I have had issues using very thin, very conductive materials in constructions, but usually my file crashes if I push the constructions too far.
I think your issue might be in your integrated roof approach. How are you modeling the thick airspace? While an airspace has some insulating properties, a thick airspace doesn't insulate significantly better than a thin airspace as air movent in the cavity allows the air to mix and easily conduct heat from one surface to the other.
When EnergyPlus models air in a zone, the default is to model it as fully mixed, so you will see very little insulation due to that airspace. I bet if you remove the airspace in the integrated roof approach you'll see results close to the 1mm metal panel. Then you'll need to go back and figure out an appropriate R-value for the airspace.
-- Karen
On Friday, April 27, 2012, smporritt <mail@xxxxxxxxxxxxxxxxxx> wrote: > > > Thanks for the thoughts. The tiles are white painted steel, emissivity 0.9, so not a low emissivity problem. Manufacturer data claims a minimum thickness of 0.5mm and a site survey estimated the thickness at 1mm. I also tried 2mm and got very similar results to 1mm and the problem exists until the thickness reaches 8mm or more.
> > The reason for wanting to have the cavity above the ceiling tiles as a separate zone is to allow night ventilation of that cavity when the tiles are replaced with PCM impregnated tiles. As I said in my original post I found the results for the two modelling methods are in reasonable agreement for plasterboard and other lower conductivity materials (inc the PCM material). The problem is comparing the retrofit results with the existing metal tile construction - which results can be believed for the metal ceiling tiles?
> > Thanks, > > Stephen > > --- In EnergyPlus_Support@xxxxxxxxxxxxxxx, "Edward G. Lyon" <eglyon@...> wrote: >>
>> Check the emissivity of the metal materials you are using. If they are low-e, there is more insulation value for the air space with low-e surfaces both sides than for an air space with higher emissivity. >>
>> Ned Lyon, P.E. (MA) >> Staff Consultant >> SIMPSON GUMPERTZ & HEGER >> 781.907.9000 main >> 781.907.9350 direct >> 781.907.9009 fax >> www.sgh.com
>> >> From: EnergyPlus_Support@xxxxxxxxxxxxxxx [mailto:EnergyPlus_Support@xxxxxxxxxxxxxxx] On Behalf Of smporritt
>> Sent: Friday, April 27, 2012 8:15 AM >> To: EnergyPlus_Support@xxxxxxxxxxxxxxx >> Subject: [EnergyPlus_Support] Problem with very thin high conductivity construction
>> >> >> >> I have come across a problem with modelling thin metal ceiling tiles and wonder if anyone can shed any light on the issue. >> >> I am modelling retrofit options for an airport terminal building, which consists of a large low zone (30m x 20m, ceiling height 2.7m), glazed along one long wall, with metal ceiling tiles (steel: thickness 0.001m, conductivity 50 W/mK, density 7800 Kg/m3, specific heat 450 J/KgK). Above the ceiling tiles is a cavity to an insulated metal deck roof.
>> >> I have tried two modelling approaches, the first being to build a single zone with a flat roof construction consisting of the metal tiles, an air gap and the insulated metal deck (called my integrated roof version). I have also built a version with a separate semi-exterior unconditioned zone above the occupied zone to represent the roof void to allow control over ventilation to the roof cavity, which I need to do later in my modelling (separate roof version). In both cases the zone volumes were kept the same.
>> >> The results for annual heating and cooling energy use using the two construction methods are very different, with the separate roof construction method using over 40% less cooling energy. However, if I replace the metal ceiling tiles with other lower conductivity materials, such as conventional ceiling tiles or plasterboard, the differences between the modelling methods are small (around 1%). Even if I reduce the thickness of the plasterboard to 0.001m the difference in cooling energy is less than 3% when comparing the integrated roof version to the separate roof version.
>> >> I have tried many tests, including building new models from scratch, EnergyPlus v6 compared to v7, CTF vs CFD, different timesteps and algorithms and reducing the roof air cavity to a very small depth to minimise differences between the modelling approaches. In all cases I am not getting any errors (e.g. convergence) reported by EnergyPlus. Using CFD I have also looked at node temperatures to see if anything odd is happening (oscillating at timestep), but apart from the temperatures being very different there is nothing odd happening.
>> >> If I increase the thickness of the metal ceiling tiles to 0.008m or more the results from the two methods converge, likewise if I reduce the conductivity of the metal tiles to around 10 W/mK or less the results converge. I also tried splitting the metal tiles into 2 x 0.0005m layers and adding a very low resistance no mass material layer between them and in that case the results were very similar for the two modelling approaches.
>> >> Is anyone aware of any issues with EnergyPlus modelling very thin high conductivity constructions? I have searched the forums and EnergyPlus documentation and can find no mention of this being an issue. Any help or thoughts gratefully received, thanks.
>> > >