Though history is rife with technological hiccups, we expect infrastructure to last and often forget the best maintenance starts with surveillance.

For Pipeline Builders, a Long Road to Understanding Rust

The Trans-Alaska Pipeline is the biggest, baddest, most regulated pipeline in America. Four feet in diameter and 800 miles long, it has a surface area of 7 billion square inches — every single one of which must be examined every three years for damage inflicted by rust using a 5-ton centipede-like machine called a smart pig, as long as a Subaru and far more expensive. Inducing magnetic flux as it scoots along, the pig measures the wall thickness of the pipe — hunting for thin spots before they turn into leaks.

WHAT I LEFT OUT is a recurring feature in which book authors are invited to share anecdotes and narratives that, for whatever reason, did not make it into their final manuscripts. In this installment, author Jonathan Waldman shares a story that didn’t make it into his latest book “Rust: The Longest War.”

When vast petroleum deposits were discovered under Alaska’s North Slope in 1968, oil companies didn’t want to build a pipeline to a port because they knew a metal tube snaking across the last frontier would be a rust nightmare. Instead, they dreamed of elaborate schemes to ship the oil southward — fleets of trains, planes, and trucks, as well as icebreaking ships, nuclear subs, even blimps. In the end, though, they did build a pipeline — one of the heaviest metal objects on the planet — outfitted it with an array of the era’s most sophisticated technology, and boldly declared it rustproof. Pipelines had been in use for over a century at that point, but the science of stopping them from developing dangerous leaks had not yet been figured out.

The first serious investigation into underground pipeline corrosion came at the behest of Congress, in 1910, when it authorized the National Bureau of Standards to look into the matter. Four men from the bureau — E.B. Rosa, Burton McCollum, George H. Ahlborn, and Kirk Harold Logan — looked at stray current, because corrosion, as Humphry Davy had shown a hundred years earlier, was really an electrochemical phenomenon. But after 10 years, they remained flummoxed. They figured there was more to it than just that.

In the 1920s, the bureau set out again to get to the bottom of things, with the help of the American Gas Association and the American Petroleum Institute. This time, the bureau investigated soil and its characteristics, with help from the U.S. Department of Agriculture’s Bureau of Soils. The bureau started a field program, eventually burying more than 36,000 samples of 1.5 -inch and 3-inch pipe. They did this at 128 test sites all over the country, from a foot and a half to six feet deep, in 95 types of soils, from Kalamazoo to Camden. They buried iron, steel, stainless steel, copper, lead, zinc, and aluminum alloys, covered in range of coatings, for a grand total of 333 varieties of materials.

Every two years, the bureau pulled out 250 samples, weighed them, figured out how much metal had been lost, and calculated the average pitting penetration. They measured the maximum penetration of rust pits, and computed a pitting factor, which was the ratio of maximum depth to average depth. Then, looking at long tables of numbers they’d collected, they determined the mathematical relationship between the pitting factor and the corrosiveness of different soils.

By 1933, some things were clear to Logan, who’d continued with the research. A ferrous pipe buried in a Ruston sandy loam would outlive one buried in Susquehanna clay, which in turn would outlive one buried in a Merced silt loam. But much remained mysterious. “There are so many diverse factors that affect the corrosion of pipes,” NIST scientists wrote, “that the planning of adequate tests and the proper interpretation of the results are matters of considerable difficulty.”

At times, the scientists sounded defensive. “Much information has been derived from the various attempts to establish the importance of individual factors,” they wrote, adding, “but the final answer and the complete understanding of the phenomenon have not yet been attained.” In other words: it’s complicated. Definite conclusions could not be drawn, except for these: 1) the most commonly used pipes corroded fastest, and 2) the bureau’s measurements were not useful for predicting long-term corrosion. So the bureau (by then NIST) did what it had always done, and embarked on a new series of tests.

While scientists struggled to understand pipeline corrosion, pipeline operators remained nonplussed:

Ray Rountree, United Pipe Line Corporation, 1932: Gas leakage from corrosion is as “positive and persistent as taxes and death, and apparently just as difficult to overcome.”

Hugo Johnson, Los Angeles Gas & Electric Corporation, 1933: “Attempts to cure rust troubles have not been entirely successful.”

Norman Hoff, Los Angeles Gas & Electric Corporation, 1933: “Fifteen years ago we knew little about soil corrosion and pipe protection. Today, after 15 years of experience, we still know comparatively little.”

Elmer Schmidt, Lone Star Gas Company, 1934: “Many valuable articles and reports have been published on the general subject of pipeline corrosion, its causes and prevention. But corrosion still occurs, and the gas industry must still pay the bill for the tremendous annual damage from this cause.”

Hoff also held some more nuanced views. He argued that the goal wasn’t to make pipes everlasting, but to give them equal lifespans, so that they could all be predictably maintained and replaced before they became obsolete. He assessed the burden imposed by corrosion economically, detailing, in three dense pages, equations relating pipe thickness, internal pressure, and cost. Then he conceded defeat. “It is not difficult to see that the use of the foregoing equations is about 90 percent good judgment and 10 percent simple algebra,” he wrote. “In fact, it reminds one of the recipe for properly preparing a carp. After taking great pains to properly clean and season the carp, it is recommended that you throw it away and buy yourself a pork chop.”

Because a cure for rust remained as elusive as an understanding of it, utilities struggled to remove it, defend against it, and deal with the leaks it caused. Until 1925, the Peoples Gas Light & Coke Company in Chicago tried forcing rust clogs out — often making things worse. Tired of shutting down gas mains and digging them up to clear them of stoppages, the utility invented a vacuum system. It consisted of a two-cylinder compressor mounted to a Ford truck.

The truck would be lifted off one wheel with a jack; its driveshaft would be connected to the compressor and the compressor to a gas line. Once the device was tied to the gas line of a house with a 50-foot hose — and the gas meter disconnected so as not to rupture it — the compressor was fired up. When the suction was good and high, someone pulled a plug, figuring that sudden suction worked better. The apparatus sucked gas into a 30-gallon sediment tank, and then through a centrifugal horsehair filter. It was so powerful that it extinguished pilot lights in appliances next door. A fellow standing at the house could hear the rust traveling through the hose. “It was not uncommon, in case of very rusty services,” wrote a district superintendent named E.G. Campbell, “to remove from five to 10 pounds of rust from a single service.” The system worked well, nearly eliminating the need to dig up pipes. The gas was vented harmlessly out the top of the truck.

Another such trap, used in California in the 1930s, was more or less a three-foot-tall water bong. Setting it up on the street, the utility company connected a main line to it, and forced gas into the bottom of the vessel, which was half-full of water. As the gas bubbled out, most of the rust got stuck in the water. At the top of the vessel, where the gas vented out, any more rust was caught in screens and wet burlap. Soon enough, the water became a soft mud-like sludge, which the utility men tried to avoid spilling into the gutter. Another rust trap, which the same utility used underneath houses, looked like a big hamster cage. A simple affair, it was just a dampened burlap bag inside a cage of wire mesh.

Already, one could be forgiven for getting the sense that our infrastructure was crumbly.


To minimize external pipe corrosion, companies coated pipes in asphalt, or encased them in wooden boxes, and filled in the boxes with asphalt. Better was a half-inch thick coating of somastic, invented in 1938. (Epoxy was discovered the same year, but not perfected for more than a generation; the epoxy coating used on the Trans-Alaska Pipeline didn’t stick.)

Before the pipe was buried, someone visually checked for holes in the coating — holidays, they were called. Beginning in 1931, coating inspectors did this job mechanically, using a brine-soaked cloth connected to a battery and a meter. If current escaped into the pipe, the meter wailed. The inspector, carrying the meter, slid the cloth along the pipe. The battery sat on the ground. Soon enough, the cloth gave way to a semicircular brush and frame, with a second fellow carrying the battery. The bottom half of a pipe could be inspected in one go, and the top half in another. Shortly thereafter, a hinged circular brush was devised, and the battery was placed in a wheeled cart beside it. Because the bristles wore down quickly, these holiday detectors were not reliable. Someone devised one with a solid copper ring, and Shell, with the help of O. C. Mudd, developed a consolidated unit, with a coiled spring, that crawled along a pipe.

By the late 40s, pipes were wrapped in two layers of fiberglass and felt, or two layers of fiberglass and asbestos, coated in coal tar. One engineer testified that such a coating would last 10 years, and that after such time he would expect 10 or 15 leaks each year for every 100 miles of pipe. Another engineer, who spent four years appraising gas and water pipelines for the Arkansas State Department of Public Utilities, said in 1951 that such a coating was indeed the best possible, but that quantifying its durability made no sense. The issue was debated in court because the Arkansas State Game and Fish Commission had built a lake atop a pipeline right-of-way, and the pipeline operator wanted compensation:

Lawyer: “From your experience, how many years would you say this wrapping would last in an area such as Stone Dam Creek?”

Engineer: “I wouldn’t hazard a guess. I don’t see how any man can pick out a particular spot or pipeline and say that it is going to leak in five or 12 years or any number of years.”

Lawyer: “Anything that any man would say about it would be purely guesswork?”

Engineer: “Entirely guesswork, and the damage to be sustained is guesswork.”

The judge, compelled by the engineer’s frankness, agreed. Not even an expert could predict, with specificity, the location or timing of pipeline leaks. More generally, though, any dummy could predict them, because they were ubiquitous. Now our infrastructure was on the record as crumbly.

Gas leaks caused by corrosion created their own symptoms. Impurities in manufactured gas (made from coal or oil) were toxic to vegetation, yellowing, shortening, or even killing shrubs and trees. In 1932, property owners encountering this effect along a pipeline right-of-way in Salem, Massachusetts, sued the utility company. To determine if gas leaks had killed the trees in question, the company hired a tree man named Milton Heath. Heath, sensing an opportunity, founded a company in 1933 devoted entirely to this field. It was the first gas leak detection company in the country.

To detect leaks in the 1930s, a man had many tools at his disposal. He could put a stethoscope to a pipeline and listen, or use an amplified microphone called a soundograph. Both were so sensitive that the sound of a flushing toilet could cause interference. More often, he used one of many brands of combustible-gas indicators. If necessary, he could dig a barhole near his pipeline, and then employ one of these detectors. But the best, most reliable method relied on eyeballs aimed at nearby vegetation. Heath specialized in vegetation surveys. In 1947, he hired his eighth employee, a man just graduated from the University of New Hampshire named Stuart Eynon. Eynon, now retired in Ashland, Massachusetts, is one of the oldest and most experienced leak detectors in the country.

When Eynon started doing leak surveys, in Michigan, on foot — just like the linewalkers of the 1870s who constantly surveyed the country’s earliest pipelines in Pennsylvania —he was discreet. He’d look for plants that seemed deprived of oxygen in the soil. He’d try not to talk to people, because utility companies were not eager to publicize their leakage problems. If he bumped into someone curious, he’d explain, vaguely, that he was doing a safety survey for the utility company. City employees and police officers who’d ask, “What’s going on here?” were best avoided. He drove around in an inconspicuous black car. He walked five miles a day, and classified the leaks he discovered as A, B, or C, the grades referring to hazard, not size. The federal government later incorporated this approach, with his help, referring to leaks as Grade 1, Grade 2, and Grade 3.

In the 50s, Eynon started inspecting rights-of-way in a Jeep. He worked all over the Midwest and the East, and picked up a few contracts in west and the south. Until the 60s, he relied on his eyes. Then, Heath and the utility giant Con Edison developed infrared detectors. Because they relied on diaphragms thinner than a human hair, they were sensitive to motion. With such sensors mounted atop Ford station wagons, and later Jeeps and Land Rovers, they drove along rights-of-way steadily and carefully. According to Eynon, a skilled pair of eyeballs could outperform the earliest instruments every time. A decade later, flame ionization detectors came along, and after that, laser-based detectors. Leak detectors today now use solid-state infrared devices, and have lost the old-fashioned skill. “It’s too bad,” Eynon told me.

Around the time Eynon started doing leak surveys (and into the mid-70s), major utilities were switching from manufactured gas to natural gas. This had two consequences, both of them good for Heath’s business. The first was a huge spike in leaks. Manufactured gas, being very wet (1,000 pounds of water vapor in a million cubic feet), had kept the pipes’ joints, stuffed with jute, healthy and virtually airtight. Natural gas being far drier (only 7 pounds of water in a million cubic feet), the jute dried out, and pipelines leaked dramatically. One utility discovered after many digs that half of its cast-iron joints were leaking. Some pipelines lost half their gas this way; about 10 percent of most utilities’ gas was unaccounted for. As utilities recognized the futility of maintaining leaky cast-iron pipes, they converted to steel, which was stronger and could be made thinner, and hence more cheaply. But that led to the second consequence: As NIST had discovered, steel pipelines corroded faster than cast-iron pipelines, especially if they were connected to cast-iron pipes or pipes encased in concrete.

As a result, it wasn’t unusual for Eynon to find a leak every mile on a 500-mile pipeline. “It astounds people to realize how many leaks are out there,” he recalled. To keep up with the work, he hired graduates from forestry schools all over the country. Many, after years of working for Heath, grew sick of the travel, and took jobs at utility companies where they could stay put and become vice presidents. Eynon liked the travel, and eventually did leak surveys in every state in the country. Of leak detection, he said what linewalkers from the 1870s to the present know: “It’s something you have to work on constantly. It needs constant surveillance.”

A dozen years after NIST published its uninspiring study, leak surveillance remained the best thing going, as an author, writing in NACE’s new journal Corrosion, acknowledged: “The main problem attacked in these studies … is as far from a solution as when the investigation was first made.” A dozen years after that, NIST concluded: “Underground corrosion that has occurred can be explained, but, even today, theory does not permit accurate prediction of the extent of corrosion to be expected.” (They still use linewalkers — and drones — on the Trans-Alaska Pipeline.)

When utilities switched from manufactured gas to natural gas, one more significant change came along. They added an odorant, because natural gas has no smell. When Los Angeles did this, in July of 1929, leak complaints spiked 20-fold — and this gave someone an idea. To detect leaks, utilities once or twice a year quadrupled the amount of odorant they used, and waited for reports to come in. (Had Texas added odorant sooner than it did, it could have avoided the catastrophic gas explosion that killed 300 New London school students in 1937.) Others tried ionizing their gas, then tried to detect it where it shouldn’t have been. A German company put ammonia into its pipe, and then looked for heavy fog. Others considered more dangerous tracers, including hydrogen, helium, neon, and acetylene, and radioactive tracers.

To detect leaks, utilities let no idea fall by the wayside. If they couldn’t get Eynon’s eyes, they went for a dog’s nose. In 1948, one company trained a bloodhound to detect gas leaks. The dog performed best in the morning, when not distracted by traffic. His services cost $10 per mile. Another company, in 1956, trained an Alsatian to detect leaks. In six months, it covered 327 miles of pipeline, and found 795 leaks. Leak-detecting dogs are not just thing of the past. Since 2007, the British energy company Penspen has used a dog named Blitzen to check pipelines for leaks. In the phrasing of the Pigging Products and Services Association, “The dog was successfully calibrated before, during and after the inspection.”

It’s easy to forget how newfangled our constructions are. The first successful pipeline was built in 1862. Corrosion theory wasn’t laid out until 1938. Smart pigs, dreamed up in the 1950s, weren’t especially useful until the 1990s. Though our history is rife with technological hiccups, we expect infrastructure to last, and forget what Stuart Eynon and Alyeska know well: the best maintenance starts with surveillance.

Jonathan Waldman, a 2017 Alicia Patterson fellow, is working on his second book, SAM, about the development of an unusual robot. His first book, “RUST: The Longest War,” was a finalist for the LA Times Book Prize.