ML13324A251

Analysis of Explosive Vapor Cloud & Missile Hazards for Rail & Highway Transportation Routes.

ML13324A251
Person / Time
Site:	San Onofre
Issue date:	07/31/1979
From:	Broadhurst R, Chang Li, Schmidt E NUS Corp
To:
Shared Package
ML13302A324	List: ML13302A323 ML13324A248 ML13324A250 ML13324A251
References
NUS-3367, NUDOCS 7910240525
Download: ML13324A251 (71)

Text

4 NUS 3367 ANALYSIS OF EXPLOSIVE VAPOR CLOUD AND MISSILE HAZARDS FOR RAIL AND HIGHWAY TRANSPORTATION ROUTES NEAR THE SAN ONOFRE NUCLEAR GENERATING STATION UNITS 2 AND 3 Prepared for Southern California Edison Company

- NOTICE THE ATTACHED FILES ARE OFFICIAL RECORDS OF THE DIVISION OF DOCUMENT CONTROL. THEY HAVE BEEN CHARGED TO YOU FOR A LIMITED TIME PERIOD AND MUST BE RETURNED TO THE RECORDS FACILITY BRANCH 016. PLEASE DO NOT SEND DOCUMENTS CHARGED OUT THROUGH THE MAIL. REMOVAL OF ANY PAGE(S) FROM DOCUMENT FOR REPRODUCTION MUST.

BE REFERRED TO FILE PERSONNEL.

inager DEADLINE RETURN DATE .. ssessment

~~'/fft Ao'~? a 4 1 RECORDS FACILITY BRANCH i 20850 7 910 240

- NLJS CORPORATION

TABLE OF CONTENTS Section and Title

1.0 INTRODUCTION

2-1 2.0

SUMMARY

3-1 3.0 ANALYTIC MODELS 3.1 General Overressure due to 3.2 Explosive 3-9 Cloud at Air Intakes 3-17 3.3-1033 FammbleVapor 3.4 3-15 3-17 3.6 Solid Explosives 3-1 3-1 3.7 Self -Proelled Missiles ACCIDENT SEEIY4-1 4.0 ACCIDENT RATES AND 3-5 ASSESSMENT Hazardous Mtra Transporting 4.1 Accident 3.77Rates for Trucks on California 1-54 4.1.1 Truck Accident Rates 4-3 Rate 4.1.2 1-5 Tank Truck Accident 4.1.3 spill Rate and Distribution4 4.1.4 Explosion Rates 4-14 of Explosives 4.1.5 Accident Rates for Carriers Cryogenic 4-14 of Compressed flammable Gases and 4.1.6 Cylinders Flammable Gases ii NUJS CORPORATI

TABLE OF CONTENTS (Continued)

Section and Title Page No.

4.2 Transportation Accidents on the Railroad Passing by 4-16 the San Onofre Nuclear Generating Station 4.2.1 Train Accident Rates for Track Passing SONGS 4-17 4.2.2 Pressurized Tank Car Loss of Lading Rate 4-17 4.2.3 Severity of LPG Loss of Lading Accidents 4-19 4.2.4 Explosives, Accident Rates and Severity of Accident 4-28 5.0 ANALYSES AND RESULTS 5-1 5.1 General 5-1 5.2 LPG 5-1 5.3 Other Hazardous Gases 5-6 5.4 Explosives 5-6 5.5 Flammable Liquids 5-6 REFERENCES iii NUS CORPORATION

LIST OF FIGURES Figure Title Page No.

3-1 Simplified Event Tree for Transportation Hazards 3-2 3-2 Event Tree for a Spill 3-3 3-3 Region Definitions 3-4 3-4 Critical Buildings to Overpressure Damage 3-8 3-5 Region II Defined by the Building Layout 3-11 3-6 Probability of Flammable Plume Ignition Versus Plume 3-16 at the Time of Ignition 4-1 Trucks - LPG & NH 3 Spill Quantity - Gal - Normalized 4-9 to Max Truck Load of 10,000 gal.

4-2 Railroad Loss of Lading Quantity - Normalized at Maximum 4-21 Car Load of 33,000 Gallons iv NUS CORPORATION

LIST OF TABLES Table No. Title Page No.

2-1 Summary of Risk Assessment 2-2 4-1 Summary of Data Supplied by California Department 4-2 of Transportation 4-2 U.S. DOT Intercity Highway Truck Accident Rates 4-4 per Mile 4-3 National Truck Accident Rates 4-6 4-4 Summary of BMCS Reports for Compressed Hydrocarbon 4-8 Gases 4-5 Truck Spill Quantity 4-10 4-6 Hazardous Material Incident Reports LPG Trucks 4-12 4-7 Nationwide Accident Rates for Explosives or Dangerous 4-15 Articles 4-8 ATSF Freight Train Operational Data 4-18 4-9 Nationwide Train Accident Rate 4-20 4-10 Railroad Quantity of Spill 4-22 4-11 Summary of Mechanical Damage Induced Loss of Lading 4-25 Accident Severity 4-12 Railroad - Compressed Flammable Gases 4-26 4-13 Loss of Lading Caused by Fire 4-27 5-1 Commodity Dependent Input Parameters 5-2 5-2 Overpressure and Flammable Vapor Cloud Input Parameters 5-3 5-3 Results - LPG 5-4 5-4 Sensitivity Study - LPG on 1-5 5-5 5-5 Other Highway Results 5-7 5-6 Highway Military Explosive Shipment Size Distribution 5-8 v

NUS CORPORATION

1.0 INTRODUCTION

The San Onofre Nuclear Generating Station is located near Interstate 5, a major eight lane highway and the Atchison Topeka and Santa Fe railways main coastal north-south route. The hazards of activities along these transportation routes was considered in detail in 1976 and an evaluation included in the FSAR Section 2.2.

Recent questions from NRC have indicated the need to revise and update the previous analysis.

The revised analysis, as was the case for the previous analysis, determines the probability of occurrence of potential accidents. Standard Review Plan 2.2.3 defines design basis events external to the station as those accidents for which a realistic estimate of the annual probability of exceeding 10 CFR 100 exposure guidelines is in excess of approximately 10-70r for which a conservative estimate of this probability is in excess of approximately 10-6.

In the present analyses, it is conservatively assumed that the 10 CFR 100 exposure guidelines are exceeded if:

1. external overpressure on a safety-related structure resulting from explosions exceeds a specified value,

2. a flammable gas mixture occurs at the air intake to a safety-related structure,

3. a missile impacts a safety-related structure.

The following sections of this report summarize the work, describe the detailed analytic models, describe the accident rates and give the results of the analysis.

1-1 NUS CORPORATION

2.0

SUMMARY

The analysis of overpressure, flammable cloud intake and missile hazards from hazardous material activities on 1-5 and the ATSF railway has considered historic and projected shipment frequencies, historical accident rates and severity factors appropriate for the conditions near the plant and consequence models and para meters selected to provide an adequate and overall conservative description of the events.

Results of the analyses are summarized in Table 2-1. As can be seen the total probability of each consequence is substantially below 10-6 per year with accidents involving most commodities well below 10-7 per year. Only LPG is a significant contributor to the total probability. A sensitivity study on major parameters provides added assurance that the probability of each of the undesirable events at a unit is less than approximately 10-6 per year. Clearly then the probability of exceeding the 10CFR100 radiological exposure guidelines as a result. of transpor tation accidents is less than 10-6 per year.

The results of the analysis are believed to be a conservative estimate of the true probability of exceeding 10CFRI00 guidelines. Specific items of conservatism are as follows:

o The criteria selected are conservative. Exceeding 3 psi, having a flammable vapor cloud at an air intake or having a missile impact the plant will not necessarily cause any radio activity release, much less one sufficient to exceed 10CFR100 guidelines. For example one floor reinforced precast concrete and reinforced masonry houses designed to comply with California earthquake codes were essen tially undamaged (except for windows and doors) when subjected to peak overpressures of 1.7 psi (corresponding to about 3.4 psi peak 42 reflected overpressure) in Nevada Nuclear Tests . The probability estimated for San Onofre is that of exceeding 3 psi peak reflected overpressure.

2-1 NUS CORPORATION

TABLE 2-1

SUMMARY

OF RISK ASSESSMENT Probability of Flammable Probability of Exceeding Vapor Cloud Being Probability of 3 psi at the Plant Missile Impact 6

6 per year per year - 10-6 per year LPG - Highway 0.51 .089 0.17 LPG - Rail 0.23 .035 0.039 LNG -- .010 -

Liquid Hydrogen 0.001 .002 -

Compressed Hydrogen - 1 0.002 .012 -

Compressed Hydrogen - 2 0.0008 .001 -

Acetylene 0.0004 .002 -

Explosives - Highway 0.020 -

Explosives - Rail 0.005 -- -

Aircraft -- --- 0.080*

TOTAL 0.77 0.15 0.29

• From FSAR Section 3.5.1.6 2-2 NUS CORPORATION

o Most significant LPG truck accidents have involved the tank truck impacting massive obstructions or structures such as rock out croppings, bridge abutments or drainage ditch structures. None of these types of hazards exist along 1-5 near the plant. In addition all accidents on I-5 have been assumed to occur on the near edge of the right-of-way. This is 120 feet closer than the median strip.

o The Department of Transportation has mandated the retrofit of all flammable gas (LPG) tank cars with shelf-type couplers and head shields which reduce the likelihood of tank car punctures. Estimates of the degree of protection range from a 50% to a 90% reduction in punctures.43 These changes will be accomplished in the next few years and will tend to reduce significantly the LPG rail hazards at San Onofre.

o An instantaneous puff release of the spilled quantity is assumed. In a significant number of accidents the spill occurs over an extended time period, thereby reducing the hazard.

o The dispersion model conservatively accounts for gravity spreading of the heavier than air vapors without dispersion, followed by atmospheric dispersion using class G stability and a low wind speed.

o The explosive yield utilized is near the maximum observed. Most unconfined fuel air explosions have considerably lower yields. This is particularly true for truck spills.

o The plant area for which a flammable vapor cloud is unacceptable was conservatively described as a circle with an area more than twice the actual area of safety related buildings. In addition no credit was taken for the air intakes being above grade. Most of the vapor clouds of concern will tend to be of very limited vertical extent.

2-3 NUS CORPORATION

3.0 ANALYTIC MODELS 3.1 General Prior work has indicated that the principal contributor to hazard probability is shipments of Liquified Petroleum Gas (LPG). Other contributors are compressed flammable gases, cyrogenic liquids, other flammable liquids, and solid explosives.

Event tree's showing those events (excluding the munitions) which contribute to either the overpressure or air intake flammable cloud hazards are shown in Figures 3-1 and 3-2. In these event tree's Region I (Circle I or O ) designates that area surrounding the plant in which an explosion would lead to an overpressure in excess of the specified value. Region II (Circle II or 011) designates that area of the plant which incorporates all safety-related air intakes. These areas are shown schemat ically in Figure 3-3.

In general, a spill of flammable material can lead to either a fire, an explosion, or a flammable vapor cloud which disperses without a fire or explosion. The fire or explosion may occur at the accident site or away from the site. The explosion will affect the plant only if it occurs within Region I. The flammable cloud will be swept into an air intake only if an unignited cloud gets to Region II before the concentration drops below the lower flammable limit. As indicated, the conse quences are a function of spill size and spill location.

The above applies only to materials which vaporize to form drifting clouds. For solid explosives or munitions, only those portions of the event trees for accident site explosions are applicable.

3.2 Explosive Overpressure due to Unconfined Vapor Cloud Explosions The probability of exposing the plant to an overpressure greater than a certain value is the sum of the contributions from accident site explosions plus drifting cloud explosions. The former is:

3-1 NUS CORPORATION

NO ACCIDENT 0

FIRE 0

ACCIDENT IN 01 X

EXPLOSION SPILL ACCIDENT OUTSIDE P(S1/A)OF0 OF 01 NO. OF P1ADRIFTS OUT OF O 0 SHIPMENTS VAPOR EXPLOSION IN 01 NSH CLOUD X DRIFTS TO 01 FIRE IN 01 O.DRIFTS TO- 04 ACCIDENT X P(S2/A)

P(A/M) OTHERWISE OTHERWISE P(S3/A)

P(S4/A)

NO SPILL 0

X : POTENTIAL HAZARD O : NO HAZARD Figure 3-1. Simplified Event Tree For Transportation Hazards

EXPLOSION XE eo EXPLOSION XEO ACCIDENT IN FIRE e 0

O1 L fo IGNITED IN 01 DRIFTING CLOUD 1-fo-eo FIRE ACCIDENT OUTSIDE 01 WIND- 011 XA XEI, XAl ALI NOT IGNITED BEFORE IT REACHES 03 OTHERWISE 0 XE2.XA2 WIND- 0 P(S1/A) FIRE O EXPLOSION DRIFTING CLOUDT REACHES 0, IGNITED IN 0 1 ALi 1-fo-eo OTHERWISE AL EXPLOSION FIREOTHEOWIERWS ENXN IGNITEDO BEFORE d O OTEWS co 0 OTHERWISE XEN* XAN NOT IGNITED BEFORE[

ALNIT REACHES 0n WIND----*001 L=7 LLTL XAl LLFL = DISTANCE TO THE LOWER FLAMMABLE LIMIT DETERMINED BY THE DISPERSION MODEL XE = OVERPRESSURE HAZARD XA = AIR INTAKE HAZARD 0 = NO HAZARD Figure 3-2. Event Tree For A Spill

HIGHWAY 6 Li OR RAILWAY REGION I = EXPLOSIVE OVERPRESSURE IS OF CONCERN REGION Il= FLAMMABLE GAS IS OF CONCERN Figure 3-3. Region Definitions

E = NSH P(A/M)

N

- P(Si/A) . P(E/S) " Li (3-1) i=l1 where:

NSH = number of shipments of the commodity being evaluated P(A/M) = conditional probability of an accident per shipment mile P(Si/A) = conditional probability of the spill size given an accident P(E/S) = conditional probability of an explosion with a significant overpressure given a spill Li = length of route along which an explosion would yield an overpressure in excess of a specified value The summation allows an historical spill size distribution to be utilized.

The size of Region I and the length of highway or railway covered by the region is obtained from consideration of geometry and an explosion overpressure range relationship.

The distance R from an exploding charge to a specified pressure is calculated by the equation(1)

R = K (W) 1/3 ft (3-2)

K = constant determined by the allowable pressure

= 34 ft/lb1 / 3 - for a peak reflected overpressure of 3 psi.

from a ground level detonation. For a different criter ion, the constant K can be obtained from figure 4-12 of the reference 1.

W = pounds of TNT 3-5 NUS CORPORATION

For vapor cloud explosions, it is common practice (2, 3, 4, 5) to utilize a TNT equivalent calculated as follows:

W= F Si QtO Hc E 500 Kcal/lb - TNT (3-3)

A F = Fraction of spill quantity involved in vapor cloud Si QP = gm mole of combustible chemicals spilled A

Si = spill fraction Q = maximum quantity of shipment in volume P = density of liquid A = molecular weight Kcal\

AH = Heat of combustion ( ),

gm mole E = Yield of explosion For liquified gases shipped at atmospheric temperature under their own vapor pressure, the fraction of spill quantity in the vapor cloud is the isenthalpic flash fraction. For compressed gases it is 1.0. These values are consistent with the conservatively assumed instantaneous puff release model. For cyrogenic liquids shipped at essentially atmospheric pressure, a 10% flash fraction was used to account for initial vaporization on mixing with warm air and boiling from the spilled liquid pool.

The entire quantity in the cloud was assumed to be involved in the fuel air reaction. The change in the quantity of vapor between upper and lower flammable limits as the cloud disperses was conservatively neglected. Analysis of a drifting puff release (7) has shown that the maximum quantity between flammable limits is on the order of 60-70% for materials of interest here and that for much of the travel distance, it is less than this amount.

3-6 NUS CORPORATION

The equivalent TNT yield was taken to be 10% (E = 0.1) based on energy. This value is near the upper end of the range of values reported in the literature 4

6)

Equations 3-2 and 3-3 give the maximum distance from any structure at which the explosion involving a particular commodity could yield the specified overpressure.

The length of highway or rail-line within this distance of a plant safety-related structure can be obtained from the geometrical plant layout shown in Figure 3-4.

This length (and the size of Region I) is spill size and commodity dependent.

For drifting cloud explosions, the probability of exposing the plant to an over pressure greater than a certain value is:

P = NSH

• P(A/M)

N P(Si/A) Il - P(F/S) - P(E/S)

M E P (Ignition O ) P (E/I) AL (3-4) j=1 where:

P(F/S) = conditional probability of a fire given a spill P. (Ignition O = probability of ignition (fire or explosion) in Region I given a spill at accident site j P(E/I) = conditional probability of an explosion given an ignition a L. = incremental length of route located a given dis tance and direction from the plant 3-7 NUS CORPORATION

1-5 RAILWAY 5605 490' 725' o u350

-- B C FUEL 1IANDLING BLDG 14-120' 4 A

TANK BLDG AUXILIARY BLDG DIESEL GENERATOR BUILDING NOT CONSIDERED SINCE IT CAN WITHSTAND 560' MORE THAN TWICE OTHER BUILDINGS Figure 3-4. Critical Buildings To Overpressure Damage

The probability of ignition in Region I is a function of where the accident occurs (inside Region I, outside Region I and distance away), the wind direction, and a probability of ignition as a function of cloud travel distance. The first summation allows for varying spill sizes, while the second summation allows for different accident sites.

For accidents inside Region I, the wind blows with a seaward component about half of the time and a landward component about half of the time. The distance traveled in Region I used for calculating the ignition probability was the maximum distance normal to the transportation route to the edge of Region I for each of the wind directions (seaward and landward).

The transportation routes outside Region I were divided into small increments. For each increment, the probability of the cloud centerline being blown through Region I was calculated utilizing FSAR site wind direction probabilities. The probability of an ignition in Region I was based on the shortest distance from the accident site to the Region I boundary and the maximum cloud path length in Region I (which is the Region I diameter). The maximum accident site to plant distance considered was the maximum downwind distance to the lower flammable limit calculated by the dispersion model described later.

3.3 Flammable Vapor Cloud at Air Intakes In order for a flammable vapor cloud to be swept into a plant air intake, the flammable cloud must intersect Region II without prior ignition.

The probability of this occurring is given by:

P = NSH P(A/M)

N

. P(Si/A) 1 - P(F/S) - P(E/S) i= I M

.I 1- P (Ignition before O ) (continued)

• j=1 3-9 NUS CORPORATION

P (wind blowing to O) ALj (3-5) where:

P (Ignition before OH) = probability that the cloud ignites before it gets to plant (OH1)

P. (Wind blowing to O) = probability that the wind is blowing toward O from a given accident site j.

The probability of ignition before the cloud reaches Region II is based on the minimum distance from the accident site to Region II. The probability of a flammable cloud intersecting Region II is based on the FSAR wind direction probabilities and a crosswind distance from cloud centerline to the lower flam mable limit based on the dispersion model described later. Region II was analytically defined as a circle enclosing plant air intakes as shown in Figure 3-5.

3.4 Dispersion Model The dispersion model is used to determine the distance that the vapor cloud has to travel to reach the lower flammable limit and the crosswind distance to the lower flammable limit. The downwind distance is used only to determine the maximum length of transportation route which must be considered. The crosswind distance is effectively added to the boundary of Region II to determine the probability of a flammable cloud intersecting Region II and being swept into a plant air intake.

An instantaneous puff dispe-rsion model modified to account for initial gravity slumping at heavier than air vapors is utilized. The diffusion equation for an instantaneous (puff) ground level release with a finite initial volume is  :

X (d) a 2 a 2 (a 2 a 72)12]

Q = 7.87 ( y + ly z + I exp 1/2 y 2 + z 2 a y2 +a ly2 a 2 +a 1z2 (3-6) 3-10 NUS CORPORATION

1-5 RADIUS 370' C a FUEL HANDLING 30' BLDG A U TANK m BLDG D

AUXILIARY Oi BLDG INTAKE STRUCTURE E

Figure 3-5. Region II Defined by the Building Layout 3-11

X (d) unit concentration at coordinates y, z from the center I of the puff, (m- 3 )

a (d), a (d) = standard deviation of the gas concentration in the hori yz zontal crosswind and vertical crosswind direction respectively, (M) d = distance from the origin of the puff release (m) a IY, a = initial standard deviation of the puff in the horizontal and vertical direction respectively, (m)

( aj y > a I z for heavier than air gas. For neutral or lighter than air gas or I, = aI z yz = distance from the puff center in the horizontal and vertical directions, respectively, (m)

For those gases heavier than air, a I y and 1 Z are determined from the Van Ulden gravity spreading model (Reference 9). The initial cloud formed at the accident site is assumed to be cylindrical in shape with the axis perpendicular to the ground and spreads according to the density difference between the cloud and the air. It is assumed that during the gravity spreading phase, the flammable vapor concentration in the cloud remains unchanged. The cloud spreads until the turbulent energy of the spreading equals the potential energy difference between the heavy gas layer and the surrounding air.

From Reference 9:

Ro 2 + 2 _g (Po - Pa) Vo R =

' " Po (3-7)

H = Ho R (3-8) 3-12 NUS CORPORATION

where:

R = cloud radius (m)

H = cloud height (m)

Ro = Ho = initial size of the cloud of the cylinder (m) 2 g = gravitational constant = 9.8 m/sec Vo = initial volume of the mixture (m3 )

Po initial density of the mixture (kg/m3)

Pa = density of the air (kg/m 3 )

equation:

The gravity spreading ends at time ts which satisfies the 2u* = uf (3-9) where:

1 (Po - Pa) Vo u fr PO (3-10) uk u* =- In (z/z) (3-11) u is a fixed wind velocity at a specified height z k is the Von Karman's constant = 0.4 zo = roughness length = 0.05 m 3-13 NUS CORPORATION

If this criterion leads to a final cloud height of less than 1 meter, then slumping is stopped when the 1 meter height is reached. One meter represents the height of terrain features within the cloud.

The distance that the cloud travels during the spreading period is as follows:

t d In / )dt s k z (3-12) 0 At the end of the gravity spreading, the concentration of the cloud is assumed to have a Gaussian distribution with the center point concentration being one (or pure vapor). The a I are obtained from:

I y'Iz aI = y R y H (3-13)

Sz =

R = Radius of the cloud at the end of the spreading H = Height of the cloud at the end of the spreading

/ 1/3

\/2 n C (3-14)

C = assumed initial puff concentration C = Gaussian cloud center point concentration = 1.0 (or pure vapor)

The equation for Y comes from equating the amount of vapor in the cylindrical cloud to that in the Gaussian cloud.

3-14 NUS CORPORATION

An initial puff concentration of 0.25 was assumed for the base case analysis to account for dilution due to turbulent mixing at the release point. Several analyses and tests have shown (10, 11) that momentum transfer of the rapidly expanding gas induces considerable mixing which if uninhibited is sufficient to reduce the resulting gas cloud concentration to below the flammable limit. Sensitivity studies of this initial concentration have shown that the results are not strongly affected by the assumed value.

The Gaussian cloud disperses in accordance with equation 3-6 starting at distance d from the spill site.

3.5 Vapor Cloud Ignition As indicated in Section 4.0, most spills of flammable vapor are ignited essentially at the accident site. For example, James(12) quotes statistics from the Associa tion of American Railroads where for 81 vapor cloud ignitions, 58% occurred from a few feet up to 50 feet, 18% between 50 and 100 feet and 24% from 100 feet to 300 feet.

A curve of integrated ignition probability as a function of distance from historical data of LPG spill accidents was published in Reference 13. The curve is shown in Figure 4-6 and is represented as a line:

logl 0 A - 1.38021 1P = Y2 +e2.45318 (3-15)

A = 0.175 r 2 (3-16) where:

r is the distance in meters.

The information in Reference 12 agrees reasonably well with this curve. The data of Reference 13 also indicates that 10.5% of the drifting cloud ignitions resulted in an explosion while 89.5% resulted in a fire. This agrees well with the data discussed in Section 4.0.

3-15 NUS CORPORATION

2 A, FLAMMABLE PLUME AREA (An a

c 0 0+

CD w n m rj

-4% ->

3 V r

00c

-- Il ID- II CA z

• 00 CC 3 00 CD

3.6 Solid Explosives For solid explosives, the probability of exceeding a certain overpressure was obtained from:

P = P(A/M)* P(E/A)

• NSH L (3-17) where:

P(A/M) = conditional probability of an accident per shipment mile

- P(E/A) = conditional probability of an explosion per accident NSHi = Annual number of shipments of size 1 L. = length of route along which explosion of shipment size would yield an overpressure in excess of a specified value.

The length of route is obtained from the geometrical configuration of the plant and the route and the overpressure range relationship of equation 3-2. NOTE that it is conservatively assumed that if an explosion occurs, it will involve all explosives in the shipment simultaneously.. Staggered explosions will not yield the same overpressure as simultaneous explosions.

3.7 Self-Propelled Missiles In some instances accidents involving compressed liquid gases have been observed to generate self-propelled (rocketing) missiles due to the effects of the boiling liquid expanding vapor explosion (BLEVE). The probability of a missile impact is given by:

3-17 NUS CORPORATION

P = NSH . P(A/M)- P(S/A)

- P (Missile/S). Pi (I/Missile)

Number of Missiles - AL (3-18) where:

P (Missiles/S) = conditional probability of missile formation given a spill P (I/Missile) = conditional probability of an Impact given a missile at accident site i AL = incremental length of route It should be noted that this relationship is valid only when the product P(I/Missile)

Number of Missiles is much less than one.

The probability of a plant impact given a missile is a function of the accident site location relative to the plant and the effective target area of the plant. For accident sites not adjacent to the plant, the missile must be propelled upward to clear ground level obstructions in order to impact the plant. For close in accident sites missiles with little or no vertical velocity can impact the plant. For truck accidents (which because of the shorter missile range will be of the latter type) the probability of impact for a given accident site is given by the ratio of the sector angle in the horizontal plane intersecting the plant to all possible angles (3600).

The summation in equation 3-18 can be replaced by an integral integrated over the length of the highway within the maximum missile range of the plant.

For railway accidents, the maximum missile range is greater and higher angled trajectories are necessary to impact the plant. For this case, the probability of impact for a given accident site is the ratio of the solid angle which would yield a plant impact to the total solid angle of a hemisphere. Again, this is integrated over all possible accident locations.

3-18 NUS CORPORATION

The above model assumes the missile direction is random (not directionally dependent). Accident reports indicate that large missiles tend to follow trajec tories along the transportation route. This would substantially reduce impact probability, particularly for accident sites near the plant.

3-19 NUS CORPORATION

4.0 ACCIDENT RATES AND ACCIDENT SEVERITY ASSESSMENT Truck and railroad traffic density and number of accidents along transportation routes adjacent to the SONGS site are analyzed to evaluate basic site specific accident rates. These local accident rates are adjusted by nationwide accident statistics, where the data base is larger, to assess for commodity classes accident rates and severity of accidents. These accident rates and severity density functions are used as input data to the analytical risk model.

4.1 Accident Rates for Trucks Transporting Hazardous Materials Truck accident rates are determined from data collected on Interstate Route 5 (

5) adjacent to the SONGS site. Adjustment factors and severity rates are determined from a nationwide data base collected by the U.S. Department of Transportation.

4.1.1 Truck Accident Rates on California I-5 Accident rates for all trucks* and commodities are determined for a 10 mile segment of 1-5 extending approximately equidistant in both directions from the SONGS site. California State Department of Transportation supplied data is summarized in Table 4-1. From this data, an observed truck accident rate of 0.566 x 10-6 accidents per truck mile is evaluated. The data given in Table 4-1 is for 1-5 from mile post R61.38 to mile post R71.38. Truck traffic rates are based on weighted sample counting and extrapolated to annual counts. Northbound and southbound data are combined. Traffic accidents are reported to the state if property damage is $200.00 or greater or there has been personal injury or death.

• Truck is defined as any vehicle 5000 pounds or more excluding pickup trucks, vans and buses.

4-1 NUS CORPORATION

TABLE 4-1

SUMMARY

OF DATA SUPPLIED BY CALIFORNIA DEPARTMENT OF TRANSPORTATION Calendar Truck Miles Number Acci gents per Year on 1-5 of Accidents 10 miles 1974 20.38 x 106 12 0.589 1975 19.88 x 106 9 0.453 1976 21.83 x 106 15 0.687 1977 22.65 x 106 12 0.530 Combined 84.74 x 106 48 0.566 4-2 NUS CORPORATION

Later in this analysis, 1-5 accident rates are combined with U.S. DOT data where the property damage threshold for reporting accidents has been increased from

$250.00 to $2,000.00. To correct for the data base inequities, U.S. DOT experience before and after the reporting threshold change is used to generate a correction 14, 15 factor. Table 4-2 presents data covering the transition period.

1973 accident rate 0.952 -x 10- 66-0.423 $2000 accidents Correction factor =

1971-72 accident rate 2.25 x 10 6 $ 250 accidents The 2.25 x 10-6 rate is the average of 1971 and 1972 rates.

This factor is applied to the 1-5 accident rates based on the assumption that California accident rates would be reduced by the same proporation as that observed on the national level. The fact that the California threshold is $200.00 vs. $250.00 for the U.S. DOT would make the correction factor a conservative assumption.

The accident rate corrected to the $2,000.00 death or injury reporting criteria for all trucks on 1-5 is:

0.423 x 0.566 x 10-6 = 0.239 x 10- 6 accidents/mile 4.1.2 1-5 Tank Truck Accident Rate The bulk of hazardous commodities carried on 1-5 past the San Onofre Site are in tank trucks.

Therefore, the I-5 tank-truck accident rates are assessed by applying a correction factor based on nationwide experience. An Authur D. Little, Inc. Report1 6 evaluated a national tank-truck accident rate of 1.33 x 10-6 per loaded tank truck mile.

4-3 NUS CORPORATION

TABLE 4-2 U.S. DOT INTERCITY HIGHWAY TRUCK ACCIDENT RATES PER MILE Accident Reported Accident 6 Injury Fatality -6 Year If Over* Rate x 10 Rate x 10 Rate x 10 1971 $ 250 2.19 1.00 0.083 1972 $ 250 2.31 0.996 0.081 1973 $ 2000 0.952 1.02 0.071

• Accident also reported if there was an injury or fatality.

4-4 NUS CORPORATION

This accident rate is based on data from 1968 through 1972 (5 years). The average number of loaded tank-truck accidents was 1650 accidents per year and the average loaded tank-truck usage was 1.24 x 10 miles per year. During the same five year period, the Bureau of Motor Carrier Safety publishedl4 data yielding an inter-city truck accident rate of 2.41 x 10-6 accidents per mile. This accident rate is the ratio of 160,347 accidents and 66,389 x 106 truck miles. (Data extracted from reference 14 for this evaluation is shown in Table 4-3.)

Nationwide truck accident statistics show that loaded tank trucks have a lower accident rate than all types of trucks combined. (1.33 x 10- 6 vs. 2.41 x 10-6 for years 1968 through 1972 with the same reporting criteria). Therefore, the I-5 accident rate for all types of trucks (0.239 x 10-6) is corrected to loaded tank truck accident rate by assuming the same relative improvement exists in California (1-5) as observed nationwide.

Loaded tank truck 0.239 x 10-6 1.33 x 106 = 0.132 x 10-6 accidents/mile accident rate on 1-5 2.41 x 10-6 4.1.3 Spill Rate and Distribution Compressed gases in the liquified state pose a hazard only if they are released from the pressurized containment and vaporize. Most accidents reported and used in determining the tank-truck accident rate did not result in a loss of lading (spill).

Therefore, to assess the potential hazard of a pressurized tank-truck carrying flammable, gases, the fraction of spills per accident (probability of spill given an accident has occurred) must be determined.

The 1-5 data base is insufficient to generate this ratio, hence the Bureau of Motor Carrier Safety (BMCS) of the U.S. Department of Transportation accident re porting system was consulted. Submission of accident report form MCS 50-T by carriers is required if the accident has property damage greater than $2,000.00 or personal injury or death. The standard annual report published by the BMCS does not analyze spill frequencies of flammable or hazardous materials. Therefore NUS Corporation requested and received computer magnetic tape records of the accident report forms for calendar years 1973 through 1977.

4-5 NUS CORPORATION

TABLE 4-3 NATIONAL TRUCK ACCIDENT RATES Calendar Total Intercity Total Intercity Accidegt Rate Year Vehicle Miles Accidents per 10 miles 1968 11704 x 106 29209 2.50 1969 12461 x 106 30672 2.46 1970 12390 x 106 33203 2.68 1971 13951 x 106 30581 2.19 1972 15883 x 106 36682 2.31 Combined 66389 x 106 160347 2.41 4-6 NUS CORPORATION

A computer program was written to select and print accidents which had the following characteristics: transporting hazardous material and tank type of body and during an over the road (intercity) trip. The magnitude of this data sort is illustrated by 1977 accident data.

30,567 accident reports were submitted 24,216 were over the road trips 3,457 had tank truck bodies 1,886 involved transporting hazardous materials 977 had all characteristics The 977 accidents were manually reviewed to further select accidents which involved compressed flammable gases, LPG, propane, butane, etc.; were on divided highways when the accident occurred; and were not on an entrance or exit ramp when the accident occurred. This reduced the number of accidents for 1977 to 33.

Two of the 33 accidents had spills. This process was performed for each year 1973 through 1977; the results are summarized in Table 4-4. As indicated 7 out of 109 accidents had spills.

A spill quantity distribution was obtained from data submitted to the Office of Hazardous Materials Operations on Hazardous Materials Incident Report forms.

The data base period was from mid 1973 through 1977. All traffic accident induced spills for LPG and bulk anhydrous ammonia were considered. Both commodities are carried in DOT MC 330 and MC 331 specified trucks. Differences in tank truck load volume is accounted for by normalizing all truck capacities to a nominal 10,000 gallon maximum load quantity. Spills which did not give both spill size and truck capacity or fraction of load spilled were censored. The distribution is generated from a total of 35 truck spills. The results are shown in Figure 4-1.

Table 4-5 gives the distribution used in the analysis.

4.1.4 Explosion Rates Given that a tank-truck accident has occurred and a loss of lading has occurred, three things can happen: (1) the spilled LPG can vaporize, expand and form a drifting cloud (the possible outcomes of the drifting cloud will be treated elsewhere 4-7 NUS CORPORATION

TABLE 4-4

SUMMARY

OF BMCS REPORTS FOR COMPRESSED HYDROCARBON GASES Accident Calander Year Total Result, Code 1973 1974 1975 1976 1977 No Spill A 14 17 17 23 31 102 Spill B I 1 1 2 2 7 Fire C 0 0 0 0 0 0 Explosion E 0 0 0 0 0 0 Total 15 18 18 25 33 109 This analysis gives the probability of spill given an accident has occurred of 7/109 = 0.064. This is for compressed LPG Truck MC 330 and MC 331, on highways typified by 1-5 passing in front of the SONGS site.

4-8 NUS CORPORATION

10k 9k 8k 7k 6k 5k 4k 3k 2k 2

0 1000 J 900 a 800 700 600 500 400 300 200 100I 0.01 0.05 0.2 0.5 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 99.9 99.99

% OF SPILLS LESS THAN ORDINATE VALUE Figure 4-1. Trucks - LPG & NH3 Spill Quantity - Gal - Normalized To Max Truck Load of 10,000 gal.

TABLE 4-5 TRUCK SPILL QUANTITY Fraction of Spill Quantity Spills Gallons 0.2 10000 0.2 9000 0.2 7000 0.4 1000 4-10 NUS CORPORATION

in this analysis) or (2) the spilled LPG can be ignited forming a fire (the size of the fire can vary from small to large) or (3) the spilled LPG can expand and be ignited to form an explosion.

Spill data have been examined to determine and classify the results of the spill into one of the three classifications above. To perform this analysis, data from the Materials Transportation Bureau, U.S. Department of Transportation has been obtained and sorted to eliminate those accidents which are inappropriate for analysis of LPG Tank-Trucks. Since mid 1973, the Materials Transportation Bureau (MTB) has required reporting of hazardous materials incidents in accordance with the Hazardous Materials Control Act of 1970. Hazardous material spills while in transit or temporary storage are required to be reported.

For this analysis, only those spills which were the result of a highway accident and where trucks were carrying LPG products are included. Analysis of incident reports from mid 1973 through 1977 revealed 34 events which satisfied the above criteria. However, 11 of these 34 incidents involved property damage of less than

$2,000.00 and to be consistent with other accident statistics in this analysis, they were censored. The 23 remaining LPG spill events plus one additional which will be explained are presented in Table 4-6 along with the resulting severity density function. None of these LPG spills occurred within one hundred miles of San Onofre.

As indicated in Table 4-6 no vapor cloud explosions were observed in the 5 year period ending in 1977. In Mexico, during 1978, there was strong evidence that a vapor cloud explosion occurred from an LPG truck spill.17 A post accident examination showed severe tree limb breakage at approximately 350 feet from the center of the blast area. Also a sonic boom was reported. This accident was included to obtain the 0.042 mean explosion rate. This is conservative since the LPG truck accidents which did not lead to an explosion have not been accounted for.

Three other accidents are discussed, those classified as Boiling Liquid Expanding Vapor Explosions (BLEVE). Although originally classified as explosions by the MTB, a investigation by NUS revealed that overpressures were not severe.

4-11 NUS CORPORATION

TABLE 4-6 HAZARDOUS MATERIAL INCIDENT REPORTS LPG TRUCKS Number of Incidents Sub- One Severity, i 1973 1974 1975 1976 1977 Total Added Total Pi Spill 1 2 4 2 3 12 -- 12 .500 Fire 1 1 0 3 3 8 -- 8 .333 BLEVE* 0 0 1 1 1 3 -- 3 .125 Explosion 0 0 0 0 0 0 1 1 .042 TOTAL 2 3 5 6 7 23 1 24 1.000

• Boiling liquid expanding vapor explosion.

4-12 NUS CORPORATION

1. April 29, 1975, Eagle Pass, Texas. A tank-truck with 8748 gallons of LP gas overturned, ruptured and was ignited. A prefab sheet metal building 200 feet from the location of the truck accident experienced no structural damage and only one broken window.18 According to the owner of the Border Diesel Service located in the building, the entire window frame was knocked out because of faulty assembly when the building was put together. One witness was exiting his pick-up truck at 150 feet from the accident when the blast occurred. The pick-up truck had the rear window knocked out, but otherwise sustained no damage.19 The observed level of damage is consistent with an explosion of approximately 100 lb of TNT.

The equivalent yield on an energy basis. is less than 0.196 which is typical of a boiling liquid expanding vapor explosion (BLEVE).

2. August 19, 1976 Flint, Michigan. A tank truck with approximately 9000 gallons of LPG, propane, impacted the guard rail on an exit ramp, slid along the rail for 258 feet and on to an overpass. The truck fell from the overpass to the pavement below. Ignition occurred on or below the overpass. Seven vehicles and occupants within 400 ft suffered burns and fire damage, but no indications of over pressure damage - there were no body panel indentations; broken windows were minimal and might have been fire-caused. The closest building, a house, at approximately 600 ft had no structural damage and no broken windows although fires were started on the property. At a distance of 1/4 mile a vibration was felt, but no sonic boom was reported.20 Overpressure evidence suggest that the accident produced a boiling liquid expanding vapor explosion (BLEVE).

3. February 7, 1977, Detroit, Michigan. A tank truck loaded with approx imately 9000 gallons of LPG, propane, collided with a guard rail on an overpass exit ramp. The truck broke through the guard rail and fell 20 feet to an embankment below, adjacent to a freeway. Ignition occurred at impact with the embankment. The closest building 300 to 400 feet away had no structural damage and no broken windows, although the tarred roof caught fire.21 A highway light fixture at approximately 80 feet had a broken globe and was burned; the next light fixture at 160 feet was intact but burned. None of the witnesses reported a concussion or sonic boom. 2 2 Evidence suggests that the overpressure was generated by a BLEVE.

4-13 NUS CORPORATION

4.1.5 Accident Rates for Carriers of Explosives Accident rates for trucks carrying explosives is evaluated by adjusting 1-5 accident rates with nationwide experience. The BMCS evaluates accident rates for various categories of commodities. One of these groups is explosives or dangerous articles.

Table 4-7 presents accident data extracted from BMCS annual reports (14, 15, 23, 24).

Nationwide truck accident statistics show that trucks carrying explosives or dangerous articles have a lower accident rate than all types of trucks (0.958 x 10-6 vs. 2.41 x 10-6 accidents per mile). Assuming that trucks carrying explosives on I-5 would experience the same relative improvement, 1-5 accident rates are corrected as follows:

1-5 Explosives 0.958 x 10-6 x 0.239 x 106 = 095 x 10- 6 accidents/mile.

Trucks Accident Rate 2.41 x 10-6 The probability that an accident would result in an explosion was determined by data provided by the Institute of Makers of Explosives on the accident statistics for commercial shipments of explosives. During the four year period of 1972-1975, there were 70 accidents reported of which three involved explosions. From this information, it is estimated that the conditional probability of an explosion given an accident is 3/70 or 0.043. Accident Reports are filed when an explosive shipment accident results in (1) fire, (2) death or injury, (3) property damage exceeding $1,000.

The accident rate of 0.095 x 10- 6 accidents per mile and the condition probability of explosion of 0.043 are used in the analytical hazard model.

4.1.6 Cylinders of Compressed Flammable Gases and Cryogenic Flammable Gases Nationwide accident rates for compressed gases in cylinders and cryogenic gases are assumed to be represented by tank truck accident rates. Therefore, the 1-5 accident rates evaluated for loaded tank trucks are assigned to trucks transporting cylinders of compressed flammable gases and cryogenic flammable gases.

4-14 NUS CORPORATION

TABLE 4-7 NATIONWIDE ACCIDENT RATES FOR EXPLOSIVES OR DANGEROUS ARTICLES Carriers Truck Number Accidents Year Reporting Miles of per 106 miles Thousands Accidents 1969 5 65046 110 1.69 1970 1 1764 1 0.57 1971 7 52971 28 0.53 1972 7 56568 30 0.53 Combined - 176349 169 0.958 4-15 NUS CORPORATION

BMCS data was searched for the period 1973 through 1977 to determine a loss of lading rate per accident for compressed gases in cylinders. Accidents included in the data were over-the-road (intercity) trips transporting hazardous materials.

Accidents occurring on individual highways and exit and entrance ramps were also included. A total of 16 confirmed accidents with cyclinders of flammable compressed gases were identified. By including non-flammable gases the number of accidents increased to 19. Two of the 19 accidents resulted in loss of lading giving a spill rate per accident of:

2 spills = 0.105 spills/accident 19 accidents Reviewing BMCS data for cryogenic truck accidents and spills as a result of accidents showed insufficient data to derive a parameter. Therefore the spills per accident for LPG tank trucks (.064 spills/accident) was assumed applicable for cryogenic tank trucks.

Reviewing the MTB data for severity of accidents involving flammable gases in cylinders and cryogenic gases revealed insufficient data to form fire and explosion rates given a spill. Therefore the parameters determined for LPG products were assumed applicable as follows:

Probability of explosion given a spill = .042 Probability of fire given a spill = .458 Probability of no fire or explosion given a spill = 0.50 These parameters were used as input data to the analytical risk model.

4.2 Transportation Accidents on the Railroad Passing by the San Onofre Nuclear Generating Station Hazardous materials transported on the Atchison, Topeka and Sante Fe (ATSF) rail line by the SONGS site are military ordnance and LPG. The ATSF Railway Company does not anticipate any other hazardous materials being shipped along 4-16 NUS CORPORATION

this track. Railroad accident rates and accident effect rates are evaluated from ATSF supplied data and national data. These accident rates are applicable to the track and materials under study.

4.2.1 Train Accident Rates for Track Passing SONGS From data supplied by ATSF for a section of Track from Fullerton to San Diego and passing by SONGS, Table 4-8 is developed. Data is from 11 years less the month of December 1978 and the track is from mile post 165.5 to 268.0, a distance of 102.5 miles. While there are sidings, there are no splits on this section of track.

From Table 4-8, the average accident rate for the 11-year period is:

10 accidents = 3.70 x 10-6 accidents per train mile (102.5 miles) x (26,378 Trains) 4.2.2 Pressurized Tank Car Loss of Lading Rate Loss of lading rate as a result of a train accident is evaluated for pressurized non insulated tank cars transporting LPG gases past the SONGS site. These tank cars are referenced by specification numbers 112A and 114A. They transport gases in the liquid state under pressure at ambient temperatures.

A loaded tank car loss of lading rate is determined in. an Association of American Railroad Report25 to be 0.152 x 10-6 loss of ladings per tank car mile. This spill rate is based on data from 1965 through 1970 (6 years) where a total of 49 loss of lading accidents were observed. During this period, the average loaded pressurized tank car traffic (flammable gases) was 5.38 x 107 miles per year (from 196 Waybill statistics). Thus, the nationwide loss of lading rate for liquified compressed flammable gases is:

49 accidents = 0.152 x 10-6 accidents 5.38 x 107 car miles . 6 years car mile year 4-17 NUS CORPORATION

TABLE 4-8 ATSF FREIGHT TRAIN OPERATIONAL DATA Number of Number Year Freight Trains of Accidents*

1968 2242 0 1969 2398 0 1970 2346 0 1971 2450 0 1972 2346 0 1973 2450 3 1974 2450 1 1975 2398 0 1976 2450 1 1977 2398 2 1978 2450 3 Total 26378 10

• 3 accidents have been excluded from the ATSF submitted data-one was a passenger train accident, 2 were minor accidents to cars where no derailment occurred.

4-18 NUS CORPORATION

Next, the nationwide pressurized tank car loss of lading rate is adjusted to the ATSF track passing by the SONGS site. This adjustment is based on the assumption that the ATSF pressurized tank car rates will show the same relative improvement over nationwide rates as ATSF train accident rates show to nationwide train accident rates.

Table 4-9 presents the annual data for the past ten years. (This is the same time period from which the ATSF train accident rate was evaluated.) The nationwide average train accident rate for this ten year period is 10.95 accidents per million train miles.

The loss of lading rate for pressurized LPG tank cars on the ATSF track in the vicinity of the plant site is:

3.70 x 10-6 x 0.152 x 10-6 = 0.0514 x 10-6 Loss of lading per 10.95 x 10- 6 loaded tank car mile This loss of lading rate is used elsewhere in this report and as an input parameter to the analytical model.

The quantity of lading spilled as a result of an accident is determined for railroad tank cars from Hazardous Materials Incident Report forms submitted to the Office of Hazardous Materials Operations, DOT. The data base period for this analysis is mid 1973 through 1977. All spills in this assessment are DOT specification 112A or 114A tank cars loaded with flammable compressed gases (LPG). Tank car volume is normalized to a nominal maximum load capacity of 33000 gallons. A distribution of spill quantity is generated from 76 tank car spills. The results are shown in Figure 4-2. Table 4-10 presents the distribution values used in the analysis.

4.2.3 Severity of LPG Loss of Lading Accidents This section presents definitions for categorizing the severity of LPG loss of lading accidents and using these definitions to analyze accident data developes a probability density function. Jones et al,26 in their evaluation of the risks of 4-19 NUS CORPORATION

TABLE 4-9 NATIONWIDE TRAIN ACCIDENT RATE Total Train Train TrainAccidgnt Miles-Thousands Accidents Rate per 10 miles Year 1968 876489 8028 9.16 1969 864081 8543 9.89 1970 838674 8095 9.65 1971 783844 7304 9.32 1972 781408 7532 9.64 831347 9698 11.67 1973 833261 10694 12.83 1974 755033 8041 10.65 1975 774764 10248 13.23 1976 750042 10362 13.82 1977 8088943 88545 10.95 4-20 NUS CORPORATION

1000 O 9 9 8 - 8 7 7 6 6 5 5 4 4 3 - 30K 32 - 20K 2*

0 2 020 O 21 O

S 100 .****- 10k

- 6 74 a

-J CLd 4

3 2

10 I I I I I I I I I I I I I I I I 1k 0.01 0.05 0.2 1 2 5 10 20 30 40 50 60 70 80 90 95 98 99 99.8 99.9

%OF SPILLS LESS THAN ORDINATE VALUE Figure 4-2. Railroad Loss of Lading Quantity - Normalized at Maximum Car Load of 33,000 Gallons

TABLE 4-10 RAILROAD QUANTITY OF SPILL Fraction of Spills Spill Quantity - Gallons 0.60 33000 0.10 30000 0.05 27000 0.05 10000 0.20 1000 4-22 NUS CORPORATION

propane rail car shipments have defined five severity categories for propane loss of lading accidents. The same five categories are applied to all compressed flammable gases; the categories are as follows:

Type I - This type of incident could be caused by a major rupture of the containment vessel resulting in a gross spill without ignition. The result would be that a very large vapor cloud would be formed. If this cloud would be ignited after an explosive fuel-air mixture had been formed, a maximum incident explosion would result. This type of incident is characterized by an unconfined fuel/air detonation.

Type II - This type of incident would be caused by a separate fire or a tank puncture resulting in a fire that would overheat the punctured propane tank or another propane tank in the near vicinity. The result would be an explosive pressure rupture of the heated tank, causing nearby overpressure damage and possible shrapnel damage from the ruptured tank. This type of incident is characterized by a propane tank explosion.

Type III - This type of incident would result from a leak or a tank puncture resulting in a large spill with ignition occurring immediately or shortly after the incident. The propane would burn uncontrollably in a large, intense fireball. No tank explosion would occur since the tank puncture would be large enough to relieve the pressure. This type of incident is characterized by a large uncontrollable fireball with no explosion.

Type IV - This type of incident would be caused by a leak, a tank puncture, a released safety valve or a bust transfer line or valve resulting in a controllable fire. The fire may be of considerable time duration and does not result in tank rupture, either due to fire control measures or protective insulation. This type of incident is characterized by a controllable fire with no explosion.

4-23 NUS CORPORATION

Type V - This type of incident would involve a leak or a puncture, either small or large, in a propane tank of loading lines which does not result in fire. If no source of ignition occurs, the propane will be dispersed in the atmosphere in a relatively short time. This type of incident is character ized by loss of lading, but no fire.

From detailed analysis of LPG, tank car accident data from the period 1965 through 1977 accidents are categorized into one of the five severity types and presented in Table 4-11. For this analysis, the differentiation between Type III and Type IV severity is not important and where the size of fire was not quickly determined, a worst case classification of Type III was assigned. To avoid misinterpretation, these two types are combined in the probability density function given in Table 4-12.

In addition to mechanical damage induced loss of lading, exposure to fire can lead to rupture by heating, loss of lading and rocketing tankage parts. Incidents with this result are categorized as Type II. Review of the University of Southern California report (26) and the AAR-RPI reports (28, 29) show that there were 17 incidents involving 49 LPG tank cars during the period of 1965-1970. These accidents can be classified as shown in Table 4-13.

The probability of a directly occurring Type II accident due to mechanical damage is .117 (from Table 4-12). The probability of a Type II event occurring due to a fire is 0.796 (from Table 4-13). Most Type II LPG tank car accidents (10 out of 12 in years 1965-1970) are caused by fires from other LPG tank cars. These other tank car fires were the result of Type II or Type III accidents which occur with a probability of 0.473 (0.117 + 0.356 from Table 4-12). The overall probability that a fire in one LPG tank car will result in a Type II event in a second LPG tank car is therefore:

0.473 x -2 x 0.796 = 0.314 12 4-24 NUS CORPORATION

TABLE 4-11

SUMMARY

OF MECHANICAL DAMAGE INDUCED LOSS OF LADING ACCIDENT SEVERITY Severity By Category Types Total Accident Number Data Period I II III IV* V of Cars Source 1965 thru 1970 0 2 20 2 26 50 25, 28, 29 1971 0 3 4 0 6 13 Note 1 1972 1 1 1 0 10 13 Note 1 1973 0 5 8 0 8 21 Note 1 1974 2 3 10 1 12 28 Note 1 1975 0 4 6 1 5 16 Note 1 1976 0 0 2 0 5 7 Note 2 1977 0 1 3 0 11 15 Note 2 TOTALS 3 19 54 4 83 163

• Where accident reports did not have a positive identification of fire size, a worst case assignment of Type III was made.

Note I - From Federal Railroad Administration Unpublished Summary List of Tank-Car Accidents - Hazardous Materials Section.

Note 2 - Data source Hazardous Material Incident Reports, MTB, DOT.

4-25 NUS CORPORATION

TABLE 4-12 RAILROAD - COMPRESSED FLAMMABLE GASES

SUMMARY

OF RESULTS - LOSS OF LADING CAUSED BY MECHANICAL DAMAGE Results Number of Probability of Event (i)

Classification Events Given Spill (loss of lading)

I Explosion 3 0.018 (Detonation)

II Heated Tank 19 .117 Violent Rupture III & IV Uncontrollable Fire( 1 ) 58 .356 and Controllable Fire V Spill 83 .509 TOTAL 163 1.00 (1) Accident reports did not have good indication of fire size, therefore, unknown fire sizes were classified as Type III events.

4-26 NUS CORPORATION

TABLE 4-13 LOSS OF LADING CAUSED BY FIRE Type Number Frequency of Occurrence I 0 0.0 II 39 0.796 III 2 0.041 IV 7 0.143 V 1 0.020 Total 49 1.00 4-27 NUS CORPORATION

Therefore the total probability of Type II event due to puncture accident of an LPG tank car is 0.431 (0.117 + 0.314). For 124 LPG rail shipments per year the annual probability for Type II occurrence due to mechanical puncture of an LPG car per AT&SF track mile is 2.75 x 10-6 (0.514 x 10-7 x 124 x 0.431).

In addition to the probability of an LPG tank car fire, there is the probability that a train fire will be initiated by means of other than a puncture of an LPG tank car.

WASH 1238 (30) states that fire occurs in about 1.5% of all train accidents.

Conservatively assuming that none of these fires are caused by LPG tank punctures, the probability of a fire per train accident is 1.5 x 10- 2 . The average accident rate of AT&SF trains is 3.7 x 10- 6 accidents per train mile (see Section 4.2.1). Using 62 LPG trains per year carrying 2 LPG cars the annual probability of a train fire is 3.44 x 10- 6 (3.7 x 10-6 x 62 x 1.5 x 10-2) per track mile.

Using an average train length of 70 cars and 10 cars involved in the accident (30) with the probability that a non-LPG tank car will cause a Type II rupture (2/12 = 0.167) and the probability that either or both of the LPG cars are involved in the accident, the annual probability for a Type II rupture due to non-LPG tank car induced fire is 6.5 x 10 8 (3.44 x 10-6 x 10/70 x 0.796 x 0.167) per track mile.

The combined annual probability for Type II rupture from all causes is therefore:

2.75 x 10-6 + 6.5 x 10 = 2.82 x 10 per mile per year 4.2.4 Explosives, Accident Rates and Severity of Accident.

Reference 27 estimated that there were 1.98 x 107 explosive train-miles per year based on statistics for a 57 year period from 1917 to 1973. The annual average train miles during this same period was 1.36 x 10 . During this 57 year period there were 35 explosions involving in-transit shipments of explosive. The national probability of an explosion due to a train accident involving explosives is 3.1 x 108 explosions per explosive train mile. The accident rate for the Santa Fe Railroad is significantly less than the national average and therefore using the ratio of Santa 4-28 NUS CORPORATION

Fe Railroad accident rate to the national railroad rate, the explosions per explosive train mile for the track in the vicinity of the SONGS site is:

3.70 x 10-6 x 3.10 x 108 = 1.05 x 108 Explosions per 10.95 x 10-6 Explosive Train Mile Reference 27 also determined a significance factor to account for those accidents which did not yield a significant explosive overpressure. This significance factor is 0.154 yielding a final explosive overpressure rate of:

1.05 x 10- x .154 1.62 x 109 Significant Explosions per Explosive Train Mile 4-29 NUS CORPORATION

5.0 ANALYSES AND RESULTS 5.1 General The Darameters used in the analysis of overpressure, flammable vapor cloud and missile hazards from highway and rail transport of hazardous materials are summarized in Tables 5-1 and 5-2. The basis for the accident rate parameters is discussed in detail in Section 4. The atmospheric dispersion parameters of Table 5 2 correspond to five percentile worst case conditions. These were used for all accidents even though they occur only infrequently.

5.2 LPG on the Atchison Liquid Petroleum Gas is transported on both Interstate - 5 and in Table 5-3.

Topeka and Santa Fe Railroad. The results of the analyses are given The missile impact probability is based on a maximum truck missile range of 1654 (37) feet as observed for LPG at Eagle Pass and 4900 feet for rail tank car missiles as observed for ethylene oxide(29). In most cases, the range of rocketing tank car average of one fragments have ranges less than 1000 feet(29). In both cases, an shipments significant missile per Type II or BLEVE accident was assumed. For I-5 of other compressed liquified gases (ammonia and chlorine) were included to determine total probability.

The sensitivity of the 1-5 LPG overpressure and flammable cloud probabilities to a number of the analytical parameters is indicated in Table 5-4. The latter case considered a drifting cloud ignition probability of the form:

P(d) = 1 - ( 1 - P)d/di I

where:

P(d) = cummulative probability of ignition at distance, d P = probability of ignition at an ignition source

- 0.1 d.1 = average distance between ignition sources

= 20 meters 5-1 NUS CORPORATION

TABLE 5-1 COMMODITY DEPENDENT INPUT PARAMETERS Highway Hydrogen Hydrogen Hydrogen Railroad LPG LNG Liquid Gas-1 Gas-2 Acetylene Exolosives LPG Explosives Number of Annual Shipments 2200 420 52 260 24 52 5923 124 8 Accidents per Loaded Truck Mile 0.132 x 106 0.132 x 106 0.132 x 106 .0952 x 106 .0952 x 106 .0952 x 10 6 .0952 x106 (1 Probability of Spill Given Accident .064 .064 .064 .105 .105 .105 .0514x 043 Probability Explosion Given Spill .042 0 .042 .042 .042 .042 .0184 Probability Fire Given Spill .458 0.5 .458 .458 .45R .458 - .473 Probability of No Explosion or Fire .500 0.5 .500 .500 .500 .500 .509 Probability of BLEVE 0.125 - - - - - .117 Vapor Cloud Probability of Ignition 4 Figure 4-6 Figure 4-6 Vapor Cloud Probability Fire Given Ignition .895 1.0 .895 .895 .895 .895 .895 Vapor Cloud Probability Explosion Given Ignition .105 0.0 .105 .105 .105 .105 .105 Maximum Shipment Quantity 10,000 gal. 9,200 gal. R,500 gal. 16,425 ft 3gas 114,000 ft3 3300ft3 <11,400 lb. 30,000 20,548 lbs.

Flash Fraction .352 0.1 0.1 1.0 1.0 1.0 - .352 (2)

Quantity of Vapor - Lbs. 14,665 3185 502 92 640 241 43,994 Equivalent Wt. of TN 2)- Lbs. 15,877 3321 1555 286 1982 262 47,631 Lower Flammability Limit - % - 2.1 5.0 4.0 4.0 4.0 2.5 2.1 Spill or Shipment Size Distribution Table 4-5 100% 100% 100% 100% 100% Table 5-6 Table 4-10 100%

1. Probability of significant explosion per explosive train mile

2. For maximum quantity spilled

3. 592 includes planned reduction In traffic after 1980. See text.

5-2

TABLE 5-2 OVERPRESSURE AND FLAMMABLE VAPOR CLOUD INPUT PARAMETERS Stability Class - G Wind Speed - 1.5 m/sec at 10 m height Initial Dilution - Air/Gas = 3/1 Minimum Cloud Height - I m.

Ambient Temperature - 780F Temperature of the Gas - Saturation Temperature at 14.7 psia for Liquified Gas

- Ambient Temperature for Compressed Gas Wind Direction Frequency - FSAR Table 2.3-22 Explosive Equivalent Yield - 10%

Based on Energy Region II Radius - 370 feet Peak Reflected Overpressure - 3 psi 5-3 NUS CORPORATION

TABLE 5-3 RESULTS - LPG 1-5 ATSF Overpressure Greater Than 3 psi Length of Route in Region I 0.297 0.505 (for maximum spill quantity) - miles Annual Probability due to Explosions 0.23 x 10-6 0.058 x 10-6 at Accident Site Total Length of Route Which Can 2.16 3.07 Affect Plant (for maximum spill quantity) - Miles Annual Probability Due to Drifting 0.28 x 10- 0.17 x 106 Cloud Explosions Total Annual Probability of 0.51 x 10-6 0.23 x 10-6 Exceeding 3 psi Flammable Vapor Cloud At Plant Annual Probability of Flammable 0.089 x 10-6 0.035 x 10-6 Cloud At Plant Missile Impact Length of Route Within Range of 0.71 1.96 Plant -Miles Average Probability of Impact Given 0.075 0.0071 A Missile Annual Plant Impact Probability 0.17 x 10-6 0.039 x 10-6 0

5-4 NUS CORPORATION

TABLE 5-4 SENSITIVITY STUDY - LPG ON I-5 Probability of Flammable Vapor Cloud Being Probability of Exceeding at the Plant 3 psi - per year - per year Basic Case* 0.51 x 10- 6 0.089 X 10-6 Initial Dilution - 1/1 0.51 x 10-6 0.089 x 10-6 Vertical Dispersion - 0.51 x 10- 6 0.092 x 10-6 0.5 Class G Drifting Vapor Cloud 0.76 x 10- 6 0.089 x 10-6 Explosion Probability - 0.2 Flash Fraction - 0.5 0.62 x 10-6 0.099 x 10-6 Ignition Probability - 0.49 x 10-6 0.36 x 10-6 10% per 20 meters

• The parameters for basic case are Initial Dilution - 3/1, Vertical Dispersion Class G a z-, Drifting Vapor Cloud Explosion Probability - 0.1, Flash Fraction -0.352, Ignition Probability - curve shown in figure 4-6.

5-5 NUS CORPORATION

5.3 Other Hazardous Gases Results of the detailed analysis for LNG, liquid hydrogen, acetylene and gaseous hydrogen are summarized in Table 5-5. Note that in accordance with references 31, 32, 33, and 34, the generation of a significant overpressure from an unconfined methane - air ignition from the LNG spill is considered not credible.

5.4 Explosives As stated in reference 35, the recent annual number of shipments of the military explosives past the site is 1411 by highway and 8 by rail. Reference 36 further states that changes in shipment routes and requirements for 911 of the current shipments will occur after 1980 so that there will be less than 10% of the 911 shipments in the future. Assuming that the remaining 500 shipments (1411 - 911) are unaffected yields a projected I-5 military explosive shipment frequencey of 500

+ 0.1 x 911 or 592.

(35)

The maximum net explosive weight is 20548 lbs by rail. For the highway shipments 994 of the 1411 shipments were further divided by weight (ref. 36, 45, 46, 47, 48, 49) as shown in Table 5-6. The overpressure to the plant resulting from small shipments less than 3400 lb. of TNT is below 3 psi.

Utilizing this information and the accident rates of sections 4.1.5 and 4.2.3 yields annual probabilities of exceeding 3.0 psi due to shipments of explosives of 0.020 x

-6 an6.09x1 10- and 0.0049 x 10- for highway and rail respectively. Note that all rail shipments were assumed to be of maximum size.

5.5 Flammable Liquids In addition to materials discussed above which (except for the solid explosives) have boiling points below normal atmospheric temperature, there are a number of materials shipped on 1-5 which are liquids at normal conditions. If a spill occurs, these materials will run out onto the ground and evaporate. The rate of evaporation will be mass transfer limited and will be rather slow.

5-6 NUS CORPORATION

TABLE 5-5 OTHER HIGHWAY RESULTS Probability of Flammable VaporCloudBeingSwept Probability of Exceeding Into Plant 3 psi - per year - per year LNG 0 0.010 x 10-6 6 0.002 x 10-6 Liquid Hydrogen 0.00 1 x 10-Compressed Hydrogen - 1 0.002 x 10-6 0.012 x 10-6 Compressed Hydrogen - 2 0.0008 x 10- 6 0.001 x 10-6 Acetylene 0.0004 x 10-6 0.002 x 10-6 5-7 NUS CORPORATION

Table 5-6 HIGHWAY MILITARY EXPLOSIVE SHIPMENT SIZE DISTRIBUTION Size Distribution for Net Explosive Highway Shipment of Weight (lbs) Shipments Explosives Frequency

( 3400 938 .944 3400-4400 11 .011 4400-5400 10 .010 5400-6400 7 .007 6400-7400 8 .008 7400-8400 6 .006 8400-9400 5 .005 9400-10400 4 .004 10400-11400 5 .005 994 5-8 NUS CORPORATION

The nature of the fire or explosion resulting from a flammable liquid spill is dependent on the chemical and physical properties of the materials. The chemicals are liquids at ambient temperature and pressure, and, in general, they have low vapor pressures and high vapor densities. Thus, the vapor formed tends to hug the ground, and only a thin vapor interface exists between the air and the liquid.

Therefore, spilled fuels are unlikely to produce an explosion with a strong blast wave but will produce a simple flash-over flame igniting the remainder of the fuel(38). All of the gasoline spills recorded in reference 3 burned in this manner.

In the Canvey Island Study(39), the only hazard of hydrocarbon liquids evaluated was the potential flow of burning liquids to populated areas.

0 The toxic hazard from these shipments has been evaluated in NUS-1942(4 ). This evaluation included an assessment of the evaporation rate following a spill. The reference 40 evaporation rate model was combined with an area source dispersion model to determine the maximum downwind distance a flammable cloud could travel and the amount of vapor which could be in the flammable range.

The dispersion model assumed the spill was divided into a series of strip and that the vapor source rate of each strip was dispersed as a point source using standard Guassian dispersion. The total concentration at the centerline is the summation of the contribution from all strips.

A review of the materials shipped indicates that gasoline has the highest vapor pressure and therefore the highest evaporation rate. The next most volatile liquid shipped is acetone. For gasoline, the downwind distance to the lower flammable limit is estimated to be less than 460 feet compared to the distance to the plant from the highway of 540 feet. This calculation assumes Class G stability, 1.43 m/sec wind speed and 1000F temperature for evaporation. For acetone, the distance to the LFL is less than 120 feet.

As discussed previously because of its high vapor density ignition of a gasoline vapor cloud would not create an overpressure. Reference 41 considered the hazards of a gasoline truck explosion (that resulting from a contained reaction of a stoichiometric mixture of air and gasoline in the truck) and concluded the distance to an overpressure of 3 psi was approximately 60 feet.

g 5-9 NUS CORPORATION

If it is assumed that an acetone vapor cloud could explode (acetone vapor density is half of that of gasoline) the TNT equivalent of the amount of vapor in the flammable range using a 10% energy yield is less than 50 pounds of TNT. The distance to a 3 psi overpressure is 125 feet from the cloud center and less than 200 feet from the accident.

From the above, it is concluded that because of their low vapor pressures, flammable liquids shipped past San Onofre do not contribute to the overpressure or flammable cloud hazards at the plant.

5-10 NUS CORPORATION

REFERENCES

1. TM 5 - 1300, "Structures to Resist the Effects of Accidental Explosions,"

Department of the Army, the Navy, and the Air Force, June, 1969.

2. Geiger, W., "Generation and Propogation of Pressure Waves Due To Unconfined Chemical Explosions and Their Impact on Nuclear Power Plant Structures," Nuclear Engineering and Design 27 (1974) 189-198.

3. Strehlow, R.A., 1973, "Unconfined Vapor-Cloud Explosions - An Over view," invited review paper, Colloquium on Fire and Explosion, in Four teenth Symposium (International) on Combustion, pp 1189-1200, The Combustion Institute, Pittsburgh, Pennsylvania, c 1973.

4. Strehlow, R. A. and Baker, W. E., "The Characterization and Evaluation of Accidental Explosions," report prepared for the National Aeronautics and Space Administration, Washington, D.C., by the University of Illinois, Urbana, Illinois, Report No. NASA CR-134779, 1975.

5. Eichler, T. V., Napadensky, H. S., "Accidental Vapor Phase Explosions on Transportation Routes Near Nuclear Plants," prepared for Argonne National Laboratory, Final Report J6405, April 1977.

6. Lannoy, A., Gobert, T., "Analysis of Accidents in Petroleum Industry Determination of TNT Equivalent for Hydrocarbons," J 10/8.

7. Butragueno, Jose L., and Costello, James F., "Safe Stand-Off Distances for Pressurized Hydrocarbons," Abstract.

8. Slade, David H., "Meteorology and Atomic Energy," U.S. Atomic Energy Commission, July 1968.

9. Van Ulden, A. P., "On the Spreading of a Heavy Gas Released Near the Ground," Royal Netherlands Meteorological Institute, De Bilt, Nether lands.

10. Hardee, H. C., and Lee, D. 0., "Expansion of Clouds from Pressurized Liquids," Sandia Laboratories, Albuquerque, New Mexico, SAND74-5210.

11. Heidzeberg, "Loss Prevention Symposium," European Fed. Chemical Indus tries, September 1977.

12. James, G. B., "Fire Protection in the Chemical Industry," National Fire Protection Association Quarterly, p. 256, Volume 41, 1947-48.

13. Simmons, J. A., "Risk Assessment of Storage and Transport of Liquefied Natural Gas and LP-Gas," National Technical Information Service, PB-247 415, November 1974.

NUS CORPORATION

14. "11971-1972 Accidents of Large Motor Carriers of Property," U.S. Depart ment of Transportation, Federal Highway Administration, Bureau of Motor Carrier Safety, May 1974.

15. "11973 Accidents of Motor Carriers of Property," U.S. Department of Transportation, Federal Highway Administration, Bureau of Motor Carrier Safety, July 1975.

16. Little, Arthur D., Inc., "A Modal Economic and Safety Analysis of the Transportation of Hazardous Substances in Bulk," report prepared for The U.S. Department of Commerce, Maritime Administration, Office of Domestic Shipping, Washington, D.C., Report No. COM-74-1 1272, 1974.

17. Discussion with Anthony H. Lasseigne, Hazardous Materials Division, National Transportation Safety Board, January 19, 1979.

18. Telephone discussion with Bob Davis, owner of building at Davis Used Cars, Eagles Pass, Texas, February 15, 1979.

19. Telephone conversation with Joe Bastalleros, owner of Borden Diesel Services, February 16, 1979.

20. Telephone conversation with William Miller (313-732-9255) Police Officer, 00 Flint, Michigan. First officer arriving at accident site (approximately 5 min. after ignition). Filed accident report, February 16, 1979.

21. Telephone conversation with Chief Phillips, Detroit, Michigan, Fire De partment, February 14, 1979.

22. Telephone conversation with Officer Robert Grohola, Detroit, Michigan, Police Department, February 6, 1979.

23. "11969 Accidents of Large Motor Carriers of Property," Bureau of Motor Carrier Safety, Federal Highway Administration, U.S. Department of Transportation, December 1970.

24. "1970 Accidents of Large Motor Carriers of Property," U.S. Department of Transportation, Federal Highway Administration, Bureau of Motor Carrier Safety, March 1972.

25. "Final Phase 02 Report on Accident Review," Railroad Tank Car Safety Research and Test Project, Association of American Railroad and Railway Progress Institute, Report No. RA-02-2-18, August 1972.

26. Jones, G. P., et al., "Risk Analysis in Hazardous Materials Transporta tion," report by the University of Southern California Institute of Aero space Safety and Management for the Department of Transportation, NTIS Report No. PWn-230 810, March 1973.

NUS CORPORATION

27. Illinois Institute of Technology Research Institute, "An Evaluation of the Sensitivity of TNT and the Probability of a TNT Explosion on the GM&O Railroad Line Past the Braidwood Reactor Site," in Amendment No. 8 to the Application for Construction Permits and Operating Licenses for Byron Units I and 2 and Braidwood Units I and 2, NRC Dockets STN 50 454, STN 50-455, STN 50-456 and STN 50-457, Commonwealth Edison, October 17, 1974.

28. "Phase 02 Report on Dollar Loss Due to Exposure of Loaded Tank Cars to Fire - 1965 thru 1970," Association of American Railroads and Railway Progress Institute, Report No. RA-02-1-10, August 11, 1972.

29. "Final Phase - 01 Report on Summary of Ruptured Tank Cars Involved in Past Accidents," Association of American Railroads and Railway Progress Institute, Report No. RA-01-2-7, July 1972.

30. "Environmental Survey of Transportation of Radioactive Materials to and from Nuclear Power Plant," WASH-1238, U.S. Atomic Energy Commision, Directorate of Regulatory Standards, December 1972.

31. Foster, 3. C., Jr., et. al., "Detonatability of Some Natural Gas-Air Mixtures," Air Force Armament Laboratory, AFATL-TR-74-80, November 1970.

32. Kogarko, S.M., et. al., "An Investigation of Spherical Detonations of Gas Mixtures," International Chemical Engineering, Vol. 6, No.3, July 1966.

33. Loesch, F. C., "Thermal Radiation and Overpressure from Instantaneous LNG Release into the Atmosphere," TRW Report No. TRW-08072-4, April 26, 1968.

34. "Safety Evaluation Report, Tennessee Valley Authority, Hartsville Nu clear Plants A and B," U.S. Nuclear Regulatory Commission, NUREG 0014.

35. Letter to Mr. D. F. Bunch, Chief, Accident Analysis Branch, Office of Nuclear Reactor Regulation, Nuclear Regulatory Commission from De partment of the.Army, Washington, D.C.

36. Letter to Mr. Duane F. Martin, Southern California Edison Company from Department of the Navy, Seal Beach, California, November 20, 1978.

37. "National Transportation Safety Board," Highway Accident Report, NTSB-HAR-76-4.

38. Personal communication with Dr. Roger Strehlow, Department of Aero nautical and Astronautical Engineering, University of Illinois, Urbana, Illinois.

39. "CANVEY - An Investigation of Potential Hazards from Operations in the Canvey Island/Thurrock Area".

NUS CORPORATION

40. "Determination of Acceptable Site Specific Frequency of Hazardous Chemical Shipments Passing San Onofre Units 2, 3,", Prepared for Southern California Edison, NUS - 1942, NUS Corporation, September 1977.

41. Iotti, R.C., Krotiuk, W. 3., DeBoisblanc, D.R., "Hazards to Nuclear Plants from on (or near) Site Gaseous Explosions," Abstract, Topical Meeting on Water-Reactor Safety, CONF-730304, March 26-28, 1973.

42. Glasstone, Samuel, Editor, "The Effects of Nuclear Weapons, Revised Edition, United States Atomic Energy Commission, April 1962.

43. "Safety Effectiveness Evaluation", Analysis of Proceedings of the National Transportation Safety Board into Derailments and Hazardous Materials April 4-6, 1978, Report Number: NTSB-SEE-78-2, National Transportation Safety Board.

44. U. S. Department of Transportation, Federal Railroad Administration, Office of Safety, "Accident/Incident Bulletin No. 146, Calendar Year 1977" August 1978.

45. Letter to Mr. D. F. Martin, Southern California Edison Company from Department of Navy, Seal Beach, California, April 5, 1979.

46. Letter from Naval Air Station North Island, San Diego, California.

47. Letter to Southern California Edison Company from Naval Weapons Station, Concord, California, March 13, 1979.

48. Letter to Southern California Edison Company from Naval Plant Repre sentative, Naval Industrial Reserve Ordinance Office, Pomona, California, March 6, 1979.

49. Letter to Southern California Edison Company from U.S. Marine Corps, Traffic Division, Camp Pendleton, California, March 14, 1979.

NUS CORPORATION