Keyboard shortcuts

Press ← or → to navigate between chapters

Press S or / to search in the book

Press ? to show this help

Press Esc to hide this help

Certify, optimize and compare designs

Three pages under STUDIES in the desktop app answer design questions by solving many calculations around the open problem:

PageQuestionCommand-line equivalent
CertifyDoes the design meet its limits once the nuclear-data uncertainty is counted, and which measurement would narrow the answer most?actinv decide
OptimizeWhich design, within the ranges you give, does best on one quantity while meeting every limit?actinv optimize
CompareHow do materials, spectra or irradiation lengths compare?actinv study run

Each page builds the input file the command reads from the open problem, runs the command in the background, and draws its result. Values are per gram of material. In the browser workbench the pages set the study up and download the file instead of solving. The schemas and limits of the commands are in Advanced workflows and the command line.

Set up the material, the irradiation and the source first. The pages that need something from the problem say what is missing and offer a button to the page that sets it.

Certify

Certify checks a list of limits against the uncertainty band of the nuclear data. A design passes a limit only if the edge of the band, not just the nominal value, is on the right side of it.

What it needs. The problem must have Uncertainty switched on under What to compute, with a cross-section covariance file chosen. The flux-only uncertainty mode cannot certify. Until both are set, the page shows a warning and an Open What to compute button. The schedule must also have at least one step.

What to fill in. Under Limits, each limit is a card:

  • a name, which must be unique;
  • the quantity: Decay heat, Gamma heat, Beta heat, Alpha heat, Activity, or Activity of one nuclide (type the nuclide, such as Co60);
  • at most or at least, and the value, in W/g for heats and Bq/g for activities;
  • the time, chosen from the schedule’s step ends and named by what they mean, such as “at shutdown” or “1 y after shutdown”;
  • the edge it is checked at: Conservative band edge, Normal band edge or Nominal value only.

The conservative edge adds the declared unmodelled share to the band. The normal edge is the band at the problem’s confidence level. Nominal ignores the uncertainty and is offered for comparison. Choose + Limit for more cards, and set how many measurements to suggest per limit (1 to 20, default 5).

Certify solves the problem once with bands, then checks every limit. Progress shows in the status bar.

What the answer shows. The verdict is Certified or Not certified, with a line such as “2 of 3 limits hold at their band edges, with the nuclear-data uncertainty at 95% confidence”. A warning counts limits that pass on the nominal value but fail with the uncertainty, which a calculation without bands would have certified. A second warning appears if the run is incomplete because some reaction or decay data were missing, in which case the band may be too narrow. The table has one row per limit: Name, Quantity and time, Value at the band edge, Limit, Margin (how far under or over the limit) and Without uncertainty. What to measure next ranks the cross sections, half-lives, fission yields or whole reactions whose measurement would narrow each band most, each as a share of the band’s variance that a perfect measurement of that one quantity would remove.

Optimize

Optimize searches within ranges you set for the design that makes one quantity as low or as high as possible while meeting the limits.

What it needs. A material with elements and a schedule with at least one step. Uncertainty is optional: without it, limits and goal are judged on the nominal value, and the page says so.

What to fill in.

  • What the search may change lists up to 6 things, each with a range. Choose + Something to vary to add one.

    • Element content: an element’s weight percent, from one value to another. The other elements are scaled so the total stays 100 wt%, so at least one element must be left unvaried. The composition must be in weight percent.
    • Flux: the total flux as a multiple of the problem’s, from a lower to a higher positive multiple. The source needs a total flux.
    • Step length: one schedule step’s length, from one duration to a longer one, such as 30 d to 2 y.

    Each element or step can be varied once.

  • Goal is Lowest or Highest of a quantity (the same list as Certify) at a time in the schedule, and, when the problem has uncertainty, judged at a band edge.

  • Limits are optional and are the same cards as on Certify. A design counts only if it meets every limit.

  • Search sets how many designs to Try spread over the ranges (default 12), and how many more around the best so far (default 8). The total is at most 64 designs; above that the page shows the count in red and Search designs stays disabled. Each design is one solve, or one solve with bands when the problem has uncertainty. The search is seeded, so the same inputs try the same designs.

Search designs runs the search; the status bar counts designs.

What the answer shows. Best design found gives the value of the goal and how many designs were tried. A table lists each varied thing with the Best design value beside The problem now. Use this design puts those values into the open problem; it takes effect like any edit, so run it from Check & run, and use Undo edit to go back. If no design meets every limit, the page says No design meets every limit: widen the ranges, loosen a limit or try more designs. Every design tried lists all designs ranked, and is open by default in that case.

Compare

Compare solves every combination of the materials, spectra and irradiation lengths you choose and ranks the results side by side, each solved as Run solves it.

What to fill in. Three columns:

  • Materials: This problem’s material, plus any library materials added with Add a library material. The library is the one on the Material page.
  • Spectra: This problem’s spectrum, plus any of the four standard shapes (D-T fusion (14.1 MeV), Fission (Watt), Thermal (Maxwellian), Slowing down (1/E)). The standard shapes take the problem’s total flux.
  • Irradiation: This problem’s irradiation, plus further lengths typed after Also irradiate at full flux for, such as 30 d, 5 y. Each irradiation is followed by this problem’s cooling.

“This problem’s” entries are unavailable, with the reason on hover, when they cannot be compared: a composition not in weight percent, an empty material, a spectrum without flux or one that is rebinned or set by wall loading, or an irradiation with power, wall-loading or library-switch steps. Each case is reported at shutdown and at the end of each cooling step. The page shows the number of combinations (materials times spectra times irradiations). Compare needs at least 2 and at most 200 combinations in the app; a study file run with actinv study run takes up to 1024.

What the answer shows. Ranking shows one quantity (Activity or Decay heat) at one time, per gram, lowest first. Only the columns that differ between cases are shown. Each row gives the value, Times the lowest and a bar for scale. A case that did not finish is listed last as “did not finish”.

Compare answers “which is lower” for the combinations you pick. It does not search between them (use Optimize) and does not apply uncertainty bands (use Certify).