Earth Potential Rise (EPR) and BS 7671 Section 442

If a DNO, principal contractor or designer has asked you for an “EPR assessment” or an “earthing study”, this explains what they are asking for and why it matters. It also covers the part almost every guide to this subject leaves out: BS 7671 has an entire section about what that rise does to the low voltage installation sitting underneath it, and it is not the section anyone expects.

The short version: EPR is how far your earthing system's voltage climbs during an earth fault — EPR = I_e × R_e. It is not the hazard itself; the hazard is the touch and step voltages that are a fraction of it, judged against BS EN 50522 and ENA TS 41-24. But that rise also lands somewhere in the LV installation, and BS 7671 Table 44.1 says where — on equipment insulation as a stress voltage, or, in one arrangement, on people as a fault voltage.

1. What the rise actually is

When an earth fault occurs on a high-voltage system, fault current flows through the site's earthing system and into the surrounding soil. Because soil has resistance, the earthing system — and everything bonded to it — rises in voltage relative to the general mass of earth far away (“remote earth”). That voltage is the earth potential rise, sometimes called ground potential rise.

A site drawn above the ground line with the ground potential plotted below it on the same horizontal scale. Above: a substation with its earth electrode driven into the soil, an earthed perimeter fence, a person with a hand on that fence, another person walking past mid-stride, and a buried metallic service running away off the right of the drawing. An HV earth fault drives the current Ie down the electrode into the soil. Below the ground line the potential is plotted against the same distance: it peaks directly at the electrode and decays away in both directions towards zero volts at remote earth. The rise equals Ie multiplied by Re, the resistance of the earthing arrangement; for 500 amperes through 2 ohms that is 1,000 volts. The person at the fence meets a touch voltage hand to feet, the walking person a step voltage across one stride, and the buried service carries the full rise off the site as a transferred potential. — Ohm's law across the earthing arrangement. A person meets a fraction of the rise — except through transferred potential, which can carry all of it somewhere nobody expects.

It is Ohm's law across the earthing arrangement, and it needs no standard: 500 A returning through soil via a 2 Ω earthing system is a 1,000 V rise. The fault lasts only until protection clears it, but during that time the voltages around the site can be dangerous.

2. Touch, step and transferred

EPR itself is not the hazard. It is the reference. What can hurt someone is the fraction of it they are actually exposed to:

Touch voltage — between a bonded metal part someone is touching and the ground they are standing on. Hand to feet.
Step voltage — between one foot and the other, as the ground potential falls away with distance from the electrode. Worst where the funnel is steepest.
Transferred potential — the rise exported away from the site by metallic services: cable sheaths, pipework, fences, rails, telecoms. This is the exception to “a fraction of it”: a person a long way from the substation can be exposed to close to the full EPR, somewhere nobody is expecting it.

Three things dominate the result, and the next section takes the first two apart: how much fault current actually returns through the soil, and the earthing system impedance — set by electrode geometry and soil resistivity, which is why a Wenner or Schlumberger soil resistivity survey is the first input any study needs. The third is the fault clearance time, which changes nothing about the EPR but sets what is permitted, because the body tolerates more for less time.

3. Where R_e and I_e actually come from

Both numbers in EPR = I_e × R_e are usually handed over as givens, and both are worth taking apart — because that is where the design decisions live.

R_e is soil times geometry

Here is something that catches people out: BS 7671 gives you no way to calculate an earth electrode's resistance at all. It recognises seven types of electrode (Reg 542.2.2 — rods or pipes, tapes or wires, plates, structural metalwork, welded reinforcement, cable sheaths, other underground metalwork), requires the design to allow for corrosion (542.2.1), requires the type and depth to survive soil drying and freezing (542.2.4), caps R_A × I_Δn at 50 V on TT (411.5.3), and requires the resistance to be measured (643.7.2). Then its own note sends you elsewhere: “Further information on earth electrodes can be found in BS 7430.”

For a first pass the classical results are enough, and they are worth knowing because their shape tells you what to change:

A hemisphere is exact — R = ρ / 2πr. Integrate the shell resistance outwards and it converges, which is why the soil in the first few metres does almost all the work and the ground a hundred metres away does almost none.
A vertical rod — R = (ρ / 2πL) × [ln(8L/d) − 1]. Note that L is outside the logarithm and d is inside it: driving a rod deeper helps, making it fatter barely does.
An earth grid — R_g ≈ ρ/4r + ρ/L_t, with r = √(A/π). The first term is the exact resistance of a disc of ground the same area as the grid, so a grid behaves like the patch of earth it claims.

The chain that produces an earth potential rise. Soil resistivity of 100 ohm metres feeds an electrode — a 625 square metre earth grid — which gives an earthing resistance Re of 2.00 ohms to remote earth; multiplied by the 500 amperes returning through the soil, that is a 1,000 volt rise. A stacked bar breaks the grid resistance into its two parts: the ground the grid covers, rho divided by four r with r of 14.1 metres, contributes 1.77 ohms or 89 per cent, while the conductor in it, rho divided by total length, contributes only 0.23 ohms. In the same soil a single 2.4 metre rod would be 40 ohms. — Area is the lever, not copper. The first term is the exact resistance of a disc of ground the same size as the grid — which is why pouring more conductor into the same footprint barely moves the answer.

Area is the lever, not copper. On the worked case — 100 Ω·m soil, a 25 × 25 m grid — the ground the grid covers supplies 1.77 Ω of the 2.00 Ω answer, and all 440 m of buried conductor inside it supplies the other 0.23 Ω. Doubling the copper in the same footprint barely moves the number. In that same soil a single 2.4 m rod would be about 40 Ω: a grid is not a bigger rod, it is a different mechanism.

The ρ that feeds all of it comes from a soil resistivity survey. A Wenner set gives it directly — ρ = 2πa·R for probe spacing a and meter reading R — and the survey sweeps a precisely because one reading describes one depth. Everything above assumes uniform soil, which no real site is.

I_e is not the whole fault current

This is the one that most often makes an EPR look several times worse than it is. Only the part of the earth fault current that returns through the ground lifts the electrode. Cable sheaths, armours and any overhead earth wire carry the rest back to the source metallically, and on a well-sheathed urban network that can be most of it.

So I_e = total earth fault current × the fraction returning through soil. That fraction depends on sheath impedances, on how many circuits leave the site and on the source arrangement — it is computed in the study (we use SES FCDIST), and there is no formula to reach for. On the worked case, 2,000 A of HV earth fault with a quarter returning through soil gives the 500 A used throughout this page.

The same left-to-right distance across both halves: the circuit above, the ground potential beneath it. Read down any vertical line and you get the conductor at that point and the voltage of the ground under it.

4. The bridge nobody mentions: BS 7671 Section 442

Everything above is the high-voltage side of the problem, and it is where most explanations stop. But you work to BS 7671, and BS 7671 has a section titled “Protection of low voltage installations against temporary overvoltages due to earth faults in the high voltage system”. That is this phenomenon, named explicitly, with two tables attached.

The link is exact. Section 442 calls the substation's earthing resistance R_e and the part of the HV earth fault current flowing through it I_e. Their product — the same R_e × I_e that is the EPR — is what appears throughout Table 44.1.

5. Where the rise lands

Table 44.1 gives three voltages for each arrangement. Two are stress voltages, across the insulation of low voltage equipment: U₁ at the substation and U₂ at your installation. The third, U_f, is a fault voltage between exposed-conductive-parts and Earth — a touch hazard, not an insulation one.

System	HV/LV earths	U₁ substation	U₂ your installation	U_f on people	Lands on
TT	connected	U₀	R_e·I_e + U₀	—	your installation
TT	separated	R_e·I_e + U₀	U₀	—	the substation
TN	connected	U₀	U₀	R_e·I_e	people
TN	separated	R_e·I_e + U₀	U₀	—	the substation

Bold is where the rise appears. “—” is the table's own “no consideration need be given”.

Read the table across and the point falls out: the earth potential rise appears somewhere in every arrangement. In three of the four rows it lands on insulation, as a stress voltage. In exactly one — TN with the HV and LV earthing arrangements connected — it lands on people, as a fault voltage on anything they can touch. The design choice is not whether the rise reaches the LV installation. It is what it reaches.

That also explains why Regulation 442.2.3's list of remedies is exactly three items: separate the HV and LV earthing arrangements, change the LV system earthing, or reduce R_e. Move the rise across the boundary, change which row of the table you are on, or make the rise smaller. There is nothing else the table permits.

6. Two quantities, two rules

The stress voltages and the fault voltage are judged completely differently, and only one of them has a number.

One stands across insulation, the other across a person. Only one of them has a number in BS 7671, and confusing the two is the easiest way to misread Section 442.

Stress voltages are limited by Regulation 442.2.2 and Table 44.2. At U₀ = 230 V that is 1,430 V where the HV fault is cleared in 5 s or less, and only 480 V where it lasts longer. The two bands correspond to how the HV system is earthed: short clearance for low-impedance earthed systems, long for isolated-neutral and resonant-earthed ones.

A bar chart of the permissible power frequency stress voltage at 230 volts line to earth. An HV fault cleared in 5 seconds or less permits 1,430 volts, which is U0 plus 1200 volts, typical of a low-impedance earthed HV system. A fault lasting more than 5 seconds permits only 480 volts, which is U0 plus 250 volts, typical of isolated-neutral or resonant-earthed HV systems. There is no sliding scale: BS 7671 gives two bands either side of 5 seconds. — The clearance time does not change the rise. It changes what is permitted — and here it is a step, not a curve.

The fault voltage has no number at all. Regulation 442.2.1 requires only that U_f “shall not exceed a dangerous level”. Its NOTE 1 offers one route: in a TN system with the earths connected, connection to a global earthing system can be considered a safety measure against dangerous fault voltages — a system formed by interconnecting enough local earthing systems that proximity removes dangerous touch voltages.

Do not confuse a stress voltage with a touch voltage. Table 44.2 limits what appears across equipment insulation. The permissible touch and step voltages — what a person may be exposed to, as a function of fault duration — are not in BS 7671 at all. They are in BS EN 50522 and ENA TS 41-24, which are the documents an earthing study is actually assessed against. Different quantities, different documents.

7. Whose job is this?

Here is the part that decides whether any of it is your problem, and it is easy to miss. Regulation 442.2.3 states that the requirements are “deemed to be fulfilled for installations receiving a supply at low voltage from a system for distribution of electricity to the public”, and that “the calculation for U₁, U₂ and U_f is normally not necessary for the LV system installer”.

Supplied at LV from the public network Deemed satisfied. The requirements are “deemed to be fulfilled”, and the calculation for U₁, U₂ and U_f is “normally not necessary for the LV system installer”. Somebody else already owns it.

Your own HV/LV substation A live duty. Responsibility falls mainly on the substation installer, owner or operator, who also has BS EN 61936-1 to satisfy. This is the case an earthing study exists for.

So on an ordinary DNO-fed job, Section 442 is somebody else's duty and quoting it at the installer is a misreading. It becomes live where the installation has its own HV/LV substation — and there responsibility falls mainly on the substation installer, owner or operator, who also has BS EN 61936-1 to satisfy. That is exactly the case an earthing study is commissioned for.

Regulation 442.1.1 makes the overlap explicit. It says Section 442 gives rules for the designer and installer of the substation, and that three pieces of information about the HV system are necessary:

the quality of the system earthing;
the maximum level of earth fault current;
the resistance of the earthing arrangement.

Those are I_e and R_e and the context around them — which is precisely what an earthing study produces. The regulation and the computed earthing model are asking for the same three numbers.

8. The rest of Section 442

Section 442 is not only about HV earth faults. Regulation 442.1 scopes four situations, and the other three are worth knowing because nobody associates them with this part of the standard:

442.3 — loss of the neutral in a three-phase TN or TT system. Insulation and components rated line-to-neutral can be temporarily stressed at the line-to-line voltage, up to √3 × U₀ (about 400 V).
442.4 — accidental earthing of an IT line conductor, with the same √3 × U₀ consequence.
442.5 — a line-to-neutral short circuit. The voltage between the other line conductors and neutral can reach 1.45 × U₀ for up to 5 s — about 334 V on a 230 V system.

All three are written as “consideration shall be given”, with no calculation and no limit attached — facts about what equipment may have to survive, rather than checks to perform.

9. What an EPR study actually delivers

A competent earthing study takes the soil survey, the site layout and the DNO fault data, builds a computer model of the electrode system (we use SES MultiFields, part of the CDEGS suite, with FCDIST for the current distribution), and calculates the EPR together with the touch, step and transferred potentials. Results are compared against the BS EN 50522 and ENA TS 41-24 limits; the site is also classified for telecoms purposes, the traditional hot/cold site assessment.

Where limits are exceeded the design is optimised — additional electrodes, grading conductors, high-resistivity surface layers such as crushed rock, or restricting access — and the final report presents the compliant design with the evidence DNOs and approval authorities expect. For how a study is scoped and what it delivers, see our earthing study design service, or the earthing pricing guide for what one costs. If you are working out which system earthing arrangement you have in the first place, the TN-S, TN-C-S and TT explainer covers that — and it matters here, because it selects your row of Table 44.1.

Frequently asked questions

What is earth potential rise (EPR)?

The voltage an earthing system reaches, relative to remote earth, while fault current flows through it into the soil. EPR = Ie x Re, where Ie is the part of the earth fault current returning through the ground and Re is the earthing arrangement's resistance. It is sometimes called ground potential rise (GPR). A 500 A return through a 2 ohm electrode system is a 1,000 V rise.

Does BS 7671 say anything about EPR?

Yes — Section 442, which almost no EPR guide mentions. It does not calculate the rise for you; it tells you what the rise does to the low voltage installation. Table 44.1 gives the resulting stress voltages U1 and U2 and the fault voltage Uf for each system type and each HV/LV earthing arrangement, and Table 44.2 gives a permissible limit for the stress voltages.

Is EPR itself the hazard?

No. EPR is the reference. The hazard is the fraction of it a person is actually exposed to — touch voltage hand-to-feet, step voltage foot-to-foot — plus transferred potential, which is the exception because metalwork can export close to the full rise a long way from the site.

Why does it matter whether the HV and LV earths are connected?

Because it decides where the rise lands. BS 7671 Table 44.1 shows the same Re x Ie appearing as a stress voltage on the substation's equipment, or on your equipment, or — for a TN system with the earths connected — as a fault voltage on exposed-conductive-parts. It never vanishes. Interconnection is the general UK practice, and BS EN 50522 is where the feasibility criteria live.

What is the difference between a stress voltage and a touch voltage?

A stress voltage (U1, U2) appears across the insulation of low voltage equipment and across surge protective devices; BS 7671 Table 44.2 limits it to U0 + 1200 V for an HV fault cleared in 5 s or less, or U0 + 250 V above that. A touch voltage is what a person can be exposed to, and its limits are not in BS 7671 at all — they are in BS EN 50522 and ENA TS 41-24. Different quantities, different documents.

Do I have to calculate U1, U2 and Uf?

Usually not. Regulation 442.2.3 says the requirements are deemed to be fulfilled for an installation taking a low voltage supply from the public distribution system, and that the calculation is normally not necessary for the LV system installer. It becomes a live duty where the installation has its own HV/LV substation, and responsibility sits mainly with the substation installer, owner or operator.

How do I work out the earthing resistance Re?

Not from BS 7671 — it gives no formula for any electrode. It recognises seven types (Reg 542.2.2), requires the resistance to survive corrosion and seasonal drying (542.2.1, 542.2.4), caps it on TT (411.5.3), requires it to be measured (643.7.2), and then points at BS 7430. For a first pass the classical results are enough: a hemisphere is rho / 2 pi r exactly, a vertical rod is (rho / 2 pi L)(ln(8L/d) - 1), and a grid is roughly rho/4r + rho/Lt with r the radius of a disc of the same area.

Why does a substation need a grid rather than rods?

Because area is the lever and rods cannot supply it. In 100 ohm-metre soil a single 2.4 m rod is about 40 ohms; a 25 m x 25 m grid in the same ground is about 2 ohms. The grid formula shows why: its first term is the resistance of a disc of soil the size of the grid, and that term is roughly 89 per cent of the answer. Adding more conductor inside the same footprint trims the rest and little else — which is also why driving a rod deeper helps and making it fatter barely does.

What is a "hot site"?

A site whose EPR is high enough to restrict what telecoms equipment may be installed and how it must be protected. The classification comes from the telecoms and DNO framework rather than from BS 7671, and an earthing study normally reports it alongside the touch and step results.

What can I do if the numbers do not pass?

Regulation 442.2.3 lists exactly three measures: separate the HV and LV earthing arrangements, change the LV system earthing, or reduce Re. Those are the three levers Table 44.1 exposes — move the rise across the boundary, change which row of the table you are on, or make the rise smaller. On the earthing design side, reducing Re means more or better electrode, and touch and step voltages can also be managed with grading conductors or a high-resistivity surface layer.

When do I need an EPR or earthing study?

When a DNO, principal contractor or approval authority asks for one — typically for a new or modified HV connection, a private HV/LV substation, or a site near an existing one. Regulation 442.1.1 asks the substation designer for the quality of the system earthing, the maximum earth fault current and the resistance of the earthing arrangement, which is precisely what a study produces.

Verify before you rely on it. EPR = I_e × R_e is Ohm's law and needs no standard, but everything past it does: how the earth fault current splits between the soil and the metallic return paths, and what the electrode geometry and soil structure make R_e, are what a computed earthing model exists to answer. The Section 442 material here reproduces Table 44.1 for TT and TN only — BS 7671 also tabulates six IT rows, which are not reproduced. And the permissible touch and step voltages are in BS EN 50522 and ENA TS 41-24; neither is reproduced or summarised anywhere on this page, because Table 44.2 limits a different quantity. Treat any figure here as a magnitude, not a study.

Been asked for an EPR assessment, or need one checked before it goes to the DNO? Get in touch.

Earth Potential Rise (EPR)