Mastering grid forming vs grid following in real-time testing

Key Takeaways

• Grid forming control should be tested as waveform authority, with focus on voltage source behaviour, current limits, and weak-grid stability.
• Grid following control works well on strong grids, but faults and low short circuit strength expose its dependence on an external reference.
• Closed-loop real time simulation gives you the clearest view of timing and interface effects before commissioning locks in costly assumptions.

Grid-forming and grid-following inverters need different test plans because they control the grid in completely different ways.

Renewable power additions reached almost 510 GW in 2023, which pushed inverter behaviour from a specialist topic into a grid stability issue. As grids add more solar, storage, and power electronic interfaces, you can’t treat grid forming and grid following as interchangeable control settings. One control mode creates the local voltage and frequency reference. The other measures that reference and injects current against it. That split should shape how you test, tune, and approve inverter controls.

“A grid forming inverter acts as a controlled voltage source that sets local voltage angle, magnitude, and frequency for the network around it.”

Grid forming establishes the local grid reference

A grid-forming inverter acts as a controlled voltage source that sets local voltage angle, magnitude, and frequency for the network around it. It does not wait for a strong grid to define the waveform. It’s the control anchor during weak-grid operation, islanding, and black start.

A battery system energizing a de-energized feeder is the clearest example. The inverter ramps voltage, establishes nominal frequency, and absorbs the first mismatch between generation and load. When a motor starts or a feeder breaker closes, the controller adjusts its output to keep the waveform coherent instead of chasing a phase-locked loop target.

That operating role shifts what you need to validate. Voltage source behaviour brings inner voltage loops, virtual impedance, current limit logic, and energy reserve management into the critical path. If those pieces are tuned poorly, the inverter can look calm in a static study and still lose synchronism once load steps, saturation, or measurement lag appear.

Grid following depends on an external voltage source

The main difference between grid following and grid forming control is that a grid following inverter measures an existing waveform and aligns its current injection to it. It depends on an external voltage source, usually through a phase-locked loop. If that reference becomes weak or distorted, performance degrades first and stability can go next.

A large solar plant tied to a strong transmission system fits this model well. Each inverter tracks grid voltage, controls active and reactive current, and lets the bulk system set frequency. That structure is efficient when short circuit strength is high and the plant is not expected to establish service after an outage.

The tradeoff appears when you move the same controller to a remote feeder or a converter-heavy microgrid. Phase-locked loops can hunt during phase jumps, voltage notches, or low short-circuit-ratio conditions. Current control still matters, but a harder question shows up quickly: who keeps the waveform steady when the reference itself starts to wobble?

Frequency regulation comes from the inverter control law

Frequency regulation in a grid-forming inverter comes from the control law that links power imbalance to a frequency response. The controller intentionally shifts frequency or phase angle when load and generation separate. Droop control does this with a simple slope, while virtual machine methods add inertial behaviour and damping terms.

A 200 kW load step on an islanded battery inverter shows the mechanism clearly. Active power rises at once, frequency dips according to the droop setting, and the controller settles at a new operating point as the source picks up the extra load. You’re not watching a measurement feature. You’re watching the inverter enforce system balance.

The important tuning work sits in the slope, bandwidth, and recovery limits. Aggressive settings will hold frequency tightly but can cause oscillation against other sources or against network resonance. Softer settings improve damping yet allow deeper frequency excursions, which can trip motors, protection, or adjacent controls if the wider system was not tuned as a whole.

Fault response reveals the biggest control difference

Fault response separates these control modes more clearly than any steady state test. A grid following inverter usually detects the sag, keeps tracking the disturbed waveform, and injects current within its limit strategy. A grid-forming inverter tries to preserve a usable voltage reference while also respecting semiconductor limits, which creates different behaviour during the first milliseconds.

Consider a three-phase voltage dip on a weak feeder. The grid following unit will often prioritize commanded reactive current and remain locked to the faulted phase angle until protection or ride-through logic changes state. The grid-forming unit can support the local bus voltage and carry adjacent loads, yet it must transition cleanly into current limiting or it will distort the waveform it is trying to support.

That is why fault testing cannot stop at pass or fail ride-through flags. You need to watch sequence components, angle recovery, current limiter engagement, and how fast the controller regains a sane operating point after the cleared fault. Protection engineers also need this detail, because relay settings that suit synchronous sources can misread converter behaviour.

Weak grids expose stability limits much sooner

Weak grids expose control limits early because voltage angle, magnitude, and fault level all move more for the same disturbance. A controller that looks calm on a stiff source can oscillate, overcurrent, or mis-detect phase when short circuit ratio drops. That is why weak-grid testing belongs near the start of validation, not near the end.

Renewables supplied 86% of global power capacity additions in 2023. That shift means more feeders, plants, and hybrid sites will operate with a higher share of converter controls shaping local electrical behaviour. A remote wind and storage hub connected through a long line is a common case where short circuit support is thin and controller interactions become visible fast.

Short circuit ratio is useful, but it is not enough on its own. Network resonance, line impedance angle, and adjacent controller bandwidths can matter just as much. You’ll get better answers from parameter sweeps that move several grid strength variables at once, because many unstable cases appear in combinations that a single short circuit ratio number hides.

Real-time simulation finds control issues offline studies miss

Real-time simulation catches timing, saturation, and interface effects that offline models smooth over. You can run the controller code against a live network model, force disturbances at precise instants, and see what the plant would do before site work starts. That makes control flaws visible while they are still cheap to fix.

A hardware-in-the-loop bench built on OPAL-RT lets you connect the actual controller, I/O, protections, and communications stack to a simulated weak grid. Jitter, sensor scaling, pulse delays, and current limiter transitions all appear in the loop. A controller that looked fine in fixed step software can show oscillation once ADC timing and network latency are part of the test.

That difference matters most near control boundaries. Fault recovery, mode transfers, and dispatch step changes often fail because several small implementation details line up at once. Offline studies still matter for design speed, but they won’t show the full closed-loop timing picture you need before commissioning.

Grid-forming inverter testing must close the loop

Grid-forming inverter testing must include the network, the controller, and the physical interfaces in one closed loop. Open-loop checks will confirm equations and basic logic, but they will not prove stable behaviour under disturbances. If you want trustworthy results, the inverter has to react to the grid model as the grid model reacts back.

A solid test sequence starts with a small set of cases that expose control structure before you spend time on large scenario matrices. The goal is to stress the parts of the controller that carry system authority. Five checks usually surface the biggest issues early.

• Start with dead bus energization and startup sequencing.
• Test load step response across several operating points.
• Force balanced and unbalanced faults through current limiting.
• Check the transfers between grid-forming and grid-following modes.
• Sweep weak-grid conditions alongside communication delays.

Each test should record more than active power and RMS voltage. Waveform angle, instantaneous current limit state, controller mode flags, and protection timers tell you why a run stayed stable or failed. You’re looking for cause and sequence, because that turns a failed test into a control fix instead of a lab mystery.

“Real time simulation catches timing, saturation, and interface effects that offline models smooth over.”

Project constraints should guide control mode selection

Control mode selection should follow project constraints, because the right answer depends on grid strength, fault duties, black start needs, and the number of sources that must coordinate. A strong transmission intertie with simple power export will favour grid following. An islanded system or a weak connection that needs voltage support will favour grid forming.

A mixed plant often lands in the middle. One battery inverter can carry grid-forming duties for energization and local stability, while solar inverters stay grid following to keep controls simpler and ratings lower. That split works only if you define authority clearly, test handoffs, and align protection with the actual fault behaviour of the plant.

The teams that get this right treat control choice as an engineering constraint. They do not reduce it to a label or a grid code checkbox. OPAL-RT fits that phase of work because it lets you judge controller behaviour under stress before site assumptions harden into field problems.