A virtual Yucca Mountain

Yucca Mountain, Nevada, is the only candidate site for a national high-level nuclear repository in the US. To date, the DOE has spent $7 million assessing the characteristics of this arid terrain for its suitability. This effort involves over 1000 experts, including geologists, materials scientists, engineers and computer scientists from the Lawrence Livermore National Laboratory, who are researching the site’s geology and testing materials for making waste storage containers to be buried in tunnels.

Scientists have to be able to accurately predict the repository’s evolution over 100,000 years. As part of more general research into the earth’s anatomy, geochemist Bill Glassley’s team at the laboratory created the perfect tool: a giant computer model that can travel in time to reveal the changing structures of rocks and mountains.

Glassley’s idea was to help geologists understand everything from groundwater supplies to earthquake prediction by creating virtual landforms, like deep-sea vents, volcanoes and geological fault lines. His prototype is a model of Yucca Mountain.

Glassley’s virtual Yucca Mountain occupies over 1400 parallel microprocessors, controlled by the Blue Pacific supercomputer. When it reaches its full potential, the mountain will be split into 40 million cells, some only centimetres across, each defined by 120 geological parameters including stress, heat, chemistry, grain size, porosity and permeability.

The computer can take the rocks forwards and backwards through millions of years, simulating every process within this portion of the earth’s crust. It has grown from equations first devised in the 1980s to describe how movement of water through rock carves geographical features. At that time, nobody had hardware or software powerful enough to crunch these equations, but Glassley’s team has taken advantage of the latest generation of supercomputers to build a code that he has proved can work.

Growing mountains

One problem was a lack of geological data, because data retrieved from surveys into boreholes and through detailed mapping of the tunnel already drilled through the mountain was not detailed enough. To compensate, researchers generated data by growing a young version of the mountain into a high-resolution structure, and used the data the computer had generated itself to fill in missing geology. This proved to be surprisingly accurate: the virtual data even created features geologists had seen but not been able to explain on the real mountain.

Next, the team drilled miles of tunnels through the virtual mountain, filled them with canisters of hot nuclear waste, and sent the mountain on voyages through time.

Surprises

Yucca Mountain is formed from a 14-million year-old volcanic substance called tuff, which geologists will use for tunneling into the mountain. The mountain is in a desolate desert where temperatures can reach 50?C, and tuff contains water in its pores, in this case around 10% of its volume. Scientists are concerned what happens to this water, as the waste packages must remain dry to avoid corrosion and prevent the contents from leaking. They are particularly worried about water getting into the rock through fractures around the emplacement tunnels.

Radioactive decay of spent fuel rods – the primary waste for Yucca Mountain – will give off intense heat, often raising the temperature to over 100?C. Glassley’s model shows that this is hot enough to drive water out of the rock surrounding the packages. The vapour produced will rise up through tens of metres of fissures and pores in the rock before reaching cooler areas where it will condense and start to run down, dissolving small quantities of the minerals in the rock as it goes. When this mineral-laden liquid reaches the hot rocks again, it will vaporise, deposit its load of dissolved minerals and boil to start the cycle again. If this happens thousands of times, it will change the structure of the rock’s pores and fractures. These changes in turn, will affect the rate at which water moves through the rock.

Over thousands of years, these processes could seal rock fractures. Just 300 years into the simulation, the cycle created dome-shaped regions in the rock above and to the sides of the tunnels. The pores and fissures became clogged, halting water flowing through the rock. As water seeping down from the surface is the main way to spread radioactive waste, these impermeable domes are probably a good thing, says Glassley.

Problems

All processes have to be coupled, and the cost of this benefit is that once the waste materials cool down, and the water seeps back into the tunnels, local variations in the rock chemistry will make it more acidic in some areas and more alkaline in others. This means that the canisters will have to be resistant to a variety of chemical attacks.

The simulation revealed another problem. Some packages released more heat than others, and the water evaporating from around the hot packages was continually condensing on their cooler neighbours, leading to wet canisters with high concentrations of corrosive chloride salts on their surfaces.

One result that puzzled researchers was that each virtual tunnel developed very differently. After a while, the team realised that the differences were due to location. The mountain rock acts as a huge heat sink, but the water behaves according to the amount of rock each tunnel has around it. This factor would not have occurred to researchers if they had not built the virtual mountain, says Glassley. Every time they run the simulation it triggers new ideas. For example it may be possible to arrange waste packages with different heat outputs to minimise pH changes and excessive moisture.

The simulations gave designers a good idea of the best places to put monitoring devices for the 100-year confirmation period. But the important timescale is much longer – the 10,000 years for which the repository must be safe. If another ice age should begin during this time, rainfall may increase. There may even be earthquakes. Yucca Mountain moves 2mm a year, which translates into 20m over the life of the disposal site. Scientists do not yet know along which faults this movement is taking place, and there is no way of knowing how it will affect Yucca Mountain. Glassley’s model does not take fault lines into account, other than as conduits or barriers to fluid movement. Nevertheless he is confident that the virtual Yucca Mountain is a powerful learning tool with many other potential applications. This sort of computing power is going to revolutionise science, he says, as it allows geologists to look forward for the first time.