Function transformer_seq_params 

pub fn transformer_seq_params(
    y_series: Complex<f64>,
    y_shunt: Complex<f64>,
    tap: Complex<f64>,
    from_status: u8,
    to_status: u8,
    winding_from: u8,
    winding_to: u8,
    sn: f64,
    uk: f64,
    s_base_va: f64,
    clock: i32,
) -> ([Complex<f64>; 4], [Complex<f64>; 4], [Complex<f64>; 4])

Computes sequence-domain (y0, y1, y2) branch admittance parameters (each [yff, yft, ytf, ytt]) for a two-winding transformer, needed for asymmetric (3-phase) power flow and for short-circuit calculation.

Positive/negative sequence reuse branch_calc_param with the tap rotated by ±clock (mirrors PGM’s asym_calc_param).

Zero sequence follows PGM’s own general algorithm (component/transformer.hpp, the sc_calc_param/calc_param_asym zero- sequence block) rather than enumerating winding pairs. Four cases, in PGM’s own order — the first three are mutually exclusive, the zigzag pair is additive on top:

• YNyn — the full two-port zero-sequence branch, z0_series = 1/y_series + 3·(z_grounding_to + z_grounding_from/k²), with the tap’s phase picking up a 180° flip for clock ∈ {2, 6, 10} (reverse-connected variants).
• YN* (from side wye_n, in service) — a ground path via the magnetizing branch, plus a path via the series impedance when the other side is delta; a one-port term on yff, scaled by 1/k².
• *yn (to side wye_n, in service) — the mirror, a one-port term on ytt. Dyn is this case, and was the only form of it gridoxide previously implemented.
• ZN*/*zn — a zigzag_n winding’s zero-sequence impedance is taken as 10% of its positive-sequence value.

Then every side without a zero-sequence path of its own (zero_seq_available, private) and in service picks up −j·1e-8·sn/s_base/uk.

Two modelling substitutions

gridoxide has no i0_zero_sequence/p0_zero_sequence fields on PgmTransformer, so the zero-sequence magnetizing admittance is taken as the positive-sequence one, y0_shunt = y_shunt. That is not an approximation this crate invents: it is exactly what PGM computes when those two fields are left unset, since it defaults i0_zero_sequence to i0 and p0_zero_sequence to p0 in that case.

Likewise there are no grounding-impedance fields, so z_grounding_from = z_grounding_to = 0 throughout (PGM defaults them to zero too). That is what collapses YNyn’s z0_series to 1/y_series — i.e. y0_series = y_series — and drops the 3·z_grounding term from the one-port cases, where it would otherwise sit in series with 1/y0.