Where to store key scenario parameters for quick search in databases (e.g. main species / plasma composition)

[olivhoenen]

Hi @imbeauf, unless we've missed an obvious place, there seem to be no placeholder (e.g. in summary or dataset_description) to store information about the plasma composition (in fact the information is scattered across several IDSs: edge_profiles/ggd/ion[:] and core_profiles/profiles_1d/ion[:]).

We have a function in idstools (https://git.iter.org/projects/IMAS/repos/idstools/browse/scripts/printplasmacompo?at=refs%2Fheads%2Fdevelop, sorry this one is still password protected) that calculates the list and amount of species, such that it can get a table like that:

  species:      H         D         T         He3       He4       Be        Ne
  a:            1.0       2.0       3.0       3.0       4.0       9.0       20.0
  z:            1.0       1.0       1.0       2.0       2.0       4.0       10.0
  n_over_ntot:  5.29e-06  0.460     0.493     7.01e-07  0.011     0.024     0.012
  n_over_ne:    4.45e-06  0.387     0.414     5.89e-07  9.58e-03  0.020     0.010
  n_over_n_maj: 1.07e-05  0.933     1.000     1.42e-06  0.023     0.048     0.024

Can we think about a way to store this info in the summary (or dataset_description)?

Are there other key scenario parameters that we are not capturing in these IDSs?

From our legacy DB at ITER, with metadata stored in plain yaml files, it appears that p_sol has no placeholder: it is currently calculated as the sum of power fluxes that go through the segments of the separatrix grid_subset of the GGD in edge_profiles. It could maybe go inside summary/scrape_off_layer.

[imbeauf]

Hi @olivhoenen. The plasma composition is already present in the summary IDS. For instance, in summary/volume_average, you have n_e, n_i_total (sum over ion species) and the n_i structure listing the density of each ion species. The only difference w.r.t. core_profiles is that we have an explicit list of ion species (e.g. "hydrogen", "deuterium", ...) , instead of arrays of structure "ion(:)" - we wanted to avoid using arrays of structure for such pulse metadata.

Regarding your second point, I think p_sol corresponds to global_quantities/power_loss "Power through separatrix"

[olivhoenen]

Thanks @imbeauf ! (for information @prasad-sawantdesai and @hrdoktorx )

[prasad-sawantdesai]

Just to add, these values are calculated for single time slice where there is a maximum central electrons density. Then extracted fraction of each element. [nspec_over_ntot, nspec_over_ne, nspec_over_nmaj ]
Please find code here.
https://git.iter.org/projects/IMAS/repos/idstools/browse/scripts/create_db_entry#444
Summary ids provides storage for each slice to store the values but should we think of storing these values as single structure?

[MireilleSchneider]

I agree that the need is to save the plasma composition as a unique set of scalar variables, to give an insight on the composition of the plasma, without the time-dependence. The idea is to quickly check if this is a H plasma, or D, DT, and if there is He3 for ICRH, W, etc.

[olivhoenen]

Ok thanks for the added comment @MireilleSchneider. So while the some quantities have placeholders in the summary IDS, they are not given in a convenient way for (one of) the foreseen usage (as one would have to identify the time of maximum density and then calculate the relative fraction for all species present at that time), is this the correct understanding?

[hrdoktorx]

The two main use cases I foresee for an eventual plasma composition structure is to be able to answer:

• What is/are the main fuel species in this scenario?
• What is the list of nuclei present in this scenario?
  I would prefer answers as strings, but (atomic number/atomic mass) pairs are also acceptable.

[olivhoenen]

@hrdoktorx @MireilleSchneider is there a more precise definition (threshold) for "main species"?

[hrdoktorx]

No. It is something that is given by the scenario definition.

[MireilleSchneider]

In the create_db_entry script, the crtieria for main species was nspec_over_ntot[ispecies] > 0.45, i.e. its density more than 45% than to total ion density.

[olivhoenen]

We discussed this topic this morning with @SimonPinches and the general through was that for filtering simulations/scenarios what would be most convenient if to have the total density as well as the ratio for each species (right now we only have density per species, but we think that ratio would be more user friendly at the summary level). This would allow users to decide their own threshold. Would that work for you?

[hrdoktorx]

When you say total density, what do you mean? Where is this density measured?

[MireilleSchneider]

The total density is calculated from the centre to the last closed flux surface, so you are out ;-).

[olivhoenen]

Averaged over the plasma volume? https://imas-data-dictionary.readthedocs.io/en/latest/generated/ids/summary.html#summary-volume_average-n_i_total?

[MireilleSchneider]

Yes

[hrdoktorx]

If we are using a threshold, I would lower it from 45% to something like 40% or even 35%. JET routinely ran DT mixtures in the 40-60 range, and ITER likely will do so as well.

[olivhoenen]

A reason to not select a given threshold and only identify species with their ratio, so user can decide the range they want for their study.

[MireilleSchneider]

It could be an option but I think there should be a default threshold to facilitate the access of the information of the main species. I think 35% should be fine: there cannot be 3 main species with 35% threshold, so it is high enough to be representative.

[imbeauf]

So, to summarize, you want to have a constant structure in the Summary IDS that gives exactly the same information as the table displayed by Olivier at the starting point of the discussion (list of elements + density ratios at the time of maximum central electron density).

Plus the list of main ion species, being defined as volume average density above 35 % of the total ion density, evaluated at the time of maximum central electron density.

Should we go for this ?

[SimonPinches]

Do we need to define what constitutes a "main species" or can this be done by the user when they make a search if we store the species and their fraction?

[hrdoktorx]

We already settled on a definition above that a main species is one that has a concentration of at least 35%.

[olivhoenen]

Storing density ratios for each species being present as well as a list of main species is redundant so I would prefer to have only the first one (at least I don't see what kind of query could be made in one case and not in the other one, do you have an example?).

[olivhoenen]

Last discussion concluded that the most relevant and flexible way to store plasma composition was to only keep n_over_ne from the opener table, and letting the user define what threshold to use for considering them to be main species.