Pipeline network via Distributed-Source Transmission-Line (DSTL) theory.
A buried pipeline of length L is modelled as a transmission line with:
longitudinal series impedance z (Ω/m) — the pipeline steel + coating
return path,
transverse shunt admittance y (S/m) — leakage to soil through the
coating,
a distributed source term equal to the longitudinal geoelectric field
E (V/m) acting along the pipe.
Following Boteler & Cookson (1986) and Boteler (1997), the pipeline is
discretised into segments of length Δl and each segment is replaced by
its equivalent-π two-port:
In the small-segment limit γ·Δℓ→0 this collapses to the
distributed-parameter model (Z_ser→zΔℓ, Y_sh→yΔℓ/2). Each
segment is emitted as a single Branch with
resistance Real(Z_ser) between end nodes, and shunt admittances are
summed onto earthing_impedance_Ohm.
LPMSolver then solves the whole thing as a
standard LP network.
Insulated (open-circuit) ends are the default: the pipeline endpoints
carry only their half-segment shunt admittance, so ends are grounded
through the coating alone.
For a straight pipeline of length L, uniform longitudinal field E,
and insulated ends, the pipe-to-soil voltage is
\[V(x) = \frac{E}{\gamma}\,\frac{\sinh(\gamma(x - L/2))}{\cosh(\gamma L / 2)},
\qquad x \in [0, L].\]
Peak \(|V|\) sits at the ends and equals \((E/\gamma)\tanh(\gamma L / 2)\).
pipe_to_soil_voltage_analytic() returns this profile;
PipelineNetwork solved via LPMSolver reproduces it to
better than 1 % once n_segments is a few tens.
Discretisation guideline: keep γ·Δℓ=γ·L/n_segments well
below unity — the equivalent-π reduces to the true distributed model
only in the small-segment limit. For typical steel pipelines
(z~3×10⁻⁴ Ω/m, y~10⁻⁶ S/m), γ≈1.7×10⁻⁵ /m; a
100-km pipeline needs n_segments≳30 to get sub-percent agreement
with the closed form.
Interior nodes carry the shunt from two adjacent π-sections
(2·Y_sh); endpoint nodes carry a single half-section
(Y_sh). earthing_impedance_Ohm is the reciprocal.