Earthing systems explained

However, this did not mean the installation was remotely close to being safe (>50Vac is electrically unsafe). Had someone stood barefoot on damp ground while touching an energised garden tap, they could have bridged that 180 V potential directly to earth. The hazard was still present, but its consequences were mitigated by high resistance and luck.

To be precise about where that protection came from: it was the dry timber piles and elevation breaking the occupants’ contact with the soil — their side of the circuit — not the absence or weakness of any earthing hardware. Anyone standing on the ground outside would have been coupled back to the transformer through their own feet regardless of what electrodes were installed. Isolation is something a person has or doesn’t have at their feet; it isn’t something you can grant an installation by leaving the rod out.

Metal cladding and EV chargers are becoming more commonplace. Earthed metal cladding or a charging EV at this property could have led to a fatal electric shock. This fault would have livened the metal frame of a charging EV, and the EV would likely be parked close to soil, increasing the chances of a person bridging the 180 V to earth and suffering a severe electric shock. EV chargers also draw a lot of current and this added load stresses supply systems leading to more failed mains connections.

The Traditional Rationale for MEN

The Fault Clearance Myth
It was once taught that the earth electrode helped trip fuses or MCBs. This was true once upon a time. In the past metal pipes not only created an excellent connection to the earth mass, but they also acted as a shared fault conductor joining installations. If a neutral was lost then resulting voltage on the earthing system would travel through these metal pipes to the earthing system of neighbouring installations, through the MEN link, and back down the neighbour’s main neutral. The result is that voltages could not build up and supply fuses could trip if too much voltage was getting dumped onto the earthing system.

But this is no longer the case. Houses are now connected with plastic pipes and electricians instead install an earth rod. A typical New Zealand electrode has an impedance of 10–100 Ω or more. This is far too high to allow the hundreds of amps needed to operate an overcurrent device fast enough to achieve safety.

Even in ideal conditions — say 20 Ω — only ~11 A would flow at 230 V. That stands no chance of clearing a 60 – 120A supply fuse to the installation. Fault clearance relies on the MEN link and neutral conductor, not on current through soil. NZS 3000 (the ‘wiring rules’) states unambiguously that the earth electrode does not usefully contribute to lowering the earth loop fault impedance (ELFI).

RCDs Are Blind to the two nasty MEN Faults
Against high-impedance neutrals and polarity reversals, sub-circuit RCDs are simply ineffective as shown in fig. 1. An RCD detects imbalance between the active and neutral conductors passing through its sensor. In a neutral fault, the diversion of current into the earth happens after the current has already passed back through the RCD’s neutral coil. To the RCD, the circuit looks balanced, so it doesn’t trip.

The only way to protect against a mains neutral fault with RCDs is to place an RCD upstream so it protects the whole installation — as is standard in TT systems in Europe. This is not done in New Zealand. Another way would be to monitor phase-to-neutral voltage. If it gets too low or too high then a fault is detected and power can be disconnected or an alert activated. This is called O-PEN or OPDD detection.

Modern 30 mA RCDs are extremely sensitive and will trip reliably even if the electrode resistance is high — to several kOhms. In practice, the difference between a pristine electrode and a poor one makes no practical difference to their operation.

The One Weakness of TT Systems

TT systems are not flawless. They carry a lesser but real vulnerability: when a neighbouring MEN installation develops a neutral fault, fault current can be driven into the soil, raising the local earth potential. A nearby TT installation within that zone can experience that same rise on its own earth electrode, even though it has no direct connection to the faulted system. This can, in theory, cause a voltage to appear between exposed conductive parts in the TT installation and true earth. Note the neighbouring MEN installation is already unsafe too, so this isn’t really a ‘TT’ issue.

It’s important to understand that this is not a flaw of the TT system itself but a by-product of MEN systems injecting current into the ground. The TT site is a victim, not the source. In practice, the resulting voltage gradients in the soil are usually small and localised, and the likelihood of harmful current through a person is far lower than the risks posed by lost-neutral or transposition faults inherent in MEN systems. Still, it’s a reminder that every earthing system exists within a wider electrical ecosystem — and that the dangers of MEN don’t stay neatly confined to the installation where they begin.

Warren Harris’s Defence of Earth Electrodes

Warren Harris has argued that earth electrodes remain critical because they:

• Provide a return path for load and fault current if the neutral is broken.
• Help limit voltage rise in open-neutral situations.
• Assist RCDs in detecting earth leakage.
• Are required by AS/NZS 3000, which confirms their importance.
• Must be properly maintained to prevent corrosion or loss of integrity.

These arguments deserve careful scrutiny.

Replying to Warren Harris’s Points

Return Path for Current and Limiting Voltage Rise
There’s more truth here than I originally allowed. A functioning electrode passes some current under an open neutral, pulling the earthing system somewhat toward earth potential — though usually nowhere near enough to make it safe, as Perth demonstrated. And the criticism stands that this same current lets appliances keep half-working, so the fault can linger unnoticed for longer; a dead installation gets an electrician called, a tingling one sometimes doesn’t. But the electrode does a second thing I under-weighted: it drags the soil immediately around the installation up toward the earthing system’s potential. Touch voltage is a hand-to-feet quantity, and if the ground someone is standing on is closer to the potential of the tap they’re touching, the voltage across their body is smaller. It’s a partial, localised reduction — but it’s a reduction in exactly the quantity that harms people. In town this barely registers, because the network neutral is already held near earth potential by hundreds of installation electrodes in parallel across the MEN network — any single rod is close to redundant, and this collective neutral-earthing job is arguably the electrode’s real stated function. On a rural spur with one consumer and few parallel earths, that one rod carries a real share of both jobs. That’s the honest version of “limits voltage rise”: a network-level function performed collectively, plus a local touch-voltage function, both of which matter most exactly where there are fewest rods.

Assisting RCDs
Warren argued that electrodes can still help by providing an alternate path that allows RCDs to detect imbalance. In an earlier version of this article I read that as a claim about shock current — that the electrode increases the current through a person so the RCD trips harder. Against that version the reply is easy: a 30 mA RCD will trip through a person even at 7,000 Ω to earth, so extra current adds risk without adding protection. But I no longer think that’s the claim he was making. The argument is about proactive tripping: giving fault current somewhere to go so the RCD trips on the fault itself, before anyone touches anything.

Picture a Class I appliance whose protective earth conductor has broken somewhere between the switchboard and the appliance. Now the phase conductor inside contacts the metal frame. The frame is live, but no fault current flows — the RCD sees equal current out on the phase and back on the neutral, so nothing trips. The fault sits there silently, and the RCD’s first chance to act is through the first person who touches the frame. If, however, the frame has some other conductive route to earth — bonded plumbing, structural steel, a damp concrete floor, a connected EV — then current leaks away: phase → frame → that path → earthing system → MEN link → neutral. Say 100 mA leaves via the earth instead of the neutral. The RCD sees 5.0 A out and 4.9 A back, and trips. The fault announces itself with nobody in the circuit. One caveat: no electrode can conjure a path to a genuinely floating frame — there has to be some conductive route from the frame to the earthing system or the soil for this mechanism to exist at all.

The case where the electrode itself is load-bearing is a phase-to-earth insulation fault during a broken main neutral. This is a different animal from the open-neutral scenario in fig. 1. There, load current returns through the RCD’s neutral coil, so the RCD stays balanced and blind — the electrode passing current changes nothing. But a genuine insulation fault onto the earthing system, with the main neutral gone, can only return via the electrode and the soil. That current leaves on the RCD’s phase coil and never comes back on its neutral coil — full imbalance — so the sub-circuit RCD trips even though the neutral is broken. Without the electrode, that fault current has no return path and the fault sits live and silent. This is the same logic TT systems are built on: the electrode is a fault-detection path, not a shock-protection path.

So there is a real mechanism here, with real limits. It only works for faults downstream of an RCD — and in New Zealand we historically haven’t RCD-protected hot water, ovens and pumps. It does nothing for the open-neutral energisation itself, which remains invisible to sub-circuit RCDs. And because RCDs trip at impedances into the thousands of ohms, a fault that reaches the soil by any route will often trip an RCD via remote earth even without a local rod — the electrode makes the detection path certain rather than merely possible.

Does the Electrode Make People Less Isolated? — Where I Changed My Mind
In earlier versions of this argument I claimed the electrode creates risk by giving the soil around the house a good conductive link back to the transformer — so that pulling the rods out would leave people more isolated and less exposed. I now think that reasoning was wrong, and the mistake is worth spelling out because I suspect it’s a common one.

Earth-return resistance doesn’t scale with distance the way a wire does. Current spreading out from an electrode passes through hemispherical shells of soil whose area grows with the square of distance, so nearly all of an electrode’s resistance is concentrated in the first metre or two around it. Beyond about ten rod-lengths you’ve reached “remote earth,” which behaves as one enormous low-resistance node. The 300 m of ground between a house and its transformer contributes almost nothing; the resistance lives at the two ends — at the electrode surfaces, or under the soles of someone’s feet.

That has two consequences. First, a person standing on soil is already coupled back to the transformer earth through their own contact patch, rod or no rod. The bottleneck in that leg is the constriction resistance under their own feet, and a rod five metres away cannot bypass it — the current still has to exit through their soles. Removing every installation electrode in the country would not buy people isolation. The isolation people actually enjoy lives in their boots, dry ground and small contact area — variables no electrode touches. Second, the return leg barely matters to shock severity anyway. With a wet, well-coupled body at 500–1,000 Ω, the difference between a 5 Ω and a 50 Ω earth return moves the shock current by a few percent, between two figures that are equally lethal. The body is the resistor that decides the outcome.

Soil type reinforces this rather than changing it. Resistivity spans three orders of magnitude — moist clay around 10–100 Ω·m, dry sand, gravel and rock in the thousands. Electrode resistance scales with local resistivity, but so does a person’s foot-to-soil contact resistance (roughly 3ρ per foot in the standard models). So in conductive soil, where rods work well, people are already naturally well-earthed with or without them; and in resistive ground, where people genuinely are somewhat isolated, the rod barely couples to anything and can’t take that isolation away. In neither regime is the electrode the variable controlling personal shock risk.

Where the electrode does genuinely add hazard is much narrower than I originally argued: it exports the MEN potential into its own gradient zone. During an open neutral, the soil within a metre or two of the rod sits closer to neutral potential than to remote earth, so someone standing there while touching something referenced elsewhere — a fence, a neighbour’s water main — sees a touch voltage that wouldn’t otherwise exist, and under fault current there is step potential across that gradient. Those effects are real, but localised, and they’re the mirror image of the same gradient that reduces the touch voltage between the installation’s own metalwork and the ground beside it. The electrode redistributes where the voltage difference appears; it doesn’t wire the whole yard to the transformer — the soil and the transformer earth had already done that.

Legislation Requires It

Warren Harris mentioned that Standards require it and suggested that the standards would only require it if was indeed important. It’s certainly true that the standards require it. But standards reflect consensus, not immutable truths. Regulation can lag behind. A lot has changed since this system was first implemented and critical analysis in light of these changes raises some questions. Quoting AS/NZS 3000 shows compliance requirements, not proof that the measure is optimum in today’s risk environment.

Integrity of Electrodes
Warren Harris is right that electrodes corrode and fail. But maintenance only matters if the function matters. If the safety contribution of electrodes is marginal at best, ensuring their “integrity” becomes an exercise in maintaining something that does not significantly enhance protection.

There’s another angle that often seems overlooked: the lack of any defined benchmark in our regulations for what constitutes a “good” earth electrode in MEN installations. In nearly every other critical safety domain, we demand precise measurable criteria — conductor cross-sections, volt drop, fault loop impedances, RCD operating times, etc. But for the earth electrode, there is no compulsory limit and no test to show the electrode is performing.

Contrast that with TT systems. In France there is a design guidance threshold of 100 Ω (or less) for the installation earth electrode under certain conditions, tied to the sensitivity of the protective RCD. In those systems, the electrode is truly part of the safety chain — it must allow sufficient current to flow so the RCD can detect and act on faults.

However, in our MEN system, removing the installation electrode would not break any fault-clearing loop — those rely on the metallic PE → MEN → neutral path — and, as argued above, it would not materially change how exposed a person standing on the ground is. What it would remove is narrower: the installation’s share of the network’s collective neutral earthing, the local touch-voltage reduction under an open neutral, and the fault-detection path that lets RCDs trip proactively when the main neutral (or a protective earth conductor) is broken. In urban Auckland each of those contributions is close to redundant. On a rural spur, they are not. Which sharpens the benchmark problem rather than softening it: if these are the electrode’s real functions, the regulations should say so and give us something to test — a maximum resistance, a verification method, an inspection criterion — as TT jurisdictions do. Requiring the hardware while declining to define or measure its function remains the worst of both worlds.