Power-Flow Analysis — Problem-Type Solving Guide
This guide reorganizes everything your professor solved in class into 5 distinct problem
types. For each type you get: what it is, the method, the step-by-step procedure, the
formulas in the order you use them, and a worked reference taken directly from the class
examples.
All examples are worked in per-unit (p.u.), and all impedances in the examples are purely
reactive (so most numbers come out as pure imaginaries like $-25j$).
Quick map of the 5 problem types
Class
# Problem type Trigger / what’s asked
examples
Off-nominal-tap
“Model this tap-changing transformer” /
1 transformer → π- Ex. 1
find its shunt & series admittances
equivalent
Build the bus admittance “What is the admittance matrix?” given Ex. 2, Ex.
2
matrix $Y_{bus}$ line/shunt data 3
Newton–Raphson load- Classify buses, find unknown
3 Ex. 4
flow solution $V,\theta,P,Q$
Per-unit base conversion Multi-transformer circuit, “convert to p.u.
4 Ex. 5
across zones and solve”
Two-bus line power $P_{12}, Q_{12}$, angle for max power / Ex. 6, Ex.
5
transfer unity PF / losses / power balance 7
A long exam problem is usually a chain of these: model the transformer (Type 1) → build
$Y_{bus}$ (Type 2) → run N–R (Type 3). Types 4 and 5 are usually standalone.
Type 1 — Off-nominal-tap transformer → π-
equivalent
What it is
A transformer whose turns ratio is not the nominal 1:1 (e.g. 1:1.05) cannot be represented by
a single series admittance. It is replaced by an equivalent π-circuit: one series branch plus
a shunt branch at each end.
Set-up
Let the transformer’s own series admittance be $y = \dfrac{1}{Z}$ and let $t$ be the off-
nominal tap ratio (the side written as $1:t$).
Formulas (use in this order)
1. Series admittance of the transformer: $$y = \frac{1}{Z}$$
2. The three π-elements:
Where it
Element Formula
connects
between the two
Series admittance $\displaystyle y_{12} = \frac{y}{t}$
buses
Shunt (primary
$\displaystyle y_{10} = y\left(1-\tfrac{1}{t}\right)$ bus 1 → ground
side)
Shunt (secondary $\displaystyle y_{20} = y\left(\tfrac{1}{t^{2}}-
bus 2 → ground
side) \tfrac{1}{t}\right)$
Procedure
1. Compute $y = 1/Z$.
2. Read off the tap ratio $t$.
3. Plug into the three formulas above.
4. Draw the π: series $y/t$ in the middle, shunts $y_{10}$ and $y_{20}$ to ground at each
end.
Worked reference (Example 1)
$Z = 0.05j$, tap $1:1.05$ so $t = 1.05$.
$$y = \frac{1}{0.05j} = -20j$$ $$y_{10} = -20j\left(1-\tfrac{1}{1.05}\right) = -0.95j$$
$$y_{20} = -20j\left(\tfrac{1}{1.05^{2}}-\tfrac{1}{1.05}\right) = +0.91j$$ $$y_{12} =
\frac{-20j}{1.05} = -19.05j$$
Note the sign flip: $y_{20}$ comes out positive imaginary even though the transformer
is inductive — this is normal for the secondary shunt of a step-up tap.
Type 2 — Build the bus admittance matrix $Y_{bus}$
What it is
Given the branch impedances/admittances of a network, assemble the matrix $\mathbf{Y}$
in $\mathbf{I} = \mathbf{Y},\mathbf{V}$, where each bus current is $\bar I_k =
\sum_{h=1}^{n} Y_{kh}\bar V_h$.
The two rules (memorize these)
Diagonal element $Y_{kk}$ = sum of all admittances connected to bus $k$ (every series
branch leaving bus $k$, plus every shunt at bus $k$): $$Y_{kk} = \sum_{\substack{h=0 \ h
\neq k}}^{n} y_{kh} \quad(\text{includes shunt } y_{k0})$$
Off-diagonal element $Y_{kh}$ = negative of the admittance directly between buses
$k$ and $h$: $$Y_{kh} = -,y_{kh}$$
If two buses are not directly connected, $Y_{kh}=0$. The matrix is symmetric
($Y_{kh}=Y_{hk}$) for a network of passive branches.
Procedure
1. Convert every branch impedance to admittance: $y_{kh} = 1/Z_{kh}$.
2. Fill every off-diagonal: $Y_{kh} = -y_{kh}$.
3. Fill every diagonal: add up all admittances (branches + shunts) touching that bus.
4. If a branch is an off-nominal transformer, first run Type 1 to get $y_{12}, y_{10},
y_{20}$, then feed those into the rules above:
$y_{12}=y/t$ acts as the series branch between the two buses,
$y_{10}$ adds to the primary bus diagonal,
$y_{20}$ adds to the secondary bus diagonal.
Worked reference (Example 2 — no transformer)
$Z_{12}=0.04j,; Z_{13}=0.25j,; y_{10}=0.05j,; y_{20}=0.05j$.
$$y_{12}=\tfrac{1}{0.04j}=-25j, \qquad y_{13}=\tfrac{1}{0.25j}=-4j$$
$$Y_{11}=y_{10}+y_{12}+y_{13}=0.05j-25j-4j=-28.95j$$ $$Y_{22}=y_{20}+y_{21}=0.05j-
25j=-24.95j$$ $$Y_{33}=y_{31}=-4j$$ $$\mathbf{Y}=\begin{bmatrix}-28.95j & -25j & -4j\
-25j & -24.95j & 0\ -4j & 0 & -4j\end{bmatrix}$$
Worked reference (Example 3 — with a tap transformer between bus 2 and
bus 4)
First the transformer (Type 1) with $Z=0.015j,;t=1.05$: $$y_{T,24}=\frac{1/0.015j}
{1.05}=-63.49j,\quad y_{T,20}=\tfrac{1}{0.015j}!\left(1-\tfrac1{1.05}\right)=-3.17j,\quad
y_{T,40}=\tfrac{1}{0.015j}!\left(\tfrac1{1.05^2}-\tfrac1{1.05}\right)=+3.02j$$
Then the diagonals that involve the transformer: $$Y_{44}=y_{40}+y_{42}+y_{T,40}=-8j+
(-63.49j)+3.02j=-68.47j$$ $$Y_{22}=y_{20}+y_{21}+y_{24}+y_{T,20}=0.05j-25j-63.49j-
3.17j=-91.61j$$
giving the full $4\times4$: $$\mathbf{Y}=\begin{bmatrix}-28.95j & -25j & -4j & 0\ -25j &
-91.61j & 0 & -63.49j\ -4j & 0 & -4j & 0\ 0 & -63.49j & 0 & -68.47j\end{bmatrix}$$
Type 3 — Newton–Raphson load-flow solution (the
centerpiece)
What it is
Solve the nonlinear power-flow equations for the unknown bus voltage magnitudes and
angles, using one (or more) Newton–Raphson iterations from a flat start.
Step 0 — Classify the buses
Bus type Known (specified) Unknown (solve for)
Slack (reference) $V$ and $\theta,(=0)$ $P,;Q$
Generator / PV $V$ and $P$ $\theta,;Q$
Load / PQ $P$ and $Q$ $\theta,;V$
Interconnection $P=0,;Q=0$ $\theta,;V$
The state vector $x$ collects every unknown $\theta$ (all non-slack buses) and every
unknown $V$ (PQ buses only). You write one $P$-equation per PV+PQ bus and one $Q$-
equation per PQ bus.
Step 1 — Split $Y = G + jB$
Take real and imaginary parts of $Y_{bus}$: $$Y = G + jB \quad\Rightarrow\quad
G=\mathrm{Re}(Y),; B=\mathrm{Im}(Y)$$
Step 2 — Power-flow (injection) equations
$$P_k = \sum_{h=1}^{n} V_k V_h\big[G_{kh}\cos(\theta_k-\theta_h)+B_{kh}\sin(\theta_k-
\theta_h)\big]$$ $$Q_k = \sum_{h=1}^{n} V_k V_h\big[G_{kh}\sin(\theta_k-\theta_h)-
B_{kh}\cos(\theta_k-\theta_h)\big]$$
Factor of 3: the course writes these with a leading $3\sum$ (three-phase power). In
per-unit the factor of 3 is absorbed by the base, so the worked examples drop it and use
$\sum$. Match whichever your professor uses — the numbers are identical in p.u.
Step 3 — Flat start, evaluate the injections
Initial guess: all unknown $\theta = 0$, all unknown $V = 1$ p.u. (PV/slack keep their fixed
values). Plug into Step 2 to get $P_k(x_0)$ and $Q_k(x_0)$.
Step 4 — Mismatch (residual) vector $\Delta$
$$\Delta P_k = P_k^{\text{spec}} - P_k(x_0), \qquad \Delta Q_k = Q_k^{\text{spec}} -
Q_k(x_0)$$ Stack them: $\boldsymbol{\Delta} = \big[\Delta P,;\Delta Q\big]^{!\top}$.
Convergence test: stop when $\max|\Delta| < \varepsilon$ (a small tolerance).
Step 5 — Build the Jacobian $J$
$$J\begin{bmatrix}\Delta\theta\ \Delta V\end{bmatrix}=\boldsymbol{\Delta}, \qquad
J=\begin{bmatrix} H & N\ M & L\end{bmatrix} =\begin{bmatrix}\partial P/\partial\theta &
\partial P/\partial V\[2pt] \partial Q/\partial\theta & \partial Q/\partial V\end{bmatrix}$$
Element formulas (with $\theta_{ik}=\theta_i-\theta_k$):
Diagonal ($i=i$): $$H_{ii}=-Q_i-B_{ii}V_i^{2} \qquad N_{ii}=\frac{P_i}{V_i}+G_{ii}V_i$$
$$M_{ii}=P_i-G_{ii}V_i^{2} \qquad L_{ii}=\frac{Q_i}{V_i}-B_{ii}V_i$$
Off-diagonal ($i\neq k$): $$H_{ik}=V_iV_k\big(G_{ik}\sin\theta_{ik}-
B_{ik}\cos\theta_{ik}\big)$$
$$N_{ik}=V_i\big(G_{ik}\cos\theta_{ik}+B_{ik}\sin\theta_{ik}\big)$$ $$M_{ik}=-
V_iV_k\big(G_{ik}\cos\theta_{ik}+B_{ik}\sin\theta_{ik}\big)$$
$$L_{ik}=V_i\big(G_{ik}\sin\theta_{ik}-B_{ik}\cos\theta_{ik}\big)$$
Step 6 — Solve and update
$$\begin{bmatrix}\Delta\theta\ \Delta V\end{bmatrix}=J^{-1},\boldsymbol{\Delta}
\qquad\Rightarrow\qquad x^{(1)} = x^{(0)} + \begin{bmatrix}\Delta\theta\ \Delta
V\end{bmatrix}$$
Step 7 — Iterate
Re-evaluate Step 2 with the updated $x$, recompute the mismatch, and repeat Steps 4–6
until $\max|\Delta|<\varepsilon$. Finally back-substitute to recover the slack-bus $P,Q$ and
any PV-bus $Q$.
Worked reference (Example 4)
3-bus system: bus 1 slack ($V_1=1,\theta_1=0$), bus 2 PV ($V_2=1.05,;P_2=0.8$), bus 3 PQ
($P_3=-1,;Q_3=-1$). Unknowns: $x=[\theta_2,\theta_3,V_3]$.
$$G=\begin{bmatrix}3&-2&-1\-2&4&-2\-1&-2&3\end{bmatrix},\quad
B=\begin{bmatrix}-30&20&10\20&-40&20\10&20&-30\end{bmatrix}$$
Flat-start injections ($V=[1,1.05,1],;\theta=[0,0,0]$):
$$P_2=1.05[-2]+1.05^2[4]+1.05[-2]=0.21,\quad Q_2=2.1$$
$$P_3=1[-1]+1.05[-2]+1[3]=-0.1,\quad Q_3=-1$$
Mismatch: $$\Delta P_2=0.8-0.21=0.59,\quad \Delta P_3=-1+0.1=-0.9,\quad \Delta
Q_3=-1+1=0$$ $$\boldsymbol{\Delta}=[,0.59,;-0.9,;0,]^{\top}$$
Jacobian (using the element formulas above): $$H_{22}=-2.1+40(1.05^2)=42,;
H_{23}=H_{32}=1.05[-20]=-21,; H_{33}=1+30=31$$ $$N_{23}=1.05[-2]=-2.1,;
N_{33}=-0.1+3=2.9,; M_{32}=2.1,; M_{33}=-3.1,; L_{33}=29$$
$$J=\begin{bmatrix}42&-21&-2.1\-21&31&2.9\2.1&-3.1&29\end{bmatrix}$$
Update: $$\begin{bmatrix}\Delta\theta_2\\Delta\theta_3\\Delta V_3\end{bmatrix}
=J^{-1}\boldsymbol{\Delta}
=\begin{bmatrix}-7.23\times10^{-4}\-0.029\-3\times10^{-3}\end{bmatrix}$$
State after iteration 1: $\theta_2=-0.000723$ rad, $\theta_3=-0.029$ rad, $V_3=0.997$
p.u.
Type 4 — Per-unit base conversion across zones
What it is
A circuit split by transformers into several voltage zones. You pick a single power base,
propagate base voltages through the transformer ratios, convert every element to per-unit
on the common base, then solve the (now single-zone) circuit.
Procedure & formulas (in order)
1. Pick the base. Choose one zone’s $V_B$ and a single system-wide $S_B$ (same in
every zone): $$S_{B1}=S_{B2}=\dots=S_B$$
2. Propagate base voltage through each transformer using its nameplate ratio:
$$V_{B,\text{new zone}} = \frac{V_{2n}}{V_{1n}},V_{B,\text{old zone}}$$
3. Base impedance and base current in each zone: $$Z_B=\frac{V_B^{2}}{S_B}, \qquad
I_B=\frac{S_B}{V_B}=\frac{V_B}{Z_B}$$
4. Convert each physical element to p.u.: $$Z_{pu}=\frac{Z_{\text{actual}}}{Z_B}, \qquad
X_{pu}=\frac{X_{\text{actual}}}{Z_B}$$
5. Re-base any element already given in p.u. on its own rating (transformer reactances on
nameplate) — base-change formula:
$$\boxed{,Z_{pu}^{\text{new}}=Z_{pu}^{\text{old}}\left(\frac{V_B^{\text{old}}}
{V_B^{\text{new}}}\right)^{2}\frac{S_B^{\text{new}}}{S_B^{\text{old}}},}$$ (“old” = the
device’s nameplate values; “new” = your chosen system base.)
6. Solve the single-zone p.u. circuit (series-add all p.u. impedances, then $\bar
I_{pu}=\bar V_{pu}/Z_{pu}$).
7. Convert results back to physical units: $I_{\text{actual}}=I_{pu}\times I_B$ (use the
$I_B$ of the zone you want the answer in).
Worked reference (Example 5)
Source $E_s=220\angle0^\circ$; zone-2 line $X_L=2,\Omega$; zone-3 load
$Z_{ch}=0.9+j0.2$; T1: 30 kVA, 240/480, $X=10%$; T2: 20 kVA, 460/115, $X=10%$.
Base: $V_{B1}=240,;S_B=30$ kVA. $$V_{B2}=\tfrac{480}{240}\cdot240=480\text{
V},\qquad V_{B3}=\tfrac{115}{460}\cdot480=120\text{ V}$$ $$Z_{B2}=\tfrac{480^2}
{30,000}=7.68,\Omega,\qquad Z_{B3}=\tfrac{120^2}{30,000}=0.48,\Omega$$
$$I_{B1}=\tfrac{30,000}{240}=125\text{ A},;; I_{B2}=\tfrac{480}{7.68}=62.5\text{ A},;;
I_{B3}=\tfrac{30,000}{120}=250\text{ A}$$
Convert: $X_{L,pu}=\tfrac{2}{7.68}=0.26$; $Z_{ch,pu}=\tfrac{0.9+j0.2}
{0.48}=1.875+j0.4167$; $X_{T1,pu}=0.1$ (rating matches base);
$X_{T2,pu}=0.1\cdot\tfrac{460^2}{480^2}\cdot\tfrac{30}{20}=0.1377$.
Total: $Z=j0.1+j0.26+j0.1377+1.875+j0.4167=1.8725+j0.9144$. With $V_{pu}=\tfrac{220}
{240}=0.9167$: $$|I_{pu}|=\frac{0.9167}{\sqrt{1.8725^2+0.9144^2}}=0.4399,\quad
\theta=\tan^{-1}!\tfrac{0.9144}{1.8725}=26.08^\circ$$
$$I_{\text{actual}}=0.4399\times250=109.98\text{ A};\angle,26.08^\circ$$
Type 5 — Two-bus line power transfer
What it is
Power flowing across one line between two buses. You’re asked for the sending/ receiving
power, or the angle that makes the flow purely active, or that maximizes power, or the line
losses, or to close a power balance at a bus.
Core formula (start here every time)
Complex power leaving bus 1 toward bus 2 across series impedance $Z=|Z|\angle\varphi$:
$$S_{12}=V_1!\left(\frac{V_1^{}-V_2^{}}{Z^{*}}\right) =\frac{|V_1|^{2}}{|Z|}\angle\varphi-
\frac{|V_1||V_2|}{|Z|}\angle(\theta_1-\theta_2+\varphi)$$
Split into real/reactive ($\theta_{12}=\theta_1-\theta_2$): $$\boxed{P_{12}=\frac{|V_1|^{2}}
{|Z|}\cos\varphi-\frac{|V_1||V_2|}{|Z|}\cos(\theta_{12}+\varphi)}$$
$$\boxed{Q_{12}=\frac{|V_1|^{2}}{|Z|}\sin\varphi-\frac{|V_1||V_2|}
{|Z|}\sin(\theta_{12}+\varphi)}$$
$\varphi$ is the impedance angle: a lossless line $Z=jX$ gives $\varphi=90^\circ$ (then
$P_{12}=\tfrac{V_1V_2}{X}\sin\theta_{12}$, the classic result); a lossy line has
$\varphi<90^\circ$ (Example 6 uses $\varphi=85^\circ$).
The reverse direction $P_{21},Q_{21}$ uses the same formulas with indices 1↔2 swapped.
Sub-question recipes
(a) Make the flow purely active ($Q_{12}=0$): set the $Q_{12}$ expression to zero and
solve for $\theta_{12}$.
Ex. 6: $\sin\varphi=\sin(\theta_{12}+\varphi)\Rightarrow \theta_{12}+\varphi=180^\circ-
\varphi \Rightarrow \theta_{12}=180^\circ-2\varphi=10^\circ$ (for $\varphi=85^\circ$).
(b) Maximum power transfer: $-P_{21}$ (or $P_{12}$) is maximized when its cosine term
$=1$, i.e. $\cos(\theta_{21}+\varphi)=1\Rightarrow \theta_{21}+\varphi=0$.
Ex. 6: $\theta_{21}=-\varphi=-85^\circ \Rightarrow \theta_{12}=85^\circ$.
(c) Line losses: active loss $=P_{12}-P_{21}$ (net of the two directional flows).
Ex. 6, part 3: with $\theta_{12}=85^\circ$, $P_{12}=10.72$, $P_{21}=9.13 \Rightarrow
P_{\text{loss}}\approx 1.6$ p.u. Reactive loss $=Q_{12}+Q_{21}$.
(d) Power balance at a bus (generation = flow out + load):
$$S_{Gk}=S_{k\to\text{line}}+S_{Dk}, \qquad P_k^{\text{net}}=P_{Gk}-P_{Dk}$$
Ex. 7 ($\varphi=90^\circ$, lossless): the net injection at bus 2 sets the line flow, which
fixes $\theta_{12}$, then closes the balance at bus 1.
Worked reference (Example 7)
$V_1=1\angle0^\circ,;V_2=1,;Z_L=j0.5;(\varphi=90^\circ)$; bus 2: $P_{G2}=0.25,;P_{D2}=1$.
Net at bus 2: $P_{21}=P_{G2}-P_{D2}=0.25-1=-0.75\Rightarrow P_{12}=0.75$ p.u.
$$0.75=\frac{1}{0.5}\cos90^\circ-\frac{1}{0.5}\cos(\theta_{12}+90^\circ) ;\Rightarrow;
0.375=\sin\theta_{12};\Rightarrow;\theta_{12}=22.02^\circ$$ $$Q_{12}=\frac{1}
{0.5}\sin90^\circ-\frac{1}{0.5}\sin(22^\circ+90^\circ)=0.1459\text{ p.u.}$$
$$S_{12}=0.75+j0.1459,\qquad
S_{G1}=S_{12}+S_{D1}=0.5+j0.25+0.75+j0.1459=1.25+j0.3959$$ Closing bus 2:
$S_{G2}=S_{21}+S_{D2}\Rightarrow Q_{G2}=0.646$ p.u.; reactive loss
$Q_{12}+Q_{21}=0.2918$ p.u.
Cross-cutting reminders
Admittance sign convention: with inductive (lagging) branches, $y=1/(jX)=-j/X$, so
series elements are negative imaginary and the off-diagonals $-y$ come out positive
imaginary — double-check signs every time.
Always split $Y=G+jB$ before any N–R step. $G$ and $B$ are what enter the $P/Q$
equations and every Jacobian element.
Angle reference: the slack-bus angle is the reference ($\theta=0$); every other angle is
measured from it. Keep angles in radians when you update the state, even if you reason
about them in degrees.
Per-unit hides the factor of 3 — generation/load powers, injections and line flows in
p.u. drop the three-phase $3$. Re-insert it only when converting back to MW/MVAr.
Build long problems as a pipeline: Type 1 → Type 2 → Type 3. The transformer model
feeds the $Y_{bus}$, and the $Y_{bus}$ feeds the load flow.