The EPA doesn't assume anything for charging losses. Tesla tests the car according to an SAE spec and provides that data to the EPA. The test involves filling the battery, driving on a dyno until it can no longer maintain the directed profile, then measuring the amount of energy required to refill the battery. In that data submission to the EPA for the 3 LR, Tesla also happened to include a comment that the usable capacity was 78,270 Wh.
So the empirical data for rated consumption (~235-237 Wh/mi based on the 2 data points mentioned above) disagrees with the rated range (310-313mi) and the expected usable capacity (78,270). One of three possibilities is correct:
- The empirical data is in error or otherwise biased and the rated consumption is actually 250 Wh/mi [78270/313]. See @Troy above. More empirical data would prove or disprove this.
- The usable capacity is less than 78,270Wh, and more like ~73000Wh [310*236], and the EPA data submission was incorrect or our interpretation of the comment is wrong.
- The actual usable range is greater than the indicated 310-313 rated miles. This would only be possible if there was a significant buffer below 0 rated miles that was accessible to the driver. Here I estimated that buffer would need to be about 7%. Driving the car until it's depleted and noting the energy used after the rated range was zero would help prove or disprove this.