Class IV, not class I: the lightning risk assessment behind the number

"Do we need lightning protection?" is not answered by a rule of thumb, and it is not answered by a class either. IEC 62305-2 answers it with a number: assemble eight risk components over every zone of the structure, sum them into a risk of loss of human life, and compare that with the tolerable risk of Table 4. Only then does a class come out — and it comes out as the lightest one that works, which is usually not the one a specification asks for. This guide runs the lightning risk assessment tool through the standard's own worked example. The worked example is a real run — every figure below was taken from the calculator, not typed in by hand, and checked back against the printed table in the standard.

What the tool is for

The risk management of IEC 62305-2:2010, and only that. Four risks:

• R1 — loss of human life or permanent injury
• R2 — loss of service to the public
• R3 — loss of cultural heritage
• R4 — economic loss

each assembled from the eight components R_A, R_B, R_C, R_M, R_U, R_V, R_W, R_Z per clause 6 and Table 6, summed over the zones, and compared with the tolerable risk. Where R4 is relevant, the Annex D economic comparison follows: the annual cost of the loss against the annual cost of the protection.

The engine holds no coefficient of the standard. Every one is read from the IEC tables in the database, and the mapping layer that turns "archive, paper store" into the Table C.5 row giving r_f = 0,1 holds no number either — it records which row it used. That separation is the point: the engine can be checked against the standard's worked examples, the mapping can be checked row by row against the printed tables, and neither can quietly invent a value because neither contains one.

The worked example, reproduced cell for cell

Annex E.3 of the standard is an office building in five zones. Load it in the tool and press Assess:

(values × 10⁻⁵, against a tolerable R_T = 1 × 10⁻⁵)

That is Table E.20 of the standard. It is also, cell for cell, what the tool returns — R_B 4,78e-5, R_V 4,87e-5, zone Z3 8,88e-5, Z4 7,12e-6, Z5 6,23e-7, total 9,65e-5 — along with the intermediate quantities: A_D = 27 470 m², A_M = 8,45 × 10⁵ m², N_D = 0,11 per year, N_M = 3,38 per year.

R1 exceeds R_T, so protection is required. And the result says where the risk is, because that decides what to do about it: 50 % in R_V and 50 % in R_B — fire, from a flash to an entering line and from a flash to the structure. Not a single percent in R_C, R_M, R_W or R_Z, the four components that failure of internal systems contributes. Which means the answer here is bonding, SPDs at the line entrance and fire measures. An assessment that reports only a total gives you a class; one that reports the components tells you what to spend the money on.

The class is an output, and it is the lightest one that works

With that breakdown the tool proposes: class IV (LPL IV) to IEC 62305-3, together with lightning equipotential bonding at the entrance with SPDs designed for LPL III–IV, and manual fire extinguishing in the occupied zones. Protected R1 falls to 6,00e-6, below the 1,00e-5 tolerable, so the case is compliant.

Class IV, not class I. It is described as the lightest class that brings every relevant risk to or below Table 4, because that is what the standard's procedure produces, and specifying class I where class IV suffices is money spent to no calculated benefit. The result also states the boundary of its own authority in the same breath:

The class above is the outcome of the risk calculation only. The protection itself is designed to IEC 62305-3, and the SPD coordination to IEC 62305-4 and IEC 61643.

The collection area is drawn, not typed

A_D for a rectangle is equation (A.2), and asking for L, W and H forces every real plan into a rectangle — which is wrong for an L-shaped building and meaningless for a PV plant. Clause A.2.1 permits A_D to be determined graphically: the outline of the structure offset outward by 3H. So the outline is drawn on a basemap or typed as coordinates, and the area is taken from it:

A_D = area(outline) + perimeter(outline) × 3H + π × (3H)²

which is the area enclosed by the Minkowski sum of the outline with a disc of radius 3H. The drawing shows the same construction — the outline stroked 6H wide with round joins is that offset — so the picture and the arithmetic cannot disagree. For a rectangle it reproduces (A.2) exactly, and the test suite checks that it does, including the Annex E.4 hospital's A_D = 2,23 × 10⁴ m².

For a strongly concave outline the offset band overlaps itself in the notch and the area is over-stated, which over-states the risk. That is the safe direction, and the page says so in its limits rather than leaving you to find out.

Where a site is made of parts of different heights — a PV plant with module rows, an MV station, a BESS container, a lightning mast — each part is offset by 3H of that part and the union is taken.

What it refuses to assume

This is the part that makes the rest trustworthy.

• N_G, the ground flash density. Clause A.1 defines it as flashes to ground per km² per year, and notes that the value is available from ground flash location networks. It has to be read from such a network, a flash density map or a national annex, and the report prints which. A total flash density is a different quantity and the tool will not convert one into the other without a cloud-to-ground fraction you state. The thunderstorm-day route, N_G = 0,1·T_D, is offered with the standard's own restriction to temperate regions attached — which for the Gulf means it is the wrong instrument and the page says so rather than producing a number.
• U_W, the equipment withstand voltage. Tables B.8 and B.9 are tabulated at 1, 1,5, 2,5, 4 and 6 kV, so only those five are offered, and the figure comes off the equipment datasheet.
• P_SPD and P_EB better than LPL I. Tables B.3 and B.7 give a range of 0,005 to 0,001 and no single value, so no single value is produced. The SPD datasheet has to supply it.
• A hazardous-area classification. The explosion rows of Table C.5 are never selected for you.
• Zones that do not add up. If the zones account for less than the whole population of the structure, the page says by how much — because the risk of what was left out is not counted.

And before anything is entered, the result panel reads: "Not assessed yet: ground flash density; the outline of the site. Nothing here is assumed on your behalf."

Which input actually decides the answer

Every number in the assessment scales linearly with N_G, and several of the coefficients span a factor of ten inside the range the standard allows. So the tool sorts its own inputs by what they do to the verdict, not to the number:

One input decides the answer. Nothing else does. Fire risk in the Archive zone (Table C.5, r_f) — adopted: high fire risk. Across its whole range: nothing that will burn, to high fire risk — the verdict changes at "nothing that will burn" and at "low fire risk".

Eight further inputs move the risk without changing the verdict; seven are treated as settled. And it names its own convention where it uses one: a span below 10 % of the current margin is treated as settled, "IEC 62305-2 sets no such threshold; it is ieccalc's own, and it decides only what you are asked about, never what is computed."

That is the difference between an assessment and a number. r_f in that zone is a property of what is actually installed — cables, oil, batteries — and until somebody has looked, the conclusion is provisional no matter how many digits it carries.

How the implementation is checked

Annex E works three cases through and prints every intermediate quantity, so all three are reproduced:

• 129 checks against the Annex E cases and the Annex D costs
• 63 checks that the plain-language layer answering the E.3 building reproduces the printed R1 and that every coefficient it chose is the standard's own row
• 27 checks of the drawn geometry against closed forms, including equation (A.2)
• 100 coefficients in the database each found in its own printed table

Three cells of the standard's own tables disagree with the standard's own totals by one in the last printed digit — Table E.21 zone Z4 of solution b, Table E.33 zone Z3, and the R_W structure total of Table E.33, which is the sum of the rounded cells rather than of the values. In each case the engine reproduces the printed grand total, and the tests record the discrepancy and the reasoning rather than loosening a tolerance to hide it.

Limits

This is the risk assessment. Nothing here sizes an air termination, a down conductor, an earth termination or an SPD — that is IEC 62305-3 for the protection system and IEC 62305-4 for the SPD coordination. The loss values of Annex C are typical mean values, and a national annex or a client may set others, so every one of them is editable.

The short version

The question "is protection required, and to what class" has a defined answer, and it depends on four things that no rule of thumb contains: how often lightning strikes the ground where you are, how much area the structure and its lines collect, what is inside each zone, and what the tolerable risk is for the loss you care about. Get those and the class falls out — often lighter than expected, occasionally much heavier, and always with a component breakdown that tells you which measure is worth buying.

What it will not do is produce a number before you have N_G and an outline. That is not an inconvenience; it is the assessment.

Assess your own site →