Consider a power system consisting of N number of buses. Buses in this power system are categorized into slack bus, PV buses and PQ buses for load flow study. The number of PQ buses is N ${}_L$. The balanced Newton-Raphson method is used to carry out load flow study in polar form. H, S, M, and R are sub-matrices of the Jacobian matrix J as shown below:

≤ft[ {*20c P\\ Q } ] = J≤ft[ {*20c δ \\ V } ] where J = ≤ft[ {*20c H&S\\ M&R } ]

The dimension of the sub-matrix M is

Power System Bus Classification for Load Flow

In power system analysis, load flow studies are performed to determine the system's operating state under normal conditions. The Newton-Raphson method is a common technique used for these studies. Buses in a power system are categorized based on the specified parameters (voltage magnitude V, voltage angle δ, real power P, reactive power Q) which influences the structure of the Jacobian matrix used in the calculation.

- Slack Bus: There is typically one slack bus. Its voltage magnitude (V) and angle (δ) are specified. It serves as a reference and balances the power mismatch in the system, including losses.
- PV Bus (Generator Bus): For PV buses, the real power injection (P) and voltage magnitude (V) are specified. The voltage angle (δ) and reactive power injection (Q) are the unknown variables to be solved for.
- PQ Bus (Load Bus): For PQ buses, both the real power injection (P) and reactive power injection (Q) are specified (representing loads). The voltage magnitude (V) and voltage angle (δ) are the unknown variables to be solved for. Let the total number of PQ buses be **N${}_L$ .

Jacobian Matrix Structure in Newton-Raphson Load Flow**

The Newton-Raphson method solves the non-linear load flow equations iteratively. The Jacobian matrix (J) is central to this process. It represents the first partial derivatives of the power mismatches with respect to the voltage angles and magnitudes.

The system of equations solved at each iteration is:

$\left[\begin{array}{c}\Delta P\\ \Delta Q\end{array}\right]=J\left[\begin{array}{c}\Delta \delta \\ \Delta V\end{array}\right]$

The Jacobian matrix J is partitioned into four sub-matrices based on the relationship between power mismatches and voltage/angle changes:

$J=\left[\begin{array}{cc}H& S\\ M& R\end{array}\right]$

Here:

- H involves derivatives of ΔP with respect to Δδ.
- S involves derivatives of ΔP with respect to ΔV.
- M involves derivatives of ΔQ with respect to Δδ.
- R involves derivatives of ΔQ with respect to ΔV.

Deriving the Dimension of the M Sub-matrix

To determine the dimension of the sub-matrix M, we need to consider the number of variables and equations involved.

Let the total number of buses in the system be N.

- Number of slack buses = 1.
- Number of PQ buses = **N${}_L$ **.
- Number of PV buses = $N-1-N_L$.

The vector of unknown state variables that the Newton-Raphson method solves for consists of:

- Voltage angles (δ): These are unknown for all buses except the slack bus. Therefore, there are $N-1$ unknown voltage angles. The dimension of the vector $\Delta \delta$ is $(N-1)\times 1$.
- Voltage magnitudes (V): These are unknown only for the PQ buses. Therefore, there are $N_L$ unknown voltage magnitudes. The dimension of the vector $\Delta V$ is $N_L\times 1$.

The vector of power mismatches that are calculated includes:

- Real power mismatches (ΔP): These are calculated for all buses except the slack bus (i.e., for PV buses and PQ buses). Thus, there are $N-1$ real power mismatches. The dimension of the vector $\Delta P$ is $(N-1)\times 1$.
- Reactive power mismatches (ΔQ): These are calculated only for the PQ buses, as the voltage magnitude is unknown for these buses. Thus, there are $N_L$ reactive power mismatches. The dimension of the vector $\Delta Q$ is $N_L\times 1$.

The overall Jacobian matrix J has dimensions determined by the total number of equations and the total number of unknowns, which are both $(N-1)+N_L$. So, J is $((N-1)+N_L)\times ((N-1)+N_L)$.

The partitioned equation relevant for M is:

$\Delta Q=M\Delta \delta +R\Delta V$

Considering the dimensions:

- $\Delta Q$ has dimensions $N_L\times 1$.
- $\Delta \delta$ has dimensions $(N-1)\times 1$.
- $\Delta V$ has dimensions $N_L\times 1$.

For the term $M\Delta \delta$ to be compatible with $\Delta Q$, and knowing $\Delta \delta$ is $(N-1)\times 1$, the matrix M must have dimensions such that when multiplied by $(N-1)\times 1$, it results in a $N_L\times 1$ vector.

This implies that the number of columns in M must match the number of rows in $\Delta \delta$ (which is $N-1$), and the number of rows in M must match the number of rows in $\Delta Q$ (which is $N_L$).

Therefore, the dimension of the sub-matrix M is:

**Dimension(M) = $N_L\times (N-1)$ **