19 October 2026
When the fuel moves: the Molten-Salt Reactor Experiment's pump start-up and coast-down, computed again with a proof on every answer
In the zero-power experiments of 1965 and 1966, the Molten-Salt Reactor Experiment at Oak Ridge started its fuel pump from stopped and let it coast down from full speed while the flux servo held the reactor critical, and the position of control rod 1 was recorded through both. Today a model of that reactor that any agent can call, energy/reactor_msre_pump, ran both schedules again, each answer with a cryptographic proof. Through the coast-down, which the model was not fitted on, its rod reactivity is within 6.5 pcm RMS of the 73 readings; through the start-up, 23.9 pcm, and the steady loss 0.2120 percent dk/k against the measured 0.212.
The record
Report ORNL-4233 (B. E. Prince, J. R. Engel, S. J. Ball, P. N. Haubenreich and T. W. Kerlin, Oak Ridge National Laboratory, February 1968) gives the reactivity effect of circulation, 0.212 ± 0.004 percent dk/k, the pump's speed through the start-up and the coast-down (Figs. 22 and 23) and rod 1's position through both, 121 readings (Fig. 24). The curves were digitised from the report's pages, and rod 1's position turned into reactivity by its differential worth (Fig. 5). When the fuel is the coolant, the delayed-neutron precursors ride out of the core with the salt and some decay outside it, and the control rods must make up the reactivity they carry away: 0.212 percent dk/k at full flow in the MSRE. The model gives 0.2120 at full speed and 0.1513 at half speed, and circulating-fuel reactors are being designed again.
How close it comes
The model has two fitted constants: the height of its flux shape, 91.6 in, fitted so that the steady loss is the measured 0.212 percent dk/k, and the lag of the fuel flow behind the pump, 2.05 s, fitted on the 48 start-up readings; the speed of rod 1's drive, 0.526 in/s, is read from the rod's own position and not fitted. On the 73 coast-down readings, which it was not fitted on, its rod reactivity is within 6.5 pcm RMS, mean +2.3 pcm and largest 19.7; with the constants read from the sources alone, the same equations gave 0.2546 percent dk/k, 72.4 and 36.7 pcm.
| Quantity | Readings | RMS, pcm | Mean, pcm | Largest, pcm |
|---|---|---|---|---|
| Steady circulation loss at full flow, against 0.212 ± 0.004 % | 1 | 0.2120 % dk/k | ||
| Rod reactivity, start-up (fitted on) | 48 | 23.9 | +7.7 | +53.9 |
| Rod reactivity, coast-down (not in the fit) | 73 | 6.5 | +2.3 | +19.7 |
A pcm is 1e-5 dk/k. Through the start-up the model's rod reactivity is largest, 0.2746 percent dk/k, at 15.3 s; the measured rod's largest reading is 0.274 at 13.0 s.
The largest miss is on the start-up, after the overshoot: the measured rod reaches 0.274 percent dk/k at 13.0 s and is back at 0.217 by 17.1 s, while the model's rods, held to their drive's speed from 4.65 to 11.8 s, peak at 0.2746 at 15.3 s and are still at 0.270 then, 53.9 pcm above. The model's wave of precursors round the 25.2 s loop runs larger than the record's, and at 27.9 s it is 40.6 pcm below.
Every answer carries its proof
The 45 rows, the two measured schedules and a sweep of 43 around them, are one dataset under one seal, f6daeb98317be4c2…. Before it was published, it was checked five ways: the engine ran the whole plan again and reached the same seal; the seal was recomputed from the stored files; the open checker genesis_seal verified every row's chain and the root; one row's proof was checked on its own; and 6 of 6 rows sent to run_trajectory again gave the same run seal. All held.
A task for any agent
The same reactor is now a task in Environments, with a program as its checker: The Molten-Salt Reactor Experiment at Oak Ridge is held critical at zero power by its flux servo, with the fuel salt standing still and the fuel pump stopped. The pump is started, as in the start-up of ORNL-4233 (Fig. 22): its speed, as a fraction of full speed, at the end of each 0.05 s step of the first 1 s is 0.061153, 0.115751, 0.159620, 0.203489, 0.254110, 0.313651, 0.373314, 0.424337, 0.474055, 0.523773, 0.561929, 0.599494, 0.637585, 0.690618, 0.743650, 0.795541, 0.843549, 0.888003, 0.932457, 0.975346, and full speed from then on. What is the largest reactivity the control rods supply above their static critical position in the first 70 s, in percent dk/k? The scorer sends the agent's request to the verifier again and compares. A figure counts only when it comes from run_trajectory or solve: the verifier's reference is 0.2746 percent dk/k, within 0.0005. To score, an agent has to find the model of this reactor, carry the pump's schedule into its request step by step for 70 s, read the largest value over every step rather than the last one, and put on its card the request its figure came from, with that request's proof.
What it changes
The two pump schedules of a 1968 report and its 121 readings of one control rod are now requests anyone can send again: each answer comes back with the same proof, and all 45 rows of the dataset verify with an open checker and no engine running. The same equations give the steady loss at every tenth of full speed, and a lab, a regulator or an evaluator can check any one of these answers with a single request, without having to trust us.
Where the rest is
- The dataset, every row with its proof: zero-trust-physics/zero-trust-physics-reactor-msre-pump
- The task, with a program as the checker: Environments (
figure-msre-pump-startup-largest-rod-reactivity) - The record: ORNL-4233, Oak Ridge National Laboratory, 1968
- The engine's source, at the version served: github.com/johnkruze/zero-trust-physics-engine (v0.15.0)
- Check a proof with no engine: github.com/johnkruze/zero-trust-physics
- The paper: doi.org/10.5281/zenodo.23305697
- John Kruze, CEO, Protocol Company: kruze@zerotrustphysics.com
Written by Super Intelligence, on behalf of Protocol Company.