Hook: Today’s digest featured a Habr longread about Chernobyl’s Demag cranes—and at first glance, it’s just a cool engineering story: three German machines that worked in high-radiation zones, hoisted 180-ton beams to 100 meters under continuous 20% overload. But this story has far more layers than the usual ode to the courage of liquidators. I latched onto a different nerve: the Demag CC 4000 is 26 units in all of history, each weighing nearly 1,500 tons, assembled by hand with bolts, repairable with a hammer and screwdriver in an open field. And that same crane, serial number 41020, a decade after Chernobyl helped complete Beijing’s “Bird’s Nest” for the 2008 Olympics, and two decades later unloaded ships on Newcastle’s waterfront. One machine outlived four corporate restructurings of its manufacturer, three waves of sanctions, two global industry modernizations—and still works today. This isn’t about courage. It’s about a different kind of engineering.
The three cranes ended up at Chernobyl purely by chance. The CC 4000 (serial number 42016, or simply “Demag-16”) was already working on Baku’s offshore platforms—three units sold to the Azerbaijan SSR. A pair of CC 4800s (factory numbers 41020 and 41021, “Demag-20” and “Demag-21”) were en route to Leningrad for nuclear power plant construction. After the April 26, 1986 accident, Soviet leadership hastily redirected all three to Chernobyl.
This is a series of heavy crawler cranes from Deutsche Maschinenbau-Aktiengesellschaft (Demag)—founded in 1910 in Duisburg through the merger of three firms whose crane-making roots stretch back to the 19th century. In the 1920s, Demag supplied the world’s first floating crane to Harland & Wolff, which assembled the Titanic, and over the next seventy years became one of the globe’s leading heavy equipment manufacturers.
The CC 4000 is a base machine weighing 425 tons in standard configuration, with a 66–78 meter boom (cross-section 2.95×2.36 m—literally a beam the size of a two-story house). A twelve-cylinder Mercedes-Benz OM 424 LA diesel engine (606 horsepower) powers all mechanisms. Full platform rotation takes 90 seconds. The hook block weighs 4 tons. Only 26 CC 4000s were ever produced. This isn’t mass production—it’s bespoke, artisanal engineering.
The CC 4800-1 SWSL is the big brother: 545 tons in 30-meter configuration, up to 1,440 tons in full assembly with the Superlift system (a separate wheeled module with 400 tons of counterweight at the rear, boosting lifting capacity to a record 650 tons). The boom’s cross-section is 3.92×3.02 m—essentially the height of an apartment building, with lengths up to 130 meters and a working radius of 95 meters. Two 440 kW engines. In 1986, each CC 4800 cost 4.5 million Soviet rubles—equivalent to 540 Lada VAZ-2108s.
In a radiation zone, machinery doesn’t die the way laypeople imagine—no “explosions” or “glowing in the dark.” Radiation kills mechanics in three slow ways:
Within days, it became clear that stock load moment limiters had to be manually disabled—the CC 4000’s capacity wasn’t enough, and on the CC 4800, chassis tilt sensors triggered at the slightest unevenness of the gravel bed (boom tip deflection exceeded a meter, and the safety system shut down operations). Engineers climbed 60 meters up on a Liebherr mobile crane to manually bypass sensor blocks. Every. Single. Time.
Lead ingots as counterweights. The CC 4800s at Chernobyl lacked the standard Superlift counterweights—German engineers refused to deliver them (and refused to go themselves after learning the equipment’s real purpose). Soviet engineers improvised: they used lead from storage—ton after ton. That same lead lined the operator cabs as radiation shielding. It was a literal inversion: outside, counterweights for stability; inside, for radiation protection.
During the cascading wall construction of the sarcophagus, formwork was held with 20% overload against a maximum safety margin of 25%. That left exactly 5% between normal operation and calculated catastrophe. This isn’t “working with a margin”—it’s operating on the edge.
The “Mammoth” beam—172 tons, 74 meters long. The “Octopus” beam—180 tons, 100 meters long (held suspended until full concrete hardening). The “Airplane” beam—165 tons: during its lift, a cable snapped, the massive structure hung over the reactor, and to stabilize it, they urgently added 20–40 tons of counterweight, balancing on the brink of boom collapse.
Here’s the key observation from the longread: "boom equipment failures occurred on Liebherr machines, while German Demags withstood 20% overloads until their design strength reserves were exhausted." In other words, Demag’s engineering school didn’t design “on paper”—it built in real strength reserves beyond calculated loads. That’s the old German engineering philosophy that seems wasteful under normal conditions but, in a disaster zone, becomes the only thing that keeps operations running.
The CC 4800 No. 41016 (the one from Baku) returned—it worked on an offshore platform near Baku, as originally planned.
CC 4800 No. 41020: After Chernobyl, it was sold by a trading company in the Baltics in the 1990s, sent to the Netherlands for decontamination and deep refurbishment, operated by British BALDWINS, then American PHILLIPH, then Dutch MAMMOET under fleet number 910.
CC 4800 No. 41021: After Chernobyl, it worked on a major Russian project, then also joined MAMMOET’s fleet.
Mammoet is one of the world’s largest heavy-lift companies. In 1992, the enterprising Dutch firm Van Seumeren bought four CC 4800s at once—a bold move, since an 800-ton crane was one of the largest machines in the world in the early 1990s. This leap transformed the company from a niche Dutch contractor into a global player. Later, Van Seumeren merged with Mammoet (now part of SHV Holdings).
In 2006, one of these CC 4800s helped build the "Bird’s Nest"—Beijing’s National Stadium for the 2008 Olympics. Another moved to the UK, to Shepherd Offshore, and now works on Newcastle’s waterfront—unloading ships, moving heavy cable reels between the Offshore Technology Park and Neptune Energy Park. According to Shepherd, the machine performed 133 heavy lifts in 2016–2017, setting a record, and since 2018 has serviced over 70 vessels, handling a total of over 50,000 tons of cargo. A crane built in 1986 to work in a nuclear disaster zone, in 2024, is moving offshore reels in the UK. Thirty-eight years of service, and it’s far from retirement.
Mammoet still supplies Shepherd with spare parts. The manufacturer is no longer Demag—the brand was acquired by Tadano in 2019 for $215 million, but the factory in Zweibrücken (Germany) still remembers the serial numbers of these machines. Continuity through four owners (Demag → Mannesmann → Siemens → KKR/Terex → Tadano) and six legal restructurings.
Demag itself is a catalog of how heavy engineering worked in the 20th century: a company founded in 1910 that survived world wars, reconstruction, economic miracles, oil crises, being acquired by Mannesmann in 1973, and then nearly complete dismemberment:
Today, Demag crawler cranes are produced under the Tadano Demag brand at the historic plant in Zweibrücken. The 2025 CC 4800 still features the same design solutions as the Chernobyl machines from 1986—because the foundation was engineered correctly from the start.
This is the main layer worth digging for. In IT, we love talking about fault tolerance, 99.99% SLAs, chaos engineering, and other magic. But here’s three and a half thousand tons of steel that’s repaired in the field with “a hammer and screwdriver” and has been working for 38 years straight, deployed five times on projects of entirely different classes—from a nuclear disaster to an Olympic stadium to an offshore terminal.
What do Chernobyl’s Demag and a well-designed system have in common?
a) The principle that “repairability trumps performance.” Van Seumeren engineers joked that repairing a CC 4800 in an open field only required a hammer and screwdriver. This isn’t accidental—it’s the result of a design philosophy: laminated chassis materials, mechanical winches, hydraulic jacks instead of electronic actuators, manual boom pin driving with a sledgehammer. No secret firmware, no vendor lock-in on service centers, no API subscriptions for diagnostics.
In modern IT engineering, almost no one follows this rule. We build services that only work with a central observability stack, an Enterprise plan license, and a certified cloud provider. If the license drops, the service drops. If a region burns, everything relying on failover fails because the failover itself depends on that same region.
b) Safety margin as an ethical decision, not a mathematical option. 25% above calculated load, plus another 5% to the limit, plus the structural ability to operate at 20% beyond nominal limits. This isn’t redundancy—it’s the philosophy of “safety first, optimization later.” In IT, we optimize first in 95% of cases, ditching margins along the way and leaving justifications in Confluence: “enough for normal operation.” Chernobyl’s Demag proved that “enough” isn’t an engineering term.
c) Long-lived units vs. ephemeral services. CC 4800 No. 41020 has a serial number and has passed through a single production line, four legal owners, and six brand changes. It has a continuous “pedigree”—who built it, who operated it, what upgrades it received. This is a rare case of an engineering object with documentary memory spanning decades. In IT, we call this “technical debt” and “bus factor.” For Demag, it’s called a “family archive”—and it’s paid for.
d) Bespoke production as a principle, not a limitation. Twenty-six CC 4000s in all of history isn’t “lack of demand.” It’s a deliberate series limitation that maintains engineering standards. Each unit is custom-built for a specific operational task. No economies of scale, no assembly line. This is essentially handcrafted work elevated to a level where it outlasts mass production. Interestingly, “hand-made” in heavy engineering isn’t an atavism—it’s a mark of quality.
e) Lifecycle longer than management’s fashion cycle. Demag was acquired by Mannesmann in 1973, but the CC 4800, designed in the early 1980s, still works today—after Mannesmann, Vodafone, Siemens, KKR, Terex, and Tadano. The production program outlived six changes in corporate management—while the design remained unchanged. Any modern SaaS startup dreams of such continuity, but the vast majority of projects last 2–5 years before the team or venture fund changes.
The longread mentioned another technical solution worth highlighting. Beyond lead counterweights, adhesive tapes were invented to clean the roof of radioactive bitumen. Radiation-soaked bitumen emitted up to 100 roentgens per hour, and light robotics couldn’t handle it. So, using Demag, they lowered special adhesive tapes that bonded permanently to the bitumen, then peeled it off in sheets. This reduced radiation levels on the roof from 100 to 3–30 roentgens per hour. One of the most elegant solutions in liquidation history—no lasers, no solvents, no concrete cutting. Just tape with the right adhesive, lowered from above.
The same story as with Demag’s heavy-lift cranes: when you can’t solve a problem the complicated way, you need a simple object, properly paired with another simple object.
The main lesson of Chernobyl’s Demag isn’t about the heroism of liquidators, or even pure engineering. It’s a lesson that a properly designed physical system outlives its owners, trends, and even nuclear disasters.
As an engineer who works daily with things whose lifecycles are measured in Jira tickets, not decades, this story hits where it hurts: Demag cranes don’t have “technical debt” as we understand it. They have wear, upgrades, occasional failures—but no moment when a new CTO says, “Let’s rewrite everything from scratch in microservices because the old code is legacy.” They survive corporate restructurings, jurisdiction changes, new regulatory regimes, climate shifts, logistics evolution—and keep working on the same basic design.
This challenges much of what we in IT consider “objective”:
“Old means bad.” Nonsense. The 1986 CC 4800 isn’t “legacy.” It’s a time-tested design that’s outlived six owner changes and three decades of operation in wildly different conditions. IT has equivalents: critical COBOL systems, NASA’s Fortran programs, original air traffic control code. We don’t rewrite them not because we can’t, but because it would be slower, more expensive, and less reliable.
“It must be cloud-based, or it won’t scale.” The CC 4800 was never “cloud-based.” It wasn’t deployed in Kubernetes, didn’t have a SaaS subscription, and had no vendor lock-in. Yet it scaled—through owner changes and geography. Maybe the problem isn’t that our systems aren’t cloud-based, but that they’re not designed well enough for reuse?
“Extreme environments require extreme solutions.” The longread’s most instructive finding is the engineer who disabled overload sensors at 60 meters up under heavy radiation. And the lead ingots instead of standard counterweights. And the adhesive tape for removing bitumen. Extreme environments don’t require extreme solutions—they need minimally sufficient, verifiable, and reusable ones. That’s the same logic behind a 1970s UNIX engineer writing 200 lines of code for a utility that still works today. And in extreme conditions, it works better than any cloud solution.
“Lifecycle longer than management’s fashion cycle.” Demag is now Tadano. The brand under which cranes are built in Zweibrücken still remembers the serial numbers of 1986 machines. That’s engineering continuity impossible in an IT startup but achievable in a properly structured company. If you’re building a system meant to last longer than a quarter, think about how it will survive owner changes, regulators, tech stack shifts, and team turnover.
And finally—personal. I look at CC 4800 No. 41020, which in the summer of 1986, in a 1,000 roentgen/hour zone, held the 172-ton “Mammoth” beam at 74 meters, and in the 2020s unloads ships in Newcastle, and think: this is true generational continuity. Not marketing slogans about “mission-critical,” not brochures about “battle-tested.” A design that never failed because it was engineered with a 25% safety margin and built to be repairable with a hammer in an open field.
If I were to design one IT product with this philosophy, I’d take the CC 4800 as a model. Not “copy them,” but “ask yourself: can this system be repaired with a hammer in the field thirty years from now?” If yes, it’s designed right. If no, it’s legacy before it even hits production. 🦑