Wiring the Blue Gene supercomputer at Livermore. It helps scientists in "stockpile stewardship," their post-Cold War mission: assessing the reliability of existing weapons without exploding them. (Lawrence Livermore / National Laboratory)
LIVERMORE, Calif. - Physicist Bruce Goodwin compares our nuclear weapons to vintage cars: 20 to 40 years old and subject to corrosion.
They weren't designed to last forever. At some point, they may no longer work. Scientists can't assess the stockpile by exploding a few warheads; nuclear testing would violate U.S. policy in place since 1992.
Solving this conundrum falls to the national weapons labs, including Lawrence Livermore National Laboratory, where Goodwin heads up weapons research.
Livermore's approach involves a combination of supercomputer simulations and experiments, the most ambitious of which will use a $5 billion laser apparatus in an attempt to create a controlled version of an exploding hydrogen bomb.
Last week, the United States agreed with Russia to cut nuclear arms 30 percent, a move the Livermore scientists say only underscored the importance of their mission. The fewer weapons we have, the logic goes, the more reliable they need to be.
But also last week, the Government Accountability Office, the investigative arm of Congress, criticized energy officials for "weak management" over the laser project and allowing its budget to soar. Outside critics go further, contending that much of the work at the weapons labs is contrived to keep billions of dollars flowing, while having little impact on our security.
The new treaty still allows the United States to deploy 1,550 warheads on missiles and submarines, down from 2,200 today. Even if they are in a decrepit state, that's more than enough to deter attacks.
Building nuclear weapons
Livermore opened in 1952, branching off from Los Alamos National Laboratory, where nuclear weapons were born. Tucked between pastures and vineyards about 80 miles east of San Francisco, Livermore specialized in making nuclear weapons smaller, lighter, and more destructive.
The first thing visitors notice is an obsession with safety. Company cars carry placards inside warning that seat belts are mandatory. Bob Hirschfeld, a public-relations officer, said workers aren't allowed to climb ladders until they've completed ladder training.
It may seem ironic to have such rules at the place where the world's most horrific weapons were designed, but there's a logic to them. Those who've made careers here believe nuclear weapons keep us safe.
That's Goodwin's philosophy. He agreed to meet in one of the lab's classified buildings. While most of the scientists at Livermore favored shorts, T-shirts, and sandals, Goodwin showed up in a suit and tie, explaining that he has to be ready to meet members of Congress and other high-level officials.
He began his career as an astrophysicist, specializing in exotic objects called neutron stars. His path changed when he was a graduate student and attended a talk by then-famous antinuclear activist Helen Caldicott.
"I realized how much of what she was saying was wrong," he said. If other countries were going to stockpile nuclear weapons, it made no sense for the United States to abstain.
He also became convinced that nuclear weapons prevented wars - at least large-scale ones like World War II.
During the 1980s, he designed nuclear weapons. In those days, they weeded out the duds by setting off prototypes in the Nevada desert. Goodwin tested five of his own designs, and all worked, he said.
But then came the end of the Cold War and, by 1992, the last U.S. nuclear test.
With that, Livermore and Los Alamos were charged with a new mission called stockpile stewardship. Instead of creating weapons, their job was to assess the reliability of existing ones without exploding them.
Another part of that mission is safety - ensuring, said Godwin, that these weapons can't explode by accident if, say, they're on a plane that gets struck by lightning and crashes. They're also designed so that terrorists stealing them couldn't set them off.
So Goodwin's job changed from weapons designer to stockpile steward.
As weapons degrade
The first nuclear weapons were as big as this, Goodwin said, pointing to the conference table made to seat at least 10 people. The biggest ones weighed 20,000 pounds.
Now they're many times smaller thanks to a process called boost, which uses a combination of nuclear fission, the splitting of big atoms, and fusion, the glomming together of smaller ones.
Conventional explosives start the whole thing, setting off a "primary" that uses fission of plutonium and fusion of some smaller atoms to quickly and violently implode a "secondary" - usually a container of additional fusion fuel.
The secondary, Goodwin said, makes the really big blast, "though everything is big when we're talking about nuclear weapons."
Millions of lines of computer code tell exactly how these weapons are supposed to explode, he said, but these codes may need to be adjusted as the weapons degrade over the years.
"On average, these weapons are 30 years old. One of our weapons just had its 40th birthday," he said. Over time, he said, the plutonium in the primary part of the weapons undergoes radioactive decay, which shoots off particles that can damage other parts of the weapon.
Beyond that, Goodwin said, some parts are made of plastic, which can crack, while other components can be damaged by chemical interactions, including ordinary corrosion.
Before some recent refurbishments, he said, "some of them were literally rotten at the core."
Other studies are allaying earlier degradation concerns. In one experiment, Livermore scientists accelerated the aging of the plutonium "pits" at the heart of our nuclear weapons. In 2006, they reported that this component would hold up for 80 to 100 years.
But other things could go wrong. The goal is to understand these weapons in a more thorough way so the scientists can predict their behavior without testing them.
This is a feat Goodwin said he didn't believe was possible back in the 1990s. "I realized we'd need a computer that was a million times better than the best machine available at the time." But then processing literally got a million times faster.
And the national labs use the fastest computers ever built.
Some of Livermore's supercomputers are in a classified area. A sign warns employees that a visitor is present and that only unclassified conversations are allowed.
The scientists realized they needed this building in the late 1990s after discovering they didn't have the power to run their latest computing behemoth, dubbed Purple.
This supercomputer needed between four and eight megawatts of power - about as much as it takes to power the rest of the town of Livermore, with a population of 80,000.
It was the fastest computer in the world for more than three years, said program director Michel McCoy, until it was surpassed by a new model, called Roadrunner, which is used at Los Alamos. If not for the weapons labs, he said, it's unlikely these computers would have gotten built.
"With 100 teraflops, it's a little passe," McCoy said about Purple. A teraflop refers to the ability to do a trillion operations in a second.
A few years ago, they achieved a quadrillion operations per second - a petaflop. They now have a half petaflop computer, called Blue Gene, based on a design created for genetics research.
Here, the main job of these computers is to check on their "weapons codes." Behind each of the approximately 100 types of nuclear weapons in our stockpile, there's a code - about a million lines of software - describing what happens if it's set off.
In the post-testing world, McCoy said, the codes must be truly predictive. The processes are complex, with primary explosions involving fusion and fission, thus triggering the larger secondaries. All this has to be simulated in three dimensions, said McCoy.
He and his colleagues are anticipating an even-more-powerful model in 2012. But it won't end there.
What they really need, he said, is to get into the "exascale," which means an additional factor of 100 improvement - to 1,000,000,000,000,000,000 operations per second. "That's what we need to be comfortable that we can certify our weapons."
They hope to get their latest dream computer by 2018.
As a kind of spin-off, they say, they could also create models of Earth's climate that better predict the effects of increased carbon dioxide and other greenhouse gases, an issue whose economic and social effects could be relevant to national security.
By the end of this year, if all goes as planned, the weapons scientists will back up their computer models with data collected in the giant laser experiment known as the National Ignition Facility.
Deep inside the apparatus, the lasers are supposed to compress hydrogen until its atoms fuse into helium, the same energy-liberating reaction that fuels a hydrogen bomb.
But after all their simulations and experiments are done, there looms the question of what should happen if they discover that some weapons have gone bad.
Some problems can be fixed with what's called life extension - replacing or repairing corroded or damaged parts. But others advocate a more controversial approach of replacing warheads with up-to-date models.
In some cases, it might not be possible to refurbish or replicate weapons, said Goodwin. Some of the older weapons use vacuum tubes instead of modern electronic components, for example. And you can't get vacuum tubes anymore, he said.
And the old weapons rely on materials that have proved hazardous to workers, especially beryllium metal.
Others worry that designing a new or even modified weapon would tempt future U.S. governments to conduct tests, thus opening the door for China, Russia, and others to begin testing and improving their nuclear arsenals.
Building new nuclear weapons won't make us safer, said physicist Arjun Makhijani of the nonprofit Institute for Energy and Environmental Research in Takoma Park, Md. Weapons don't have to be perfect to deter an attack.
"If deterrence is the sole purpose of your arsenal, it doesn't matter," he said, "if they give you 90 percent of the yield or 10 percent."
He said much of the work at Livermore was more about preserving the weapons labs than protecting citizens. "Like any bureaucracy, it wants to perpetuate itself," he said.
"I think it's important to have people who know about the weapons we have - dismantling them and maintaining them," he said. "But those are engineering problems and don't require physicists."
Princeton University physicist Frank von Hippel said there was still important work for the Livermore scientists to do, mostly involving disarmament.
The United States has more than 1,000 nuclear weapons on what he calls hair-trigger alert - ready to be deployed within 15 minutes. The Russians have thousands of missiles aimed at us and are ready to launch equally fast.
That makes accidental nuclear war a bigger concern, he said.
"Obsessing about whether there's a 1 percent chance they might not work is just crazy."