The first step is to define energy losses. When I hear that term regarding electrical grids, I typically assume they are referring to the $P=I^2R$ losses of the lines. This is because this particular loss is something that can be changed (by planning higher voltage lines).
There are also reactive loads. If a system is out of balance, with too much capacitance or too much inductance, power is lost charging those. That is a loss factor which matters greatly for electrical companies. They tend to have banks of capacitors which can be switched in to balance the loads (modern power consumption tends to be inductive on average because electric motors are inductive).
If you are accounting for loses in the production system, you're going to have to carefully define what the energy losses are with respect to. For these things, we are typically talking about losses with respect to the theoretical energy load of the fuel. In this sense, a typical coal plant is around 37% efficient. It can also be compared to the efficiency of an ideal thermodynamic cycle. Known as the Carnot efficiency, this is a value that is dependent on temperature, but tends to be around the 60% to 70% range.
So define losses based on what is most useful, and then go from there. What is useful really depends on what you think you can optimize. Tailor your definition of loss to best capture the variables you wish to improve upon.
