Hz to mHz Converter for Grid Frequency Deviation

On a power system, the interesting part of the frequency is the last three digits

A synchronous grid never actually sits on 50.000 or 60.000 Hz. It hunts around it all day, a few hundredths of a hertz either way, and every rule written about that hunting — dead bands, containment thresholds, load-shedding stages, response payments — is expressed in millihertz. Grid codes, on the other hand, quote absolute frequencies in hertz. Move a threshold from one form to the other by entering the deviation on the left; the millihertz figure that trending screens and settlement reports use appears immediately.

Conversion factor: mHz = Hz × 1 000. A 0.05 Hz excursion from 50.000 Hz is 50 mHz, and the 0.8 Hz worst case Continental Europe designs against is 800 mHz.

Why the smaller unit took over

Frequency is the balance signal

There is no separate meter for “too much generation”. Surplus power speeds the machines up and shortfall slows them down, so a deviation of tens of millihertz is the whole system telling you, continuously and everywhere at once, how far supply and demand have parted.

The action happens in the third decimal

Normal operation is contained within ±0.05 Hz. Writing every threshold as 0.010, 0.015, 0.036 and 0.050 Hz invites a misplaced decimal on a settings sheet; 10, 15, 36 and 50 mHz does not.

Rates are quoted the same way

Rate of change of frequency turns the same figures into a slope. A 1 000 MW loss on a system holding 200 GVA·s of kinetic energy starts the frequency falling at 0.125 Hz/s — 125 mHz every second, before any reserve arrives.

Turning a grid-code threshold into a relay or trending setting

Enter the deviation, not the absolute frequency

Subtract nominal first. A code clause reading “49.800 Hz” on a 50 Hz system is a deviation of 0.2, and 0.2 is what belongs in the field — feeding in 49.8 gives a perfectly correct but useless 49 800 mHz.

Line it up against the service thresholds

With the deviation in millihertz you can put a unit’s dead band, the reserve activation point and the statutory limit in one list and see instantly which one bites first, without three different decimal conventions in the same column.

Copy the plain figure into the settings sheet

The copy button hands over the number by itself — no unit and no thousands spaces — so it drops straight into a relay settings file or a SCADA alarm limit. Ctrl+C with the cursor in a field does the same.

Go the other way when the spec arrives in millihertz

Tender documents and response product definitions are written in mHz while your protection settings are entered in Hz. The swap arrows (↔) reverse the direction, and typing in the right-hand field does it without any click at all.

Control thresholds across three synchronous areas

Each row gives the frequency a rule is written at, the deviation from nominal that implies, and the same deviation in the unit dispatchers and settlement systems read. Continental Europe and Great Britain run at 50 Hz, the North American interconnections at 60 Hz.

System	Threshold	Frequency	Deviation in Hz	Deviation in mHz
Continental Europe	Dead band and insensitivity cap for containment reserve	—	±0.010 Hz	±10 mHz
Continental Europe	Standard frequency range	49.950–50.050 Hz	±0.05 Hz	±50 mHz
Continental Europe	Containment reserve fully deployed	49.800 / 50.200 Hz	±0.2 Hz	±200 mHz
Continental Europe	Maximum instantaneous deviation designed for	49.200 / 50.800 Hz	±0.8 Hz	±800 mHz
Great Britain	Dead band on the fast frequency response products	—	±0.015 Hz	±15 mHz
Great Britain	Operational limits	49.8–50.2 Hz	±0.2 Hz	±200 mHz
Great Britain	Statutory limits	49.5–50.5 Hz	±0.5 Hz	±500 mHz
Great Britain	First stage of low-frequency demand disconnection	48.8 Hz	−1.2 Hz	−1 200 mHz
North America	Recommended maximum governor dead band	59.964 Hz	−0.036 Hz	−36 mHz
Eastern Interconnection	Typical first load-shedding stage	59.3 Hz	−0.7 Hz	−700 mHz
Western Interconnection	Typical first load-shedding stage	59.5 Hz	−0.5 Hz	−500 mHz

Dead bands compared without a decimal shift

A 10 mHz insensitivity cap and a 36 mHz governor dead band are hard to weigh against each other while both are still buried in the third decimal of a hertz.

Ramp rates converted the same way

Rate of change of frequency uses the identical factor, so 0.5 Hz/s becomes 500 mHz/s and a relay setting can be read against a recorded event slope directly.

Resolution finer than a measurement window

Eight decimals are kept, which is more than a phasor measurement unit reporting to a tenth of a millihertz will ever need from a threshold calculation.

Questions from the frequency control room

Why is grid deviation reported in millihertz rather than hertz?

Because the numbers that matter would otherwise all live behind two leading zeros. Ordinary operation on a 50 Hz system stays inside ±0.05 Hz, and the thresholds people argue about — a 10 mHz insensitivity limit, a 15 mHz product dead band, a 36 mHz governor setting — differ from each other by hundredths of a hertz. Rewriting them as whole millihertz gives everyone integers to talk about, removes an entire class of decimal-point error from settings sheets, and matches the resolution that modern measurement equipment reports anyway.

How wide is a governor dead band, and why is there one at all?

Narrow, and it exists to stop machinery chasing noise. A steam or gas turbine governor that reacted to every millihertz would spend its life opening and closing valves against a signal that is wandering anyway, wearing the actuator and unsettling the plant for no useful gain. North American practice recommends no more than ±36 mHz on a 60 Hz base, which means the machine can legitimately ignore anything above 59.964 Hz. European rules on units providing containment reserve are much stricter, capping both the intentional dead band and the response insensitivity at 10 mHz — the price of being paid for the service is reacting almost immediately.

What does ROCOF measure and where are the relays set?

Rate of change of frequency is the slope rather than the level — how fast the system is falling in the first instants after a loss, before any reserve has arrived. It is used to detect that a piece of network has islanded, since a disconnected fragment with mismatched generation and load slides away far faster than the main system ever does. Great Britain historically set loss-of-mains protection at 0.125 Hz/s, or 125 mHz/s, but as inertia fell that became sensitive enough to trip embedded generation during ordinary system events; a national programme retuned the fleet to 1 Hz/s with a 500 ms delay.

Why does a converter-dominated system fall faster after a trip?

Spinning mass is what buys the first few seconds. The initial slope is roughly the lost power times nominal frequency, divided by twice the stored kinetic energy: lose 1 000 MW on a 50 Hz system carrying 200 GVA·s and the frequency starts down at 125 mHz/s, but halve the inertia and the same loss produces 250 mHz/s. Wind and solar connected through converters contribute nothing to that store unless they are specifically configured to emulate it, so as they displace synchronous plant the nadir arrives sooner and lower, and reserve has to be both faster and deeper to catch it.

What happens when the frequency reaches its statutory limits?

Reaching them is not itself the emergency — it is the point at which automatic schemes stop waiting for the market to solve the problem. In Great Britain the statutory band is 49.5–50.5 Hz, so 500 mHz either way; below that, further decline brings the first stage of low-frequency demand disconnection at 48.8 Hz, where a block of load is shed by relays in a fraction of a second. North American schemes work the same way with different numbers, typically starting to shed at 59.3 Hz in the east and 59.5 Hz in the west. Overfrequency has its own staircase, tripping generation instead of load.