IEC 62305-2: what changed in the 2024 edition

1. One risk, not four

The 2010 edition asked for up to four risks — loss of human life, loss of service to the public, loss of cultural heritage, and economic loss — each judged against its own tolerable value. The 2024 edition combines loss of human life and loss due to physical damage into a single risk R. Every risk component now carries a subscript naming which of the two it belongs to, and the two shares sum into one figure. Loss of cultural heritage is folded into the physical-damage loss rather than assessed on its own.

The economic risk is gone outright, and the cost–benefit annex that supported it went with it. That has a practical use: it is the fastest way to tell which edition a tool is really computing. A results screen that still offers a fourth risk, or offers to weigh the cost of protection against the cost of loss, is implementing the 2010 method whatever its version notes say.

2. Frequency of damage — a second instrument, not a risk

Loss caused by internal systems failing did not disappear when the risks were consolidated; it moved. The 2024 edition introduces the frequency of damage F — a count of damaging events per year, assembled from the same four sources of damage as the risk, and compared against its own tolerable frequency rather than against R_T. It is run in addition to the risk calculation, where failure of the internal systems would unacceptably impair the service the structure provides.

Two things about it catch people out. The tolerable value is not a single number — the standard separates systems whose failure is critical from those where it is not, and the two thresholds are an order of magnitude apart. And the standard is explicit that where a failure carries consequences for the environment, or involves safety-related critical equipment, it belongs in the risk calculation and not here. Frequency of damage is the availability instrument, not a softer route to a pass.

A tool that offers only R against R_T cannot answer the availability question at all. Ours computes F alongside the risk and prints it against the threshold that applies.

3. Strike-point density N_SG, not just flash density

The event rates start from the lightning density at the site. The current edition works from the ground strike-point density N_SG — the quantity a modern lightning location system reports (per IEC 62858) — rather than the ground flash density N_G alone. Where only N_G is available, it is converted with a factor; where you have satellite data, that is a last resort. If your old assessment plugged a flash-density map value straight in, that step now has an extra rung.

4. Thunderstorm warning systems, and people caught outdoors

This is the change most likely to be missing from a tool claiming the current edition, because it adds an input rather than moving an existing number. The 2024 edition recognises temporary preventive measures triggered by a thunderstorm warning system conforming to IEC 62793: where people can be moved to safety or a process shut down on warning, several risk components are reduced by a factor the assessment carries explicitly.

It is a real reduction with a real condition attached. The warning system has to conform to that standard, and the provisions it triggers have to be appropriate to the case — a written procedure nobody executes is not a protection measure. Claiming the credit obliges you to say what actually happens when the alarm sounds.

Alongside it, the edition introduces the risk of a direct strike to people in open areas — car parks, platforms, sports grounds, yards — which the previous edition did not assess at all. If people are outdoors within the structure's curtilage, that is a risk component an old calculation never contained.

5. Risk is evaluated per zone

The biggest structural change is how the verdict is formed. Under the 2024 edition, each risk zone is assessed on its own and compared against the tolerable risk R_T; the structure needs protection if any zone exceeds its limit. There is no single structure-wide risk figure to compare — the per-zone truth is the assessment.

Practically, this means the way you divide a building into zones matters more than it used to. A small, high-consequence zone — a server room, an operating theatre — can drive the whole result even though it is a fraction of the floor area, because its risk is no longer averaged into the rest of the structure.

Our sample assessment is that case, and it is worth looking at because nothing about it is contrived — it is a three-storey retail block with a sales floor and a storeroom:

The storeroom is occupied 200 hours a year against the sales floor's 3 000 — someone is in it about one fifteenth as often — and it still governs the building, because it carries the high fire-risk classification and the sales floor does not. Nobody has to agree with that zoning for the mechanism to be clear: the verdict is the worst zone, not a blend of the two, so a space you are barely ever in can decide what the whole structure needs. That is also where the change in section 6 bites hardest, and the two compound.

6. Occupancy scaling has been removed

In the 2024 edition a zone's loss values are relative to that zone's own population: each of L_T, L_D, L_F1 and L_O1 is defined as a ratio to “the total number of persons in the risk zone or the structure” (§C.2), not to the building's total. A zone's risk therefore stands on the danger present in that zone, and not on how many of the building's people happen to be in it. The only presence term left is P_P = t_z/8 760, the probability that a person is in the dangerous place at the time (§B.11) — a duration, not a headcount share. If your previous method diluted a zone's loss by its share of the building's occupants, expect numbers to move upward.

7. Loss values were reorganised

The typical loss values were restructured. A zone's losses are chosen by loss category — very high, high, normal, low — and the hazards that most affect the outcome select the category directly: Table C.2's top row is for zones with a risk of explosion, with life-saving electrical equipment, or where failure of an internal system would injure people or endanger the environment. Where the damage would reach surrounding structures or the environment, an additional loss L_E is added on top (eq. C.1 and C.2, evaluated via Annex E). Physical-damage losses are treated as loss of property that is at risk whether or not people are present: the presence factor P_P = t_z/8 760 is defined as the probability that a person is in a dangerous place (§B.11), so it does not weigh them. The upshot for a practising engineer is that the loss side of the calculation is read differently from the 2010 tables — one of the reasons a like-for-like comparison against an old worked example will not line up exactly.

What this means for your numbers

Put together, these changes mean a building assessed under Ed. 3 can land on a different verdict than the same building under the 2010 edition — most often a higher zone risk, because the occupancy dilution is gone and the consequence sits with the zone. That is not a tool disagreeing with a spreadsheet; it is two different editions of the standard. When you reconcile a new result against an old one, expect the per-zone structure, the occupancy treatment and the loss values to be where they diverge.