[Strawbale]Re: convection in porous thermal insulation layers
jhollan at amper....muni.cz
Thu Apr 21 21:55:43 CEST 2005
a good news for the Earth Day:
there is a way to get U=0.1 W/m2K standard bale walls, reaching the
theoretic limit. I'm on the verge of finishing the needed technology
(dividing the bale by gypsum layers into thirds, perpendicularly to the
heat flow). I hope I will be able to show it as a remark on our poster at
9. Passivhaustagung next Friday in Ludwigshafen.
Some theory at first:
Thick layers insulate proportionally to their thickness whenever the so
called Nusselt number is close to 1. This number is a ratio of real
thermal flux (increased due to convection) to that without convection
(i.e., to the thickness/lambda value).
To arrive at Nusselt number, modified Rayleigh number Ra for porous layers
is a base. Ra is proportional to the height of cavity H, temperature
difference T and permeability K. For the usual winter temperature in a
cavity and usual materials with lambda = 0.04 W/(m.K) it is about
Ra = 0.7 (H / 1 dm) (T / 10 K) (K / 0.01 mm^2)
As I wrote on Mar 10, horizontal layers can be without convection even in
case of upward heat flux. Ra=40 is the threshold (or 25, if the layer is
open to the free air at the top). Above this threshold,
Nu = 1 + 0.04 (Ra - 40).
Vertical layers with horizontal heat flux are always subject to
convection. But Nusselt number remains below 1.1 if Ra is below 10.
If the convection cell would have a square cross-section (it can be
roughly the case for bale walls, due to the perturbation by inhomogeneous
volume density) then crudely
Nu = 1 + Ra/100 below Ra=15,
Nu = 1 + Ra/60 15 < Ra < 40,
Nu = 1 + Ra/45 40 < Ra < 100.
In another words, Ra=20 compromises the insulation properties by one
third, Ra=90 thrice. (These are step-values guessed from a smooth graph
in Bejan's book on convection).
Straw has K = 0.1 mm2 -- that's a lot. I measured this value too, with
help of a long polyethylene bag giving a overpressure of about 1 Pa (my
bales were about 75 kg/m2).
Such a very porous material is to be either made less permeable, or
divided into more layers.
1) dividing the layer
In the roof-space, dividing the layer into halves (e.g., by paper, cardboard
or any old foil with a lot of holes) is enough. Ra diminishes four times.
Two layers of bales (total height 80 cm) would have Ra about 175 at 30 K
temperature drop. After separating the bales horizontally, Ra diminishes
to tolerable 45. Excellent U value down to 0.05 W/m2K can be achieved this
One layer of standard bales in roof-space has up to U=0.3 W/m2K in these
conditions however, and two unseparated layers have probably no less than
In a wall, the solution is not that simple. Just separating the bales
horizontally cannot help (I was wrong assuming that). A 80-cm double layer
with vertical convection barrier in the middle gives U below 0.1, but most
people prefer thinner walls.
Loose fill insulation could be divided horizontally into layers which are
just fifth of the width of the cavity, then the convection contribution
stays below ten per cent even for permeability K = 0.1 mm2. However,
vertical barriers (cardboard) are more efficient, two are always enough.
Ra (computed from horizontal dimension instead of height) drops nine
times, from painful 90 to very tolerable 10.
Within bales, such vertical barriers may be made from gypsum most easily,
as it is floats very well. A comb from thin pipes (4 mm to 6 mm outer
diameter needles, 10 mm to 16 mm modulus) can serve for the injection,
starting in mid-depth and going to the surface (starting at the opposite
side does not work, the tips of needles become too dispersed to per bale
are needed. 10 kg of gypsum per bale should suffice to create two vertical
gypsum boards, dividing the bale into thirds. The final price of such
bales will be however reach at least a fourth of the commercial insulation
price, due to the gypsum costs.
I think mud (or mud-gypsum mixture) or cement could be used for that
purpose, but this is not tested yet (I have no proper clay at home).
Large viscosity may be a problem for a simple injection device (a slim
wooden box suffices for gypsum).
Two inner boards would enhance fire resistance of the wall to almost
infinity. Twice or thrice thicker boards could serve as a load-bearing
2) lowering the permeability
Commercial cellulose fibre infill might be added to the bales during
their production, to lower their permeability to the needed 0.01 mm2. I'd
be pleased if somebody would test this option (I'll describe the
field measurement of permeability before summer, or sooner when asked...).
I hope that injection of such infill with a pipe comb might be possible
with a proper pneumatic device, creating two barriers of less permeability
which could divide a single convection cell into three, similarly as solid
barriers do. Then the bales would remain as elastic as ordinary ones.
Event giant bales should be divided by convection barriers. Without them
they can't achieve U=0.1 W/m2K, being rather near to 0.2 in the strong
frost periods (a bad news...). With three barriers however, the
non-spoiled insulation limit of U = 0.05 W/m2K can be restored!
Straw can and surely will become a reliable passive-house building
material. Not in its natural state, but as a composite.
More information about the Strawbale