1977_美国_PAR7702.pdf

8 〔图〕插图 〔图〕插图 〔图〕插图 〔图〕插图 〔图〕8\n

NATIONAL TRANSPORTATION SAFETY BOARD

WASHINGTON, D.C. 20594

PIPELINE ACCIDENT REPORT

UGI CORPORATION, NATURAL GAS EXPLOSIONS AND FIRES

ALLENTOWN, PENNSYLVANIA

AUGUST 8,1976

REPORT NUMBER: NTSB-PAR-77-2

UNITED STATES GOVERNMENT

〔图〕ORAATIONALSAFETYBOARD\n TECHNICAL REPORT DOCUMENTATION PAGE

CONTENTS

Page

SYNOPSIS 1

INVESTIGATION. 1

The Accident 1

Injuries to Persons . . . . . . . . . . . 5

Damage to Pipeline . . . . . . . . . . . . 5

Other Damage . . . . . . . . . . . . 5

Pipeline System · · · · · · · · · · 8

Meteorological Information. 8

Fire 9

Survival Aspects . . . . . . . . . . . . . . 9

Tests and Research $\dots$ 9

Other Information. . . . . . . . . . . . 10

ANALYSIS $\ldots$ 14

CONCLUSIONS . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Findings. . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . . .

Probable Cause. . . . . . . . . . . . . . . 18

APPENDIXES

Appendix A - Electro Chemical Series of Metals . . . 21

Appendix B - Metallurgical Report No. 77-1 . . . 22

Appendix C - Calculations of Cast-iron Pipe Strength. 24

NATIONAL TRANSPORTATION SAFETY BOARD WASHINGTON, D.C. 20594

PIPELINE ACCIDENT REPORT

Adopted: May 18, 1977

UGI CORPORATION

NATURAL GAS EXPLOSIONS AND FIRES

ALLENTOWN, PENNSYLVANIA

AUGUST 8, 1976

SYNOPSIS

At 2 p.m., on August 8, 1976, a gas explosion destroyed a house at 1127 Oak Street, Allentown, Pennsylvania. The gas migrated from a break in a 4-inch cast-iron main under the street through loose soil under a concrete sidewalk and into the building through cracks and openings in its foundation.

At 2:26 p.m., another house exploded across the street from the site of the first explosion. The front brick wall collapsed into the street and trapped two firemen. The street then caved in directly in front of the trapped firemen. The cast-iron gas main within the sinkhole broke into several pieces; flames from the broken main were more than 10 feet high and hampered rescue of the trapped firemen. Two firemen were killed, 14 persons were injured, 4 buildings were destroyed, and several buildings were damaged.

The National Transportation Safety Board determines that the probable cause of the accident was the ignition, by unknown sources, of natural gas which had leaked from a broken cast-iron gas main under the street in front of the houses. The gas main had been weakened structurally by localized graphitization and was undermined by a sinkhole. The failure to check the second house for an explosive atmosphere and to shut off its gas supply resulted in two fatalities and numerous injuries.

INVESTIGATION

The Accident

At 2 p.m., on August 8, 1976, an explosion destroyed a house at 1127 Oak Street, Allentown, Pennsylvania. The exterior walls of the large frame house were blown out into the street and yard and the rafters fell into a second-floor room and trapped a man. (See figure 1.) Firemen from the Allentown Fire Department arrived at 2:04 p.m. and rescued the man. Other firemen, who had been securing an area about a mile away where a malfunctioning district gas regulator had been venting a large amount of gas into the atmosphere, also responded. The firemen then began fighting small fires that had erupted in the debris left by the explosion.

〔图〕2oo 112701Dooto1370HXNS\n

Figure 1. Destroyed house at 1127. Oak Street.

• 3 -

At 2:10 p.m., a UGI Corporation (UGI) gas company serviceman and his supervisor arrived at the scene after repairing the malfunctioning district regulator. The supervisor told the serviceman to shut off the gas to all houses at the meters and to evacuate persons in the 1100 block of Oak Street. The supervisor returned to his vehicle to radio the gas company for more help.

At 2:15 p.m., firemen entered the basement of the burning house and shut off the gas flow at the meters to the house's two apartments. The firemen saw flames coming from cracks and openings in the concrete and stone foundation wall of the building; this indicated that gas was still flowing freely into the building through underground pathways from breaks in either the gas service line or main located outside of the foundation. Sewer drains in the basement floor, a possible means of gas entrance to the building, were not checked. Firemen and policemen began to evacuate the residents of the 1100 block of Oak Street.

The firemen told the gas serviceman that they would shut off the meters in nearby houses. The serviceman then began to shut off meters and ventilate buildings down the block from the destroyed house. No record was made of meters that had been shut off or of buildings that had been entered for evacuation purposes and to check for accumulations of gas. The sidewalks outside of buildings were not marked with paint or chalk to indicate where meters had been shut off. No odors of gas were detected in the basements that the firemen or serviceman entered to shut off the gas at the meters. Combustible gas indicators (CGI) were not used to check the building atmospheres for gas; they were on vehicles parked some distance away.

The meters at 1128 and 1130 Oak Street were not shut off however, and the pilot lights on appliances in the houses remained burning.

While the residents were being evacuated and the meters were being shut off, several firemen contained the fire in a small area near the front foundation wall of the destroyed house. The firemen were allowing the fire to burn until all the gas supply was consumed because they had been taught that "burning gas will not explode." 1/

At 2:26 p.m., an explosion occurred at 1130 Oak Street, directly across the street from the site of the first explosion. Several firemen who were standing within 5 to 15 feet of the two-story brick rowhouse were seriously injured by flying debris. The front wall of the building fell and trapped two firemen under brick rubble. At the same time, a 4-foot by 10-foot area in the center of the street caved in and 10-foot-high flames leaped from the opening. (See figure 2.)

.-4-

〔图〕1132 OAK STCOLLAPSED LATERE1130 OAK ST.OAKSTREETSINKHOLE1135FIREMEN TRAPPEDOAK STREETUNDER RUBBLE25-F00T-WIDEPAVEMENTS1n

Figure 2. View of accident site as seen from Amanda Street.

-5-

At 2:30 p.m., a third fire alarm was sounded for more help to combat new fires and to rescue the many injured firemen. After the injured firemen were taken to the hospital, police and firemen evacuated the remaining residents in the area. Gas company servicemen shut off 94 gas services and approximately 600 people were evacuated from their homes.

At 4:43 p.m., a gas company construction crew excavated and capped the ends of a 4-inch, cast-iron, low-pressure gas main in Oak Street. All of the gas-fed fires that had been allowed to burn were then extinguished. Service to ll customers was interrupted for several days until the extent of the failures on the gas main could be determined.

At 10:30 p.m., a break in a 3-inch cast-iron water main caused water service to be shut off in the 1100 block of Oak Street.

Injuries to Persons

Only one person was in the house at 1127 Oak Street when it exploded. Although he was pinned in bed by fallen rafters and was slightly bruised, he was not seriously injured.

Two firemen died and four firemen were seriously injured as a result of the second house explosion. One of the dead had been buried under the brick wall and the other had been severely cut and burned; both were near the gas fire that was burning in the opening in the street. (See figure 3.) Nine other firemen were injured.

Damage to Pipeline

The 4-inch cast-iron gas main was found to have five complete circumferential breaks within 39 feet. Two broken 5-foot-long sections within the cave-in had dropped 2 feet. (See figure 4.) The main was damaged beyond repair and was replaced with a coated and wrapped steel main.

Other Damage

The explosions completely destroyed the houses at 1127 and 1130 Oak Street. The rowhouse at 1128 Oak Street was damaged extensively and was later razed. The building supports for 1132 Oak Street were weakened by the explosion and the building fell of its own weight later in the day. Walls in adjoining rowhouses at 1126 and 1134 Oak Street cracked, and flying debris damaged other buildings. Window breakage was widespread within a one-half block radius.

〔图〕dOd.233GAS MAIN CUT & CAPPED107 INWVA3C.I.WATERPARTIAL DAMAGE4"C.I.GAS1126OAK STCRACKED WATER MAIN2'DEEPEXTENSIVELY DAMA

Figure 3. Plan of accident site.

-77

〔图〕1127 |OAK STREETV31V H3MS..92#SOH 0A83S1331S#1127'WATER SERVICEGAS MAINBREAKTO HOUSEA1V9ST4"SPLIT SLEEVEINSTALLED 2/15/3 〔图〕100PLAN VIEW9'10"-6'LENGTH OF SINK HOLETOP OF CURBSTREET.SURFACEv9v/CBRICK.8/G9EPLUG39 3/4..07RUBBLE643/4"COPPER WATER S

Figure 4. Plan view and side view of sinkhole and damaged utilities.

• 8 -

A 1/4-inch-wide crack was found on the bottom of a 3-inch, cement-joint, cast-iron water main in Oak Street. The water main was installed in 1896 with 3 feet of cover. It had never broken previously.

Part of a bell joint of the 8-inch vitrified-clay sewer line in Oak Street was broken. The bell joint had dropped 3 1/2 inches below the spigot end. The sewer was installed with 7 feet of cover in 1931. (See figure 4.)

Pipeline System

The 4-inch cast-iron gas main with bell and spigot joints was installed in 1915 in 12-foot lengths with 3 feet of cover. Cement and jute were used to seal the joints. There are no records of its specifications. The 1975 American National Standards code for Gas Transmission and Distribution Piping Systems indicates that a 4-inch, cast-iron, pitcast pipe with a 0.40-inch wall thickness for transporting gas can withstand a working pressure of 150 pounds per square inch, gage (psig). However, 49 CFR 192 limits the allowable operating pressure to 25 psig because of unreinforced bell and spigot joints. UGI company operating practice further limits the maximum allowable operating pressure (MAOP) to 1 psig. There are no pressure test records for this line.

Most of the original steel gas service lines were tapped directly into the cast-iron main every 10 to 20 feet without the use of insulating unions. Because steel is anodic to cast-iron, the steel service lines cathodically protected the cast-iron main. (See appendix A.)

During construction of the sewer line in 1931, the gas main, which was located 2 1/2 feet from the sewer centerline, was struck and broken at a point 15 feet east of the nearest break caused by this accident. (See figure 4.) Two years later, the gas main was broken at a point 22 feet west of the 1931 break and near the breaks revealed by this accident. The break was presumably caused by settlement over the sewer line but could also have developed because of a sinkhole. The two breaks were repaired using cast-iron, cement-joint split sleeves. These were the only breaks or leaks in this cast-iron gas main reported before this accident.

Meteorological Information

The weather at 2 p.m. was cloudy and mild with a temperature of 71° F. There had been recent periods of extremely heavy rainfall. More than 1 inch of rain fell the day before the explosion. Rain continued intermittently on the day of the accident and eventually accumulated to more than 2 inches. Some roads were flooded and a flash flood warning was issued on the day following the accident.

• 9 -

Fire

Small, non-gas-fed fires that resulted from the explosions were easily contained and extinguished. However, the gas-fed fires in the street were not extinguished because firemen did not want to leave unignited gas that could later explode.

Survival Aspects

Efforts to rescue the two firemen trapped under the bricks were hampered by the fire at the cave-in. It took about 5 minutes to rescue one fireman who was seriously injured. The other fireman, who was dead on arrival at a hospital, had been completely buried under the bricks. The hands and faces of rescue workers were burned as they moved the hot bricks to reach the firemen.

Tests and Research

On August 10, 1976, the gas main was carefully excavated. A 5-foot-long segment of the main was located under 3 feet of brick rubble and 19 inches of wet clay. It had sheared off at the location of a plugged 1 1/4-inch tap and had dropped straight down 9 1/2 inches.

The outlines of the sinkhole were indistinguishable at the 4-foot depth. However, a cavity at least 8 feet in length had existed under the gas main before the street collapsed on it. The two ends of the main on each side of the sinkhole were 11 feet apart and appeared to be supported by an undisturbed clay foundation. However, inspection of the cover over the ends revealed that one end was 8 inches lower than the other.

When the main was uncovered in each direction from the sinkhole, two more fractures were discovered. The pipe to the east of the sinkhole was laying at an angle and one end of a 7 1/2-foot-long broken segment was 7 inches lower than the other end. (See figure 4.) It was found that this piece had also been undermined, and that the distance between the two points on the main that were still in undisturbed soil was approximately 18 feet.

At the east end of the 7 1/2-foot-long piece of broken main, a 3/4-inch copper water service line to 1127 Oak Street was found to be in contact with the cast-iron gas main and a 3/4-inch-wide break was found in the main. The copper line had been installed at some time to replace the original lead water service line. A black metallic scale was found on the bottom of the 3/4-inch copper line where it crossed and was in contact with the cast-iron main. This scaling indicated that the cast-iron main was cathodically protecting the copper line. (See appendix A.) There was no evidence to indicate that the positions of any of the pipe at this location shifted after the initial break.

• 10 -

While uncovering the gas main, two l-foot-diameter sinkholes were found under the main. (See figure 3.) One was 2 feet deep and the other was 4 feet deep.

Metallurgical consultants for the city of Allentown visually examined five pieces of cast-iron pipe that were removed from the accident site. However, the fracture faces of the pieces were allowed to corrode before they could be analyzed with a scanning electron microscope.

The Safety Board's metallurgical examination indicated that the gas main was pit-cast pipe (cast vertically with dry sand molds and cores) with wall thicknesses varying from a minimum of 0.33 inch to a maximum of 0.50 inch at the fractures. The average thickness was 0.41 inch.

The Safety Board's hardness measurements, microstructure examination, and X-ray energy dispersive analysis indicated that the pipe material was typical for a gray cast-iron pipe. Most of the fractures appeared to be typical of overload breaks in gray cast-iron pipe and appeared to have occurred at approximately the same time.

Small amounts of graphitization (corrosion) were found on the gas main in several places. The only area of heavy graphitization was where the 3/4-inch copper water service line to 1127 Oak Street touched the top of the gas main. At this fracture, a porous structure of iron oxide and carbon extended 0.15 inch into the 0.38-inch-thick wall of the main; this reduced the effective wall thickness by 40 percent. (See figure B3 of appendix B.)

Other Information

Events Preceding the Accident. -- About 12:30 p.m., on August 8, 1976, Allentown firemen and UGI gas service personnel were called to 19th and Tilghman Streets where natural gas was coming out of a vent line. They found that an underground district gas regulator vault had flooded from the heavy rainfall and that the regulator had failed.

An automatic relief valve, which was located downstream of the regulator and which was set to operate when the pressure increased to 15 inches water column (w.c.) $\frac{2}{7}$, opened as a result of the regulator failure and vented most of the gas to the atmosphere through the vent line. However, the increased pressure did cause the six adjoining district regulators to stop supplying gas into the system. The adjoining regulators had been set to feed gas into the system at 7 inches w.c. If the pressure exceeded this, the regulators would automatically close until the pressure returned to 7 inches w.c. The pressure recorded at a district regulator at Plum and Liberty Streets, 10 blocks from the malfunctioning regulator,

2/ One pound per square inch (psi) is equivalent to a column of water 27.71 inches high at 60° F. Low-pressure gas is normally distributed at 7 inches water column or one-quarter pound (4 ounces).

• 11 -

increased to about 9 inches w.c. (See figure 5.) The accident site was all blocks from the malfunctioning regulator and 3 blocks from Plum and Liberty Streets.

UGI Emergency Plan and Liaison with Fire Officials. -- The UGI written emergency plan states that personnel responding to an emergency should act immediately to protect the public from injury. The senior employee should identify himself to fire department personnel on the scene and review with them the assistance that the gas company can provide. In the case of an explosion, the first UGI employee on the scene should assess the situation and report this to UGI area headquarters. Emergency personnel would then be sent to the scene by the area headquarters. The emergency plan does not assign emergency responsibilities to gas service-men and does not indicate what tools and equipment, including CGI's, a serviceman is required to carry when responding to emergencies. UGI's emergency plan did not provide adequate guidelines to promptly isolate leaking gas mains in low-pressure systems.

UGI maintains liaison with fire and police officials and met with them earlier in 1976. The company published a guide for fighting a gas facility fire, and a course based on this information was attended in April 1971 by many of the firemen that responded to this emergency. The firemen were told: "Burning gas will not explode," and "If natural gas is burning outside, the firemen should make no attempt to extinguish a gas fire." 3/

The training course did not include field exercises such as checking buildings for gas accumulations with a CGI, or the extinguishing of natural gas fires with chemical (dry powder) fire extinguishers.

Sinkholes. -- Sinkholes are depressions of the ground surface caused by karst development. Karst is an area where earth materials have undergone solution erosion, usually in limestone, dolomite, gypsum, or salt.

All States have some karst features, but the Great Valley region of Pennsylvania is one of four areas of major karst development in the United States. A limestone formation from 1. to 4 miles wide and about 1,500 feet deep underlies Allentown. The Allentown formation generally consists of dense bluish-gray-to-gray magnesian limstones (dolomite) in beds of from 6 to 18 inches thick. The dolomite and interbedded limestone can be found from the surface to 100 feet deep with the average depth being 20 feet.

Rainfall contributes to karst development. Allentown has moderate rainfall of approximately 48 inches per year. Most of the limestone under Allentown is so thick that it cannot be penetrated by water. However, when water percolating through the soil encounters the moderately

• 12 -

MALFUNCTIONING REGULATOR AT 19th & TILGHMAN STREETS REPAIRED

GAS MAIN BREAK CREATED EXPLOSION AT 1127 OAK ST.

〔图〕NOON8/8/76GAS MAIN BREAKSUNDAAT. TIME OF12y391236EXPLOSION AT1130 OAK ST6SATURDAY3MONIDAY123~GRAPIIC COETRGIS COAIUNINCH

Figure 5. Gas pressure chart, in inches water column, from regulator vault at Plum and Liberty Streets.

• 13 -

soluble limestone rock near the surface, it makes channels through the thinly bedded layers of limestone. When underground openings are formed under a thin limestone bed, the highly fractured and eroded "roof" often collapses and a "collapse sink" occurs.

A collapse sink occurs instantaneously and exhibits steep-sided and abruptly descending folds. Topographically, a sinkhole is a depression that varies in depth from a few feet to 30 feet and varies in area from a few square yards to several acres. The most common sinkhole form is funnel-shaped and broadly open upward, but collapse sinks may have different configurations.

Most sinkholes in Allentown are of the doline variety, i.e., they are round and are formed slowly downward by solution beneath a soil mantle without rock disturbance. There may be hundreds of doline sinkholes over a square mile area in a karst region.

Because limestone underlies only 15 percent of the continental United States, sinkholes are considered a local problem and have not generated national interest.

History of Broken Gas Mains in Allentown. -- Because this accident had many similarities to past accidents, the following incidents at nearby locations are listed. All of these accidents were not definitely attributed to sinkholes or cast-iron gas mains. However, since almost all of the gas mains in Allentown are cast-iron, it is likely that the gas mains involved in these accidents were cast-iron.

(1) On October 13, 1925, at 7th and Chew Streets, a gas explosion destroyed a residence, killed three persons, and injured two persons.

(2) On December 2, 1949, at 8th and Liberty Streets, a 10-foot-deep and 50-foot-long crater caused a gas main to fracture, which caused an explosion and fire.

(3) On December 13, 1958, the Mountainville Inn was destroyed by a gas explosion. Seven persons were killed and 22 persons were injured while they watched firemen fight flames from a broken cast-iron gas main in the nearby street. (The break was later attributed to damage by construction equipment.)

(4) On December 13, 1968, a gas main in the 700 block of North Railroad Street was broken by a street cave-in. Fifteen homes were evacuated as a precaution.

• 14 -

ANALYSIS

The pressure increase of from 2 to 8 inches w.c. in the distribution system that occurred when the district regulator at 19th and Tilghman Streets malfunctioned did not cause any material failures or create interruption of gas service to any customers. A new 4-inch, cast-iron, pit-cast pipe with a 0.40-inch wall thickness could have withstood an internal hoop stress more than 300 times greater than that exerted when the district regulator malfunctioned.

Although the slight pressure increase could not fracture the gas main, it did, however, "back-off" adjoining district regulators and kept them from supplying gas into the distribution system.

The pressure-recording chart taken from the regulator at Plum and Liberty Streets (see figure 5) illustrates what time the events took place that affected the gas main at the accident site 3 blocks away. After the regulator at 19th and Tilghman Streets was repaired, the pressure became static in the distribution system and leveled off at approximately 8 inches w.c. The first sudden drop of 0.3 inch w.c. indicates that the first gas main break occurred shortly after 1:30 p.m., about 30 minutes before the first explosion.

The other four breaks occurred shortly before 2:30 p.m. and are shown on the pressure-recording chart by the sudden 0.7-inch pressure drop. This drop, which caused the district regulator to start supplying gas again, indicated massive main breaks such as would have occurred when the street caved in after the second explosion at 2:26 p.m.

The gas main failure was caused by the presence of the ll-foot-long and 8-foot-deep sinkhole. The 3 inches of rain and a rising water table probably caused some sloughing off of the roof of the sinkhole. This greatly enlarged the sinkhole and undermined the gas main and the sewer line which were probably over a water channel opening in the limestone.

The Safety Board calculated the strength of a 4-inch cast-iron main with a 0.40-inch wall thickness and found that such a main can span a void of 11 feet without breaking, if the main is well supported by soil at each end. Excavation of the main involved in this accident revealed that the distance between two points on the main that were still in undisturbed soil was approximately 18 feet. The calculations showed that such a main cannot span this distance without breaking. (See appendix C.)

In this case, the main sagged over the sinkhole and broke where it had been weakened by graphitization where the water service line to 1127 Oak Street contacted the main. The graphitization resulted from a galvanic cell between the two dissimilar metals which was created with

• 16 -

With the main completely broken at one point and with the inadequate soil support, the main became cantilevered over the sinkhole. In this position it could only withstand one-sixth as much loading as a main that was intact and supported at both ends. The shock of the explosions and the weight of the bricks striking the pavement 3 feet away from the heavily stressed main caused the cave-in which resulted in the other four main breaks.

It is difficult to detect the presence of sinkholes underground and if detected, there is no standard method of repairing such voids. The Safety Board analyzed underground detection equipment and reported its findings in a 1973 Special Study. 5/ Because antennas used with the "downward-looking" (ground-penetrating) radar equipment were designed to be more sensitive to detecting voids that have a longitudinal axis, such as pipes and sewer tiles, it has been concluded that vertical axis sinkhole voids, especially those underneath a signal-attenuating gas main, would be difficult to detect with commercially available equipment.

Radar equipment is now available for the detection of underground utilities. Modification of this equipment might enable it to detect the presence of sinkholes under pavement. Using this type of equipment to periodically monitor sinkhole enlargement would allow the pipeline operator to take preventive action before a situation becomes critical. The use of such equipment in Allentown should be considered together with UGI's planned use of other experimental methods, such as acoustical emission, for detecting other sinkholes.

This accident emphasizes the need to develop a downward-looking, vehicle-mounted radar unit that can detect sinkholes under pavement and other limestone cavities. The Safety Board believes that such equipment can be perfected with existing technology.

The police, firemen, and gas serviceman did an excellent job of evacuating nearby buildings and can be credited with saving many lives. However, the field liaison between the firemen and the gas serviceman lacked a clear assignment of responsibilities. Several persons gave orders but no one recorded whether or not the orders were carried out. In the confusion, the gas supplies to 1128 and 1130 Oak Streets were not shut off. This oversight left pilot lights burning on appliances in the buildings and thereby provided the presumed source of ignition for the second explosion. If a positive field procedure, such as the marking of sidewalks, had been used to indicate where meters had been shut off, none of the buildings would have been overlooked.

Although UGI maintains liaison with the city, the gas fire training program for Allentown firemen has not been updated in 5 years. The UGI Guide for Fire Fighters' statement, "Burning gas will not explode," is

• 17

misleading. It would be more accurate to state that "Burning gas will not explode but, if the source of the gas is underground, do not assume that all escaping gas is being consumed by the fire. Check for gas accumulations in nearby buildings and sewers."

The guide's instruction, "If natural gas is burning outside, the firemen should make no attempt to extinguish a gas fire" was followed by the firemen at this accident even though the presence of the gas-fed fire at the cave-in hampered rescue of the firemen trapped nearby. The guide should add to its instruction: "However, if life is in jeopardy, the gas fire should be extinguished with a dry chemical and water-fogging equipment

The firemen's and gas serviceman's lack of use of combustible gas indicators also indicates a lack of proper training. These devices, when used in the vicinity of a gas leak, provide a positive check to insure safety of emergency personnel.

The Safety Board has often emphasized the need for better emergency training and increased liaison between gas companies and fire departments. Additional requirements, which could have prevented some of the injuries and fatalities in this accident, have been added to 49 Code of Federal Regulations 192.615. Unfortunately, these requirements did not become effective until October 1, 1976, 2 months after this accident, and cannot be fully implemented for some time. However, the Safety Board is pleased to note the new requirements and will continue to make recommendations to gas companies which, if also implemented by other gas companies with similar systems, will further improve gas industry safety.

CONCLUSIONS

Findings

1. An unusually heavy rainfall of 3 inches in 2 days caused a sinkhole under the gas main to enlarge.

1. The first break in the gas main occurred more than 45 minutes before the other four breaks, but all were caused by settlement of the gas main over the sinkhole.

1. The main first broke where it had been weakened by graphitization where a 3/4-inch, copper water service line to 1127 Oak Street contacted the main. The graphitization resulted from a galvanic action between the two dissimilar metals.

1. Breaks in the sewer and water lines were caused by the sinkhole.

1. Continued gas supplies to 1128 and 1130 Oak Streets left possible sources of ignition for the second explosion.

• 18 -

1. Neither the gas company personnel nor firemen used combustible gas indicators at the accident site and the atmosphere in the second house that exploded had not been checked for a possible explosive mixture. Specific training in and conscientious use of combustible gas indicators might have prevented the second explosion.

1. The trapped firemen and rescue personnel might have been injured less if the gas fire in the street had been extinguished with dry chemicals.

Probable Cause

The National Transportation Safety Board determines that the probable cause of the accident was the ignition, by unknown sources, of natural gas which had leaked from a broken cast-iron gas main under the street in front of the houses. The gas main had been weakened structurally by localized graphitization and was undermined by a sinkhole. The failure to check the second house for an explosive atmosphere and to shüt off its gas supply resulted in two fatalities and numerous injuries.

RECOMMENDATIONS

As a result of its investigation of this accident, the National Transportation Safety Board made the following recommendations: — to the UGI Corporation:

"Revise its 1968 Guide for Fire Fighters and training program by incorporating instructions on how to deal with events similar to this accident with particular emphasis on the proper use of combustible gas indicators. (Class III, Longer Term Followup) (P-77-1)

"Revise its emergency plans to incorporate the revisions that became effective October 1, 1976, to 49 Code of Federal Regulations 192.615, and to insure emergency response coordination with fire and other public officials. Particular emphasis should be placed on the availability and the proper use of computible gas indicators. (Class III, Longer Term Followup) (P-77-2)

"Expedite, in conjunction with equipment manufacturers, the development of a survey unit that could be used to detect the location of sinkholes in the vicinity of cast-iron gas mains. "Downward-looking" radar equipment should be investigated as one possible means of surveying for sinkholes. (Class III, Longer Term Followup) (P-77-3)"

-- to the Office of Pipeline Safety Operations of the Materials Transportation Bureau of the U.S. Department of Transportation:

"Encourage, coordinate, and monitor development of equipment which could be used to detect the location of sinkholes in the vicinity of underground utilities. (Class III, Longer Term Followup) (P-77-4)"

• 19 -

-- to the Gas Piping Standards Committee of the American Society of Mechanical Engineers:

"Develop guidelines to assist operators during emergencies to promptly isolate leaking segments of pipe in a low-pressure system. (Class III, Longer Term Followup) (P-77-5)"

BY THE NATIONAL TRANSPORTATION SAFETY BOARD

May 18, 1977

• 21 -

APPENDIX A

ELECTRO CHEMICAL SERIES OF METALS

35 KNOWN METALS IN THIS SERIES

THOSE METALS AT THE UPPER END OF THE SERIES SACRIFICE THEMSEVES PROTECTING ALL THOSE METALS BELOW THEM WHEN CONNECTED TOGETHER IN AN ELECTROLYTE.

• 22 -

APPENDIX B Metallurgical Report No. 77-1

〔图〕。87。2\n

Figure Bl. Pipe as received and after chipping away most of the external soil adhering to the pipe.

No.'s 1, 3, 4, 5 and 8: fractures

No's 2, 6, 7 and 9: intentional cuts

SL: sleeve at a service location

〔图〕插图

Figure B2. Fracture surface identified as "1" in Figure 1. Bracket denotes dark spot at the top of the pipe. Ultrasonically cleaned.

APPENDIX B

• 23 - Metallurgical Report No. 77-1

〔图〕FR-号\n

Figure B3. Longitudinal section through the top of fracture "1" close to the point of deepest corrosion penetration. Arrow "FR" indicates the fracture. (Unetched X10)

〔图〕插图

Figure B4. Closer view of graphitized zone/matrix boundary. (Unetched X50)

• 24 -

APPENDIX C

Calculations of beam stress and deflection for 4-inch cast-iron pipe with a wall thickness of 0.4 inch.*

Beam stress and defection are calculated from the following two formulas:

$$f = \frac{15.28 \text{ W D L}^2}{D^4 - d^4} \quad \text{and} \quad y = \frac{30 \text{ f L}^2}{DE}$$

in which, $f$ = beam stress (14,000 psi maximum) and $y$ = deflection at mid-span (in.)

$W_e$ = external weight of earth in lbs/ft (324# for 3 1/2 of cover) (Table 1-8)

L = distance between support (ft)

D = outside diameter d = inside diameter

$E$ = modulus of elasticity (15,000,000 psi)

Trial 1 (ll-foot span across sinkhole)

Trial. 2 (18-foot span across sinkhole)

$$ f = \frac{15.28 \times 324 \times 4.8 (11)^2}{(4.8)^4 - (4)^4} $$

$$ f = 15.28 \times 324 \times 4.8 (18)^2 (4.8)^4 - (4)^4 $$

f = 10,462 psi

f = 28,018 psi

$$ y = \frac{30 (10,462)(121)}{4.8 \times 15 \times 10^6} $$

$$y = \frac{30 (28,018) (324)}{4.8 X 15 X 10^6}$$

$$ y = 0.527'' \text{ (or } 1/2\text{-inch)} $$

$$y = 3.78 " \text{ (or } 3 \text{ 3/4-inches)}$$

Conclusions: Trial 1

Trial 2

Main would deflect ½-inch but would not break unless there was external loading from vehicle being driven over the main.

Main would deflect over ½-inch (theoretecially 3 3/4-inches) until the beam stress would exceed 14,000 psi and the main would break.

• Formulas and weights from USA Standard for Thickness Design of Cast Iron Pipe (USAS A 21.1 - 1967)