Grid-following GFL

What it is (precise)

In a grid-following PCS an internal phase-locked loop continuously estimates the angle, frequency, and magnitude of the voltage at its terminals. The control system then regulates output current, usually in a rotating dq reference frame with inner current loops that respond in milliseconds, so the commanded active power (P) and reactive power (Q) flow between the battery racks and the grid.

Picture a fast, obedient current source that borrows its voltage timebase from the network. This is the mirror image of a Grid-forming (GFM) inverter, which imposes its own voltage magnitude and angle and behaves as a voltage source behind an impedance.

The current-source behavior has two hard consequences worth memorizing. First, dependence on grid strength: GFL control assumes a relatively stiff voltage reference, so as the short-circuit ratio (SCR) at the POI falls, the PLL tracks a weaker, noisier signal, control loops begin interacting with the grid impedance, and stability margins shrink.

Second, limited fault current: the power semiconductors cap the contribution at roughly 1.1 to 1.2 per unit of rated current, versus 5 to 7 per unit from a synchronous machine. And with no live voltage to lock onto, a GFL plant cannot perform a Black start.

Why it matters in a real grid-scale project

Mode selection is a commercial and Interconnection decision, not just a controls detail. GFL is the default because it is mature, cheaper to study and commission, and it satisfies most existing interconnection agreements and grid codes, which were largely written around grid-following behavior. For a typical container-plus-PCS BESS feeding a strong transmission node, GFL clears the grid code and energizes the project with the least schedule and study risk. That is why it dominates the installed fleet and why it is the safe assumption on any strong-grid site where SCR is comfortably above 3.

The risk appears on weak grids and high-renewable networks. As synchronous generation retires, system strength drops, and clusters of GFL plants electrically close together can exhibit subsynchronous oscillations or control instability that surface in interconnection studies as required mitigations, output limits, or expensive restudy loops.

System operators in Australia and Texas already require or strongly encourage grid-forming capability for new BESS. GFM is often, though not always, the same PCS hardware plus a controls and firmware change, sometimes needing extra current or energy headroom, so confirm early whether a candidate PCS is GFM-capable or field-upgradeable before a future grid-code mandate strands the asset.

Numbers and standards worth memorizing

The dominant U.S. rules for GFL behavior are IEEE 1547 for distribution-level interconnection and IEEE 2800 for bulk-system inverter-based resources; both define Ride-through envelopes, frequency and voltage response, and reactive-power requirements, and IEEE 2800 is being amended to address grid-forming capability.

Inverter type-testing follows IEEE 1547.1 and UL 1741 SB, the supplement covering grid-support functions. Harmonics at the POI are held to IEEE 519 guidance, on the order of 5 percent total demand distortion. In Europe the binding rules are the connecting-country grid codes built on the ENTSO-E Requirements for Generators (RfG) framework.

Grid strength is captured by SCR: the three-phase fault MVA at the POI divided by the plant MW rating. Rule of thumb, above about 3 the grid is strong; between roughly 2 and 3 it is weak and GFL control turns marginal toward the low end, where GFM or dedicated weak-grid tuning is recommended.

PLL bandwidths run in the tens of hertz. Frequency response from a GFL BESS is usually a 2 to 5 percent droop, with fast frequency response products delivering full output in well under a second; ERCOT FFR, for example, requires full response within about a quarter-second, roughly 250 milliseconds, of the frequency trigger.

Mode is independent of battery chemistry: whether the racks are LFP, the stationary default, or NMC, the GFL-versus-GFM choice lives in the PCS controls, not the cells. Watch the reference point, too.

P-Q capability is quoted at the low-voltage inverter terminals, while grid-code obligations apply at the Point of Interconnection, net of the medium-voltage Transformer and collection-system losses, so the same plant shows two different P-Q pictures depending on whether you stand at LV or at the POI. Keeping MV / LV / HV straight is what stops an interconnection meeting from arguing the wrong number.

How it shows up in specs, studies and contracts

On a PCS datasheet, grid-following behavior appears as a P-Q capability chart, Four-quadrant operation limits, reactive capability at zero active power (statcom mode), and declared compliance with IEEE 1547 or IEEE 2800 ride-through curves.

Read the fine print: does the kVA rating and reactive range hold across the full ambient and DC-voltage window, is capability stated at the inverter terminals or referred to the POI, and what minimum SCR does the vendor require for standard control settings? Many datasheets quietly state a minimum SCR of 2 to 3, below which the vendor demands a project-specific stability study before it will warrant performance.

In interconnection studies, GFL plants on weak or converter-dense grids increasingly trigger a requirement for EMT models, typically PSCAD, on top of the standard RMS models, plus SCR screening and subsynchronous-oscillation studies; late or generic vendor models are a common critical-path risk.

In market rules, GFL is what qualifies a BESS for Frequency response and FFR products that measure and react to frequency, not for services requiring inherent voltage-source behavior such as black start or true inertia. Procurement questions worth asking: is the hardware GFM-capable, what current headroom exists, what is the firmware upgrade path, and which models arrive under NDA and by when?

Common pitfalls

The classic field failure is PLL loss of synchronism during deep voltage dips or phase-angle jumps. Ride-through rules require the plant to stay connected through faults, but a GFL inverter riding through a fault is tracking a distorted, depressed voltage, and a poorly tuned PLL or current loop can trip the plant or drive an oscillatory recovery.

This is why weak-grid projects insist on EMT studies run with the vendor's real control code rather than generic library models, and why commissioning includes staged fault-ride-through and setpoint-step tests before the plant is allowed to energize commercially.

Two contractual traps recur. First, treating GFM as a checkbox: a bid stating the PCS "supports grid-forming" may mean an untested future firmware option that has never been type-tested or studied at your SCR, so require evidence, not a roadmap.

Second, counting fast frequency response as inertia: FFR is a measured-and-react service that lags the event by the detection and response time, while inertial response is instantaneous by physics. Interconnection agreements or capacity contracts that promise "inertia-like" services from a GFL plant are promising something the control mode cannot physically deliver.