DC load flow replaces the nonlinear power flow equations with a linear system, P = B·θ, solved in one step. It assumes negligible resistance, small angles and 1 pu voltages. It returns active power flows, not voltages or losses. It works on meshed transmission networks and fails on distribution feeders, where resistance weighs as much as reactance.
Short answer
DC load flow is a linear version of the power flow. Instead of solving nonlinear equations by iteration, it assumes three things: line resistance is negligible against reactance, the angle differences between buses are small, and every voltage is 1 pu. With those three assumptions, the active power on a line is the angle difference between its ends divided by its reactance, and the whole system is solved with a single matrix operation.
The result is flows in megawatts and bus angles, with no voltages, no reactive power and no losses. It is useful for checking many contingencies quickly on meshed high-voltage networks. It is not useful on distribution networks. For the full study with voltages, see the load flow study.
What the approximation assumes
It starts from the full equation for the active power between two buses, which depends on the voltage magnitudes, the line impedance and the sine of the angle difference. Three assumptions simplify it:
- Negligible resistance. Only the reactance x of each line is used. The line susceptance becomes b = 1/x.
- Small angles. If the angle difference is a few degrees, its sine is almost equal to its value in radians.
- Voltages of 1 pu. Magnitudes are not calculated; they are fixed at their nominal value.
The active flow from bus i to bus j becomes Pij = (θi − θj) / xij. Taken together, the net injections of the buses (generation minus load) are related to the angles by P = B′ · θ, where B′ is the susceptance matrix of the network. The linear formulation with each flow limited by line capacity is the one used in dispatch and planning models.
How it is solved
- Choose a reference (slack) bus with zero angle, which absorbs the power balance. See the bus types in a load flow.
- Build B′: on the diagonal, the sum of the susceptances of the lines connected to each bus; off the diagonal, the susceptance of the line joining each pair, with negative sign. Remove the row and column of the reference bus.
- Solve B′ · θ = P for the angles of the other buses.
- Calculate each line flow with the formula above and compare it with its limit.
- The reference bus supplies whatever is missing to balance the system.
All data is converted to per unit on the same base before the matrix is built.
Worked example
Example data (assumed values, not measured): three-bus network, 100 MVA base. Reactances: line 1-2, 0.10 pu; line 1-3, 0.20 pu; line 2-3, 0.10 pu. Bus 1 is the reference. Bus 2 has 60 MW of generation (0.6 pu) and bus 3 has a 120 MW load (−1.2 pu). Line 2-3 is limited to 70 MW.
1. Susceptances. 1/0.10 = 10 pu on lines 1-2 and 2-3; 1/0.20 = 5 pu on line 1-3.
2. Reduced matrix (buses 2 and 3). B′₂₂ = 10 + 10 = 20; B′₃₃ = 5 + 10 = 15; B′₂₃ = −10. The system is: 20·θ₂ − 10·θ₃ = 0.6 and −10·θ₂ + 15·θ₃ = −1.2.
3. Angles. The determinant is 20·15 − 10·10 = 200. θ₂ = (15·0.6 + 10·(−1.2)) / 200 = −0.015 rad. θ₃ = (10·0.6 + 20·(−1.2)) / 200 = −0.090 rad.
4. Flows. Line 1-2: (0 − (−0.015)) / 0.10 = 0.15 pu = 15 MW. Line 1-3: (0 − (−0.09)) / 0.20 = 0.45 pu = 45 MW. Line 2-3: (−0.015 − (−0.09)) / 0.10 = 0.75 pu = 75 MW.
5. Check. At bus 3, 45 + 75 = 120 MW comes in, equal to the load. At bus 2, 0.6 = −0.15 + 0.75. The reference supplies 15 + 45 = 60 MW.
6. Result. Line 2-3 carries 75 MW and its limit is 70 MW: it is overloaded by 5 MW.
Distribution factors (PTDF)
A PTDF tells what fraction of 1 MW injected at a bus, and withdrawn at the reference, flows through a line. In the example it is obtained by solving the same system with a unit injection. Injecting 1 pu at bus 2: θ₂ = 15/200 = 0.075 and θ₃ = 10/200 = 0.050, and the flow on line 2-3 is (0.075 − 0.050)/0.10 = 0.25. The PTDF of bus 2 on line 2-3 is 0.25: a quarter of whatever bus 2 generates goes through that line, and the rest goes straight to the reference.
To relieve the 5 MW overload it is enough to lower the generation at bus 2 by 5 / 0.25 = 20 MW, from 60 to 40 MW. Check: with 0.4 pu, θ₂ = −0.03 and θ₃ = −0.10, and the flow on line 2-3 is 0.70 pu = 70 MW. With the PTDF table of the whole network, a redispatch or a line outage can be evaluated without rebuilding the system.
Where it fails: distribution networks
The first assumption requires resistance to be much smaller than reactance. On high-voltage lines the R/X ratio is usually small. On a medium-voltage feeder it can be 1 or more, and the three consequences show immediately.
Example data: a 1.0 pu source feeds a 50 MW, 10 Mvar load through one line (100 MVA base). Case A, transmission line: r = 0.01 and x = 0.10 pu (R/X = 0.1). Case B, feeder: r = 0.10 and x = 0.05 pu (R/X = 2). The full power flow (solved iteratively) and the DC flow give:
| Case A (R/X = 0.1) | Case B (R/X = 2) | |
|---|---|---|
| Load angle, full power flow | −2.86° | −0.91° |
| Load angle, DC flow (−P·x) | −2.86° | −1.43° |
| Load voltage, full power flow | 0.984 pu | 0.941 pu |
| Losses, full power flow | 0.27 MW | 2.93 MW |
| Voltage and losses, DC flow | 1 pu and 0 | 1 pu and 0 |
In case A the DC method gets the angle right, and losses are under 1 % of the load. In case B the DC angle is 57 % larger than the real one (1.43° against 0.91°), the voltage drops almost 6 % and the losses are almost 6 % of the load, and the method sees neither. Also, in a radial network the flow in each section is the sum of the loads downstream: there is no need to solve angles, and what decides the design is voltage drop, which only a full power flow calculates.
When to use it and when not to
- Yes: meshed high and extra-high voltage networks, fast N-1 contingency screening, planning and dispatch with many cases, and as a starting point for a full power flow.
- No: voltage regulation, reactive compensation, losses, radial networks and distribution feeders, or any network where R/X is not small. For low voltage in a real case, see how to correct low voltage in a load flow.
Mistakes to avoid
- Applying the method to a medium-voltage feeder and trusting its flows and voltages.
- Forgetting that PTDFs depend on the reference bus chosen.
Checklist
- Meshed network and small R/X ratio on every line in the study.
- Reactances of all lines on the same base.
- Reference bus defined and power balance closed.
- Flows compared with the limit of each line.
- N-1 contingencies checked with the PTDFs.
- Critical cases confirmed with a full load flow.
- Voltages and losses taken only from the full load flow.
Related guides
- Load flow study: what it delivers and when it is required
- Gauss-Seidel and Newton-Raphson in load flow
- Bus types in a load flow: swing, PV and PQ
- Correcting low voltage in a load flow
- AC voltage drop in cables
- Per-unit system in a short-circuit study
References
- Calculations and data: our own, with assumed data for the example.
Need this study?
Full load flows, contingency reviews and checks of loading and voltages on your network are part of our engineering design and studies service. We do it for projects in any country. Write to us.
Frequently asked questions
What does DC load flow return, and what does it leave out?
It returns the voltage angle at every bus and the active power flow on every line. It does not return voltage magnitudes, reactive power or losses, because the three assumptions of the method remove them. That is why it cannot be used to check voltage regulation or reactive compensation.
What is a PTDF and what is it used for?
It is the share of one megawatt injected at a bus, and withdrawn at the reference bus, that flows through a given line. With a table of PTDFs you can calculate how every flow changes when generation or load is moved, without solving the system again, and you can see which generator to redispatch to relieve an overloaded line.
Why does DC load flow fail on distribution feeders?
Because the first assumption, resistance negligible against reactance, does not hold: on a distribution feeder the resistance is of the same order as the reactance or larger. Also, in a radial network the flow in each section is already known by adding up the loads downstream, and the real problem is voltage drop, which the DC method does not calculate.