The assumption is that one has to operate in hybrid mode (EV only battery capacity is depleted) and that one would not use charging mode simply to fill the plug-in HVB, In hybrid operation, all energy comes from the fuel Thus, one wants to minimize the fuel cost between wall charging. It’s not about comparing cost of wall power vs fuel cost. ICE has to be used to burn the fuel to operate the vehicle in hybrid mode. There’s no other way even though converting the energy in the fuel with ICE is only maybe 35-40 % efficient.
There are many times when my hybrid battery is virtually charged full by ICE (like when on level roads or going down a longer slight downgrade) and one runs in negative split mode. Negative split mode is not a preferred mode as one is not reaping the benefit of hybrid operation. ICE rpm drops by spinning generator to supply rpm to keep car at same speed but with very little load on ICE, ICE is still likely out of its most efficient operating range. Yes, one can back off pedal and drive in EV mode depleting charge and start the process again. But, is that the best use of EV? If one had more battery storage, ICE could pick up some charging load instead of operating in negative split mode and perhaps operate more efficiently. The added EV would then be used where one would get the most benefit where ICE otherwise might have had to be used very inefficiently.
There have been several research papers written on optimization of a plug-in by operating ICE to charge the HVB, propel car, and provide traction motor power to assist propelling car without using HVB - perhaps using ICE for all at the same time. The issue is that current algorithms currently have no look ahead, predictive function. So, assumptions have to be made in attempting such optimization. By adding a charging mode available to the driver, the driver can attempt to use his predictive skills in the optimization process.