Vapour Resistance of OSB

An Eccentric Anomaly: Ed Davies's Blog

In the past there's been a lot of confusion in discussions on the Green Building Forum (GBF) about the ability of OSB to impede or allow the transfer of water vapour. This is quite significant as OSB is a handy and cheap material for wall and roof constructions where movement of water vapour is a significant concern (although by which modes and in which directions is also a bit controversial).

Somebody using the pseudonym “A.L.” on the Navitron Forum sent me some references and discussion regarding my concerns about my roof build up following on from discussion on the GBF: Ventilation gap under sarking board which might throw some light on the inconsistency of quoted OSB vapour performance values.


The basic measurement of concern here is the vapour resistance of a sheet of material. In SI units it's measured in GN·s/kg - giga newton seconds per kilogram. A typical value might be 20 GN·s/kg. That means that you have to push with a force of twenty billion newtons for one second to push a kilogram of water vapour through the sheet. Alternatively, you could push with one newton of force for 20e9 seconds (634 years and a month or so) or any combination which multiplies to the same amount.

An alternative way of writing this unit is as MN·s/g which is numerically the same but doesn't follow the SI convention of using base units in the divisor (the base unit of mass being, bizarrely, the kilogram).

Going with resistance is permeance which measures the ability of the material to transmit water vapour. This is just the reciprocal of the resistance and so is expressed in kg/(GN·s), g/(MN·s) or, best, µg/(N·s).

Resistance and permeance are properties of a particular layer, say 18 mm of OSB. Following a naming convention which is sufficiently bizarre that I always have to stop and think about it, thereby sometimes getting it wrong, the properties of the material which are independent of thickness are called resistivity and permeability.

The units of resistivity are GN·s/(kg·m) (or MN·s/(g·m)). A resistivity of 750 GN·s/(kg·m) for a material means that you have to push with a force of 750 billion newtons for a second to push a kilogram of water vapour one metre through the material, or 2 kg half a metre or whatever. Permeability units are the reciprocal, as you'd expect, though with the multiplier expressed canonically they're µg·m/(N·s) - if you push with a force of one newton for a second you can shove one microgram of vapour one metre through the material (or 2 µg half a metre, etc).

In principle, you can work out the resistance of a particular layer by multiplying the thickness of the layer (in metres) by the resistivity of the material. If you've got 18 mm of a material with resistivity 750 GN·s/(kg·m) then the resistance would be 750 × 0.018 = 13.5 GN·s/kg. This doesn't seem to work for OSB; see below for the point of this article.

Sometimes resistivities are given as “µ” values. These are the resistivities of materials relative to that of still air which can be taken as 5 GN·s/(kg·m).

“A.L.” sent a link to this document with a description of the conversions involved. One which previously made the subject a bit clearer for me was Breathability In Buildings by Neil May.


Another link “A.L.” sent was to the BBA Certificate for Norbord OSB/3 which I hadn't previously seen and which makes interesting reading (for a rather narrow definition of “interesting”).

In section 5 it gives two sets of µ values for OSB/3: 30 (wet cup), 50 (dry cup) as per BS EN 12524 : 2000 but also a table of experimental values found in accordance with BS EN ISO 12572 : 2001 (wet cup):

Panel thickness (mm) Water resistance factor (µ)
9 219
15 147
23 107

This is getting really weird; µ values are supposed to be properties of the material and aren't supposed to change with thickness. Presumably there's something a bit odd going on so that simple multiplication of the resistivity by the thickness doesn't give the actual resistance of the layer. “A.L.” speculates briefly in passing that there's some sort of surface effect causing this. I think he's probably right though it does seem a bit odd.

If we take these µ values at face value then multiplying them by the thickness of the panels gives us the thickness of a layer of still air with the equivalent resistance. Multiplying that by the resistivity of still air (5 GN·s/(kg·m)) gives us the resistance of the panel:

Thickness (mm) µ Equiv. air Resistance
(mm) - (m) (GN·s/kg)
9 219 1.971 9.85
15 147 2.205 11.025
23 107 2.461 12.305

Doing a rough fit to these numbers gives an approximate formula:

R = (8.3 + 0.175·t) GN·s/kg

where R is the resistance of a layer of OSB and t is its thickness in millimetres.

In other words, the relationship between thickness and vapour resistance looks to be roughly linear but not proportional; the intercept on the y-axis is not zero and is actually a significant proportion of the resistance provided by layers of typical thicknesses used in building.

The coefficient of the thickness term in this formula (0.175 GN·s/kg·mm or 175 GN·s/kg·m) corresponds to a µ value of 175 / 5 = 35. This is in reasonable agreement with the BS EN 12524 : 2000 wet cup µ value of 30.

I think this constant offset probably accounts for a large part of the wide range of resistivities quoted for OSB (e.g., 100 to 300 given by Neil May in the paper linked above).

[Update: 2014-08-30: turns out I'm not the only one who finds OSB and water vapour puzzling, somebody's done a PhD on the subject.]