OSE551 Reliability and Risk Analysis for Offshore Plants Daejun CHANG (djchang@kaist.edu) Division of Ocean Systems Engineering Fire and Explosion - Explosion Risk Analysis -1- Ocean Systems Engineering Prof. Daejun CHANG Fundamentals Fundamentals -2- Ocean Systems Engineering Prof. Daejun CHANG Fire (Explosion) Triangle Fire (Explosion) Triangle Fuel Ignition source Air (oxygen) Fire/Explosion Sparks, flames, static electricity, heat Since air always exists for open-air explosion, we focus on the coexistence of the fuel and ignition source. -3- Ocean Systems Engineering Prof. Daejun CHANG Concept of Explosion Risk Assessment Concept of Explosion Risk Assessment Risk = Consequence x Frequency Consequence = overpressure Frequency f = f cloud x f ign f cloud : Frequency that the cloud exists at the point. f ign : Frequency that the ignition source exists at the point. -4- Ocean Systems Engineering Prof. Daejun CHANG Some real Some real--world issues world issues The cloud size is changing with time. Leak Dispersion Cloud formation Dilution by air ESD (process isolation) and EDP (blowdown) changes the leak rate. The ignition frequency is changing with time. Ignition frequency depends on the number of equipment, electrical instrument, hot work etc. Upon detection of the gas, the ESD system stops the electrical supply to the system (electric isolation). In consequence, the explosion risk changes with time. -5- Ocean Systems Engineering Prof. Daejun CHANG Time Dependence Time Dependence -6- Ocean Systems Engineering Prof. Daejun CHANG Leak rate with time Leak rate with time Time, s Leak rate, kg/s On set of leak Gas detection & Process isolation Emergency depressurization (blowdown) -7- Ocean Systems Engineering Prof. Daejun CHANG Gas volume with time Gas volume with time Time, s Leak rate, kg/s Gas volume, m 3 On set of leak Gas detection & Process isolation Emergency depressurization (blowdown) Dilution by ventilation -8- Ocean Systems Engineering Prof. Daejun CHANG Ignition density with time Ignition density with time Time, s Leak rate, kg/s Gas volume, m 3 Ignition density On set of leak Gas detection & Process isolation Emergency depressurization (blowdown) Dilution by ventilation -9- Ocean Systems Engineering Prof. Daejun CHANG Explosion frequency with time Explosion frequency with time Time, s Explosion frequency Gas volume, m 3 Ignition density Small because of low ignition density Small because of low cloud size -10- Ocean Systems Engineering Prof. Daejun CHANG Cloud Size Estimation Cloud Size Estimation -11- Ocean Systems Engineering Prof. Daejun CHANG Cloud size estimation Cloud size estimation Do we have to estimate the cloud size for all leak rates? 8 representative leak rates by NORSOK Standard Z-013: 0.75, 1.5, 3, 6, 12, 24, 48, 96 kg/s Do we have to simulate all the leak rates? Usually, some of them are simulated and the others interpolated Simulated: 0.75, 1.5, 3, 6, 12, 24, 48, 96 kg/s Interpolated: 0.75, 1.5, 3, 6, 12, 24, 48, 96 kg/s What situation do we have to simulate? All the scenarios including ESD and EDP? Numerous simulation cases Frozen cloud assumption! -12- Ocean Systems Engineering Prof. Daejun CHANG Frozen cloud assumption Frozen cloud assumption Time, s Leak rate, kg/s Gas volume, m 3 On set of leak Leak rate The cloud size is just dependent on the leak rate at the moment. That implies the cloud size is independent of its history. Is it justifiable? Cloud size -13- Ocean Systems Engineering Prof. Daejun CHANG Effect of Wind and Leak Direction Effect of Wind and Leak Direction -14- Ocean Systems Engineering Prof. Daejun CHANG Combined effects of leak direction and wind Combined effects of leak direction and wind The leak has direction as well as rate. Leak to the inside vs. Leak to the outside The former is the more destructive. Wind has both magnitude (speed) and direction High wind speed - High dilution rate - Wider dispersion Wind direction - The effect of the wind direction depends on the leak position. -15- Ocean Systems Engineering Prof. Daejun CHANG An approach to the combined effects An approach to the combined effects As the leak rate, we cannot simulate all the cases depending on Leak rate Leak direction Leak position Wind direction Wind speed -16- Ocean Systems Engineering Prof. Daejun CHANG An approach to the combined effects An approach to the combined effects Approaches There are N leak,pos leak positions. For each position, there are N leak,dir leak directions. For each leak direction, there are N leak,rate reference leak rates. For each rate, there are N wind,dir wind directions. For each wind direction, there are N wind,spd wind speed. For example: Total simulation cases = N leak,pos x N leak,dir x N leak,rate x N wind,dir x N wind,spd = 4 2 8 3 5 = 960 = 4 2 4 3 2 = 192 if interpolation is used based on the frozen cloud assumption -17- Ocean Systems Engineering Prof. Daejun CHANG An approach to the combined effects An approach to the combined effects Simulation and interpolation Leak position: Deck 1 (D) Leak direction: North (N) Wind Direction: North-South (NS) Frequency 0.071 0.0064 0.005 0.0036 0.0022 0.0008 0.0003 0.0001 Probability Leak Rate Wind Speed 0.75 1.5 3 6 12 24 48 96 0.03 1.5 DNNS11 DNNS12 DNNS13 DNNS14 DNNS15 DNNS16 DNNS17 DNNS18 0.09 4 DNNS21 DNNS22 DNNS23 DNNS24 DNNS25 DNNS26 DNNS27 DNNS28 0.05 6 DNNS31 DNNS32 DNNS33 DNNS34 DNNS35 DNNS36 DNNS37 DNNS38 0.02 8 DNNS41 DNNS42 DNNS43 DNNS44 DNNS45 DNNS46 DNNS47 DNNS48 0.01 12 DNNS51 DNNS52 DNNS53 DNNS54 DNNS55 DNNS56 DNNS57 DNNS58 -18- Ocean Systems Engineering Prof. Daejun CHANG An approach to the combined effects An approach to the combined effects Simulation and interpolation Leak position: Deck 1 (D) Leak direction: North (N) Wind Direction: North-South (NS) Frequency 0.071 0.0064 0.005 0.0036 0.0022 0.0008 0.0003 0.0001 Probability Leak Rate Wind Speed 0.75 1.5 3 6 12 24 48 96 0.03 1.5 DNNS11 DNNS12 DNNS13 DNNS14 DNNS15 DNNS16 DNNS17 DNNS18 0.09 4 DNNS21 DNNS22 DNNS23 DNNS24 DNNS25 DNNS26 DNNS27 DNNS28 0.05 6 DNNS31 DNNS32 DNNS33 DNNS34 DNNS35 DNNS36 DNNS37 DNNS38 0.02 8 DNNS41 DNNS42 DNNS43 DNNS44 DNNS45 DNNS46 DNNS47 DNNS48 0.01 12 DNNS51 DNNS52 DNNS53 DNNS54 DNNS55 DNNS56 DNNS57 DNNS58 S: Simulated cases S S S S S S S S -19- Ocean Systems Engineering Prof. Daejun CHANG Explosion Simulation Explosion Simulation -20- Ocean Systems Engineering Prof. Daejun CHANG Is the cloud fixed at the position for which the dispersion analysis is done? The cloud can be moved by the wind. It can also travel on its own momentum. If the leak position is changed, the cloud position will change. The cloud position is possible at any allowable place of the installation. Position of gas cloud Small cloud (low category): 5 - 9 positions Large cloud (high category): 2 - 3 positions Position of gas cloud Position of gas cloud -21- Ocean Systems Engineering Prof. Daejun CHANG Position of gas cloud Position of gas cloud Small cloud Large cloud -22- Ocean Systems Engineering Prof. Daejun CHANG It is known that the overpressure varies with the ignition position within the cloud. Ignition point within the gas cloud Small cloud (low category): center Large cloud (high category): 2 - 3 positions Ignition position within the cloud Ignition position within the cloud -23- Ocean Systems Engineering Prof. Daejun CHANG Continuous and Discrete Ignition Continuous and Discrete Ignition -24- Ocean Systems Engineering Prof. Daejun CHANG Ignition density Ignition density Without ignition, there is no explosion. Ignition density determines the explosion frequency. Time, s Explosion frequency Gas volume, m 3 Ignition density -25- Ocean Systems Engineering Prof. Daejun CHANG Naturally there are two types of ignition sources. Continuous: the ignition source is constantly active. Discrete: the activity of the ignition source is intermittent. The source can be either of the two (continuous or discrete) Both of the two (continuous and discrete at the same time) TDIIM Time-Dependent Internal Ignition Model Developed by a JIP program led by DNV Continuous and discrete ignition Continuous and discrete ignition -26- Ocean Systems Engineering Prof. Daejun CHANG TDIIM (Time TDIIM (Time--Dependent Internal Ignition Model) Dependent Internal Ignition Model) Continuous ignition Conditional probability that the gas cloud explodes if it touches the continuous ignition source. Discrete ignition Probability that the gas cloud explodes which contains the discrete ignition source -27- Ocean Systems Engineering Prof. Daejun CHANG Discrete ignition Discrete ignition Discrete ignition Probability that the gas cloud explodes which contains the discrete ignition source The ignition source is active and inactive intermittently. The explosion probability is proportional to the contact time between the cloud and the ignition source. The cloud ultimately explodes if left in contact with the source. f dis = Pr dis,total x V cloud /V deck x Dt f dis : discrete ignition frequency Pr dis,total : sum of all the discrete ignition sources V cloud : cloud volume V deck : deck or space volume Dt: residence time of the cloud -28- Ocean Systems Engineering Prof. Daejun CHANG Discrete ignition Discrete ignition Discrete ignition Probability that the gas cloud explodes which contains the discrete ignition source The ignition source is active and inactive intermittently. The explosion probability is proportional to the contact time between the cloud and the ignition source. The cloud ultimately explodes if left in contact with the source. f dis = Pr dis,total x V cloud /V deck x Dt f dis : discrete ignition frequency Pr dis,total : sum of all the discrete ignition sources V cloud : cloud volume V deck : deck or space volume Dt: residence time of the cloud -29- Ocean Systems Engineering Prof. Daejun CHANG Time, s Explosion frequency Gas volume, m 3 Ignition density Category 4 (3,100 m 3 ) t 1 t 2 t 3 t 4 Discrete ignition Discrete ignition Ignition probability = Pr dis,total x V cloud /V deck x Dt Pr dis,total : Sum of all the discrete ignition sources = (pump + compr + ) * 1600m2 V cloud : cloud volume = 3,100 m3 V deck : deck or space volume = 12,500 m3 Dt: residence time of the cloud = t 2 - t 1 + t 4 - t 3 -30- Ocean Systems Engineering Prof. Daejun CHANG Continuous ignition Continuous ignition Continuous ignition Conditional probability that the gas cloud explodes if it touches the continuous ignition source. As soon as the cloud touches the source, it will explode. Continuous ignition is possible only on the boundary. f con = Pr con,total x Q cloud /V deck f con : continuous ignition frequency Pr dis,total : sum of all the continuous ignition sources Q cloud : cloud volume growth rate (m3/s) V deck : deck or space volume -31- Ocean Systems Engineering Prof. Daejun CHANG Time, s Explosion frequency Gas volume, m 3 Ignition density Category 4 (3,100 m 3 ) t 1 t 2 t 3 t 4 Continuous ignition Continuous ignition f con = Pr con,total x Q cloud /V deck f con : continuous ignition frequency = (pump + compr + ) * 1600m2 Pr dis,total : sum of all the continuous ignition sources Q cloud : cloud volume growth rate (m3/s) [V(t 2 ) V(t 1 )]/(t 2 -t 1 ) V deck : deck or space volume Q cloud -32- Ocean Systems Engineering Prof. Daejun CHANG Frequency Combination Frequency Combination -33- Ocean Systems Engineering Prof. Daejun CHANG Frequency combination for Category 1 Frequency combination for Category 1 Position ID Probability Position ID Probability Type ID Probability C 0.5 D 3.41106E-06 2.13192E-07 C 1.48149E-05 2.06506E-06 D 3.41106E-06 C 0.5 D 3.41106E-06 2.13192E-07 C 1.48149E-05 2.06506E-06 D 3.41106E-06 C 0.5 D 3.41106E-06 2.13192E-07 C 1.48149E-05 2.06506E-06 D 3.41106E-06 C 0.5 D 3.41106E-06 2.13192E-07 C 1.48149E-05 2.06506E-06 D 3.41106E-06 C 0.5 D 3.41106E-06 2.13192E-07 C 1.48149E-05 2.06506E-06 D 3.41106E-06 C 0.5 D 3.41106E-06 2.13192E-07 C 1.48149E-05 2.06506E-06 D 3.41106E-06 C 0.5 D 3.41106E-06 2.13192E-07 C 1.48149E-05 2.06506E-06 D 3.41106E-06 C 0.5 D 3.41106E-06 2.13192E-07 C 1.48149E-05 2.06506E-06 D 3.41106E-06 C 0.5 D 0 0 C 0 0 D 0 Overpressure, barg Ignition Position (C:Center, E: Edge) Ignition Type (D:Discrete, C:Continuous) E Cloud Position Cloud Class No of Cloud Positions Integrated Probability E E E E E E E 8 E 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.5 0.13 0.5 0.13 1 2 3 4 5 6 7 0.13 0.13 0.13 0.13 0.13 0.13 1 8 9 0.13 -34- Ocean Systems Engineering Prof. Daejun CHANG Frequency combination for Category 5 Frequency combination for Category 5 -35- Ocean Systems Engineering Prof. Daejun CHANG Exceedance Curve Exceedance Curve -36- Ocean Systems Engineering Prof. Daejun CHANG Overpressure detection Overpressure detection The overpressure is function of time and position. Consequently, one exceedance curve is about one detection position. An average can be taken over several detection positions. -37- Ocean Systems Engineering Prof. Daejun CHANG Exceedance curve Exceedance curve Cumulative frequency The frequency of the higher over pressure is negligible compared to the lower frequency. 1.0E-10 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-05 1.0E-04 0 1 2 3 4 5 6 7 Drag load, bar C u m u l a t i v e
f r e q u e n c y ,
/ y r Level 11.8m Level 13.4m Level 19.4m Level 22.3m Level 28.6m -38- Ocean Systems Engineering Prof. Daejun CHANG Miscellaneous Miscellaneous -39- Ocean Systems Engineering Prof. Daejun CHANG Some points not explained Some points not explained Ventilation study To simulate air change rate of the installation The initial condition of the dispersion study is the results of the ventilation study. Cloud volume Only the volume with concentration higher than the LEL is effective for the explosion. Windrose data The probability distribution of the wind direction and speed. FLACS supports these tasks. -40- Ocean Systems Engineering Prof. Daejun CHANG Review Review -41- Ocean Systems Engineering Prof. Daejun CHANG Explosion The most catastrophic accident Inherent to ocean plants handling flammable gas within congested space Goal To design structure against the explosion with a given frequency (once in 10,000 years (10 -4 /yr) or once in 100,000 years (10 -5 /yr)) Task To estimate the explosion load with the threshold frequency If the explosion load exceeds the structure strength, change the design for - Structural strength - Safety systems configuration and reliability - Spatial congestion (or equipment arrangement) - . . . Explosion Risk Analysis An Example None of the design changes is easy to implement. None of the design changes is easy to implement. Precise detection in the early stage is the key. Precise detection in the early stage is the key. -42- Ocean Systems Engineering Prof. Daejun CHANG Continuous ignition Ignition sources are exist. Inherent to ocean plants handling flammable gas within congested space Goal To design structure against the explosion with a given frequency (once in 10,000 years (10 -4 /yr) or once in 100,000 years (10 -5 /yr)) Task To estimate the explosion load with the threshold frequency If the explosion load exceeds the structure strength, change the design for - Structural strength - Safety systems configuration and reliability - Spatial congestion (or equipment arrangement) - . . . Continuous Ignition and Discrete Ignition -43- Ocean Systems Engineering Prof. Daejun CHANG Leak Gas Cloud Explosion Factors affecting dispersion Factors affecting dispersion -- Leak rate & direction Leak rate & direction -- Wind speed & direction Wind speed & direction -- Spatial congestion Spatial congestion Factors affecting explosion Factors affecting explosion -- Cloud position within the facility Cloud position within the facility -- Ignition density Ignition density -- Ignition position within the cloud Ignition position within the cloud -- Spatial congestion Spatial congestion Affecting safety systems Affecting safety systems -- Gas detection system Gas detection system -- Emergency shutdown system (ESD) Emergency shutdown system (ESD) -- Power shutoff system isolating ignition sources Power shutoff system isolating ignition sources How many conceivable cases? Explosion Risk Analysis - Mechanism Dispersion Explosion -44- Ocean Systems Engineering Prof. Daejun CHANG 1. Consequence Analysis - 3D geometry model construction - CFD simulation for ventilation, dispersion, and explosion 2. Explosion Frequency Estimation 3. Risk Presentation: Explosion overpressure vs. Probability 4. ALARP Demonstration -45- Ocean Systems Engineering Prof. Daejun CHANG Main deck Mezzanine deck 1. Consequence Analysis 3D Model Open volume = 87.6 % Open volume = 89.5 % -46- Ocean Systems Engineering Prof. Daejun CHANG Ventilation One wind velocity (4m/s) and 12 directions Dispersion Wind direction (3) Wind Speed (5) Leak Rate (8) Leak Position (4) Leak Direction (2) = 960 Scenarios 112 scenarios are simulated and the rest are interpolated. Explosion Cloud Size (7) Cloud Position (3~9) Ignition Point (2~4) = 128 scenarios are simulated 1. Consequence Analysis CFD Simulation -47- Ocean Systems Engineering Prof. Daejun CHANG OFON Wind rose 0 5 10 15 20 25 0 30 60 90 120 150 180 210 240 270 300 330 1.5 4 6 8 9 10 14 Ventilation Simulation -48- Ocean Systems Engineering Prof. Daejun CHANG Ventilation Study Results The volume fraction of air change rate greater than 12 per hour is 99 %. Well ventilated 0 10 20 30 40 50 60 70 80 90 100 0 100 200 300 400 500 Air changes per hour C u m u l a t i v e
f r e q u e n c y ,
/ y r 0 78 155 0 30 60 90 120 150 180 210 240 270 300 330 -49- Ocean Systems Engineering Prof. Daejun CHANG Leak Points for Dispersion Simulation Seg11 (TEG contactor inlet cooler) Seg7 (MP compressor suction cooler) Mezzanine deck Seg9 (HP compressor suction scrubber) Seg1 (HP fuel gas scrubber) Main deck -50- Ocean Systems Engineering Prof. Daejun CHANG Example: Dispersion from a Leak 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 0 100 200 300 400 500 600 700 800 900 1000 Time, sec. E q u i v a l e n t
S t o i c h i o m e t r i c
C l o u d ,
m 3 0 5 10 15 20 25 30 L e a k
R a t e ,
k g / s Cloud volume Leak rate ESD Blowdown Leak at Segment 1 in Main Deck at 24 kg/s Wind from the south -51- Ocean Systems Engineering Prof. Daejun CHANG Explosion scenario with cloud category 1 (800m 3 ) Explosion scenario with cloud category 7 (11,350m 3 ) Gas Cloud and Ignition Position for Explosion Simulation -52- Ocean Systems Engineering Prof. Daejun CHANG -2.0 -1.0 0.0 1.0 2.0 3.0 4.0 5.0 0.0 0.2 0.4 0.6 0.8 1.0 1.2 Time, sec. O v e r p r e s s u r e ,
b a r g Main deck floor Mezzanine deck floor Blast wall Example: Explosion of a Cloud -53- Ocean Systems Engineering Prof. Daejun CHANG Discrete Gas Age Maintenance Manning Technology Module Adjust Total Pump 2.10E-07 0.90 0.85 1.00 0.60 25 0.46 2.41E-06 Electrical eq. * 2.70E-08 0.90 0.90 1.00 0.60 5089.5 0.49 6.68E-05 Other equipment * 2.10E-09 0.90 0.90 1.00 0.60 5089.5 0.49 5.19E-06 Other ** 1.70E-08 1.00 1.00 1.00 1.00 2544.8 1.00 4.33E-05 Personnel * 4.00E-08 1.00 0.95 0.60 1.00 5089.5 0.57 1.16E-04 * per m2 exposed to gas SUM 2.34E-04 ** per m2 exposed to gas - Only one deck level Continuous Gas Age Maintenance Manning Technology Module Adjust Total Hot work (# hours per 365*24h) 0.00E+00 - - - - - - 0.00E+00 Pump 9.60E-05 0.90 0.85 1.00 0.60 25 0.46 1.10E-03 Electrical equipment * 2.60E-06 0.90 0.90 1.00 0.60 5089.5 0.49 6.43E-03 Other equipment * 2.60E-06 0.90 0.90 1.00 0.60 5089.5 0.49 6.43E-03 Other ** 1.30E-06 1.00 1.00 1.00 1.00 2544.8 1.00 3.31E-03 Personnel * 3.00E-06 1.00 0.95 0.60 1.00 5089.5 0.57 8.70E-03 * per m2 exposed to gas SUM 2.60E-02 ** per m2 exposed to gas - Only one deck level 2. Explosion Frequency Estimation Ignition intensities Function of state and number of ignition sources -54- Ocean Systems Engineering Prof. Daejun CHANG 1.0E-10 1.0E-09 1.0E-08 1.0E-07 1.0E-06 1.0E-05 1.0E-04 0 1 2 3 4 5 6 7 Drag load, bar C u m u l a t i v e
f r e q u e n c y ,
/ y r Level 11.8m Level 13.4m Level 19.4m Level 22.3m Level 28.6m 3. Risk Presentation -55- Ocean Systems Engineering Prof. Daejun CHANG Conclusions Conclusions -56- Ocean Systems Engineering Prof. Daejun CHANG Conclusions A lot of assumptions and interpolations Still persuasive Rooms for improvements Difficult to verify Only the assumption are observable. But, the detailed process is hidden. Quality control is important. Compared to fire risk analysis Explosion risk analysis is more systematic and quantitative Need to apply a similar approach to fire risk analysis