Snow drift — exceptional load

Local build-up of wind-blown snow: against taller structures, behind parapets, at projections and in valleys (EN 1991-1-3, Annex B).

Configuration

Where the drift forms The lower roof, abutting or close to a taller block. It also applies to nearby structures that do not abut; as a rule, when they are less than 1.5 m away.

Data

s_k — ground snow load [kN/m²] The characteristic ground snow load for the site. The same value as on the snow page.h — height difference [m] The height of the adjacent structure above the roof considered.

b₁ [m]

b₂ [m]

b₁ = width of the lower roof, b₂ = width of the taller structure. The expression uses b = max(b₁, b₂), but l_s is measured on b₁.α — pitch of the lower roof [degrees] Table B.1 distributes μ₃ over the two faces according to the pitch. Above 30° μ₁ vanishes; above 60° μ₂ vanishes as well.

▸ Fill in the data on the left and press Calculate

Tool information

What this page computes

The exceptional snow drift load to EN 1991-1-3, Annex B: the wind-blown snow that builds up locally, in four configurations — against a taller structure, behind a parapet, at a projection, and in the valley of a multi-span roof.

The ordinary snow page covers section 5.3: the shape coefficients of the roof itself. This is a different thing. The snow does not appear out of nowhere behind the obstacle — it is taken from the rest of the roof and deposited there, in quantities several times the uniform load.

Two things that get done wrong, both with numerical consequences

C_e and C_t do not apply

The expression everyone knows is

\(s = \mu_i \, C_e \, C_t \, s_k \qquad \text{(5.1)}\)

but that is for persistent and transient situations. Annex B enters through a different expression:

\(s = \mu_i \, s_k \qquad \text{(5.3)}\)

No exposure coefficient and no thermal coefficient. On an exposed site, where \(C_e = 0.8\), applying (5.1) out of habit would underestimate by 25%, on the unsafe side. It is logical: the exposure coefficient describes how much snow the wind removes from the roof — but here the wind-removed snow is precisely the subject.

It does not add to the ordinary snow load

B1(2) is explicit: for this load case, any other snow load on the roof is disregarded. It is an accidental load, checked in the accidental combination, not added on top of the section 5 distribution.

Whether Annex B applies is a nationally determined parameter. The NOTE to 5.2(2) states that the National Annex may recommend its use for sites where the snow melts completely between winters and where moderate or high wind speeds can occur. The page computes the coefficients; whether they apply is a design decision.

The common structure

Three of the four configurations use the same form:

\(\mu = \min\left(\frac{2h}{s_k},\ \frac{2b}{l_s},\ \text{cap}\right)\)

Each of the three terms has a physical meaning:

\(2h/s_k\) — how much fits behind the obstacle. A 3 m obstacle cannot hold more snow than a 3 m drift.
\(2b/l_s\) — where it comes from. The deposited snow is taken from the feeding area; if the roof is small, there is nowhere for it to come from.
the cap — the code limit: 8 for a taller structure and a parapet, 5 for a valley and a projection.

The page shows all three terms and marks the one that governed. It matters in practice: if the cap governs, the geometry would have demanded more than the code allows, and enlarging the obstacle changes nothing further.

The drift length is \(l_s = \min(5h,\ b_1,\ 15\ \text{m})\) — with the exceptions below.

The four configurations

Against a taller structure (B3). Here Table B.1 enters: the base coefficient \(\mu_3\) is distributed over the two faces according to the pitch of the lower roof. Above 30° \(\mu_1\) vanishes, and above 60° \(\mu_2\) vanishes too — at steep pitches snow is no longer retained. It also applies to structures that are close without abutting; as a rule, when the gap is under 1.5 m.

Behind a parapet (B4). The same expression as for a taller structure, with the same cap of 8. That is no coincidence: the physical mechanism is identical, snow building up at the foot of a vertical obstacle.

At a projection (B4). Here \(\mu_i = \min(2h_i/s_k,\ 5)\), with a coefficient on each side, and the length is \(l_{s,i} = \min(5h_i,\ b_i)\) — without the 15 m cap. Below 1 m² of vertical face the effect is negligible, but the page computes it anyway, so that what was neglected stays visible. For canopies over doors with a span under 5 m an additional cap applies, \(2b/l_{s1}\).

In a valley (B2). The cap is 5, not 8, and the lengths are directly \(l_{s1} = b_1\) and \(l_{s2} = b_2\), without the 15 m limit. The feeding term uses \(b_3\), the length of three successive spans.

A neighbouring check, easily confused

§5.3.2(2): where the roof edge has a parapet, snow fences or other obstacles, the ordinary roof shape coefficient must not be taken below 0.8. That is a section 5 requirement, entirely separate from the calculation here — snow can no longer slide off the roof, so the uniform coefficient no longer reduces.

What it does not cover

§6.2, §6.3 and §6.4 — local effects in persistent situations: snow overhanging the eaves and loads on snowguards. A different chapter, with different expressions.
B2(5) — the simultaneous-loading check across several valleys, where the total over a one-metre strip must not exceed the product of the ground snow load and the building length. The page gives the coefficients for one valley.
Cylindrical roofs and non-uniform geometries, where choosing \(b_3\) requires judgement.
Where the sliding snow ends up at steep pitches — the page says it must be checked, but does not compute it.

The values are transcribed from EN 1991-1-3:2005, pages 26–30, and each expression has a test that reproduces it by hand.

Snow — the section 5.3 shape coefficients, for the ordinary distribution.
EN 1990 combinations — the accidental combination, which the result here feeds.
Wind — the other climatic action, with its own roof zones.