The previous post handled already about natural convection in mineral wool. We found the following paper about a simulation of an attic insulated with mineral wool. It is written by scientists from Saint-Gobain and “Ecole Nationale Supérieure de Chimie et de Physique de Bordeaux, France”. The abstract is as follows:

Convection phenomenon inside the insulation materials that are representing porous media may be responsible for important increase of the heat flow density through the building element, when the climatic conditions are very cold. A numerical study based on CFD software simulations is performed in order to find out the conditions for which the convection onsets and furthermore to quantify its influence on the Uvalue of the building element – that is a loose-fill attic. The geometry of the grid is very close to a 2D cut in a real attic. The most significant heat transfer phenomena present in the attic – conduction in solid elements, convection in porous medium, convection in air spaces are taken into account. This is done exclusively by using the mass, momentum and energy conservation equations. The insulation materials analyzed are highly permeable mineral wools. The results are expressed in terms of critical Ra* numbers and variations Nu* against Ra*. A critical Ra* value of 28 is found. Three products, (2 –in glass wool, 1 – in stone wool) are studied – all showed absence of convection for outdoor temperatures higher than –40°C. The stone wool product is the most exposed to convection onset.

• The insulation on the attic has a thickness of 0.5m.
• Glass and stone wool were tested (Saint-Gobain produces glass wool).
• The thermal conductivity is assumed to be 20% larger parallel compared to perpendicular.
• The natural convection effect starts at -40°C for stone wool and colder for glass wool.
• The thermal conductivity becomes the double at -80°C.

I have not any doubt about the simulations and it is realistic that more permeable wool is more sensitive for natural convection. But I am surprised about the assumed thickness: 0.5m. I doubt that many attics in France have 0.5m thermal insulation. I guess what can happen at lower thicknesses. It means I have to study Nusselt and Rayleigh numbers.

In the previous post, it was shown that creating a vacuum in mineral wool improves the thermal conductivity with a larger value than the thermal conductivity of still air. I am sure that moderate temperatures and thicknesses were used. I doubt that extreme temperatures are needed to induce the effect.

I found a PhD thesis about this topic “Heat transfer in Rockwool, modelling and method of measurement”, written by Susanne Dyrbøl in 1998, describes the phenomenon in detail. In this work, the effect of natural convection is described and measured in a vertical insulation layer.

To conclude, I mention a recent paper, which calculates the thermal conductivity in mineral wool. Natural convection is not taken into account … .

The more I study, the less I understand why cellular glass has only 1% of the thermal insulation market. Probably the recent price drops of cellular glass, initiated by GLAPOR will change that situation.