[natecostello]

I've been reading old forum discussions on optimal insulation methods. I decided to put together a quasi-1D thermal analysis of the roof/ceiling to allow some sensitivity studies (e.g. XPS vs Thinsulate) and evaluation of where the major losses are (ribs vs bulk roof). The primary intent of this model wasn't quantitative determination for losses, but instead comparative evaluations.

The analysis is here.

On that page you'll see a description of the model, and a link to a google spreadsheet with the parameters (and basis) used to determine the resistance values. I wouldn't consider myself heavy on thermal analysis so any feedback or suggested improvements (or glaring errors) are welcome.

Tentative Conclusions:
XPS (1.5") on the ceiling doesn't provide a massive benefit over SM 600 (1.65") thinsulate (15% delta in loss).
25-30% of ceiling heat loss occurs via the ribs.

Notes: Bulk insulation is XPS or Thinsulate (between ribs). Over top of this is 5mm coroplast panels with the outward (cargo-space-facing) side covered in 1/8" neoprene. Model also allows for neoprene on backside (one case examined). Outside of the panels there are two lengths of L-track that run almost the full length of the cargo area. L-track includes a FR4 thermal break from the ribs. L-track helps retain the coroplast panels.

 

[jondrums]

Please take a look at coating the metal surfaces with a thermally insulative paint - such as lizard skin or pretty much any paint with ceramic spheres mixed in. For starters assume 1mm thick and 0.2 W/mK bulk conductivity. That should really cut down on the losses through the ribs

 

[sportcoupe]

Back from the dead.

Some people put way too much effort in a conversion van.

 

[Vanpackr]

I've come to understand that these vans are as much a medium of self expression as they are a utility. And folks tend to pour themselves into their passions.

 

[Tonwagen]

A resistor network, as you have, is the framework I would start with, too. The hard part is estimating the values of some of those resistors. So long as they represent uniform sheet thermal resistance (over a known area) between the inside and outside ambient temperatures they can be estimated accurately. But when convection and radiation are taken into account it gets harder. For example, I believe a vertically-oriented, thin, thermally-conductive membrane in still air (imagine a thin copper sheet or even a piece of window glass) has an R-value of about 1 due to the boundary layers alone (.5 on each side). And the skin of the van when sitting in the sun (or under a clear night sky) is at nowhere near the outside ambient temperature due to radiative heat transfer. I'm not dismissing your effort, just pointing out that at some point additional detail is not likely to improve the overall estimate unless complex models of some of the resistances are used.

--Frank (not a thermal engineer)

 

[jondrums]

modelling the radiative heat transfer is only necessary for the outer skin - and yes, you are correct that the radiative effect from the sun or a cloudless night sky is significant. However, the radiation going on inside the insulation materials and inside the van can be ignored without any concern <%1.

Every time someone wants to talk about using CAD or doing thermal analysis, there is the predictable backlash of "why bother". Those aren't helpful comments, please just keep it to yourself and ignore the thread. We're having fun over here.

 

[natecostello]

Please take a look at coating the metal surfaces with a thermally insulative paint - such as lizard skin or pretty much any paint with ceramic spheres mixed in. For starters assume 1mm thick and 0.2 W/mK bulk conductivity. That should really cut down on the losses through the ribs

Thanks. I updated the analysis to add a sensitivity to lizard skin using your values:

Lizard Skin was looked at in two different cases. In one case, the effect of Lizard Skin was examined by modeling it as insulation on the backside of panels, which would be analogous to applying it to the full interior surface of the ceiling (including the ribs). In the second case, the effect of Lizard Skin was examined by modeling it as rib insulation, which would be analogous to applying it to just the ribs. Total loss for Lizard Skin applied to the full interior, the ribs only, and no application are 203.8W, 204.4W, and 205.4W respectively. All of the above cases used Thinsulate for bulk insulation as the base case.

This isn't a big surprise. Keep in mind my design (and model) includes a 5mm layer of coroplast with 1/8" of neoprene adhered to the face that covers the ribs (and rest of the ceiling) already. One of the original trades studies was adding another 1/8" layer of neoprene on the backside of the coroplast and that only changed things by roughly 15%.

Also...if you want to dig in more: All the models as well as a ton of other stuff in a repo I'm using to gather my design related thoughts: https://github.com/natecostello/van_two_point_oh

 

[natecostello]

A resistor network, as you have, is the framework I would start with, too. The hard part is estimating the values of some of those resistors. So long as they represent uniform sheet thermal resistance (over a known area) between the inside and outside ambient temperatures they can be estimated accurately. But when convection and radiation are taken into account it gets harder. For example, I believe a vertically-oriented, thin, thermally-conductive membrane in still air (imagine a thin copper sheet or even a piece of window glass) has an R-value of about 1 due to the boundary layers alone (.5 on each side). And the skin of the van when sitting in the sun (or under a clear night sky) is at nowhere near the outside ambient temperature due to radiative heat transfer. I'm not dismissing your effort, just pointing out that at some point additional detail is not likely to improve the overall estimate unless complex models of some of the resistances are used.

--Frank (not a thermal engineer)

Thank you for the feed back. Agree on all counts. No one should take the absolute watts as accurate, but the relative watts between cases is a good metric for comparison (and should hold decently well across different radiative situations). I considered not including the boundary effects at all, but did so in the end. The basis for the horizontal exterior and interior boundary layers came from here and total to an R-value of 0.78 (pretty close to 1).

 

[jondrums]

thanks! no surprise, the lizard-skin is only going to do anything if there is bare metal showing. If there is already some insulation, I can't imagine it does anything

 

[orton]

Since you are doing an insulation comparison you might include Aerocel closed cell foam in your review. Widely used to insulate beer tanks and piping. I used two 1" layers in the cubby holes above and below the window indents, curved roof over the cab and on the curved roof above the windshield. Can be folded to fit through the cubby hole openings. Better "R" value than Thinsulate and better noise rating than polyiso. Easier to install compared to Polyiso.

Aerocel® AC Sheet & Roll - Aeroflex USA

www.aeroflexusa.com

Walls | Orton Travel Transit

www.ortontransit.info

It would be interesting to know how the Aerocel compares to your other two choices. Suspect it might be better than either choice.

 

[Jacek]

Did you use Aerocel with the adhesive backing or a separate glue?

 

[orton]

Did you use Aerocel with the adhesive backing or a separate glue?

I used adhesive recommended by Aeroflex: Aeroseal contact adhesive. No issues with the adhesive. Did not use the adhesive backed choice.

 

[natecostello]

Better "R" value than Thinsulate and better noise rating than polyiso. Easier to install compared to Polyiso.

From a thermal conductivity view: better than Thinsulate, worse than XPS, identical to neoprene. Like the idea of having the adhesive included, thanks.

 

[harryn]

Interesting work - thanks.

I believe that you might be under estimating the delta T, especially for the coatings.

I have been in a Transit when it was full sun and 110 F outside. The surface was far too hot to touch.

Inside of the van, coated only with a layer of (cannot remember which one) sound insulation spray the inner surface was very comfortable to the touch. Might have been lizard or it's equivalent.

What was surprising is how hot the piece of sound dampener was that had been attached to the roof interior even though it was coated as well. (something similar to dynamat)

My goal at least is to have it 70 F inside when it is 120 F outside. (yes it really did get to 118 F this year).

Assuming that the surface is 20F hotter than ambient, then 140 F - 70 F is a 70 F delta.

Please continue on with the work - it is quite useful.

Harry

 

[natecostello]

I believe that you might be under estimating the delta T, especially for the coatings.

Thanks for the encouragement. Two notes regarding the delta T (apologies if this was clear and I'm missing your point):
1. These calculations are all in metric, so I am looking at a 30C delta (so equivalent to 86F internal while at freezing outside).
2. The delta T is just assumed and imposed as a boundary condition. The resulting watts transferred for that delta T is really the metric for comparing different materials/configurations.

What was surprising is how hot the piece of sound dampener was that had been attached to the roof interior even though it was coated as well. (something similar to dynamat)

Maybe due to the higher specific heat capacity of the elastomer in the dynomat sustaining temperature while sourcing heat into your fingers? Kind of the opposite of how aluminum foil doesn't feel hot after it leaves an oven.

 

[harryn]

You are right - not sure what I was thinking, of course they are metric. Even in the late 70s when I studied chemical engineering all calculations were done using the metric system.

One of the interesting things about the insulations that are routinely used in vans is the effect of temperature on their insulation value (resistance) and here is why I think about this aspect.

At 0 C it really isn't all that cold out so really any insulation can work.

When a van really needs to be insulated against low outdoor temperatures is when the outdoor temperature is in the range of (-20C) to (-40 C).

On the other end of the scale, the need for insulation is when the outdoor temperatures are in the +30 to + 50 C range, and the van exterior skin temperature can easily be 20 - 30 C hotter than this, so really you are insulating against a steel skin that can be at 80 C.

30 C is not that comfortable for being inside of a van in either winter or summer, so I perhaps consider 20 or max 25 C as an interior air temperature?

If a person just uses the typical published value for an insulation, it is not valid for either of these temperature ranges, but instead at a more moderate temperature range. At these more important exterior temperature conditions, the values are quite different from the typical published information and can easily vary 3x over the total range.

I wondered if you can somehow add this temperature dependent insulation value capability to your code, at least a fixed value that reflects the effect of temperature in these more important ranges?

BTW - I have dabbled with some much more simplistic approaches to this and it is surprising how well 18mm thin lamination baltic birch competes when covering this entire temperature range vs a number of materials. I had originally never considered it as an option but I have had to re-think my preferred synthetic vs BB for some applications.

 

[natecostello]

If a person just uses the typical published value for an insulation, it is not valid for either of these temperature ranges, but instead at a more moderate temperature range. At these more important exterior temperature conditions, the values are quite different from the typical published information and can easily vary 3x over the total range.

I wondered if you can somehow add this temperature dependent insulation value capability to your code, at least a fixed value that reflects the effect of temperature in these more important ranges?

The simplest way would just be to recalculate the "resistor" value using the R-value or kappa (thermal conductivity) for the temperature of interest using the spread sheet, and then rerun the LT spice model. Granted you are making an assumption about bulk temperature throughout that insulating component thickness. To make it account for the gradient would be a fair amount of work.

 

[harryn]

The simplest way would just be to recalculate the "resistor" value using the R-value or kappa (thermal conductivity) for the temperature of interest using the spread sheet, and then rerun the LT spice model. Granted you are making an assumption about bulk temperature throughout that insulating component thickness. To make it account for the gradient would be a fair amount of work.

I agree. Spice certainly is a flexible enough platform to deal with these non linearities, although I am more of a user of the output than knowledgeably about using it.

It isn't always so easy to find these physical properties over the extended temperature range so sometimes you just have to use what you can find.

The ceramic coatings temperature value would be somewhere near the steel skin surface temperature, which I think is what causes the greatest off set in real world experience vs the usual calculations, along with the challenge of accounting for the emissivity effects.

Since the insulation is typically trapped between the steel skin and an interior wall, for the higher temperatures, any physical property value that can be found in the 50 - 80 C range could be used.

For the low temperature end, if we can find physical properties near (-20 C) that is probably adequate as a calculation results enhancement.

 

[harryn]

Maybe due to the higher specific heat capacity of the elastomer in the dynomat sustaining temperature while sourcing heat into your fingers? Kind of the opposite of how aluminum foil doesn't feel hot after it leaves an oven.

I am not sure of the cause, but that could be it. The result was so dramatic that I really question the use of these materials in a van. Obviously a 5 second "touch with the finger test" is not really "data" but it was not a small effect either.

 

[harryn]

I had a chance to look at that Transit again yesterday. It turns out that my memory was incorrect - the (dynamat or equivalent) was not covered by the spray on sound deadener. It was lizard skin sound spray on material on the rest of the interior, except for a few ceiling ribs.

Rather than > 100 F, this time is was about 60 F and anything not coated felt very cold to the touch.

There is no question in my mind that these coatings have a significant impact.