Foundation
A transmission line is, strictly speaking, a distributed-parameter system: every infinitesimal segment along its length has its own resistance, inductance, capacitance, and conductance. For lines short relative to the electromagnetic wavelength, this distributed system admits a simple lumped equivalent — the π-model — consisting of a series impedance flanked by two shunt admittances. This page derives the π-model from the telegrapher’s equations and explains why it is the universal building block of bus admittance matrices in power-flow software.
Consider a uniform transmission line of length ℓ with per-unit-length parameters: series resistance r, series inductance ℓ, shunt conductance g, and shunt capacitance c. At position x along the line, let v(x,t) and i(x,t) denote voltage and current. Applying Kirchhoff’s laws to an infinitesimal segment dx and taking the limit dx → 0 gives the telegrapher’s equations:
These are coupled first-order PDEs in space and time, describing how voltage and current waves propagate along the line. The series term r + jωℓ captures conduction losses and magnetic energy storage; the shunt term g + jωc captures dielectric leakage and electric energy storage between the conductor and ground.
In sinusoidal steady state at angular frequency ω, the time derivatives become multiplication by jω. The telegrapher’s equations reduce to ordinary differential equations in x for the voltage and current phasors V(x), I(x):
where z = r + jωℓ is the series impedance per unit length and y = g + jωc is the shunt admittance per unit length. Differentiating once more and substituting yields the wave equation:
The complex constant γ is the propagation constant, and Zc = √(z/y) is the characteristic impedance. The solution along a line of total length ℓ relates sending-end and receiving-end phasors through the ABCD matrix:
This is the exact two-port description of the line in steady state. The hyperbolic functions encode wave reflections, voltage rise along uncompensated lines, and frequency-dependent behavior.
For short and medium-length lines (typically under 250 km for 60 Hz transmission), the hyperbolic terms can be expanded and truncated, giving an exact two-port equivalent. This lumped equivalent takes the form of a π-shaped circuit: a series admittance Ys connecting the sending and receiving ends, and two equal shunt admittances Ysh/2 at each end. The exact formulas are:
For short lines where |γℓ| ≪ 1, using sinh(x) ≈ x and tanh(x/2) ≈ x/2:
The series admittance is simply the reciprocal of the total line impedance Z = R + jX, where R = rℓ and X = ωℓℓ. The shunt at each end is half the total line admittance. This is the standard π-model that appears in every MATPOWER case file and every textbook power-flow problem.
With the π-model fixed, the branch’s contribution to the bus admittance matrix follows directly. Let bus i be the sending end and bus j the receiving end. The currents at each end, in terms of bus voltages Vi and Vj, are:
Collecting into matrix form, the 2×2 branch admittance matrix is:
This block is what the Ybus construction assembles into the full network matrix. Multiple parallel branches simply sum their Ybranch blocks at the corresponding bus pair.
A transformer modifies the π-model in two ways: the tap ratio T (the off-nominal turns ratio, including phase shift if present) and the magnetizing impedance. For practical OPF, the magnetizing branch is usually neglected and only the tap ratio matters. With a complex tap T on the from-side, the branch admittance matrix becomes:
For a unit tap T = 1, this reduces to the ordinary line π-model. Phase-shifting transformers introduce an imaginary part in T, breaking the symmetry of Ybranch but preserving the underlying complex-power balance. Tap-changing under load (TCUL) transformers are modeled by treating T as a decision variable in some advanced OPF formulations, though Lirion’s centralized models treat it as a fixed parameter.
For lines longer than approximately 250 km at 60 Hz, the truncation sinh(x) ≈ x introduces errors above 1%, and the full hyperbolic π-model should be used. Voltage rise on lightly-loaded long lines (the Ferranti effect) becomes significant — receiving-end voltage can exceed sending-end voltage by 10% or more on a 500-km uncompensated line. In such cases, either compute Ys and Ysh/2 from the exact hyperbolic formulas, or split the line into multiple shorter π-sections.
For HVDC links, the π-model is replaced entirely by current source converter or voltage source converter models that capture the rectifier/inverter behavior at each terminal. These are outside the scope of standard AC OPF and require dedicated multi-terminal HVDC formulations.