How To Build A Nuke Plant

[Jaglavak]

Several years ago I read an interesting report from the IAEA about operational experience from unconventional nuclear power plants that have been built around the world, which basically means molten salt or liquid metal cooled reactors. One thing that stood out was the number of problems with pumps and heat exchangers. In fact with one exception, every sodium cooled plant ever built has had a leak with resulting molten sodium fire. And with one exception, every piece of equipment in the loop has had a leak and fire. The exception being the reactor vessel itself, none of which have ever leaked. While little to no radiation was released, this is beyond bad optics.

With that in mind, and with the added goals of greatly reduced fissionable inventory and an order of magnitude reduction in size and complexity, there is an interesting option.

Back around the early 1900's, internal combustion engines took over the market from steam engines primarily due to about a 50 times reduction in size and weight. While the engines themselves were not much lighter, internal combustion engines don't need a boiler. Instead they get the heat transfered by direct contact between the fire and the working gas. Applying the same notion to nuclear power requires a very large surface area between the hot nuclear fuel and the working gas.

Translated to hardware it looks like this: A molten salt reactor with an output of around 750 MW at 1300 F. Molten salt passes upward through the reactor and into the float bowl of a large updraft carburetor with a 3 ft diameter venturi. Approximately 700 cubic feet per second of argon from a centrifugal compressor picks up a 20 cubic foot per second spray of salt passing through the venturi. The salt is then removed by a panel of vortex tubes similar to the ones used on helicopter and tank turbines. The still liquid salt then flows into a downcomer back to the bottom of the reactor with no circulation pump required, while the heated argon passes through a centrifugal turbine for about 250 MW of power production. The cooled and expanded argon then passes through a cycle cooler and back to the compressor inlet. While a 1% slipstream of salt spray does carry over through the cyclone tubes, the primary loop is built to handle it.

A radial flow compressor and turbine are used due to the very simple and rugged shape, and because of the bearings. Multi-nozzle aerostatic bearings supplied by high pressure argon can easily handle the load and speed of a turbine this size. While this type of bearing does require several thousand horsepower of pumping power, that is acceptable at less than 1% of the power output. Shaft seals are non contact labrynth seals using argon flush gas. Power takeoff is via a second argon compressor wheel mounted between the bearings. So it looks like an extra wide 12 foot high turbocharger. The only materials involved are fuel salt, austenitic stainless steel and hastelloy-n, and argon. And nothing touches the turbine and shaft except gas, which means the service life can be measured in centuries.

The power takeoff compressor passes about as much flow as the primary loop, and delivers power to a remote turbine in the form of pressure times flow rate. The power takeoff flow also picks up heat from the cycle cooler on the way, cooling the primary loop while acting as the bottoming cycle for a significant increase in thermal efficiency. This allows use of a simple compact single stage primary loop while still providing good fuel efficiency.

A power plant of this type could operate with a very small amount of fuel, potentially less than 100 lb of fissionables. The neutron counter types can design any arrangement of reactor they like as long as it puts out 750 MW of hot salt. Molten salt reactors tend to be constant temperature load following devices, so nothing special is required there. Every component is a well understood time tested design that has been extensively used in similar applications. Every part of the plant can be accessed for maintenance or replacement with nothing buried in concrete. And the overall plant size is dramatically smaller and cheaper than existing designs. Of course everyone's head would explode at NRC and it would take about 100 years to get so much as a test rig approved. But it would work, and do it efficiently and safely.