One big problem with discussing TEC's and efficiency is deciding what "efficiency" refers to.
One commonly used measure of "efficiency" is called the Coefficient Of Performance. (COP) It is the heat pumped (Q) divided by the power supplied to the TEC. (I * V)
Here is an example of the Kryotherm simulator showing a system with a COP of 5.56:
In this case the 172 Watt TEC is powered with 3V, and is cooling a 25 Watt heatload. (Note that Tob, the temperature of the object being cooled, is only 0.2 °C below ambient.)
]JR[,
It appears to me, that your calculator is off by a factor of 10 in this case.
Here's an example of a more practical setup:
In this case, the TEC is powered with 18V, (73% of Vmax) and the TEC is cooling a heatload of 70 Watts. All the other input parameters are the same as before. The calculated results show that the heatsource is cooled to 13 °C and the TEC draws 8.87 Amps for a power consumption of 159.7 Watts and a COP of 0.482.
The following simulation is identical to the last except in this case the heatsource is only 35 Watts.
Notice that the current draw (I) is less than in the previous image, even though the voltage is the same in both cases. In this case the object is cooled to 0.2 °C, and the COP is 0.358.
Finally a simulation with the voltage to the TEC at Vmax:
Notice that the inputs are identical to those of the second image, except for the applied voltage. Also notice, that even though the voltage is higher, (24.6V) the temperature of the heatsource is actually higher than when the TEC was powered with 18V. (14.1 °C vs 13.0 °C)
I don't know if this will make anything clearer to anyone, but hopefully it gave those interested in thinking about this stuff something to thing about.