The design of this heat exchanger is based on my ideas only. I do not know how to
perform calculations on heat exchangers, so basically the thing is my vision on what's
right. But still, a lot of decisions had to be made regarding the shape of the
evaporator, the size of the evaporator and the materials to use.
The evaporator
A realistic figure for the load on the cooling system is about 200W. This incorporates
a heavy overclocked Athlon (120W), a heavy overclocked video card (50W), RAM+chipset (10W)
and thermal losses (20W). If we want to get rid of this heat with a compressor running
at 50% duty cycle, the evaporator must be able to handle a load of 400W. The ability of
the evaporator to consume heat is a function of the following variables:
- delta-T between coolant and boiling refrigerant.
- Contact area between fluid and evaporator coil.
- Evaporation area inside the evaporator.
Let's assume a target coolant temperature at -20 °C, and R134a as the
used refrigerant. If we set the evaporation temperature at -24 °C (quite pessimistic), this leaves
us with a maximal loss of 4 °C at 400W load. This translates to a heat resistance
between evaporating refrigerant and coolant of about 0.01 K/W. Based on my experience
with waterjackets, this is no problem, even with a modest evaporator coil surface area
and only 500L/hr coolant flow.
But let's see how many refrigerant has to change phase per second. The latent heat
of vapourisation of R134a is 217kJ/kg of refrigerant. This means that you need 217kJ of
energy to change one kilogram of refrigerant from a liquid to vapour. Thus, for a 400W (400 J/sec) load
on our evaporator, we need to evaporate 0.0018kg of refrigerant per second (400 / 217000 = 0.0018).
In a real system we need somewhat more since the refrigerant entering the evaporator is not already
at -24 °C. Therefore we just assume that we need 2 grams (0.002 kg) of refrigerant per
second. Since the density of R134a is 1378 kg/m3, 2 grams of R134a equals to 1.45
milliliters of refrigerant per second. This is not much. Some experiments with liquefied butane in a copper
pipe showed that even 20 cm2 of surface area would be enough. So, the first idea was to use a
simple straight copper pipe as the evaporator inside a thicker pipe carrying the coolant.
This idea is shown in the picture below.
A very simple heat exchanger design
(Note: in this image the flow is wrong. Reversing either the coolant flow or refrigerant flow would improve results)
I could have built this, and probably it would have worked well. But I was worried that, in this
design, boiling refrigerant would blow liquid refrigerant straight into the suction line. Thus,
I decided to use a coiled evaporator inside a somewhat larger tube. The calculations above and the
experiments I did made clear that the length of the coil is not that important, so I guessed 1.5 meter
of 12mm tubing would be more than enough for the evaporator coil. I have chosen 12mm tubing so I can use the
heat exchanger in another project when the entire phase-change project fails somehow.
The entire evaporator coil must be made of copper to allow silver soldering of the joints. Silver soldering
the joints is required in the refrigerant loop. Nobody ever told me why, but I guess it is required
to withstand pressure and to prevent the small refrigerant molecules from migrating through the joints.
The evaporator housing
The evaporator must be submerged into the flowing liquid. The easiest method is to just submerge the
coil into a coolant container. I chose not to do so, but to put the coil into a larger pipe equipped
with coolant in/out nipples. This allows me to use the heat exchanger as an inline fluid cooler. This
is useful during experimenting, and the heat exchanger can serve as a direct replacement for the
standard radiator used in my current watercooling system.
Material candidates for the outer pipe are steel, stainless steel, PVC and copper. Steel is easy to weld,
and it is possible to solder it together with the copper evaporator coil. But steel rusts, so it is not
a good candidate. Stainless steel is better, but difficult to weld without special equipment, and impossible
to solder. PVC would be the perfect material to use, but it cannot handle the cold. Thus, copper is the
material of choice.
| 
