Ok, thanks for the files.

It was easy to see where the problem was. The documentation says the following (bold emphasis added):

2.2.1. Endogenous partial load efficiency losses

It should be noted, though, that the endogenous modelling of partial load efficiencies requires that the process has its efficiency modelled through the ACT_EFF parameter (on the shadow side of the process).

3.2.7. Partial load efficiency losses

It should be noted that the modelling of endogenous partial load efficiency losses requires that the

process has its efficiency modelled by the ACT_EFF r,v,t,p,s parameter (on the shadow side), while the

penalty cost approach can be used for any process, regardless of how the process efficiency has been

modelled.

5.4. USAGE NOTES FOR BASIC PARTIAL LOAD EFFICIENCY MODELLING

The ACT_LOSPL(r,y,p,'FX') parameter can only be used for processes that have their efficiency

modelled by the ACT_EFF parameter (on the shadow side). The process efficiency will then be

endogenously modelled according to the actual load level in each timeslice.

You have not defined ACT_EFF on the shadow side. The shadow side is at the opposite of the PCG, and your PCG for the testelsr_new technology is HYGNtcs. You are defining the efficiency on the PCG:

Thus, it is was immediately clear that partial load efficiencies could not be working at all for the technology testelsr_new. I wonder why on earth you are now defining the efficiency on the PCG, while you earlier defined it in the normal way (using EFF in VEDA)? It is very unusual to define ACT_EFF on the PCG, and is meant only for the special case where you have multiple commodities in the PCG, and you want to have different efficiencies for each of the outputs.

I tested by fixing the ACT_EFF, by raising the starting point for the losses higher, ACT_LOSPL('UP')=0.8, and the process then worked as expected, having large partial efficiency losses in the night timeslices.

It was easy to see where the problem was. The documentation says the following (bold emphasis added):

2.2.1. Endogenous partial load efficiency losses

It should be noted, though, that the endogenous modelling of partial load efficiencies requires that the process has its efficiency modelled through the ACT_EFF parameter (on the shadow side of the process).

3.2.7. Partial load efficiency losses

It should be noted that the modelling of endogenous partial load efficiency losses requires that the

process has its efficiency modelled by the ACT_EFF r,v,t,p,s parameter (on the shadow side), while the

penalty cost approach can be used for any process, regardless of how the process efficiency has been

modelled.

5.4. USAGE NOTES FOR BASIC PARTIAL LOAD EFFICIENCY MODELLING

The ACT_LOSPL(r,y,p,'FX') parameter can only be used for processes that have their efficiency

modelled by the ACT_EFF parameter (on the shadow side). The process efficiency will then be

endogenously modelled according to the actual load level in each timeslice.

You have not defined ACT_EFF on the shadow side. The shadow side is at the opposite of the PCG, and your PCG for the testelsr_new technology is HYGNtcs. You are defining the efficiency on the PCG:

Code:

`PARAMETER ACT_EFF /`

REG1.2030.TESTELSR_new.HYGNtcs.ANNUAL 0.017627

REG1.2015.TESTELSR_new.HYGNtcs.ANNUAL 0.017288462

REG1.2050.TESTELSR_new.HYGNtcs.ANNUAL 0.01798

/;

I tested by fixing the ACT_EFF, by raising the starting point for the losses higher, ACT_LOSPL('UP')=0.8, and the process then worked as expected, having large partial efficiency losses in the night timeslices.