(navigation image)
Home American Libraries | Canadian Libraries | Universal Library | Community Texts | Project Gutenberg | Children's Library | Biodiversity Heritage Library | Additional Collections
Search: Advanced Search
Anonymous User (login or join us)
Upload
See other formats

Full text of "Philadelphia Gas Works natural gas pipeline rupture, explosion, and fire, Philadelphia, Pennsylvania, May 11, 1979"







NATIONAL 



TRANSPORTATION 

SAFETY 

BOARD 




^7YBO^ 



WASHINGTON, D.C. 20594 



PIPELINE ACCIDENT REPORT 

PHILADELPHIA GAS WORKS 
NATURAL GAS PIPELINE RUPTURE, 
EXPLOSION, AND FIRE 
PHILADELPHIA, PENNSYLVANIA 
MAY 11, 1979 



For Reference 
Do Not Take NTSB-PAR-79-3 

From the Library 




UNITED STATES GOVERNMENT 



^L^t*^ / 



TECHNICAL REPORT DOCUMENTATION PAGE 






1 . Report No. 

NTSB-PAR-79-3 



2. Government Accession No. 



k. Title and Subt i tie Pipeline Accident Report- 
Philadelphia Gas Works Natural Gas Pipeline 
Rupture, Explosion, and Fire, Philadelphia, 
Pennsylvania, May 11, 1979 



3 .Reci pient ' s Catalog No. 



5. Report Date 
September 27. 1979 



6. Performing Organization 
Code 



8. Performing Organization 
Report No. 



7. Author(s) 



9. Performing Organization Name and Address 

National Transportation Safety Board 
Bureau of Accident Investigation 
Washington, D.C. 20594 



10. Work Unit No. 
?7^3 



1. Contract or Grant No. 



1 2. Sponsor i ng Agency Name and Address 

NATIONAL TRANSPORTATION SAFETY BOARD 
Washington, D. C. 20594 



13-Type of Report and 
Period Covered 

Pipeline Accident Report 
May 11, 1979 



1 '4 . Sponsor i ng Agency Code 



1 5 . Suppl ementary Notes 



16. Abstract 

At 3:05 p.m., e.s.t., on May 11, 1979, two almost simultaneous explosions and an 
ensuing fire destroyed three buildings near the intersection of Tacony and Margaret 
Streets in Philadelphia, Pennsylvania. Seven persons, including a Philadelphia Gas Works 
(PGW) employee, were killed, 19 persons were injured, and several adjacent rowhouses 
were damaged. The explosion also caused a section of Margaret Street to cave in, 
exposing a large cavern under the paved surface. 

The National Transportation Safety Board determines that the probable cause of the 
accident was the sagging and breaking of an 8-inch, cast-iron gas main, due to the 
undetected erosion of the soil support under it, resulting in the migration of leaking gas 
into adjacent buildings where it was ignited by an undetermined source. 



17. Key Words 

Cast-iron gas main, low-pressure gas, 8-inch 

pipe, 6-inch water main, road cave-in, 

evacuation, liaison with fire department, 

prompt gas shutoff, "greased off mains and services. 



19-Security Classification 
(of this report) 
UNCLASSIFIED 



20. Security Classification 
(of this page) 
UNCLASSIFIED 



lo.Di stri but ion Statement 
This document is available 
to the public through the 
National Technical Information 
Service, Springfield, Virginia 
22151 



21 .No. of Pages 



2A 



22. Price 



NTSB Form 1765.2 (Rev. 9/74) 



CONTENTS 

SYNOPSIS . '. 1 

INVESTIGATION 1 

The Accident 1 

Injuries to Persons 6 

Damage to Pipeline 6 

Other Damage 6 

Pipeline System 8 

Meteorological Information 8 

Fire 8 

Medical and Pathological Information 14 

Survival Aspects 14 

Tests and Research 14 

Other Information 15 

ANALYSIS 18 

CONCLUSIONS 19 

Findings 19 

Probable Cause 20 

RECOMMENDATIONS 20 

Appendix A - Investigation 21 

Appendix B - Calculation of gas loss 22 



NATIONAL TRANSPORTATION SAFETY BOARD 
WASHINGTON, D.C. 20594 

PIPELINE ACCIDENT REPORT 

Adopted: September 27, 1979 

PHILADELPHIA GAS WORKS 

NATURAL GAS PIPELINE 

RUPTURE, EXPLOSION, AND FIRE, 

PHILADELPHIA, PENNSYLVANIA 

MAY 11, 1979 

SYNOPSIS 

At 3:05 p.m., e.s.t., on May 11, 1979, two almost simultaneous explosions and 
an ensuing fire destroyed three buildings near the intersection of Tacony and 
Margaret Streets in Philadelphia, Pennsylvania. Seven persons, including a 
Philadelphia Gas Works (PGW) employee, were killed, 19 persons were injured, and 
several adjacent rowhouses were damaged. The explosion also caused a section of 
Margaret Street to cave in, exposing a large cavern under the paved surface. 

The National Transportation Safety Board determines that the probable cause 
of the accident was the sagging and breaking of an 8-inch, cast-iron gas main, due 
to the undetected erosion of the soil support under it, resulting in the migration of 
leaking gas into adjacent buildings where it was ignited by an undetermined source. 

INVESTIGATION 

The Accident 

From 8:20 a.m. to 2:25 p.m., on May 11, 1979, a PGW serviceman was 
routinely replacing gas meters at buildings near the intersection of Margaret and 
Tacony Streets in Philadelphia. During part of this time, the serviceman's company 
vehicle was parked in front of 2302 Margaret Street. (See figure 1.) The 
serviceman crossed the intersection many times, but he did not detect any gas 
odors and no one reported any gas odor to him. The serviceman left the area after 
completing his work. 

At 2:30 p.m., a gas pressure recorder which was connected to the gas system 
that ran under Margaret Street, and located 1 1/2 blocks from the intersection, 
indicated that the pressure had dropped from 6 to 3.6 inches of water column. (See 
figure 2.) This information was telemetered into the PGW distribution department 
dispatcher office where an alarm sounded. A pressure specialist was dispatched to 
the location 1 1/2 blocks from the accident site. He checked the pressure chart 
which indicated a failure somewhere on the gas main. The chart showed the 
pressure drop in inches of water column and the time of the drop in pressure. The 
pressure-recording chart was found to be 7 minutes slow, which meant that the 
indicated 2:23 p.m. pressure drop actually occurred at 2:30 p.m. While the 
specialist was checking out the equipment, the explosion occurred. 



- 2 - 




- 3 - 



4 PM 



2 PM 



12 N 



2:30 P.M. 



i2:23 P.M. 



INCHES OF 
WATER COLUMN 



2 Mil I 3 I II 4| I 5 



(CLOCK 7 MINUTES SLOW) 



Figure 2. Pressure-recording strip chart. 



- 4 



About 2:45 p.m., a bus driver stopped his bus on Margaret Street near the 
intersection to discharge passengers. As he opened the bus doors he detected a 
heavy odor of gas that caused his eyes to water. He closed the doors, drove the bus 
from the intersection, and stopped in the middle of the next block. He left the bus, 
alerted some area residents, and asked if someone would call the gas company. 
The driver then returned to the bus and drove out of the area. 

At 2:47 p.m., the telephone service section of PGW received the first 
telephone caU reporting a strong odor of gas in the 4700 block of Tacony Street 
near the intersection. This report was transmitted immediately to the PGW 
customer service department. Another PGW serviceman, who had been working 9 
to 10 blocks from the site, was contacted by radio at 2:54 p.m. and ordered to 
investigate the report. By 3 p.m., the serviceman arrived at the intersection, 
parked his truck across from the tavern on Tacony Street and hurried to take a gas 
reading in a manhole located in the southwest corner of the intersection. (See 
figure 1.) Immediately after taking the reading, he looked north and saw two PGW 
supervisors arriving on Tacony Street. He ran to meet them and reported a 100- 
percent lower explosive limit (LEL) reading on his combustible gas indicator (CGI). 
The supervisors told him to check the houses for gas leaks. Following the 
instruction, he entered the tavern and seconds later the building exploded. 

The explosion, followed rapidly by a smaller explosion, occurred at 3:05 p.m., 
18 minutes after PGW was first notified of the gas odor. It destroyed the building 
containing the tavern and an apartment; an adjacent rowhouse; and a garage behind 
the buildings. All of the buildings caught fire. The roof of the building that 
contained the tavern was blown up and off, slid into the street, and hung over a 12- 
foot-high pile of debris that partially covered the sidewalks. A section of Margaret 
Street beside the destroyed buildings caved in from the weight of the debris. 

Immediately after the explosions, the PGW personnel began alerting and 
evacuating persons in the area. Heavy gas concentrations were evident in some of 
the evacuated buildings. PGW personnel broke windows and opened doors in nearby 
buildings to free the gas. 

At 3:07 p.m., the Philadelphia Fire Department (PFD) received the first fire 
alarm followed by a second alarm at 3:14 p.m. Units of the PFD arrived at the site 
at 3:12 p.m., and immediately began fighting the fire. (See figure 3.) PGW 
personnel discussed the situation with the firefighters and planned the method of 
extinguishing the fire. The firefighters were careful not to flood the basement of 
the tavern because of the possibility of drowning anyone trapped in the debris; this 
action delayed extinguishing the fire. To control the gas-fed fire, the PGW 
crewmen started injecting grease 1/ into the gas main and service lines at 3:40 
p.m., and at 4:50 p.m. the flow of natural gas at the intersection was stopped and 
the fire went out. 



1/ A procedure where grease is injected into the gas service lines or gas mains to 
shut off the flow of gas in low-pressure systems. "Greasing off" service lines and 
mains is a faster process of stopping gas flow than the inflated bag method and is 
used generally in emergencies. After the emergency, depending on the amount of 
grease used, the affected gas main or services are abandoned, replaced, or blown 
clean. The inflated bag method takes longer to accomplish, but it does not leave 
any residues in the lines afterwards. 



5 - 




- 6 - 



During this time, the rubble from the destroyed buildings was removed by a 
crane and deposited in the hole in Margaret Street. This was done rapidly in an 
attempt not to cover the entrance to the tavern and to allow rescuers to reach any 
persons still alive in the building. The weight of the rubble caused more of the 
road to collapse; the hole eventually became 44 feet long, 20 feet wide, and 10 feet 
deep. (See figure 1.) The cavern was approximately 4,780 cubic feet in volume. 

At 5:12 p.m., PGW personnel opened the sewer manhole at the intersection of 
Melrose and Margaret Streets and found that a large volume of water without mud 
or debris was flowing through the sewer. The water clarity and composition 
indicated that the sewer was intact; no sewer wall breaks were visible. 

At 7:30 p.m., while the search for survivors continued, PGW customer service 
crews continued the gas odor survey of all houses near the accident site. At 9:45 
p.m., the PGW distribution crew completed cutting and capping all natural gas 
mains leading to the intersection. Six "greased off" locations were capped. (See 
figure 4.) 

At 12:01 a.m., on May 12, 1979, the PGW crews completed the installation of 
the 2-inch bypasses necessary for the restoration of gas service to the area 
residents. Gas service to 57 customers had been interrupted as a result of the 
accident. 

By the next morning seven bodies had been removed from the rubble. Six of 
these persons, including the PGW serviceman who had been checking for gas odors, 
had been trapped in the tavern at 4701 Tacony Street. The seventh person was 
crushed by debris outside the tavern. No one was in the rowhouse at 4703 Tacony 
Street at the time of the accident. 

Injuries to Persons 

Operating 
Injuries Personnel Other 

Fatal 1 6 

Nonfatal 19 

Damage to Pipeline 

When the hole was cleared of debris and the water was pumped out, investi- 
gators found that the 8-inch, cast-iron gas main below the street was broken in 
three places. (See figure 5— Profile.) 

Other Damage 

A 6-inch, cast-iron water main in the hole was broken in five places. (See 
figure 6.) The water main had been installed in 1941 with 4 feet 3 inches of cover. 
The deepest break in the water main occurred where the main crossed above an 8- 
inch sewer lateral from a rowhouse. At this point one end of the broken main had 
dropped 6 feet below its original horizontal position, and had severed the 8-inch 
sewer lateral connection with the sewer main. (See figures 6 and 7.) The oval- 
shaped, 3-foot by 2-foot brick sewer was very old, but was found to be intact in the 



- 7 - 




O 
.1 



ai- 



> 




CO 


z 




t- 


o 




_ "J 


u 




Oa: 


^^: 




>-< 


cm"- 


OOC 


o u> 
Oh- 


-J 

z 

UJ 


UJCM°- 
t/5 ^ O 


z < 


"H 


-.fMO 




UJ 


DC .■* 




OC 


lOH 




o 


l-Z< 



o 
u 



area. A television camera inserted into this sewer and moved along its interior 
confirmed its integrity. Several of the 8-inch sewer laterals were later found to be 
broken off where they entered the sewer main; however, this was not detected by 
the television camera. 

The explosions and fire destroyed the two buildings at 4701 and 4703 Tacony 
Street and a garage. Thirty other dwellings and several vehicles were damaged by 
flying debris; window breakage was widespread within a 1-block radius of the 
accident. The explosion was powerful enough to blow a heavy metal cellar door 
that had been located on the sidewalk in front of the tavern, 80 feet diagonally 
across the intersection. 

Pipeline System 

The 8-inch, cast-iron gas main pipe with bell and spigot joints was installed in 
1899 in 12- and 16-foot lengths and was buried with approximately 3 feet of cover. 
The joints were sealed with cement and jute. There are no records of the pipe 
specifications. The main was operating at an average pressure of 6 inches of water 
column (about 1/4 psig) when the accident occurred. This type of pipe, with 
unreinforced bell and spigot joints is limited by Federal regulation (49 CFR 192) to 
an operating pressure of 25 psig. There were no pressure test records for this pipe. 
Most of the original gas service lines were tapped directly into the main. There 
had been no reports of breaks or leaks in the main in this area before this accident. 
The pressure recorded on the chart had been constant. At the accident site, the 
low-pressure, integrated, multiflow gas system was fed by mains from four 
directions. 

In the last 5 years, 215 pressure regulator alarms were activated on the PGW 
system. However, not aU of these alarms indicated pipe failures; some represented 
vandalism, some equipment malfunctions, some electrical failures, and some 
telephone line problems. 

PGW is a municipal gas utility wholly owned by the city of Philadelphia and 
operated by the Philadelphia Facilities Management Corporation, a nonprofit 
corporation recognized and existing under the laws of Pennsylvania. PGW has a 
policy of assisting fire departments at the scene of a fire or explosion involving 
natural gas. PGW personnel worked with the PFD at the site of this accident to 
coordinate the evacuation and ventilation of nearby buildings. 

Meteorological Information 

At the time of the accident it was raining and the winds were lightly blowing 
from the east. The temperature was in the mid 80's. 

Fire 

The fire was intense, fueled by escaping natural gas and the dry timber of the 
old buildings. Firefighters did not use aU of the fire hoses available to extinguish 
the fire immediately because they did not want to cause anyone trapped in the 
rubble to be drowned. The fire was not extinguished until 4:50 p.m. when the flow 
of gas was shut off. 



PRC 



100 

99 

98 

97 

96 

95 

94 

93 

92 

91 



PIPE SAW CUT 



12'-11' 



0'-41/2"^ 



\ 



\ 



\ 



$ 



LENGTH 12'-ir 



5 



A. 



1.3^5' 



k 



FIGURE 5. 
PROFILE VIEW LOOKING NORTH 8"CAST IRON GAS MAIN 

ELEVATIONS TAKEN 5—12—79 



li 



\ t 



FORMER TOP OF GRADE ALONG NORTH CURB MARGARET ST. 



f^ 




PLAN VIEW OF DAMAGED GAS MAIN 




=r-^^E:^3 



MARGARET ST. 



PRO 




fe^^'f^ y ^8" SEWER 
^-"^ W^ "SEE DET 



SEE DET/ 



FIGURE 6. 
PROFILE VIEW LOOKING NORTH 6" WATER & 3' x 2' SEWER 

ELEVATIONS TAKEN 5—13—79 




x: 



)/ 8" SEWER LATERAL (DOTTED CIRCLE) 
'■ "SEE DETAIL" FIGURE 7. 



o 



3' X 2' BRICK SEWER 




- 13 







.1 

fa 



- 14 - 



Medical and Pathological Information 

All seven persons died from burns and shock. Nineteen persons were injured 
by flying debris and broken glass. The more seriously injured were taken to the 
nearest hospital and treated. 

Survival Aspects 

Efforts to rescue the seven trapped persons were hampered by the fire. At 
3:08 p.m., a man with burns over 30 percent of his body crawled out of the 
wreckage. Firefighters assisted him to an ambulance, and then cleared debris from 
the area where the victim emerged to seek others who might still be alive. The 
rubble was carefully removed from the basements of the building during the search 
for survivors. The last body was recovered at 4:15 a.m. on May 12, 1979. 

Tests and Research 

When the gas main was uncovered, the pipe to the east was lying at a slight 
angle from the horizontal. Excavation to the west exposed two breaks in the line, 
one end of this broken section was 3 feet 9 inches below its former horizontal 
position. (See figure 5— Profile.) Further excavation to the west revealed a third 
break about 11 feet west of the second break. The gas main had shifted laterally a 
total of 1.325 feet from its original position. (See figure 5— Plan View.) The 
broken main ends and segments of the water main were cut off and visually 
examined by a Safety Board metallurgist. The pipe was also sent to the National 
Bureau of Standards for further metallurgical examination. 

Metallurgical examination of the gas main revealed the following: 

1. The fracture face was coated with corrosion products, some of which 
were removed when cleaned ultrasonically for 10 minutes with a detergent. Most 
of the remaining corrosion was removed with buffered hydrochloric acid. The 
corrosion products on the fracture face appeared to be superficial, indicating that 
the fracture was recent. 

2. Very small amounts of graphitization were detected on the outside pipe 
surface next to the fracture. 

3. The 10 Rockwell K hardness (RKh) measurements that were made were 
essentially aU the same. The average RKh value was 92.2. This value is the 
approximate equivalent of a Brinell hardness number (Bhn) of 156, which is 
considered to be a reasonable value for gray cast iron. 

4. The pipe appeared to have been pit cast, not centrifugally cast. 
CentrifugaUy cast pipe was not available when the pipe was manufactured. 

5. The scanning electron microscope examination of the fracture revealed 
the primary fracture mode was "cleavage," indicating a low ductility or brittle 
fracture. This type of fracture is common in gray cast iron. 



- 15 - 



Meteillurgical examination of the 6-ineh, cast-iron water main revealed the 
following: 

1. The three fracture faces submitted were covered with corrosion 
products. The fractures were cleaned ultrasonically with a detergent followed by 
ultrasonic cleaning with buffered hydrochloric acid. Some of the corrosion product 
adhered tightly to the fracture face, and much of it remained on the surface after 
the cleaning process. 

2. Because of the difficulty in removing the corrosion product from the 
fracture and because of the corrosive attack of the fracture features, it appeared 
that the the pipe had been fractured for some time. 

3. After cleaning, part of one of the fracture face was examined by the 
scanning electron microscope. Corrosion had eliminated many of the fracture 
features, but the fracture mode appeared to be cleavage. 

4. Ten RKh measurements were taken on transverse sections through the 
pipe. The average of these 10 measurements was RKh 92.8. This value is the 
approximate equivalent of a Bhn of 159, which is considered to be a reasonable 
figure for gray cast iron. 

Other Information 

Street cave-in. — Examination of the caved-in section of the street in front 
of the destroyed buildings revealed the following: 

1. The road surface on Margaret Street was asphalt-tar over a layer of 
cobblestone or Belgian block over a concrete layer which at one time had been the 
street's surface. 

2. TroUey tracks had been installed in the cobblestone some years ago, and 
the two tracks spanned the open ditch in an east-west direction slightly south of 
the centerline of the excavation. 

3. The cavity beneath the road appeared to have been in existence for a 
considerable time before the road collapsed. The trolley track apparently had 
supported the road surface and all the vehicular traffic on the road up to the time 
of the cave-in. (See figure 8.) 

4. The position of the gas main and the water main at the east and west 
ends of the cave-in indicated that they had been hanging in the cavity unsupported 
from beneath for an indeterminate time. 

5. The condition of the sewer main was good with no breaks or leakage in 
the area excavated. 

6. Six sanitary sewer laterals were excavated and examined. These 
laterals served the rowhouses and tavern adjacent to the cave-in. In general, the 
laterals were not in good condition; three of them appeared to have been 
disconnected from the sewer main for some time. (See figure 9). The ground 
around these laterals contained sewer sludge. 



16 - 




^ c 

> q; 

<u -a 

O Q, 

> CO 

> ed 

.2 ^ 
a, -^ 



00 



- 17 




'> a) 

> «-i 

T 00 

Q) •— I 

> t- 

CO <u 

CO u> 

'■5 >^ 

is s 



5 


w 


(U 


"O 


> 


bn 


c 


eo 


S 


e 


Oh 


CO 




XJ 



18 - 



7. Westbound buses passed the intersection of Tacony and Margaret 
Streets at approximately 20-minute intervals before the explosion. 

Water pressure. — A check of the water and sewer records indicated that 
there had been no complaints of blocked sewer drain or low-water pressure from 
the immediate area residents. However, the tavern did have low water pressure 
on May 10 and May 11, 1979. A part-time bartender in the tavern and a postman, 
who were in the building the day before and the day of the explosion, said that the 
tavern was having problems with its water pressure. The bartender had notified 
the city water department. On May 11, 1979, personnel from the department 
checked the problem and left, telling the bartender that they would return. The 
explosion and fire occurred in the afternoon. 

Calculations of gas loss . — Computerized calculation of gas flow showed a 
total of 152.76 MCF 2/ of gas lost from 2:30 p.m. to 4:50 p.m. (See appendix B.) 
From 2:30 p.m., the time of the failure, until 3:05 p.m., the time of the explosion, 
a total of 50 MCF of gas entered the cavity beneath the street. It took 
approximately 3 minutes to fill the estimated 4,780-cubic-foot cavern at a rate of 
1,619 cubic feet per minute. 



ANALYSIS 



The examination of the cave-in revealed that the large cavern had been in 
existence for a long time eind had been created by a slow erosion process. The poor 
condition of the three sewer laterals and of the soil around them leads the Safety 
Board to conclude that seepage from these laterals contributed to the erosion that 
created the cavern. The openings in the sewer main caused by the broken laterals, 
allowed sewer backup during periods of heavy rain to escape and erode the 
immediate area. Water escaping from the fracture in the water main, which tests 
indicated had existed for some time, also probably contributed to the erosion. The 
break in the water line was not complete; it was probably more of a fissure which 
let water escape under pressure and helped to erode the cavity. Had the water 
pipe broken completely, water pressure to the area would have been disrupted, 
residents would have complained of no water or low water pressure, and the break 
would have been detected and repaired within a few hours. 

The 8-inch, cast-iron gas main, although installed 80 years ago, was found to 
be in good condition, without extensive corrosion, and within definition for gray 
cast iron of that era. The light corrosion on the fractured pipe face, the ease with 
which it was removed, and the lack of corrosive attack on the fractured faces of 
the pipe itself indicated a very recent break. Cast iron when exposed to the 
elements wiU often show traces of rust in a matter of hours. The pipe, in 12- and 
16-foot sections, had hung unsupported for a long time in a span until it snapped 
and broke, due to a combination of factors such as the pipe's own weight, traffic 
vibrations, pavement settling, or other unknown outside forces. No reports of gas 
odor had been received recently from this area, no gas pressure reductions had 



2/ MCF is an industry abbreviation for 1,000 feet. 



19 - 



been noticed, and the pressure-recording chart 1 1/2 blocks away had been constant 
in the recent past. Therefore, the Safety Board concludes that the gas main 
collapsed at 2:30 p.m. on May 11, 1979, as indicated by the rapid drop from 6 to 3.6 
inches water column shown on the pressure-recording chart. 

After the main broke, gas at approximately 6 inches water column pressure 
began flowing into the cavern under Margaret Street at an estimated gas flow rate 
of 1,619 cubic feet per minute. The cavity was filled with gas in approximately 3 
minutes. Once filled and unable to escape rapidly because of the paved road above, 
the gas began to build up pressure and migrate through the foundation of the tavern 
and also up through the storm drain to the road above. At approximately 2:45 p.m., 
15 minutes after the break, enough gas had gained access to the street level to 
alert the bus driver. 

The liaison between PGW and the Philadelphia Fire Department resulted in 
prompt evacuation of the area and an effective lowering of the lower explosive 
limit in the adjacent houses by ventilation. These actions prevented secondary 
explosions which could have caused additional damage and loss of life. 

A bag inserted in the gas main and inflated is a customary method of stopping 
gas flow in low-pressure systems. However, this method is time-consuming 
because it requires excavation of the main, cleaning it, installing a pipe tapping 
unit, cutting out the top, and inserting and inflating the bag. The effective 
"greasing off" of the multiflow gas system at the leak site was a better procedure 
in that it saved valuable time. 



CONCLUSIONS 



Findings 



The large cavern under Margaret Street, which contained the gas, 
sewer, and water mains had been created over a period of time by soil 



2. Seepage from the openings in the main sewer caused by the broken 
laterals contributed to the erosion that caused the cavern. 

3. Water escaping from the water main also probably contributed to the 
erosion. 

4. The water main probably had been cracked some time before the 
accident as revealed by metallurgical analysis which showed adhesive 
corrosion products on the fracture faces and corrosion of the fracture 
face itself. Complete severance of the water pipe occurred at the time 
of the gas main failure. 

5. The gas main had been undermined by the soil erosion and had been 
hanging unsupported for a long time before it broke. 



- 20 - 



6. The gas main failed at 2:30 p.m. when it broke into three pieces from 
one or a combination of factors such as the pipe's own weight, traffic 
vibrations, pavement settling, or other unknown outside forces. 

7. Liaison between PGW and the Philadelphia Fire Department resulted in 
the prompt evacuation and effective ventilation of affected houses in 
the accident area. 

8. Prompt "greasing off" of the gas mains by PGW prevented additional 
migration of natural gas to the affected area. 



Probable Cause 

The National Transportation Safety Board determines that the probable cause 
of the accident was the sagging and breaking of an 8-inch, cast-iron gas main, due 
to the undetected erosion of the soil support under it, resulting in the migration of 
leaking gas into adjacent buildings where it was ignited by an undetermined source. 

RECOMMENDATIONS 

As a result of its investigation of this accident, the National Transportation 
Safety Board recommendated that the American Gas Association: 

"Advise its member companies of the circumstances of this accident and of 
the prompt and effective coordination betwen the gas company and the fire 
department and urge them to review their emergency practices and 
procedures, particularly those concerning evacuation and liaison with fire and 
police departments to insure that coordination is planned adequately for 
similar accidents. (Class H, Priority Action) (P-79-59)" 



BY THE NATIONAL TRANSPORTATION SAFETY BOARD 

/s/ JAMES B. KING 
Chairman 

/s/ ELWOOD T. DRIVER 
Vice Chairman 

/s/ FRANCIS H. McADAMS 
Member 

is/ PATRICIA A. GOLDMAN 
Member 

/s/ G.H. PATRICK BURSLEY 
Member 

September 27, 1979 



- 21 - 



APPENDIX A 



INVESTIGATION 



The National Transportation Safety Board was notified of the accident at 
4:50 p.m. on May 11, 1979. The Safety Board immediately dispatched two 
investigators from its Washington, D.C., headquarters to the accident site. Upon 
arrival at the scene, the Safety Board investigators took charge of the 
investigation. 



22 



APPENDIX B 

Excerpts from PGW Report on the 

Calculation of Gas Loss on May 11, 1979 

from 2;30 P.M to 4:50 P.M. 

A "skeletonized" version of the low-pressure system in the vicinity of Tacony 
and Margaret Streets was modeled on a digital computer using Dr. M.A. Stoner's 
"Gas Steady State (GASSS) Network Analysis Program through time-sharing 
facilities of the National Computer Software Systems (NCSS). The Pole Low 
Pressure Gas Equation was used in the modeling of the system. 

This "skeletonized" system consisted of the principal large low-pressure 
mains transporting gas to and from the area together with the network of smaller 
mains in the immediate vicinity of Tacony and Margaret Streets. Since May 11, 
1979 was a warm day (85° F recorded at 2 p.m. at Richmond Plant) and considering 
the time of day (2 p.m.) no large gas demand such as heating or cooking would be 
expected. We, therefore, concluded that inclusion of small gas mains (6-inch and 
smaller), except in the immediate vicinity of the break, would have no meaningful 
effect on the problem. 

The first computer run simulated conditions in the system at 2:25 p.m., 
immediately before the break. Matching the actual recording gauge pressures at 
the outlet of the regulator stations and at the test points in the study area 
established the gas flow and loading in the system. 

The second computer run simulated conditions at 2:30 p.m. the time of the 
break. A load point was established at the location of the main break and the load 
at this point, which would be the same as the gas escaping from the broken main 
was calculated by the computer when all the actueil pressures were matched. 

The third computer run simulated conditions at 3:30 p.m., which was the time 
when the Kensington and Torresdale Streets regulator loaded to 7.6 inches water 
column because the control point pressure at Melrose and Orthodox Streets dropped 
to 3.9 inches water column. The flow of gas from the main break was at its 
greatest at this time and was calculated by the computer when all the actual 
pressures were matched. 

The fourth computer run simulated conditions at 3:55 p.m., immediately after 
the flow of gas from the 6-inch main on the east side of Tacony Street was stopped 
by the injection of grease north of Meirgaret Street at 3:40 p.m. and south of 
Margaret Street at 3:45 p.m. The injection of grease into the 8-inch main on 
Margaret Street east of the break point at 3:47 resulted in a partial stoppage of gas 
flow. The resistance to the flow of gas from the main break because of the grease 
injection into the 8-inch main was represented by a short length of small-diameter 
main. By varying the diameter of this main until all the actual pressures were 
matched, the computer calculated the volume of gas lost at this time. 

The fifth computer run simulated conditions at 4:07 p.m., immediately after 
grease was injected into the 8-inch main west of Tacony Street. This grease 



23 - APPENDIX B 



injection along with the previous injections in the 6-inch main on the east side of 
Tacony Street stopped virtually all gas flow from the west. The small amount of 
gas still coming from the break again was calculated by the computer in the same 
manner as the fourth computer run. 

In our judgment, there was no need for further computer runs to calculate the 
additional loss of gas between 4:07 p.m. and 4:50 p.m. when complete shutdown was 
accomplished, since this could be established from the time-flow plot of the 
compter runs made. (See figure 10.) The time-flow plot shows a total of 152.76 
MCF (1,000 cubic feet) of gas lost from 2:30 p.m. to 4:50 p.m. 



- 24 - 
FLOW MCFH 



100 



® 
® 
® 
(2) 

TOTAL GAS LOSS = 




PLOT OF GAS FLOW THRU MAIN BREAK (FLOW RATE) 

(?) TO d) NETWORK AT STEADY STATE CONDITIONS 
^ GAS LOSS AT TIME OF MAIN BREAK, 2:30 P.M. 

GAS LOSS AT PEAK FLOW, 3:30 P.M. (KENSINGTON & TORRESDALE AT 7.6 W.C.) 

REDUCED GAS LOSS DUE TO GREASE INJECTIONS, 3:55 P.M. 

FURTHER REDUCTION IN GAS LOSS - ADDITIONAL GREASE INJECTIONS, 4:05 P.M. 

GAS FLOW STOPPED - SYSTEM RETURNS TO STEADY STATE, 4:50 P.M. 

152.76 MCF = CROSS HATCHED AREA UNDER CURVE OF FLOW RATE 



5:00*^ 



Figure 10. Time-flow plot of gas loss. 





DATE DUE 











































































'iSSH^ 



TN 880.5 .U58 79-3 
Philadelphia Gas Works 
natural gas pipeline 
rupture, explosion, and 



TN 880. 5 .U58 79-3 
Philadelphia Gas Works 
natural gas pipeline 
rupture, explosion, and 



/\ NATIONAL EMERGENCY 
^ TRAINING CENTER 
LEARNING RESOURCE CENTER 
16825 SOUTH SETON AVENUE 
EMMITSBURG, MO 21727 

NETC LRC 



?3SW 



001917 



g'^SVY 



NATIONAL TRANSPORTATION SAFETY BOARD 
Washington, D.C. 20594 



Official Business 

PENALTY FOR PRIVATE USE. $300 



POSTAGE AND FEES PAID 

NATIONAL TRANSPORTATION 

SAFETY BOARD 




C) 



DT-40F 

MONTANA DAKOTA UTIL CO 

PO BOX 1066 
SHERIDAN WY 82801