drysdale D., Fire Dynamics, J. Wiley and Sons, 2 nd Edition, 1999  Beychok M., Fundamentals of Stack Gas Dispersion, 3 rd Edition, 1994  Yaws C., Handbook of Chemical Compound Data for Process Safety, Elsevier Science & Technology Books, 1997