Load Flow

Power-flow analysis finds every bus voltage and every line flow for a given generation and load. Learn bus types, the Y-bus, the power-flow equations, Gauss–Seidel and Newton–Raphson methods, and the DC approximation.

Load flow (power flow) is the most-used calculation in power engineering. Given what each generator produces and what each load consumes, it solves the steady-state voltages and angles at every bus, and from those the flows and losses on every line.

Planners use it to check that new connections won't overload lines or push voltages out of limits; operators run it continuously for contingency analysis.

An analogy

A load-flow study is like modelling water pressure in a city's pipe network: you know where water enters and leaves, and you calculate the pressure at every junction and the flow in every pipe.

Bus types

Each bus has four quantities (P, Q, |V|, θ); two are specified and two are solved.

• Slack (swing) bus: |V| and θ = 0 fixed; it supplies whatever makes up losses.
• PV (generator) bus: P and |V| fixed (the AVR holds voltage); Q and θ solved.
• PQ (load) bus: P and Q fixed; |V| and θ solved.

The Y-bus

Line admittances are assembled into Y_bus: each diagonal element is the sum of admittances connected to that bus, and each off-diagonal element is the negative of the admittance between two buses. Y_bus is very sparse for real grids, which the solvers exploit.

Why it's non-linear

Power is V·I*, and the current depends on voltages, so the equations are non-linear in |V| and θ. They must be solved iteratively: guess voltages, compute the power mismatch ΔP, ΔQ at each bus, correct, repeat until mismatches are below a tolerance (e.g. 10⁻⁴ pu).

Gauss–Seidel vs Newton–Raphson

Gauss–Seidel updates one bus voltage at a time; it is simple but converges slowly for large systems. Newton–Raphson linearises all equations with the Jacobian each iteration and converges quadratically, typically in 3–6 iterations regardless of system size — the industry standard. Fast-decoupled load flow approximates the Jacobian (P–θ and Q–V decoupled) for speed.

DC load flow

Assuming flat voltages (|V| = 1 pu), small angles and negligible resistance, P_ij ≈ (θ_i − θ_j)/x_ij. The problem becomes a linear system B′θ = P. It is non-iterative and fast, and used for contingency screening, market clearing (LMP) and transmission planning.

Formulas

• Real power at bus i: P_i = Σ_k |V_i||V_k|(G_ik cos θ_ik + B_ik sin θ_ik)
• Reactive power at bus i: Q_i = Σ_k |V_i||V_k|(G_ik sin θ_ik − B_ik cos θ_ik)
• Newton–Raphson step: [Δθ; ΔV] = J⁻¹ · [ΔP; ΔQ]
• DC load flow: P = B′ · θ, P_ij = (θ_i − θ_j) / x_ij

Worked example: Two-bus angle

Given: Bus 1: V = 1.0∠0° (slack); Bus 2: V = 1.0 pu, P_load = 1.0 pu; Line reactance x = 0.1 pu

Find: Angle at bus 2.

• P = V₁V₂ sin(θ₁ − θ₂)/x → 1.0 = sin(0 − θ₂)/0.1
• sin(−θ₂) = 0.1 → θ₂ = −5.74°

Answer: θ₂ ≈ −5.74°

Worked example: Three-bus DC load flow

Given: x₁₂ = 0.1, x₁₃ = 0.2, x₂₃ = 0.2 pu; Bus 1 slack (θ₁ = 0); P₂ = −1.0 pu (load), P₃ = +0.5 pu (generation)

Find: Angles and line flows.

• B′ (buses 2, 3): [[15, −5], [−5, 10]] · [θ₂, θ₃] = [−1.0, 0.5]
• det = 150 − 25 = 125 → θ₂ = (−10 + 2.5)/125 = −0.06 rad; θ₃ = (7.5 − 5)/125 = 0.02 rad
• P₁₂ = (0 + 0.06)/0.1 = 0.6; P₁₃ = (0 − 0.02)/0.2 = −0.1 (flows 3→1); P₃₂ = (0.02 + 0.06)/0.2 = 0.4
• Checks: bus 2 receives 0.6 + 0.4 = 1.0 ✓; slack supplies 0.6 − 0.1 = 0.5 ✓

Answer: θ₂ = −0.06, θ₃ = 0.02 rad; flows 0.6, 0.4 and −0.1 pu

Common mistakes

• Wrong: Using DC load flow to study voltage problems. — Right: DC load flow assumes 1.0 pu voltages; use AC load flow (Newton–Raphson) for voltage and reactive power.
• Wrong: Forgetting a slack bus. — Right: Losses are unknown until solved, so one bus must balance real power.

Applications

• Connection studies: Every new wind farm, data centre or large load gets a load-flow study to check thermal and voltage limits before connection is approved.
• Contingency analysis: Control-room software re-solves the grid with each line removed in turn (N-1) every few minutes.
• Electricity markets: Locational marginal prices are computed from DC optimal power flow, so prices reflect congestion.

Key takeaways

• Slack, PV and PQ buses define which quantities are known.
• Power-flow equations are non-linear; Newton–Raphson solves them in a few iterations.
• Y_bus captures the network and is highly sparse.
• DC load flow is a fast, linear approximation for real-power flows.

Going further

• Optimal power flow: OPF adds an objective (minimum cost or losses) and inequality constraints (line ratings, voltage limits, generator limits) to the power-flow equations. Solved with interior-point methods, it underpins market clearing and security-constrained dispatch.
• Voltage stability and PV curves: Continuation power flow traces bus voltage as load increases until the 'nose' of the PV curve, where no solution exists — the voltage-collapse limit. Operating margins are measured to that point.
• Probabilistic and time-series load flow: With variable renewables, planners run thousands of load flows over hourly profiles or Monte Carlo scenarios to capture how often limits are breached, not just the worst case.