One earthing system, two ways to be killed

A substation kills people in two ways, and they are designed by different engineers. One is the power-frequency earth fault: a few thousand amperes into the grid, a potential rise of several kilovolts, and someone standing next to a structure that is suddenly not at the potential of the ground under their boots. The other is lightning: thirty or a hundred kiloamperes down a mast or a down-conductor, with a step voltage around its footing and a side flash to whatever metal happens to be close.

The two hazards share one earthing system, and that is the point most designs miss. The earthing calculation is done for the fault current; the lightning protection is drawn later, often by another discipline; and nobody checks that the same electrode arrangement satisfies both. This guide walks the personnel-safety problem end to end, in the order the standards set it out, and shows where calculator #003 does the arithmetic.

What actually hurts a person

Not voltage — current, for a duration, along a path through the body. IEC TS 60479-1 is the physiological base, and every earthing limit in the IEC world traces back to it:

• I_B(t_f) — the body current the standard's curve c2 allows for a given duration, at which the probability of ventricular fibrillation stays below 5 %;
• Z_T(U_T) — total body impedance, which falls as the touch voltage rises, so the limit has to be found iteratively;
• HF — the heart-current factor for the path: 1,0 left hand to feet, 0,8 right hand to feet, 0,4 hand to hand;
• BF — the body factor: 0,75 hand to both feet, 0,5 both hands to feet.

IEC 61936-1, Annex B assembles them into the permissible touch voltage:

U_Tp = I_B(t_f) x (1/HF) x Z_T(U_T) x BF

and, where the person is standing on a resistive surface or wearing gloves and boots, the prospective form:

U_vTp = I_B(t_f) x (1/HF) x [ Z_T(U_T) x BF + R_H + R_F ]

Two things follow immediately. First, the additional resistances R_H and R_F are the cheapest safety you can buy on a site — a crushed-rock surface layer is not landscaping, it is the R_F term. Second, the limit is a function of clearing time: halve the fault duration and the tolerable voltage rises, which is why protection settings belong in an earthing study.

The same standard offers a second, equally normative route. Annex C of IEC 61936-1 gives the permissible touch voltage according to IEEE 80 — but read its notes: that curve is drawn for 100 Ω·m soil, a 0,1 m surface layer of 1 000 Ω·m, a 50 kg person and a gravel surface. The IEC Annex B route and the IEEE 80 route are both acceptable; mixing halves of them in one report is not. Pick one, state it, and use its own body weight and surface-layer assumptions throughout.

The power-frequency case, in the order IEC 61936-1 sets it out

Clause 10.3.1 of IEC 61936-1 is an explicit sequence, and it is worth following literally rather than jumping to a mesh calculation:

1. collect the data — earth-fault current, fault duration, layout;
2. make an initial design from the functional requirements (clause 10.2.2: carry and discharge the fault current within thermal and mechanical limits, based on backup protection time, and keep integrity for the installation's life against corrosion);
3. determine whether the site is part of a global earthing system — if it is, the design is complete at that point;
4. if not, determine the soil characteristics, layer by layer;
5. determine the current actually flowing into the soil from the earthing system;
6. determine the overall impedance to earth from layout, soil and parallel paths;
7. determine the earth potential rise;
8. determine the permissible touch voltage;
9. if EPR ≤ U_Tp and the HV/LV requirements of Table 5 are met — done;
10. if not, calculate the actual touch voltages inside and around the site and compare;
11. check transferred potentials outside and inside;
12. check the stress voltage on low-voltage equipment;
13. check circulating neutral currents between parts of the earthing system.

Steps 5 to 10 are what #003 computes: the split factor and decrement factor that turn the fault current into the grid current, the grid resistance, the GPR, and then the mesh and step voltages against the tolerable limits. The earth-fault current and the X/R ratio that feed step 1 come out of #002 — and at a Dyn transformer the single-phase fault is often the larger one, so take I″_k1, not only I″_k3.

The early exit in step 9 is the one people misuse. If GPR is below the permissible touch voltage, the standard lets you stop: nobody can be exposed to more than the whole potential rise. But GPR is typically several kilovolts and the limit a few hundred volts, so in practice you almost always continue to the full analysis. A high GPR is a trigger, not a violation.

Where the HV site meets the LV world

This is the part that reaches beyond the fence, and it is normative. Table 5 of IEC 61936-1 sets the minimum requirements for interconnecting low-voltage and high-voltage earthing systems where there is no global earthing system:

with F typically 2 — and the standard's own caution: in soils with a high resistivity contrast where the top layer is the more resistive, the touch voltage can exceed 50 % of the EPR, so the factor cannot be taken on faith. Where the LV system is entirely inside the area covered by the HV earthing system, the two shall be interconnected regardless. And for installations where HV equipment sits outside closed electrical operating areas — an industrial plant, a process area — the standard points to a global earthing system so that touch voltages from an HV fault stay within the low-voltage limit of IEC 60364-4-41 (50 V).

The lightning case

IEC 61936-1, clause 8.6 is deliberately open: the analysis method is agreed between supplier and user, the user selects the protection level, and the standard points to IEC 62305 (or IEEE Std 998 for the rolling-sphere/electrogeometric work on open-air switchyards). Two requirements are not negotiable: lightning rods and shield wires shall be earthed, and shield wires shall be connected to the steel structure or the earthing conductor so the current reaches earth. A steel structure that is itself a suitable path needs no separate earthing conductor.

For the buildings on site — control room, warehouse, auxiliaries — IEC 62305-3 applies directly, and its geometry follows the class of LPS:

The protection angle for each class comes from Figure 1 of that standard and depends on the height of the air termination — and above the marked heights the angle method simply does not apply; only rolling sphere and mesh do.

Separation distance: the requirement that gets skipped

Where the LPS is not bonded to a metal part, it has to be far enough away that the lightning current cannot flash across:

s = (k_i / k_m) x k_c x l      (IEC 62305-3, Equation 4)

k_i : 0,08 (class I), 0,06 (II), 0,04 (III and IV)
k_m : 1 for air, 0,5 for concrete, brick, wood
k_c : 1 for one down-conductor, 0,66 for two, 0,44 for three or more
l   : length along the conductor from the point considered
      to the nearest equipotential bonding point or the earth termination

On a substation this is the rule that catches cable trays, conduit and instrument tubing routed up a structure that carries an air termination. In a structure with a metallic or electrically continuous reinforced framework no separation distance is required — everything is bonded by construction — but that is a conclusion to verify, not to assume.

Earth termination for the LPS

IEC 62305-3 recognises two arrangements:

• Type A — horizontal or vertical electrodes at each down-conductor, not forming a loop, minimum two in total. Each electrode is at least l₁ long if horizontal, or 0,5 l₁ if vertical, with l₁ read from Figure 3 against soil resistivity and class. Classes III and IV are independent of soil resistivity.
• Type B — a ring conductor in contact with the soil over at least 80 % of its length, or a foundation earth electrode forming a closed loop, with mean radius r_e ≥ l₁. Where r_e falls short, add electrodes of length l_r = l₁ − r_e (horizontal) or l_v = (l₁ − r_e)/2 (vertical), at least as many as there are down-conductors.

Practical details from the same clause, all of which show up in site inspections: ring electrodes buried at least 0,5 m deep and about 1 m clear of the walls; type A electrodes with their upper end at least 0,5 m down and distributed as uniformly as possible; extension of electrodes to reduce resistance being worthwhile up to roughly 60 m; and above 3 000 Ω·m soil, type B or an earthing-enhancing compound. The minimum lengths of Figure 3 may be waived if the earth-termination resistance is below 10 Ω — measured at a frequency away from power frequency and its harmonics, so the reading is not polluted by the station's own return currents.

A substation earthing grid designed to IEEE 80 will normally satisfy these requirements comfortably. Saying so in the design note, with the numbers, is what turns "normally" into evidence.

Where the two systems have to be designed together

Clause 10.3.3 of IEC 61936-1 states it plainly: the HV earthing system shall form part of the lightning protection system, and additional earthing conductors may be required at the injection points. The reasoning is that the two duties have different frequency content:

• lightning is a high-frequency injection — what matters is electrode density right where the current enters, because inductance, not resistance, governs the first microseconds;
• an earth fault is a low-frequency event — what matters is the extent of the electrode system, because that is what sets the resistance and the potential gradient.

A grid optimised only for the second can be a poor lightning earth at a mast footing, and a lightning earth optimised only for the first does nothing for step voltage during a 50 Hz fault. Both are the same copper; only the design checks differ. Where a site has more than one building or location, their earthing systems shall be interconnected — and even then, the standard warns, a stroke will produce a large potential difference between them, so protective measures for sensitive equipment are still required. That is where the coordinated SPD system of IEC 62305-4 belongs.

Protecting people around down-conductors

This is the clause that most directly addresses personnel, and it is short enough to know by heart. IEC 62305-3, Clause 8 treats the vicinity of a down-conductor as hazardous even when the LPS is fully compliant. The hazard is reduced to a tolerable level if any one of these holds:

• under normal operating conditions no person is within 3 m of the down-conductors;
• a system of at least 10 down-conductors complying with 5.3.5 is used;
• the contact resistance of the soil surface layer within 3 m of the down-conductor is not less than 100 kΩ — a note offers 5 cm of asphalt, or 15 cm of gravel, as generally sufficient.

If none of them holds, then against touch voltage the standard requires insulation of the exposed down-conductor giving a 100 kV, 1,2/50 µs impulse withstand — for example at least 3 mm of cross-linked polyethylene — and/or physical restrictions and warning notices to ISO 3864-1. Against step voltage the measure is physical restriction and signage over the same 3 m radius.

Read those three conditions as a design menu. On a manned substation, "no persons within 3 m under normal operating conditions" is usually false for at least one down-conductor — the one on the wall beside a door, or the mast next to a walkway — and that single location is what drives the gravel, the insulated section, or the barrier.

---

Worked example: a 132/11 kV substation and its control building

One site, one earthing system, two checks. The switchyard is 80 × 50 m; the control building is 30 × 12 m and 8 m high, in the corner of the yard; the yard is protected by lightning masts 25 m tall. Soil is two-layer — 120 Ω·m to 2 m, 400 Ω·m below — with 100 mm of crushed rock at 3 000 Ω·m over the whole yard.

The electrical data comes from the fault study (#002): earth-fault current 12 kA, X/R = 12, backup clearing time 0,7 s, and a split factor of 0,55 — 45 % of the fault current returns through the overhead earth wires and cable screens rather than through the soil.

Part A — the power-frequency check

The grid as first drawn: 7 m × 7 m mesh, 12 mm copper, buried 0,6 m, 20 perimeter rods of 3 m.

The check by hand, so the tolerable voltage is not a black box:

Cs      = 1 - 0,09 (1 - rho/rho_s) / (2 hs + 0,09)
        = 1 - 0,09 (1 - 120/3000) / (2 x 0,10 + 0,09) = 0,7021

E_touch = (1000 + 1,5 Cs rho_s) k / sqrt(ts)
        = (1000 + 1,5 x 0,7021 x 3000) x 0,157 / sqrt(0,7) = 780,5 V

E_m     = rho K_m K_i I_G / L_M
        = 120 x 0,7447 x 2,0751 x 6 778 / 1 335,3 = 941,3 V      -> fails by 21 %

The fix is buried length inside the same area, not a lower resistance. Tighten the mesh to 5 m and take the rods to 28 × 4 m:

Adding 40 % more copper moved the resistance by 3,5 % and the mesh voltage by 27 %. If you had spent that copper chasing the resistance figure in the specification, you would still be failing the check that decides whether a person survives.

Part B — the same site, checked for lightning

The control building takes an LPS of class II (the class comes from the risk assessment of IEC 62305-2, not from habit). From Table 2 and Table 4 of IEC 62305-3, that fixes the geometry:

rolling sphere radius r = 30 m
roof mesh                10 m x 10 m
down-conductor spacing   10 m typical

Down-conductors. Perimeter = 2 × (30 + 12) = 84 m, so 84 / 10 = 8,4 → at least 9 down-conductors, with one at each exposed corner where practicable.

Air termination over the yard. The rolling-sphere construction gives the protected radius directly. A sphere of radius r resting on the mast tip at height h touches the ground at a radius √(2rh − h²) from the mast; at a height h_x the protected radius shrinks by √(2rh_x − h_x²):

x = sqrt(2 r h - h^2) - sqrt(2 r h_x - h_x^2)

25 m mast, class II (r = 30 m), equipment at 8 m:
x = sqrt(2 x 30 x 25 - 625) - sqrt(2 x 30 x 8 - 64)
  = sqrt(875) - sqrt(416) = 29,58 - 20,40 = 9,18 m

So one 25 m mast covers a radius of about 9 m at the height of a transformer bushing, and about 30 m at ground level. That collapse with height — 9,2 m at 8 m, 5,6 m at 12 m — is why masts multiply on a switchyard and why a protection-angle sketch drawn at ground level flatters the design.

Separation distance. A metal cable tray runs up the building wall 6 m above the nearest bonding point and is not bonded to the LPS. With class II, air as the insulating medium and three or more down-conductors:

s = (k_i / k_m) x k_c x l = (0,06 / 1) x 0,44 x 6 = 0,16 m

Through a concrete wall (k_m = 0,5) the same geometry needs 0,32 m; with only two down-conductors (k_c = 0,66) it needs 0,24 m in air. Either bond the tray or hold the distance — and note that in a structure with an electrically continuous reinforced framework the requirement falls away entirely, which is a conclusion to verify on the drawings rather than assume.

Earth termination. The building gets a type B ring: mean radius r_e = √(A/π) = √(360/π) = 10,7 m, to be compared with l₁ from Figure 3 of IEC 62305-3 for class II at 120 Ω·m (classes III and IV are independent of soil resistivity). The ring is buried 0,5 m deep, about 1 m clear of the walls — and it is bonded into the station grid, because IEC 61936-1, 10.3.3 requires the HV earthing system to form part of the lightning protection system.

People around the down-conductors. Now apply Clause 8 of IEC 62305-3 to this building:

Two clean routes out, and they cost differently. Either extend the crushed rock to a 3 m band around the building — which the yard surfacing largely does anyway — or add a tenth down-conductor and satisfy condition b), which also lowers the current per conductor. If neither is done, the standard's fallback is an insulated down-conductor section withstanding 100 kV at 1,2/50 µs (at least 3 mm of cross-linked polyethylene) plus physical restriction and signage to ISO 3864-1.

Note what just happened: the crushed rock that made the touch-voltage check pass is the same layer that satisfies the lightning clause. One material, two standards, two justifications — and one maintenance obligation.

Part C — three ways this design could have been signed off wrongly

The same model, run with one assumption changed each time:

• Primary instead of backup clearing time turns a failing grid into a passing one. The tolerable voltage scales with 1/√t_s, so 0,4 s buys 32 % more headroom — headroom that does not exist on the day the primary protection fails to trip and the backup clears at 0,7 s. IEC 61936-1, 10.2.2 says it in as many words: the thermal and mechanical design is based on backup protection operating time.
• Without the surface layer the tolerable touch voltage falls from 780 V to 221 V — a factor of 3,5. That single line is the whole argument for specifying, inspecting and maintaining the crushed rock, and for not letting it be replaced by paving or covered with spoil.
• Taking S_f = 1,0 nearly doubles the mesh voltage, because all of the fault current is forced into the soil. It is the conservative assumption, and on this site it would have driven a much denser grid than the physics requires — expensive rather than unsafe, but it is the same error in the other direction: a factor asserted rather than derived. S_f comes from the earth-wire and cable-screen arrangement; #004 is where the screen paths are quantified.

What to verify, and when

IEC 62305-3, clause 7 sets the inspection regime, and it maps cleanly onto a substation project:

• during construction — the embedded electrodes, before anything is buried and invisible;
• after installation of the LPS;
• periodically, at intervals set by the nature of the structure, its corrosion environment and the class of LPS;
• after alterations or repairs, or after a known strike.

What is inspected: deterioration and corrosion of air terminations, conductors and connections; corrosion of the earth electrodes; the earth-termination resistance; the condition of bonds and fixings. And a rule that is easy to write into a method statement and easy to forget on site: inspection, testing and maintenance shall not be carried out when a thunderstorm threatens.

From IEC 61936-1, the commissioning side adds the verification of the earthing system to the list of tests, alongside clearances, protective barriers, safety signs and emergency exits — and clause 12 requires an operation manual with the safety instructions, up-to-date drawings and single-line diagrams on the premises, and the emergency routes and phone numbers displayed where people can see them.

The failure modes worth looking for

From reviewing this kind of package, the recurring ones:

• "R_g ≤ 1 Ω" quoted as the safety criterion. It is not one. A grid can meet a resistance target and still exceed the touch voltage limit, because safety is set by the gradient inside the grid, not by its resistance to remote earth. The resistance decides the GPR; the mesh geometry decides whether a person survives it.
• Earthing designed for the fault, LPS added afterwards. The two are checked against different criteria but share electrodes. If the lightning design arrives after the grid is fixed, the mast footings usually end up with no local electrode density at all.
• Transferred potential not traced. Every conductive path leaving the site — a fence, a pipe, a telecom pair, an LV neutral, an armoured cable to a remote pump — carries site potential outwards. Table 5's stress-voltage limits exist for exactly this.
• Surface layer specified, then lost. The crushed rock that produces R_F and the C_s factor is a design element with a maintenance obligation. Once it is silted up, contaminated or driven over, the calculation that justified the design no longer describes the site.
• Separation distance never calculated. Cable trays, lighting conduit and CCTV poles routed beside a down-conductor with nobody computing s — and then a side flash puts lightning current into the secondary system.
• Clearing time taken from primary protection. The body limits use the duration a person is actually exposed to, which means the backup clearing time when the primary protection fails to operate. The functional thermal check in IEC 61936-1 says the same thing explicitly.

Where the calculators fit

• #003 Substation Grounding does the earthing-system arithmetic of steps 5 to 10: grid current from the fault current with the split and decrement factors, grid resistance for the actual geometry and soil model, GPR, the tolerable touch and step voltages including the surface-layer factor, and the mesh and step voltages to compare against them.
• #002 Short-Circuit supplies the earth-fault current, the X/R ratio and the clearing-time context that the earthing study consumes.
• #004 Cable Ampacity covers the cable screens and armour that carry part of the fault current back to the source — which is where the split factor comes from, and why it is rarely 1,0.

Personnel safety on a substation is not a document. It is a chain: a fault current you have actually calculated, a soil model you have actually measured, a grid geometry that grades the potential, a surface layer that stays where it was specified, a lightning system bonded into the same earth, and inspections that confirm all of it is still true five years later. Break any link and the rest of the chain still looks perfectly compliant on paper.

---

Standards referenced in this guide

• IEC 61936-1:2010+AMD1:2014 — Power installations exceeding 1 kV a.c.: clause 8.6 (protection against direct lightning strokes), 10.2 (earthing system requirements and HV/LV interconnection, Table 5), 10.3 (design sequence, lightning and transients), 11 (inspection and testing), 12 (operation and maintenance manual), Annex B (permissible touch voltage), Annex C (permissible touch voltage according to IEEE 80), Annex D (design flow chart).
• IEC 62305-3:2010 — Physical damage to structures and life hazard: Table 2 (rolling sphere, mesh, protection angle), Table 4 (down-conductor spacing), 5.4.2 (type A and type B earth arrangements, Figure 3), 6.3 (separation distance, Tables 10–12), clause 7 (inspection and maintenance), clause 8 (protection against injury to living beings).
• IEC 62305-4:2010 — Electrical and electronic systems within structures: coordinated SPD systems.
• IEC TS 60479-1:2005 — Effects of current on human beings and livestock: body current curve c2, body impedance, heart-current factor.
• IEEE Std 80-2013 — the alternative route for tolerable touch and step voltages, referenced normatively by Annex C of IEC 61936-1 and implemented in calculator #003.