Documentation for $SET OBLONG, and its effect on lump-sum investment cost re

[KristofferEML]

For your information, I had a look at this issue again while preparing a new release with some other small improvements.  And, I decided now to implement a small correction for the LUMPINV reporting when ILED is defined but with ILED<1, because your issue disclosed that the reporting for such small ILEDs was not consistent in either case (with or without MID_YEAR). Some elaboration in more detail is given below.

I remain with my position that under MID_YEAR the LUMPINV values for Cases I.2.a/b should in fact be reported differently than without MID_YEAR.  This is a direct consequence of the difference in the assumed timing of the payments with and without MID_YEAR, which results in different amounts of interest during construction.  These cases are both correctly accounted in the objective function, but the LUMPINV reporting for Cases I.2.a/b has thus far been kept according to the original design, regardless of MID_YEAR, because MID_YEAR basically does not affect these cases in the objective function, and to maintain backwards compatibility.

However, your issue disclosed that currently there is a clear inconsistency of the reported values when ILED<1, which is due to the generalization of the IDC calculation.  Originally, only integral ILED values (1,2,3,...) were supported by the design, which was a notable limitation, and it effectively prevented using any ILED<1 for Cases I.2.a/b.  Later, the ILED handling was generalized, based on a simple geometric series formula, to support any fractional values as well, and then also using ILED=0 was allowed for Cases I.2.a/b, when using OBLONG (or some alternative objective formulation). The resulting generalization was fully consistent with the original design for any integral values of ILED, but it results in a half-year’s IDC even with ILED=0, which caused the confusion with your issue.  While in the objective function that half-year’s IDC accounting is indeed also fully consistent with mid-year discounting, I strongly think it should not be reported as IDC, because there is no meaningful interpretation of a zero ILED leading to such an interest. Instead, the IDC should only be interpreted as the time-shift related to mid-year discounting, and therefore it should be eliminated from the LUMPINV values.  This is the correction I have now implemented, and it affects only the ILED<1 cases (which were not supported in the original design of TIMES).  So, for now the reporting still otherwise remains unchanged between MID_YEAR = YES/NO, which I understood being your preference.

I hope you can agree that this small correction is justified, but if you have any reservations, please let me know.

Hello Antti,

Thank you for looking into this again, and for implementing the correction. I agree with your reasoning: there's no meaningful interpretation of a zero ILED producing interest during construction, so eliminating that spurious half-year IDC from LUMPINV for ILED<1 makes sense, while leaving MID_YEAR=YES/NO reporting otherwise unchanged. Good to know this turned out to be a genuine edge case in the generalized IDC formula rather than just a gap in my understanding :-)

Thanks again for all the time you put into this,
Kristoffer