Beer Fridge – A Case Study in Thermal Design. Part 1 – A Gift

My boss, Roland, relocated from Germany to the UK a couple of years ago and has taken to life in England with alacrity. As a gift for the Mechanical Analysis product development department (Hampton Court) he bought a little fridge which has been busy ever since cooling the beer in it that everyone has been too polite to drink. A few weeks ago it stopped working. Some inquisitive minds and a couple of screwdrivers later it was in pieces with the conclusion that it was the thermoelectric cooler (TEC) that had given up the ghost. For a product whose raison d’etre is thermal I thought it would make a great case study in the application of FloTHERM for (retrospective) thermal design.Over the coming weeks I hope to show how FloTHERM can be used to design a better beer fridge (the thermal equivalent of a better mouse trap?), taking time to showcase some of the classic FloTHERM features that ensure it remains the #1 CFD based tool used for electronics thermal analysis.

[As an aside it’s also an excellent opportunity for me to take a more user-real approach to acceptance testing the forthcoming version of FloTHERM, V9.2…]

The fridge works by having a thermoelectric cooler (TEC) pump heat using the Peltier effect from inside the fridge down to a heatsink where a fan then blows cold air over it so that the heat is convected away from vents in the lower portion of the fridge housing. Such a flow of heat should ensure the space inside the fridge remains at a low controlled temperature. Much less noisy than the classic evaporator/condenser cycle approach used in most domestic fridge/freezers (though sometimes not as reliable!).


Such an application is considered a ‘bread and butter’ case for FloTHERM. Using nothing but a ruler I measured up the main constituent parts of the fridge and had created a 3D representation in FloTHERM in just over an hour. What took a little longer was obtaining the characteristic information for the TEC and the fan. Such objects are not modelled explicitly per se, a so called ‘compact modelling methodology’ is applied where their key physical behaviour is retained but without modelling the exact physics of their operations. For the TEC, parameters such as the the current required to pump a certain number of watts against a temperature difference at two (hot side) temperatures are used as input to FloTHERM TEC ‘SmartPart’ object (this particular TEC was not a Marlow or Melcor TEC SmartParts in the installed libraries). For the fan, the fan curve that relates the pressure drop over than fan to the amount of air that it can shift was required. Thanks to the beauty of Google (other search engines are available) such information was only a part number away.

exploded1The Union Jack was simply a matter of using FloTHERM’s texture mapping capability, using a .jpg image taken by my phone camera. The ‘fly-by’ animation shown on the left here is created by setting a few choice view points locations which are then traversed in sequence and output as an .avi (I used Corel to convert to an animated GIF).

Any competent user of simulation software should have at least a grounding in the theory behind the physics being simulated to the extent whereby the results generated can be perceived as either being realistic or not. Just a fancy way of saying that I expected the air within the fridge to be cold. Here are the surface temperatures on the inside of the fridge, with the side of the fridge hidden for clarity.

inside_surface_tempNot wanting to cast dispersions on the purchasing power of my boss but by all accounts this looks to be a real cheap and nasty fridge, not at all providing a uniformly cool interior. The TEC cools a little coldplate at the bottom of the inside of the fridge, sucking heat away from the air in the lower portion of the fridge but, as hot air rises, the cold air simply stays at the bottom.

In terms of the effectiveness of the fan, a good design would have the fan induce cool air from around the underside of the fridge, blow it onto the heatsink, the air would heat up, then the vents should be arranged so that this warmed air vents back into the room. One thing you wouldn’t want is for the warm air to recirculate back into the underside intake of the fan to be blown back into the heatsink. O dear….


Going back to the TEC, one thing I did have to assume was at what current it was set to operate at. For this model I guessed at 2.5A. TEC performance is very sensitive to this, I wonder what an optimal TEC operating current would be for this fridge. More on that next time!

26th November 2010, Ross-on-Wye

Read the Rest of this Series


9 thoughts about “Beer Fridge – A Case Study in Thermal Design. Part 1 – A Gift
  • Fantastic case study Robin! Perhaps before the next purchase decision is made you could do an analysis of the model to make sure it’s a good purchase decision.

  • Superb Robin! I appreciate that you help me to make better purchases in the future. Unfortunately – I got a spare one of exactly that fridge already…. Mmh.

  • Nice study.
    I have one small question: how did you model the air inside and the air outside? I guess the temp difference is approximate 25°C between outer air maximum and the 0°C bottom plate inside the fridge? Did you use: cutouts, sub-domains, cuboids with air properties or what?
    The answer could help me finishing my study: the effect of using expanded polystyrene for thermal insulating an apartment (where I have -30°C outside air temperature and 22°C inside room temperature).
    Thank you in advance.

  • Gabriel, the air was modelled explicitly both inside and outside the fridge. Although the air speeds inside the fridge, due to the stable thermal stratification, were extremely low. The only inputs were the ambient air temperature outside the domain and the TEC driving current (that did result in sub zero degC temperatures in the little cold plate attached to the TEC). If this were a building services application we’d be talking U-Values, please see this other blog series on that issue:

  • nice work Mr.Robin Bornoff. i am doing some analysis in flotherm for temperature control using TEC. but i am getting problem over it.
    the real problem is during parameter value entrance (like hot side1 temperature, hotside 2 temperature) in smart part TEC in flotherm 9.2.
    can you please help me for this..?
    i have TEC1-12704 to use in my analysis, operating current is 3 amp.. i am little confused with hotside1 and hot side2 (in TEC construction menu of flotherm). as TEC have only one value for max current, max vaoltage, delta T and delta Q.but flotherm TEC condtruction menu asking for both sides??

    thanks in advance.

    • Salman, thank you.

      The TEC requires characteristic information at two hot side temperatures due to the temperature dependence of its operation. It is your TEC suppliers responsibility to provide enough characterisation information. This is provided by the larger suppliers. I would advise trying to get this (quite standard) data from your supplier.

Leave a Reply

This article first appeared on the Siemens Digital Industries Software blog at