I have finally created a prototype of energy harvester for my project. Let's start from the beginning. I managed to get a solid copper brick with a hole drilled through. Unfortunately, the diameter of this hole was too large to tightly fit the battery pipe, inside which hot water runs. The difference between diameters was approximately 4 mm. An idea came to mind to use thermopad. It is the flexible and squeezable small piece of mat which conducts the heat pretty well. So, I used it to feel the void between the brick and the pipe. You can check how it looks in the photos.
The brick has some screws and one can disassemble it. I unsoldered the TEG with radiator from the kit and put in against the brick. At first, I could not get better values on my meter than with heat from a hand, so I have decided to use some silver thermo paste in order to create better contact. Just the same I use for my PC's CPU. That did the trick! When water started to flow, I had over 110 mV with open circuit. With load I got approximately half of that.
I did check the temperatures. After a while – steady state – the radiator temperature was approximately 45 deg C and water coming from the tap was almost 55 deg C. I assume that the temperature of the Peltier module's hot side was little over 50 deg C, because of the heat loss across the thermopad, brick and paste. Ambient temperature was 26 deg C.
I believe this method of harvesting is a win for me. Now, I have to create a device based on EFM32 for my project, which would run for 24h straight after storing the energy in supercaps for some time: a shower or two and doing the dishes 
And now the funny part. The steady state and steady temperatures occur after only a minute of two. But when you turn off the tap, the heat in the pipe (water) still allows to get going and to store some of this heat energy in the supercaps, thanks to LTC's chip. Water had been running for some time while the STK temperature demo had been connected to the harvester board, then I turned off the hot water tap and the voltage started to drop. But it took 17 minutes, to get only 26 mV from the TEG and after this time the LCD started to faint.
But guys, the truth is that after you turn off the tap, you still have to pay for the hot water that is between the meter and the battery. Because, next time when you turn it on, this water will be cold (and you pay for it as it was hot). This is the method to partially use this heat, which otherwise would be wasted.
As always stay tuned for my next posts. I will try to create my project demo based on STK and give you some details soon.
