[Strawbale]convection in porous thermal insulation layers

Jan Hollan jhollan at amper....muni.cz
Fri Mar 11 00:25:56 CET 2005

In short:

the resistance of strawbales to airflow is to be reduced at
least twice to prevent their properties being spoiled by convection.

The problem evidently concerns many non-foam insulation materials.

In long:

At last I found a paper dealing with convection in porous
insulation materials. It appeared in the web-based

       The Nordic Journal of Building Physics (Acta physica aedificiorum)
       http://www.byv.kth.se/avd/byte/bphys/  :

   Mihail Serkitjis, PhD,
   Carl-Eric Hagentoft, PhD, professor Chalmers University of Technology

   The Influence of Natural Convection on the Heat Transfer
 in an Air Filled Porous Medium Bounded by an Air Layer

 The abstract is

 amd the very paper linked to from it is

The paper deals with convection within the layer on the attic floor, or
generally, in a case where heat flows vertically upwards (ceiling to a

Onset of convection can be determined from the paper, providing that
permeability of layer to air flow is known.

Taking airflow permeability of strawbale from page 43 of that outstanding
   Resultater 033.pdf

 as available from the http://by-og-byg.dk page

as equal to 1.0 E-7 m2,

it follows that convection in 40 cm thick insulation would start by a
temperature difference of some 15 K (for pure conduction lambda of 0.044,
or even 12 K, if the non-convecton lambda would be below 0.038). It means
that the ``modified Rayleigh number'' Ra_m would be 40 then.

Ra_m is proportional to the thickness of the layer, to the airflow
permeability and to the temperature difference. It is inversely
proportional to lambda.

Ra_m = 60 would cause increase of heat flux by 1,7 compared to
non-convection case (Nusselt number, Nu). This would happen at a
temperature difference of 22 K. For larger Ra_m the paper gives no data,
but an extrapolation indicates that the Nu would be over 2.5. This fully
explains the ,,effective lambda`` following from most measurements being
some 0.1.

I am not sure how much the case of a wall differs, with the heat flux
being horizontal. A remark in the Danish report says that convection
should start later for a horizontal layer. I also guess that indeed
convection could be worse in a vertical wall.

The vital parameter of any porous layer to insulate well in cold winter
and layers over 20 cm is a low airflow permeability.  It is a most
important parameter to be measured, for all fibrous insulation materials.
In fact, they all insulate almost equally well, if there is no convection,
this is to be expected after all (they absorb LW infrared well, conduction
by fibres is low).

We have to investigate options how to make it at least twice smaller for
strawbales.  Penetrating them with some fine stuff like ``cellulose''
(Isofloc, Climatizer) or simply sawmill may help.  Admixture of very thin
fibres might help too (hemp, flax, wool or even some grass?).

Mixing of straw with finer stuffs might be easy when a loose straw would
be used.

Air permeability of bales at the level of 5 E-8 m2 (for temperate zones)
or even 3 E-8 (for -25 Celsius outside) is to be achieved for ceilings.

Perhaps 2/3 of these values for walls (should we ask Serkitjis and
Hagentoft to compute this situation as well?).

Another option is to make the thickness (or generally, size) of the layers
small. For loose straw it's easy but perhaps even bales could be
subdivided in some extent with e.g. thin wooden sheets (even with

Of course, separating bales at least well by paper or cardboard is a must,
when the wall is to insulate thermally well in winter ``northwards from

I sincerely hope we will be able to achieve U below 0.12 for 40 cm layers,
to allow full implementation of straw for passive houses.

Please try to measure the airflow permeability (are there any more data
available?) of straw and experiment with the ways of reducing it.


More information about the Strawbale mailing list