Transatomic Power

This cropped up on HackerNews a few days ago and I immediately got nerd-sniped.

A few initial impressions:

What’s actually going on here?

Their product is a proposed Gen IV reactor design, and Transatomic is taking VC money to try and build it. MSR’s were tested throughout the 1960’s in Oak Ridge, and everything Transatomic says about them is pretty much true. They can make much more efficient use of fuel than a typical LWR, and since the fuel can exist as a liquid at atmospheric pressure, there’s no need for a pressurizer. But 50 years ago in Oak Ridge, they came to the conclusion that the energy density of MSR’s was just too small to make them economically viable.

There are things about the Transatomic paper that stick in my craw, like:

As these byproducts are gradually removed, a small amount of fuel (either SNF or low-enriched fresh fuel) is regularly added to the primary loop. This process maintains a constant fuel mass, and allows the reactor to remain critical for decades.

“Remain critical for decades” sounds awfully nice, but maintaining criticality for a long time has never really been “the problem” with modern reactors. The reason that modern reactors are shut off every 18 - 24 months is to perform maintenance on all the other pieces that the loop needs to make power: valves & actuators, the steam turbine, piping, etc. In a highly-corrosive MSR loop, these problems would be even worse. I know the people at Transatomic know this; why the hyperbole? You’ll notice that in their paper, they don’t attempt to take a guess at the maintenance interval of this design.

But nitpicking aside, how does the actual engineering of this thing look? What is Transatomic’s innovation? In their own words,

The main technical change we make is to change the moderator and fuel salt used in previous molten salt reactors to a zirconium hydride moderator, with a LiF-based fuel salt. During operation the fuel in the salt is primarily uranium. Together, these components generate a neutron spectrum that allows the reactor to run using fresh uranium fuel with enrichment levels as low as 1.8% U-235, or using the entire actinide component of spent nuclear fuel (SNF). Previous molten salt reactors such as the ORNL Molten Salt Reactor Experiment (MSRE) relied on high-enriched uranium, with 33% U-235

So far, so good. In order to keep the fuel salt in a liquid state, their outlet temperature of the reactor vessel is around 650 ºC. The lowest possible temperature that the salt can still be liquid is somewhere around 500 ºC. By comparison, the typical outlet temperature of today’s PWR is around 300 ºC.

Transatomic also says that in order to help with the corrosion problem, they’ll be making the piping out of one of the Hastelloys, such as Hastelloy-N, which was used in the original MSRE at Oak Ridge. This iron-nickel alloy has been tested to have extremely good corrosion resistance, especially to fluoride salts, in the temperature range from 704-871 ºC2. The outlet temperature of the MSRE at Oak Ridge was around 550 ºC3 as opposed to Transatomic’s 650 ºC but that’s still probably OK. I do not have any commentary as to whether this lower margin of safety would be more or less acceptable in this power reactor than in ORNL’s reasearch reactor.

The aspect of this design that is most worrying is the zirconium hydride moderator. Zirconium hydride (ZrH2) is used today in reactors everywhere as a cladding material for fuel cells. When not properly cooled, the zirconium sheds its hydrogen molecules, and this pressure buildup can cause an explosion. This is what happened at Fukushima 1 in 2011 when the roof was blown off. Standard-fare ZrH2 would be completely unsuitable for use at the temperatures required of an MSR, so Transatomic say that they are researching the use of a material such as TRIGA fuel cladding, which is UZrH. The temperature resistance of this is good, but still shaky. Transatomic says that up to 750 ºC is safe, and reference this paper from the IAEA.

The problem is that I can’t find a single source where anyone says they’ve actually ran a TRIGA-fuel reactor at a temperature of 750 ºC. The cells are designed for small research reactors, and their typical operating temperature is somewhere in the ballpark of 210 ºC. At 210 ºC, 750 ºC is a totally reasonable margin of safety. At Transatomic’s outlet temperature of 650 ºC, it’s a tough sell.

Best of luck, but they’ve got a long road ahead of them.

  1. Source: I am in the industry.

  2. Source: Hastelloy N alloy spec sheet

  3. Source: MSRE Design and Operations Report, Part I, January 1965

Highball by Studio Neat

New app from Studio Neat. Recommended.

The sharing feature may be the only good use of QR codes I’ve ever seen:

My Vesper recipe. Save this image to import it into Highball
My Vesper recipe. Save this image to import it into Highball

Secure Passwords with Ruby

Last week, Slack had a “security incident” which forced all users to change their passwords and encouraged them to turn on 2-factor authentication. 2FA is by far the best way to secure web-based accounts, but configuring it all for my Slack account made me remember that my other login passwords1 could use some work as well. This timely post on passwords by Dr. Drang gave me a good start.

TL;DR version of the good Doctor’s post: Generate pass phrases using the Diceware method, but their wordlist is cumbersome and has no character. Create your own wordlist by generating it from books found in the Gutenberg Project.

I wanted to make the following changes to Dr. Drang’s method:

To avoid installation of new software, I used Ruby. This allows me to use the realrand gem to generate truly random numbers. Install this gem with

gem install realrand

then inclue it using the following:

require 'random/online'

In my script, I’m getting random numbers from but the gem does provide other options2.

To create my word list, first I downloaded Heart of Darkness by Joseph Conrad as a plain-text file. Then I created a text file with every unique word on a new line using this command:

tr -cs A-Za-z '\n' < heartofdarkness.txt  | tr A-Z a-z | sort | uniq > wordlist-heartofdarkness.txt

After creating wordlist-heartofdarkness.txt I wrote the following Ruby script to generate pass phrases:

On the first run, I got

resist very fishing irritation worrying negligence

That’s much better than anything you’d get from the Diceware list.

Just for fun, I did the same thing with the complete works of Edgar Allan Poe (Vol. 1):

estimate witness knew stiff intent threatened

as well as Prince by Machiavelli:

void captain oppose enlisted provided murderer

These phrases are less secure if the attacker knows your word list, but it’s already worlds better than something you could make up.

Lore Podcast

Lore Podcast

If Night Vale is getting tiresome, give Lore a try. 15 minutes per episode, and all true stories. So far it’s only on Episode 2. Good time to jump in.