Grain Processing Corporation Plant II - Executive Summary

| Accident History | Chemicals | Emergency Response | Registration | Source | Executive Summary |

EXECUTIVE SUMMARY 
A. Grain Processing Corporation (GPC II) uses every means at it's disposal to operate a safe workplace.  This includes provisions of CFR 29 1910.119 (OSHA PSM) and 40 CFR part 68 E.P.A. RMP) which are systemic approaches for chemical process hazards management designed to eliminate or minimize the potential for major catastrophes caused by the release of highly hazardous chemicals.  GPC will employ these measures to ensure that the means for preventing a catastrophic release, fire, or explosion are understood and to ensure the necessary preventative measures and lines of defense are installed and maintained.  Every person on site is included in these measures and management will ensure they are properly established and utilized. 
B. GPC is located in Daviess County, Washington Indiana.  The facility is built on an area that covers 800 acres.  GPC is a corn wet miller.  GPC produces beverage and denatured alcohol, animal feed, starchs, maltrodextrins and corn syrup soli 
ds.  GPC facilities include storage tanks and processes typical of a corn wet milling plant.   
 
[The primary plant operations of concern include:] 
7 Railcar and tank truck unloading areas. 
7 Ethylene oxide storage and transfer systems. 
7 Sulfur Dioxide (Anhydrous) unloading, storage and transfer system. 
 
Use of regulated substances. 
Ethylene oxide is used in the process to produce various starches. 
   Sulfur Dioxide is used in the corn steeping process. 
       Quantities handled or stored. 
   Ethylene Oxide 
       Minimum inventory        -0- 
       Maximum inventory        231,000 lbs. 
   SO2  
       Minimum inventory        16,850 lbs. 
       Maximum inventory        198,130 lbs. 
 
C. The general accidental release prevention program and chemical-specific prevention steps at GPC includes but are not limited to: 
7 All elements of CFR 29 1910.119 (OSHA PSM) 
7 NFPA (National Fire Protection Association) 
7 E.P.A., A.P.I. (American Petroleum Institute) 
7 ASTM (American Society for Testing of Material) 
7 NBIC (National Board of Boiler and Press 
ure Vessel Inspections. 
7 ASME (American Society of Mechanical Engineers) 
 
    
Offsite consequence analyses have been done for two chemicals.  These chemicals are ethylene oxide (CAS No. 75-21-8), sulfur dioxide  
   (CAS No. 7446-09-5).  Two scenarios were considered for each substance.      These two scenarios are the "worst case release" and the "alternative scenario".      EPA defines the worst case release to be that..."the maximum quantity in the     largest vessel.... is released as a gas over 10 minutes," due to an unspecified     failure.  The alternative scenario is defined as more likely to occur than the worst     case release scenario.  The alternative scenarios were developed by reviewing     the Process Hazard Analyses for these substances (which were done as part of     the OSHA Process Safety Management (PSM) planning) as well as talking to     area supervisors and operators. 
 
EPA requires that "one worst case release scenario be submitted to represent all regulated toxic substances held above th 
e threshold quantity and one worst case release scenario to represent all flammable substances held above the threshold quantity."  The owner/operator must also submit "information on one alternative release scenario for each regulated toxic substance held above the threshold quantity and one alternative release scenario to represent all flammable substances held above the threshold quantity." 
 
Accordingly, one worst case scenario is being submitted for sulfur dioxide.  There are no flammables to consider.  In addition, alternative release scenarios are being submitted for sulfur dioxide and ethylene oxide.  Explanations of all release scenarios (worst case and alternative case) for both of the regulated substances may be found in the Worst Case Scenario and Alternative Case Scenario for both Ethylene Oxide and Sulfure Dioxide of this document. 
 
   The worst case release scenario for sulfur dioxide involves the rupture of the     storage tank and subsequent release of its entire contents as 
a gas over a  
   10 minute period as required by EPA.  There would be a substantial aerosol     fraction which would increase the bulk density of the vapor cloud. 
 
Atmospheric dispersion modeling was done to determine the distance traveled by the sulfur dioxide vapor until the concentration decreases to the "toxic endpoint" selected by EPA of 7.8 mg/m3 which is also the Emergency Response Planning Guideline Level-2 (EPRG-2).  This is defined by the American Industrial Hygiene Association (AGHA) as the "maximum airborne concentration below which it is believed that nearly all individuals could be exposed for up to one hour without experiencing or developing irreversible or other serious health effects or symptoms which could impair an individual's ability to take protective action."  A circle with a radius equal to the distance to the toxic endpoint must be drawn and the potentially exposed population within that circle must be determined. 
 
   EPA mandated the use of extremely stable meteorol 
ogical conditions for     dispersion modeling analyses.  These EPA default values were used:  Stability     Class F, 1.5 m/s wind speed, 77oF ambient temperature and 50% relative     humidity.  The latest version of the EPA approved guideline model Degadis was     used.  This is appropriate for a dense gas such as sulfur dioxide.  The model     results show that the distance to the toxic endpoint is 4.75 miles and  the     potentially affected population is 14,300. 
 
   The alternative release scenario involves a break in the one-inch transfer line     between the sulfur dioxide tanks and the sulfonator.  The use of active and     passive mitigation techniques allowed by EPA would limit this release to two     minutes (at an absolute maximum) and the release would total 29.52 pounds.      rate would be 14.7 lb/min.  The rate would be 0.418 lb/sec for 15 seconds     followed by vaporization of the SO2 in the broken line at 0.194 lb/sec for 1     minute 45 seconds.  These are conservative estimates which maximize the     potenti 
al loss of SO2.  There are difficulties evaluating this scenario using either     the EPA look-up tables or dispersion models due to the short release duration of     this scenario.  The look-up tables give a distance to the endpoint of 0.08 miles     while dispersion modeling gives a distance of 0.40 miles.  However, this is an     inappropriate use to the Degadis model.  It is not appropriate to compare a two     minute release to a one hour averaging time.  The 7.8 mg/m3 ERPG value is     based upon a one-hour averaging time.  The endpoint concentration can be     modified using available exposure/concentration relationships to match the two     minute release duration.  This yields a value of 42.7 mg/m3.  Using this as an     endpoint gives a distance 0.19 miles.  The plume never crosses the plant     fenceline so there are no offsite consequences. 
 
The alternative release scenario for ethylene oxide involves a break in the one inch line to the transfer line while ethylene oxide is being transferred.  The use o 
f active and passive mitigation techniques allowed by EPA would limit this release to five minutes (at an absolute maximum) and the release rate would be 718 lb/min.  The same averaging time difficulties occur with this scenario as with the alternative release scenario for sulfur dioxide.  The look-up tables give a distance to the endpoint of 0.77 miles while dispersion modeling gives a distance of 0.92 miles.  However, this is an inappropriate use of the Degadis model.  The 90 mg/m3 EPRG value is based upon a one hour averaging time.  The endpoint concentration can be modified using available exposure/response relationships to match the five minute release duration.  This yields a value of 312 mg/m3.  Using this endpoint and a five minute averaging time gives a distance to the endpoint of 0.14 miles.  The plume never crosses the plant fenceline so there would be no offsite consequences.   
 
In addition policies and procedures are set forth in GPC's corporate safety and PSM manuals out 
lining both normal and emergency procedures. 
 
GPC has conducted as part of the Process Safety Management rule, a process hazard analysis (PHA) of PSM covered areas, critical processes and mitigation systems to ensure safety systems are maintained for regulated chemicals.  Based on the findings and recommended actions of the P.H.A.'s changes to improve safety will be implemented as deemed necessary. 
 
D. The facilities emergency response program is coordinated with the city of Washington's Local Emergency Planning Committee (LEPC). 
Safety committees consisting of salaried and hourly operating personnel will evaluate and implement, as deemed prudent, safety recommendations.  In addition, plant-wide safety meeting for all GPC personnel will be held.  The Production Superintendent and the Plant Safety Director will meet on a regular basis to insure recommendations are being addressed in a timely and practical manner. 
 
Ethylene Oxide 
Worst Case Scenario 
 
   Ethylene Oxide has a boiling point  
of 51 oF.  It is stored as a subcooled liquid at about 45 oF.  The worst case scenario would be a tank rupture creating an instantaneous spill of the entire tank contents.  The tank capacity is 40,000 gallons.  Administrative controls are present which stipulate that the tank contents will not exceed 85% of capacity or 34,000 gallons (258,400 lbs).  It is assumed that since ethylene oxide is stored as a subcooled liquid it would be released as a liquid and would not flash immediately to a vapor.  This is born out by evaporation equations. 
 
EPA Technique: 
 
The evaporation rate is determined from the following equation. 
 
   QR = 1.4 x LFB x A 
 
   Where:    QR = evaporation rate (lb/min) 
           LFA = Liquid Factor Boiling = .12 (Exhibit B-2, Appendix B,  
               p. 86 EPA Guidance) 
           A = Diked area = 1200 ft2 
 
   QR = 1.4 x 0.12 x 1200 = 201.6 lb/min 
 
   258,400 lb/201.6 lb/min. = 1282 minutes for total evaporation 
 
   Use the lookup table (Reference Table 6, p. 29 EPA guidance) to find the distance to the t 
oxic endpoint of 3.3 miles. 
 
EPA Technique with Modeling: 
 
   The scenario can be modeled as a continuous ground level gaseous release using the EPA derived release rate of 201.6 lb/minute.  This should be modeled as a non-isothermal release to account for heat transfer between the gas cloud and the surrounding air and ground.  This gives a distance to the toxic endpoint of 0.79 miles.   
 
SB Equation Modeling Approach: 
 
   The SB Equation allows for the calculation of the evaporation rate of boiling liquids on land.  This value can be inserted into the model in lieu of the EPA derived value.  The SB Equation is designed for liquid pools that are relatively shallow which would not be the case if the EO tank ruptured and filled the dike.  The calculation will be done for comparison purposes. 
 
 
The SB Equation is as follows: 
   m = 4Xks(Ta-Tb) x (2 x 3.14159 x g x Vo)0.5 x t.50/(dHv x 3.14159 x as)0.5 
 
   Where:    m = mass flux rate (evaporation rate), kg/sec 
           X = surface roughness correction fa 
ctor = 1 for concrete 
           ks = thermal conductivity of surface = 0.92 for concrete 
           Ta = Ambient Temperature = 25 oC = 298 K 
           Tb = Liquid boiling point = 283.77 K 
           g = 9.8 m/s 
           Vo = Volume of Liquid Released = 128.7 m3 
           t = time after spill (sec) 
           dHv = latent heat of vaporization = 569,000 j/kg 
           as = thermal diffusivity of surface = 4.16 x 10-7 for concrete 
 
   m = 4 x 1 x 0.92 x (298 - 284) x (2 x 3.14149 x 9.8 x 128.7)0.5 x t0.5 
           (569,000) x (3.14159 x 4.16 x 10-7)0.5 
 
       = 3.68 x (14) X (7924.7)0.5 X t.5 
       569,000 x (1.31 x 10-6)0.5 
 
       = 4586.4 x t.5 
       650.5 
 
       = 7.05 x t.5 kg/sec 
 
   Solving this equation on a spreadsheet indicates that the entire tank contents would be evaporated in about 840 seconds (14 minutes).  This is two orders of magnitude less than the EPA's equation predicts due to the fact that the SB Equation includes no term for the surface area of the pool.  Degadis predicts that an unrestricted pool would cover 138,536 ft2 (to a depth of 0.1 cm) a 
nd take 2552 seconds (42.5 minutes) to evaporate.  This is comparable to the SB Equation results.  Restricting pool formation with a dike reduces the surface area of the liquid exposed to the air and greatly reduces the evaporation rate.  The SB Method grossly overestimates the evaporation rate because it fails to account for surface area and is thus rejected. 
 
Pure Modeling Approach: 
 
   The Degadis model will calculate the evaporation rate internally and apply it to calculate the distance to the toxic endpoint.  Unfortunately, it can only do this for a continuous release of liquid which is not the case.  The model can be run and the evaporation rate determined, then input this value into the model again as a continuous gas release (because the release is much longer than the averaging time used).  The evaporation rate determined in this manner is 69.6 lb/min which would take 3713 minutes (61.9 hours) for the EO to evaporate completely.  This should be modeled as a continuous, ground le 
vel, isothermal release.  Degadis indicates that the toxic endpoint (90 mg/m3) extends 0.47 miles for this case. 
 
Summary: 
 
   The last method is the most realistic approach although it probably overstates the extent of the toxic endpoint.  However, this conservative approach is acceptable for the worst case scenario.  The distance to the toxic endpoint of 0.47 miles should be used. 
Alternative Case Scenario 
 
   Ethylene Oxide has a boiling point of 51 oF.  It is stored as a subcooled liquid at about 45 oF.  The worst case scenario would be a tank rupture creating an instantaneous spill of the entire tank contents.  The tank capacity is 40,000 gallons.  Administrative controls are present which stipulate that the tank contents will not exceed 85% of capacity or 34,000 gallons (258,400 lbs).  It is assumed that since ethylene oxide is stored as a subcooled liquid it would be released as a liquid and would not flash immediately to a vapor.  This is born out by evaporation equations. 
   The mo 
st likely release scenario would occur from a break in the 1 inch line from the transfer pump..  It is likely that this release would occur inside the building which would reduce the release rate of EO.  A deluge system is also in place which would further mitigate the release.  The presence of these active and passive mitigation devices make calculation of the release rate to the atmosphere difficult.  It is possible that the break could occur within the dike or outside the dike and outside the building.  The "worst" alternative case would be the latter which would form a liquid pool on the ground which would evaporate more quickly than if it occurred in the dike (due to pool spreading.) 
   The first step is to determine the quantity of ethylene oxide which would be lost in this scenario.  If the pipe ruptured during transfer the release rate can be determined by using the equation of Crowl and Louvar. 
 
   Q = CD x pl x A x (2gc(Ps-Pa)/pl + 2gh)0.5 
 
   Where:    Q = Discharge rate, lb/sec. 
            
CD = Discharge coefficient = 0.61 
           pl = density of liquid = 7.6 lb/gal = 56.85 lb/ft3 
           A = Area of discharge 1" dia. = 0.00545 ft2 
           gc = Acceleration of gravity = 32.2 ft/sec2 
           Ps = Storage pressure = 35 psig = 49.7 psia = 7157 psf 
           Pa = Ambient Pressure = 14.7 psia = 2117 psf 
           h = Liquid head in tank = 10 ft. 
 
   Q = 0.61 x 56.85 x .00545 x (2 x 32.2(7157-2117)/56.85 + 2 x 32.2 x 10)0.05 
 
       = 0.189(79.7) 
 
       = 15.1 lb/sec 
 
   258,400 lb/15.1 lb/sec = 17,153 seconds = 286 minutes (4.8 hours) to empty tank 
 
   A release of this type would be detected by the sensors monitored by the gas chromatograph at South Starch.  Detection would be nearly immediate and the valve could be closed in less than 5 minutes.  Therefore, the total release would be: 
   15.1 lb/sec x 5 min. x 60 sec/min = 4530 lbs 
The evaporation rate using EPA techniques is: 
 
   QR = QS x 2.4 x LFA x DF 
 
   Where:    QR = Quantity released to air (evaporation rate), lb/min 
           QS = Quantity released = 4530 lbs 
           LFB = Liqu 
id Factor Boiling = .12 
           DF = Density Factor = .55 
 
   QR = 4530 x 2.4 x .12 x .55 = 718 lb/min 
 
EPA technique: 
 
   Using the look-up table (Reference Table 15, EPA Guidance p. 60) the distance to the toxic endpoint is 2.4 miles.   
 
EPA Technique with Modeling: 
 
   The scenario can be modeled as a continuous ground level gaseous release using the EPA derived release rate of 718 lb/minute.  This should be modeled as a non-isothermal release to account for heat transfer between the gas cloud and the surrounding air and ground.  This gives a distance to the toxic endpoint of 0.92 miles.   
 
SB Equation Modeling Approach: 
 
   The SB Equation allows for the calculation of the evaporation rate of boiling liquids on land.  This value can be inserted into the model in lieu of the EPA derived value.  The SB Equation is designed for liquid pools that are relatively shallow which is appropriate for this scenario. 
 
The SB Equation is as follows: 
 
   m = 4Xks(Ta-Tb) x (2 x 3.14159 x g x Vo)0.5 x t.50/(dHv x 
3.14159 x as)0.5 
 
   Where:    m = mass flux rate (evaporation rate), kg/sec 
           X = surface roughness correction factor = 1 for concrete 
           ks = thermal conductivity of surface = 0.92 for concrete 
           Ta = Ambient Temperature = 25 oC = 298 K 
           Tb = Liquid boiling point = 283.77 K 
           g = 9.8 m/s 
           Vo = Volume of Liquid Released = 79.7 ft3 = 2.26 m3 
           t = time after spill (sec) 
           dHv = latent heat of vaporization = 569,000 j/kg 
           as = thermal diffusivity of surface = 4.16 x 10-7 for concrete 
 
   m = 4 x 1 x 0.92 x (298 - 284) x (2 x 3.14149 x 9.8 x 2.26)0.5 x t0.5 
           (569,000) x (3.14159 x 4.16 x 10-7)0.5 
 
       = 3.68 x (14) X (7924.7)0.5 X t.5 
       569,000 x (1.31 x 10-6)0.5 
 
       = 607.8 x t.5 
       650.5 
 
       = .93 x t.5 kg/sec 
 
      = 123.3 x t.5 lb/min 
 
   Solving this equation on a spreadsheet indicates that the entire tank contents would be evaporated in about 200 seconds (3.3 minutes).  This rate is higher than the EPA's equation predicts (average of 1318 lb/min vs. 718 lb/min) but both met 
hods yield evaporation rates which indicate near instantaneous evaporation.  Therefore, a gaseous release will be modeled of 5 minute duration. 
   Degadis has difficulty doing this analysis because the release resembles more of a puff than a continuous release..  It simply tracks the isopleth representing 90 mg/m3, the toxic endpoint.  This dissipates after about 12 minutes but the 90 mg/m3 value is based on a 1-hour averaging time.  However, the instantaneous concentration could exceed a much larger value for a short time and, even if it went to zero for the remainder of the hour, the 1-hour average would still exceed the toxic endpoint.  For example, if the plume reached a concentration of 900 mg/m3 (using a 6 minute average) after six minutes and then went to zero for 54 minutes, the 1 hour concentration would be 90 mg/m3.  EPA suggests looking at this situation with shorter averaging times when the anomalous results are obtained. 
   The highest results are usually obtained for an avera 
ging time that is close to the duration of the release.  A 5 minute averaging time would be used for this case.  There is a commonly used method for converting pollutant concentrations between different averaging times.  The formula is: 
 
   C12t1 = C22t2 
 
   Where:   C1 = concentration for averaging time 1 = 90 mg/m3 
 
           t1 = averaging time 1 = 1 hour = 60 minutes 
 
           C2 = concentration for averaging time 2 
 
           t2 = averaging time 1 = 5 minutes 
 
   Solving for C2 you get 312 mg/m3 
Conducting the modeling analysis using these parameters gives a distance to the toxic endpoint of 0.26 miles.   
 
Summary:   
 
   Modeling the alternative case using the results obtained from the SB Equation is the most realistic approach.  This yields a distance to the toxic endpoint of 0.26 miles. 
 
Sulfur Dioxide 
 
Worst Case Scenario 
 
   Sulfur Dioxide has a boiling point of 14 oF.  It is stored as a liquid at about 50 oF and 75 psig in two tanks.  The worst case scenario would be a tank rupture or broken lin 
e creating an instantaneous release of the entire tank contents.  The tanks' capacities total 21,195 gallons.  Administrative controls are present which stipulate that the tank contents will not exceed 82% of capacity or 17,380 gallons (198,130 lbs). 
   EPA allows consideration of passive mitigation techniques in worst case scenario analyses.  Passive mitigation techniques are present in the form of a dike and excess flow valves.  A building around the tanks would also be considered a passive mitigation technique and would be installed if it proved beneficial.  The dike will provide no benefit for modeling purposes because EPA techniques assume that gases stored as liquids by refrigeration and pressure will be released as gases and will not form a pool.  This is a conservative assumption and may not be true for SO2.  The excess flow valve serves to isolate the two tanks and permits considering the release of the contents of only the larger of the two tanks for the worst case scenario.  T 
his volume is 9,332 gallons or 106,380 pounds. 
   The toxic endpoint for sulfur dioxide is 7.8 mg/m3 or about 3 ppm. 
 
EPA technique: 
 
   EPA requires that gases stored as liquids above their boiling points but under pressure be evaluated as gases.  This makes the EPA technique straightforward.  It is assumed that the SO2 will be released over 10 minutes at a rate of 10,638 lb/min. 
   Use the lookup table (Reference Table 6, p. 29 EPA guidance) to find the distance to the toxic endpoint is much greater than 25 miles. 
 
EPA Technique with Modeling: 
 
   The scenario can be modeled as a ground level gaseous release using the EPA derived release rate of 10,638 lb/minute.  This type of release would include liquid SO2 aerosol as well as SO2 vapor.  Therefore the bulk density of the release must be calculated to properly assess the dispersion.  The modeling software calculates a flash fraction of 12.5% which gives a bulk density of 23.4 kg/m3 for the gas cloud .   
   The modeling gives a distance to the 
toxic endpoint of 4.75 miles which is reached in 8.3 hours.  No consideration is given to the fact that meteorological conditions will certainly change over this long period which exaggerates the distance to the endpoint.  A review of the met data for Evansville, Indiana (the closest source of NOAA met data) shows that extremely stable conditions such as those modeled have not occurred for more than 5 consecutive hours in the past five years.  This is still a great improvement over the 25+ mile distance predicted by the lookup table.   
 
Other Passive Reduction Techniques: 
 
   If a building were built around the SO2 storage tank credit could be taken for a reduction in emission rate.  EPA Guidance (p. 13) states that the release rate would be reduced by 55% but would continue for 1.8 times as long so the total amount of contaminant would still be released.  The release rate would then be: 
 
   10,638 x 0.55 = 5850.9 lb/min 
 
   This would persist for 18 minutes and would be comprised entirely  
of SO2 vapor (greatly reducing the density of the gas cloud). The toxic endpoint for this scenario is 8.99 miles.  Thus the impact is more severe even though the release is spread out over a greater time.  This is due to the fact that the SO2 aerosol would likely flash to vapor and greatly reduce the gas cloud density.  Furthermore, it is likely that a catastrophic release of SO2 would occur with enough force to destroy the building and negate any effect it may have.  A building is not planned around the SO2 tanks nor is one recommended. 
 
Other Considerations: 
 
   This specific situation with sulfur dioxide is not easily characterized.  EPA requires that a release of a gas stored as a liquid under both refrigeration and pressure be dealt with as a gaseous release.  That is the scenario described above.  However, an instantaneous release of the entire tank contents would consist of SO2 vapor, liquid aerosol and a substantial pool of liquid SO2 which would subsequently boil away.  The pool 
would fill the dike to a depth of 1-2 feet which would hinder evaporation due to the limited surface area.  Some of this liquid would fill a sump within the dike.  No significant difference in modeled results was found when the evaporation time was extended to 1 hour (a more reasonable assumption) although the time to reach the toxic endpoint was extended to 12 hours. 
   The evaporation of the SO2 will initially form a very cold vapor cloud.  It is likely that ambient moisture will be condensed in the cloud.  Some water vapor will actually freeze.  This will have several effects.  It will increase the bulk density of the cloud.  This will cause increased dispersion in the x direction due to slumping of the cloud.  Only a slight decrease in distance to the endpoint was found when increased plume density was modeled.  The temperature difference between the cloud and the ambient air will tend to cause increased dispersion at the margins of the plume.  Modeling is unable to account for this 
effect.  It should be minor because the effect should not penetrate to the center of the plume and it would exert its most powerful effect in the early stages of the release when the temperature difference between the plume and the ambient air is the greatest.  Modeling indicates that this occurs within the first 5 seconds of the release.  Subsequent rapid expansion of the plume brings the temperature close to ambient. 
   This rapid cooling and vigorous boiling of the pool of SO2 would cause a localized pocket of turbulence which would enhance the initial dispersion of the plume.  Nearby buildings also make the micro environment resemble urban land use as opposed to rural.  Large numbers of heat generating sources enhance this effect.  Previous modeling done in Muscatine indicates that an intermediate roughness length is appropriate to use near a facility such as this.  A slightly more urban treatment (roughness length of 0.25 meters vs 0.1 m for rural and 1.00 m for urban) was chosen f 
or this application.  The combination of buildings, crops, heat sources and experience justifies a roughness length of 0.25 meters.   
   Finally, visualization of the plume indicates that it forms a "bubble" relatively rapidly that moves outward from the source.  This bubble moves rapidly and, in fact, never exceeds 1.2 miles in width.  The plume could never encompass the entire city of Washington.  EPA defines the potentially affected population as all residents within a circle whose radius is equal to the distance to the toxic endpoint.  While all these people are potentially affected, in reality, only those downwind of the source following a release would actually be affected.  This drastically reduces the actual number of people who could be affected. 
 
Summary: 
 
   The modeling approach is the most realistic method to use.  It is appropriate to use a roughness length of 0.25 meters for this case.  A ten minute gaseous release was chosen in accordance with EPA guidance.  The effects of  
delayed release due to evaporation of liquid SO2 were not considered.  No credit was taken for increased plume density due to condensation of ambient moisture.  The worst case impact would encompass a circle around the sulfur dioxide storage tank with a radius of 4.75 miles.  It is important to understand and communicate to those who review the consequence analysis that the total number of people EPA defines as potentially affected is always greater than the actual number of persons who would actually be affected.  This difference can be very large.  This is particularly important when there is one large population center within the area being analyzed.  The only way a large number of people could be affected would be for the "worst case" release to occur (which is very unlikely) at the right time of day (usually the early pre-dawn hours) when the winds are extremely light and out of the right direction. 
   Furthermore, the proper meteorological conditions must persist, unchanged for at  
least six hours in order for the plume to reach the outskirts of the city of Washington.  There were only six occasions from 1987 - 1991 that the stability/wind speed combination required to produce the worst case conditions occurred for at least six hours.  The wind direction during these six events would not have allowed the plume to affect the city of Washington. 
 
 
14,300 
 
Alternative Case Scenario 
 
   Sulfur Dioxide has a boiling point of 14 oF.  It is stored as a liquid at about 50 oF and 75 psig in two tanks.  The worst case scenario would be a tank rupture or broken line creating an instantaneous release of the entire tank contents.  The tanks' capacities total 21,195 gallons.  Administrative controls are present which stipulate that the tank contents will not exceed 82% of capacity or 17,380 gallons (198,130 lbs).  The most likely alternative scenario would involve a ruptured line between the tank and the sulfonator housed in a building just north of the SO2 tanks. 
   It is assume 
d that a ruptured line would release SO2 for 15 seconds until the excess flow valve shut off the flow.  This is very conservative because the valve should close nearly instantaneously (less than 3 seconds).  The excess flow valve closes at 2.2 gpm so this maximum flow will be used.  The entire contents of the line from the tank to the sulfonator would also be released.  This is approximately 50 feet of 1" pipe.  The loss would therefore be: 
 
   2.2 gal/min x 0.25 min. x 11.4 lb/gal = 6.27 lb SO2 
 
   6.27 lb SO2 / 15 sec = 0.418 lb/sec (0.1896 kg/sec) 
 
   Volume of 50 feet of 1 inch pipe = ((1"/12in/ft)/2)2 x pi x 50 = 0.2727 ft3 
 
   SO2 released from pipe = 0.2727 ft3 x 11.4 lb/gal x 7.48 gal/ft3 = 23.25 lb SO2 
 
   Assume this is released over 2 minutes or 23.25 lb/120 sec = 0.1938 lb/sec     (0.0879 kg/sec) 
 
   Thus the release would be 0.2775 kg/sec for 15 seconds followed by 0.0879 kg/sec for 1 minute 45 seconds.  
 
EPA Technique: 
 
   The EPA technique requires a constant emission rate.  If the tota 
l SO2 release were averaged over 5 minutes the lookup tables can be used.  The lookup table (Reference Table 14, p. 59, EPA Guidance) gives a distance to the toxic endpoint of 0.08 miles (422 feet).  This method discounts the effect of the relatively higher emissions during the first 15 seconds of the release.  It probably underestimates the circle of influence. 
Modeling Approach: 
 
   This can be modeled as a transient release (since the release varies with time).  The bulk density of the cloud would be 23.4 kg/m3 due to the presence of aerosols.  The release radius Is calculated as above in the worst case scenario.  It varies as the release rate falls. 
   Model results indicate that the toxic endpoint would extend 0.40 miles (2112 ft.) from the source of the break.  This result depends on the fact that the toxic endpoint is based on a 1 hour averaging time.  In fact, the model predicts the peak to occur after only 370 seconds which is far less than the averaging time for the standard of  
3600 seconds.  The use of shorter averaging times and higher concentrations indicates that a one hour average concentration of 7.8 mg/m3 is unlikely to occur beyond about 0.1 miles.  Modeling guidance and EPA suggest that for very short duration releases (such as this) we should look at an averaging time closer to the duration of the release.   
   Pollutants can be measured (and modeled) over any chosen averaging time.  The choice of an averaging time should be reflective of modeling conventions, EPA guidance and the effects of the pollutant of concern.  Every standard is set with an averaging time in mind.  In the case of the toxic endpoint the averaging time is presumed to be 1 hour.  That is, a citizen would suffer the effects of the pollutant if exposed to a concentration exceeding the toxic endpoint for an hour.  A method exists to convert standard concentrations from one time basis to other time bases.  The relation is: 
 
   C12*t1 = C22*t2 
 
   Where C1 = concentration for time basis 1  
= 7.8 mg/m3 
        t1   = time basis 1 = 1 hour = 60 minutes 
        C2 = concentration for time basis 2 
        t2   = time basis 2 = 2 minutes 
 
   C22 = C12* t1/t2 = 7.8 * 7.8 * (60/2) 
 
       = 1825.2 
 
       C2 = 43 mg/m3 
 
   Modeling can now be done with a two minute averaging time which corresponds to the release duration.  Choosing a 2 minute averaging time for the toxic endpoint moves the toxic endpoint to 0.19 miles (1003 feet).   
 
 
 
 
 
Passive Reduction Techniques: 
 
   If a building were built around the SO2 storage tank credit could be taken for a reduction in emission rate.  EPA Guidance (p. 13) states that the release rate would be reduced by 90% but would continue for 10 times as long so the same total amount of contaminant would still be released.  The release rate would then be 0.02248 kg/sec for 2.5 minutes then 0.00352 kg/sec for another 47.5 minutes.  There is a building around the sulfonator so if a break occurred in this building this would be the appropriate technique to use. 
   Modeling shows t 
hat the distance to the toxic endpoint is 0.15 miles (792 feet) using this scenario.  This information is useful to show that a broken line in the sulfonator building is of less concern than a release outside the building.  However, since the line outside the building is exposed to more hazards and there is more pipe outside the building than inside this passive reduction technique will not be used for consequence analysis. 
 
Summary: 
 
   The simplified EPA technique gives better results because it is unable to account for transient emissions.  The modeling approach is the appropriate choice for this scenario.  Since SO2 is a respiratory irritant and no one can criticize taking a more conservative approach, the use of a 2 minute averaging time is recommended which yields a circle of influence extending 0.19 miles (1003 feet) from the source.  No offsite consequences would occur from this scenario.
Click to return to beginning