Network Tools > Decoupling Library

Decoupling Element in Line Mode

When decoupling is required, this mode is used to model line sections that are too short to use regular constant parameter (or distributed parameter) lines. It simulates a PI section with L and C values corresponding to a propagation delay of exactly one time step. User lust supply R and L values of the line, and a C value is automatically calculated and added to artificially create a propagation delay of one time step.

Tprop is the propagation delay, Ts is the time step and d the length of the line.

This model introduces error due to the addition of either a stray inductance or capacitor as compared to the original circuit. Using a decoupling line model introduces additional pole and non-characteristic oscillations that can appear during switching transients. The added L or C can also affect the steady-state reactive power flow in the simulation. Overall, the solution needs to be assessed and, as any decoupling technique, it requires engineering judgment to ensure acceptable compromise on the simulation results. Anyway, some cases using the line type decoupling may not affect the simulation results much. Indeed, using a line type decoupling is generally more stable than using I/V type decoupling.

Decoupling Element in V/I Mode

When decoupling is required, this mode is used to replace large inductors in series in the network. For example, smoothing reactors or transformers leakage inductors are good candidates for decoupling. This model relies on the fact that the current of an inductor cannot vary instantaneously. Although a time step delay error on the current is present, the simulation error becomes insignificant if L is large.

The V side of the component must be on the inductance side (voltage source at the inductor representing the system voltage) while the I side of the component must be on the system side (current source representing the inductor current).

Similarly, this type of decoupling can also be used to decouple through a large capacitor. In that case, I side of the component must be on the capacitor side, as the voltage cannot vary instantaneously.

This mode is the most precise if and only if the component on which the decoupling is applied (L or C) is large enough to make sure that the time constant of current/voltage variation is much higher than the time step of the simulation.

Adding this component makes the model more prone to numerical instability, especially when a fault is applied close to the decoupling element or if a breaker is operated in the vicinity of this decoupling. If so, one can try the decoupling element in “Line mode”. Again, engineering judgment is key when applying numerical decoupling.

Decoupling Element in V/V Mode

When decoupling is required, this mode is rarely used and is recommended only for particular cases where other techniques may fail to give acceptable results based on engineering judgment.

It duplicates the voltage and impedance of the decoupling component from one side to the other.

It is known to be more stable than the I/V decoupling technique in the presence of discontinuities but less precise because of the delay on the voltages on both sides.

Network Transformers > 3-winding, w/ sat + tap + dec 

This is basically the same as having a 3-winding, w/ sat + tap and adding a decoupling element on the primary winding.

Signal Routing > Transceiver, Float OR Transceiver, Integer

This is used to decouple the control signal part. It is important to add transceiver blocks if the goal is to decouple the model at a particular location. Even though a decoupling element is present between 2 networks, HYPERSIM will not create 2 tasks if there is at least one control signal exchanged between those networks without the transceiver block.

This decoupling device also includes one time step of delay.

Additional Notes

No matter which decoupling element, one time step of delay is applied. So it can create divergence of the simulation and may or may not be used on a case by case basis.

Line type decoupling is much less prone to numerical instability because an actual “continuous” pole is introduced in the simulation. Other techniques are adding a delay without actual damping at the corresponding numerical pole.

Always validate that adding a decoupling element doesn’t affect much the simulation results by creating faults in the network with/without the decoupling element… And by validating the steady-state results, especially when using line type decoupling.