A computer simulation of a derailment accident Part II - sample simulation

Journal of Hazardous Materials, 25 (1990) 149-165 Elsevier Science Publishers B.V., Amsterdam

A COMPUTER

SIMULATION

PART II -SAMPLE

149

OF A DERAILMENT

ACCIDENT

SIMULATION

A. M. Birk’ R. J. Anderson’ A. J. Coppens 1. 2.

Department of Mechanical Engineering Queen’s University, Kingston, Ontario, Canada W.R. Davis Engineering, Ottawa, Ontario, Canada

ABSTRACT DERACS

(Derailment

developed to simulate involving shows

fires and accidental releases

the results

DERACS

of two sample

program.

Reference simulations

Technical

commodities.

that were performed

details of the DERACS

program

program

accidents

The

present

using the

paper

prototype

are presented

of the prototype

version

only approximate

of DERACS

the Mississauga

rail accident of November

have not been validated in any way and should

accurate representations illustrated,

of dangerous

simulations

is a computer

of train derailment

of the limitations

presented

discussion

Simulation)

in

[I].

Because results

Accident Computer

some of the consequences

purposes

of the Mississauga

accident.

only so that future applications

and areas within DERACS

requiring

further

They

the sample 1979.

The

not be considered

were generated

of the DERACS development

program

as

for could be

could be identified.

INTRODUCTION The

DERACS

derailment

computer

accidents

and toxic plumes.

where releases

The

program

valuable tool for analysis on-line tool for assisting The organization described

and training response [l].

0 -

was developed as a tool for analysis of commodities

have resulted

in fires,

of train explosions

was developed with the hope that it could become a purposes,

personnel

and the technical

in Reference

0304-3894/90/$03.50

program

DERACS

and possibly

it could become an

in the future.

basis for the various is based on a series

models

in DERACS

of sub-models,

1990 Elsevier Science Publishers B.V.

were

all of which

150

is either available in the open literature or was developed for DERACS. These sub-models

consider the following processes:

0 ii)

derailment mechanics

iii)

liquid and vapour release

iv)

flammable liquid spill growth

v)

plume and puff dispersion of heavy gases

vi)

liquid pool fires and effects on surroundings

vii)

fire impingement

viii)

blast and thermal effects of explosions/BLEVES

derailment impact induced rupture of tank cars

on tanks and thermal ruptures

The overall DERACS program consists of a pre-processor, and a post-processor. of the area surrounding

The pre-processor

a simulation module,

is used for data input which includes a map

the derailment, the derailment point, the train consist and its

condition at the point of derailment, and any pertinent atmospheric data at the time of the derailment. The simulation module carries out the time integration of the accident from the time of derailment to the time when the last tank has emptied safely or BLEVE’d (boiling liquid expanding vapour explosion). Further technical details of the various sub-models used in the simulation module can be found in References [I] and [2]. The post-processor

is used to generate a graphic summary of the derailment

accident simulation. This graphic summary shows a colour, plan view of the accident site including vehicle placement, vehicle puncture status, spills, fires, blast and thermal

m

CAUSTIC

m

PROPANE

Lzz9

S-WRENE

m

TOLUENE

0

CMLORINE

SODA

FLATTENED

METAL

I::::J BOXCAR APPROXSCALE

HYDRO EiXfENSION

Figure 1

Sketch of Mississauga

PARKS &ILDING

Derailment Scene (from Ref. [3])

151

Derailed

Car Number

Contents

1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 16 19 20 21 22 23 24

Table

1

Insulated

Punctured

Yes Yes

Yes Explosion

NO

Explosion Explosion

toluene box car caustic soda caustic soda caustic soda caustic soda chlorine propane styrene styrene styrene propane propane propane toluene toluene propane propane propane propane propane propane propane toluene

Summary

hazard distances post-processors

NO

Yes ?

of Derailed Cars and their Contents

and toxic plume hazard distances. can be found in Reference

As stated earlier, the purpose

Further

details of the pre-and

[2].

of this paper is to present

a sample simulation

the DERACS

computer

program.

The specific example considered

Mississauga

derailment

of 1979.

The results

using the prototype

THE

MISSISSAUGA

version

DERAILMENT [3], the Mississauga

IO, 1979 and involved a train consisting the 33rd car in the consist caustic soda, and styrene. determined

specifically

100 km/hr.

Figure

of 106 cars.

dangerous

The derailment

goods including

1 presents

and estimates

on November

was initiated by

chlorine,

propane, toluene,

a sketch

was not

place the speed at between 70 and

of the derailment

scene from Reference

[3].

of the derailed cars and their contents.

The 7th derailed car contained chlorine chloride

occurred

The train speed was at the time of the derailment

but witnesses

a listing

derailment

and a total of 24 cars were derailed. A total of 21 were tank

cars with 19 of these carrying

1 presents

here is the

in this paper were generated

of DERACS.

As reported in Reference

Table

presented

using

reaction inducted puncture

0.76m

and this car suffered

an impact or ferric

in diameter. As a result

of this rupture,

152

chlorine was released into the environment. Other cars sustained impact induced ruptures resulting in the release of flammable liquids. Of the derailed cars, those cars with bottom fittings (all of the type Ii 1 cars except car #24) had the fittings sheared off during the accident and their contents were released as a result. From the photographs

shown in Reference [3], an intense fire was located at or

near the 15th deraited car which contained toluene.

This and other fires were likely the

cause of the thermal ruptures of the 8th, 12th and 13th derailed cars which carried propane.

The explosions of these propane cars resulted in extensive damage to

neighbouring

properties.

At least one of the explosions occurred very early on, which

leads to the conclusion that it was an impact-induced

rupture. The explosions also

resulted in large pieces of the tanks being thrown long distances (over 2km). The above facts were used to formulate the input to the sample simulations performed using DERACS. Because of certain limitations of DERACS, some simplifying assumptions were required.

SIMULATION

INPUT

At present DERACS is capable of simulating the derailment mechanics of up to 40 cars; where 8 can be tank cars carrying chlorine, propane or hexane.

In order to

perform a simulation of the Mississauga derailment, it was necessary to extract a subset of the actual accident. This subset was selected as a reasonable representation

Figure 2

Approximate Map of Mississauga Area Created by DERACS Pre-Processor

153

of the Mississauga version

accident

while also fitting within the constraints

of the present

of DERACS.

The Mississauga approximate DERACS

derailment

occurred

map of the surrounding

pre-processor

at Mavis Road in the city of Mississauga.

area (10 km by i0 km) was entered

and is shown in Figure 2. This map does not contain

details of the area but does show the approximate inhabited

areas as well as roads and rail lines.

used to enter more detail than shown DERACS

graphics

The wind speed

was reported

atmospheric

the North West.

An assumed

With DERACS,

directly.

can be

simulation

and validation

DERACS

option,

mechanics

simulation

where the location

of vehicles

could be run to give a possible

Graphics

Key Page

from

using the derailment

it is possible

and input the vehicle placement

is needed.

coming

of 10 deg C was used.

the appropriate

of the derailment

DERACS

[3] to be 28 km/hr. This was entered

to predict the vehicle placement

the next few hours in the accident. However,

Figure 3

pre-processor

of 101.3 kPa and wind direction

air temperature

team at the site of an accident In such a situation

in Reference

By selecting

mechanics The bypass

development

of water, forest and

The DERACS

in the global view of Figure 2. Figure 3 shows the

pressure

it is possible

simulation.

derailment

all of the

key page which defines various symbols.

with an assumed

mechanics

location

An

using the

to bypass

and rupture

is practical

the

status

for a response

and punctures

is known.

view of what to expect

before this can be done, significant

for

further

154 Two simulation mechanics

cases were considered.

simulations

while the second

The first case uses the derailment bypasses

the derailment

mechanics

simulation.

SIMULATION

CASE

For Case

1

1 (derailment

mechanics

consist of 36 cars was used including in Table 2. toluene. consider

In the sample

simulation,

This was necessary chlorine,

Derailed

propane

Car Number

used to predict vehicle placement),

a train

8 tank cars.

up as shown

the commodity

because

the present

and hexane. Simulation

The consist was made

Hexane

Car Number

hexane

was substituted

version of DERACS

can only

was used as a substitute Simulation

Car

for

for toluene Protection

Tvpe

3 4 5 6 7

1 2 3 4 5 6 7 8

chlorine

insulated

8 9 10 11

9 10 11 12

propane

insulated

12 13 14 15 16 17 18 19 20 21 22 23 24

13 14 15 16 17 18 19 20 21 22 23 24

propane propane propane hexane hexane propane

1 2

25 26 27 28 29 30 31 32 33 34 35 36

Table 2

Summary of Consist Used in DERACS Mississauga Derailment

Simulation

of

155

because,

like toluene,

or toluene

hexane is a flammable

are spilled that could result

that they were filled to 90% The derailment

for certain. speeds.

to be straight

car #4 was assumed

As stated in Reference As a result

pressure.

was 10 deg C.

of Mavis Road and the rail line.

track at the derailment

to be the derailment

point.

The first truck of

initiator.

[3], the train speed at the point of derailment

a number

of simulation

runs

were performed

was not known

using

It was found that a train speed of 100 km/hr gave a reasonable

scene, with very good correlation, considered

important

If hexane

in a pool fire. For all tank cars, it was assumed

capacity and their initial temperature

point was located at the intersection

The rail line was assumed simulation

liquid at atmospheric

to the actual derailment

scene.

to be able to predict the correct number

different train derailment

In this case it was

of derailed cars, and the

total area covered by the derailed vehicles.

SIMULATION

CASE

For this simulation the placement Figure

2 case, the derailment

of vehicles

4 presents

and their puncture

the assumed

mechanics status

vehicle placement

shown

on Figure

resting

places of the derailed cars in the Mississauga

simulation

as based on Reference

1, 24 cars have been placed in positions

which approximate

accident.

tank cars among the derailed cars but, because of the present only 8 tank cars are shown

Figure

4

in Figure

4.

was bypassed

and

was input using the pre-processor.

For impact-induced

Assumed Vehicle Placement for Case 2: (derailment mechanics simiulation bypassed

There

As

the actual

were actually 21

limitations

ruptures,

{3].

of DERACS,

it was assumed

156 that car 8 (chlorine) sustained

was punctured

and cars 9 (propane),

16 and 17 (hexane)

total loss of containment.

SAMPLE

SIMULATION

As stated

above,

RESULTS two sample

simulation

of the derailment

placement

and ruptures

simulations

while the second

mechanics

as input.

were performed.

The first case included

a

case had the vehicle

The results of these two simulations

are presented

separately.

Case

1 - Simulation

of Derailment

Mechanics

This case involved the simulation simulation

of the resulting

derailment

scene

module

The important

agreement

mechanics

5 and 6 shows

module

of DERACS.

as well as the a plan view of the The DERAIL

shows that 24 cars have left the

with the actual Mississauga is approximately

derailment.

140 m which is also in very

two of the propane

it was predicted

cars

and one

that three impact

of the

hexane

cars.

punctures This

Figure 6.

Figure 5

The

with the actual accident.

As a result of the simulation, including

Consequences

[3].

result to note is that the prediction

total length of the track involved

good agreement

Figures

by the DERAIL

in detail in Reference

track which is in excellent predicted

of the derailment

consequences.

as predicted

is described

and Resulting

Predicted

Vehicle

Placement

for Case

1: (zoom

level 2)

resulted

is indicated

in

157

Figure

6

Predicted

Vehicle

Note that this simulation very significant results

The

dispersion assumptions [I].

present

problem

version

explosion

of bottom fittings

accident. The puncture

of DERACS

handles

will result.

consequences,

a total loss

The simulation

main simulation

program

in heavy gas

For further

details

of the basic

the reader is directed to Reference

of containment Following

result

a leak of propane as a heavy gas

then it is assumed

did not predict the puncture

in the actual accident.

which had a

of the hexane car

liquid and the propane car punctures

used in determining

the DERACS

(zoom level 3)

and not as a local fire or explosion.

If a propane tank suffers

which occurred

for Case 1:

cannot account for the loss

effect in the Mississauga

in a spill of flammable

plumes.

Placement

the derailment

was used and the results

that an

of the chlorine mechanics

car

simulations,

can be summarized

follows:

The hexane Two plumes

spill resulted

issued

from the punctured

the fire slowly burned fire-impinged were predicted.

in a fire that engulfed propane

one unpunctured cars.

As the simulation

itself out and the pool fire diameter

tanks were no longer

impinged

The heavy gas plumes

propane

decreased

progressed,

in size until the

by fire. As a result, no thermal

dispersed

downwind.

car.

ruptures

as

158

For this simulation the most significant event was the release of the propane gas as both an explosion and toxic hazard.

Figures 7 to 11 show the derailment scene at

various times during the simulation.

Case 2 - input of Vehicle Placement and Simulation of Resulting Consequences In this sample simulation, the vehicle placement was provided as input along with the location of punctures.

Figure 4 shows the assumed placement and the assumed

punctures were described in an earlier section. The DERACS simulation software was used to simulate the consequences

of the assumed initial conditions.

The results of

this simulation can be summarized as follows. The tota! loss of containment of the propane car resulted in an explosion and fire ball. Because the loss of containment occurred early on in the accident,

the propane

lading had not reached a high enough temperature for a BLEVE to be predicted. As a result, the total loss of containment resulted in a comparatively small explosion and fire ball. The resulting thermal and blast loads resulted in the destruction of nearby buildings as indicated by the hazard zones. The puncture of the chlorine car resulted in the release of chlorine gas into the environment.

This plume drifted downwind for some distance at lethal concentrations.

The total loss of containment of the hexane car resulted in a large pool fife that initially

Figure 7

Case 1 Simulation Results at 0.3 min. (zoom level 3)

159

Figure 8

Case 1 Simulation Results at 3.3 min. (zoom level 2)

Figure 9

Case 1 Simulation Results at 3.3 min. (zoom level 1)

160

Figure 10

Case

1 Simulation

Results at 9.7 min. (zoom

Figure 11

Case

1 Simulation

Results at 20.7 min. (zoom level 1)

level 1)

161

engulfed

or partially

pool fire diameter were reduced.

engulfed

decreased

However

cars experienced area. Figures

several

propane

cars.

As the fire burned

and slowly the number

at 24 minutes

a BLEVE.

of tank cars impinged

into the simulation,

This BLEVE caused

72 to 16 present

views of the

itself out, the

thermal

simulation

by fire

one of the uninsulated

and blast loads at different

tank

to a large

times.

DISCUSSION The previous DERACS.

examples

As expected,

considering

the limitations

interesting.

The derailment

total number

show some of the capabilities

there are differences of the present mechanics

This

However,

the predicted final resting

Reasons

was a function

in less

interaction

for these

The derailment

mechanics

are very

was very accurate at predicting

the

differed considerably

routines

and this

of the simulation.

development

were not very successful

observed

and validation.

at predicting

in the Mississauga

derailment.

and no predicted thermal

mechanics.

routines

routines

It should

require further

the

in predicted consequences

that

For example,

ruptures

for the

be noted that the present

are very crude such that they could run quickly

The impact rupture mechanics

further

and this resulted

of chlorine

the derailment

routines

were not in very good

The actual cars were spread out farther

model requires

from those

including

impact rupture

train speed at the time of the derailment.

place of the vehicles

of punctures

there was no predicted release

machine.

simulation

the results

of

However,

are not clear at this time and it is evident that the

simulation

location or the numbers

derailment

of DERACS,

between the cars during the remainder

discrepancies

mechanics

simulation

version

of the assumed

with the actual accident.

would result derailment

version

of cars derailed and the length of the track region affected by the derailed

vehicles. agreement

of the prototype

between reality and simulation.

development

if more accurate and realistic

results

on the IBM/AT

as do the are to be obtained

in the future. When the vehicle placement simulation

was more realistic.

as input naturally

results

vehicle placement

since adding more information

it should

were input.

All of these variables

be recalled that

No data pertaining were handled using

only the to release the

sub-models.

It should accounting

also be remembered for the small numbers

is developed,

that only 8 tank cars were included of spills,

fires,

explosions

it will be able to handle a larger number

events occurring number

were entered as input the

result

However,

of punctures

were input.

locations

is an obvious

in better output.

and locations

rates, or pool fire sizes DERACS

and puncture This

in the simulated

of possible

more commodities

accidents

vehicle interactions.

As DERACS

of tank cars and the number

will increase

Future

and additional tank types.

in this simulation,

and releases.

versions

rapidly because of DERACS

of

of the larger

could also include

162

Figure 12

Case 2 Simuiati3n

Results at 0.3 min. (zoom level 3)

Figure 13

Case 2 Simulation

Results at 0.3 min. (zoom level 2)

163

Figure

14

Figure 15

Case

2 Simulation

Case 2 Simulation

Results at 0.7 min. (zoom

level 2)

Results at 5.7 min. (zoom level 2)

164

Figure

16

Case 2 Simulation

Results

at 24.7 min. (zoom level 2)

COt&L”SIONS A prototype The program

derailment

accident computer

has been used to simulate

The derailment

mechanics

derailed and the approximate accurately result

the Mississauga development

has been developed.

parts of the Mississauga

train derailment.

track area affected by the derailment.

simulation

of vehicles It was not able to

or the location of impact-induced

of the accident consequences

accident. However, is required

program

module was able to predict the number

predict vehicle placement

the remaining

simulation

the results

if accurate predictions

ruptures.

As a

did not agree well with

were not unrealistic.

Further

of vehicle p!acement and rupture

are to

obtained in the future. As a predictive routines

tool, DERACS

were bypassed.

the actual Mississauga

The

gave better results

resulting

consequences

accident. The simulation

an explosion

of a ruptured

car resulting

in a BLEVE.

results

number

of this development

As a minimum, simulator

included a toxic chlorine

by the present

in creating accident scenarios or future accident response

of response

rupture

cloud,

of a propane

version

of DERACS.

concept is feasible.

accident simulator. team training.

Such

a

It could also

which could be used to test response

expert systems.

with

than the actual accident

work indicate that the DERACS

can be used as a derailment

mechanics

were in close agreement

of events was smaller

could be valuable for the purpose

prove useful procedures

DERACS

results

propane car, pool fires and a thermal The

because only 8 tank cars could be considered The

when the derailment

165

REFERENCES 1.

A Computer Simulation of a Derailment Accident: Part A.M. Birk Engineering,

2.

The Development

I - Model Basis. W.R. Davis Engineering Ltd.,

October 1988.

of a Derailment Accident Computer Simulation Model, Transport Canada Report TP

9254E. March 1988. 3.

Report of the Mississauga Railway Accident Inquiry. Hon. Mr. Justice S. G. M. Grange, Supreme Court of Ontario, December

1980.