change calculation into financial/fiscal year

I've been doing my calculations/written my code based on calendar year, but realised now that I should have done it based on financial year i.e. 1st April to 31st March. Is there an simpler way to do this without having to change a lot of the lines. To simplify, say I have 2 matrices:

Power = [year month day hour data1]

Speed = [year month day hour data2]

I've changed the year month day into timestamp using datenum and match the timestamp of Power and Speed to give a relative magnitude of wind speed to power:

Power curve = [year month data2 data1]

In a lot of my subsequent calculations, I've done a lot of loops based on calendar year using 'unique year' and therefore it would be a pain to change a lot of that. Just to give you an idea, I have hourly Power data for 17 power stations, each with varying year of operations, and hourly speed data from 1985-2018.

So my question is, is there a way to manipulate my initial data (Power and Speed) so that the 'unique year' corresponds to 1st April to 31st March, rather than 1st Jan to 31st Dec?

Thanks!

[dpb]

Have to run at the moment, but it strikes me the easiest thing to do would be to generate a new year vector which is the analysis year based on the calendar date.

Have to think about the most efficient way to do so, but that hopefully will give you an idea to pursue.

While you're at it, I'd suggest at least considering converting from datenum to datetime; many more functions are aware of it now and more flexibility is available.

[dpb]

"... I have hourly Power data for 17 power stations, each with varying year of operations, and hourly speed data from 1985-2018."

I spent nearly 30 yr working for/with consulting for the electric utilities first from the vendor side (nuclear) then moving over to the consulting role prior to returning to family farm almost 20 yr ago now.

Shortly after that move, a new wind farm came on line not far from here; in what should be ideal locations; it's within 30 mi of Dodge City, the highest mean wind speed location DOE identified in its survey of the US back in the early '90s.

I have since collected generation data from it and compared output to determine average capacity factor--over the nine years I've done, annual is only about 35% with a maximum monthly average of just barely over 50% and some months (mid-summer and winter doldrums) are in the 20% range.

I also compared output to average wind speed to try to determine if output was being controlled by dispatch rather than generation limit; over that time period the correlation was >80% being highly indicative of the output being input limited rather than not being online or demand-limited.

It seems difficult to justify the investment for such low capacity factors when it requires keeping standby generation available 24/7 for replacement when unavailable...

[NMans]

Seems like you've done a very interesting analysis. I'm doing this analysis for offshore wind farms, therefore the capacity factor is higher. Also, wind turbine technology is progressing rapidly - they're building bigger wind turbines these days.

But you are right that the output is input limited. I'm also interested in how much storage is needed to ensure security of supply and what kind of storage technology is suitable.

[dpb]

Just a bigger eyesore is all... :) Grey County covers some 12,000 A to produce 112 MWe nameplate MW; Wolf Creek facility itself occupies less than 640 A (altho they own about the same total) to produce 11,000 MWe and has multiple runs of 100% capacity factor for an entire year being on roughly 18-month fuel cycle.

It will take the yet-to-be-developed storage technology to be able to ever reduce baseload or standby capacity (which is sucking up NG at prodigous rate which will become far more valuable for other purposes than generation) but they'd still have to cover the whole planet surface because the input energy density is just so low.

I shouldn't have got of on the sidebar, but it's a sorepoint when I have flashing red lights on the horizon visible out of the bedroom windows instead of unbroken horizon and stars...used to be able to see Milky Way clearly; now it's almost totally obliterated.

[NMans]

Ah fair enough! Sorry to hear that..

[dpb]

We're fortunate that they are still (at least so far) on the far horizon but they're some of the newer monsters and it's almost 60 mi away and they're still on the horizon from the second floor window.

I can't imagine the problem when they're right in the neighborhood, especially if it's your neighbor getting the royalty check and all you have is the noise and visual pollution.

It's one of these things that is much better in theory and for some than others in practice.

Answer 1

[dpb]

Actually, it's pretty simple to accomplish. Given the year and month vectors, create a "financial year" by

yrFin=yr;                % initialize to existing year
ix=(mo<4);               % logical index to Jan-Mar
yrFin(ix)=yr(ix)-1;      % belongs to previous financial year

I would still recommend looking seriously at datetime instead of datenum and in particular then investigate the use splitarray, varfun and findgroups to simplify the computations by groups where here for starters the grouping variable would by yrFin

Also NB: that the names year, month, etc., are Matlab builtin functions used with datetime class so I've not used those as variables.

[NMans]

Thanks dpb - this works very well but for incomplete data (less than a year), it means that the data is merged into the year before. I just have to delete incomplete data so it doesn't affect the year before.

[dpb]

Having hard time envisioning what wouldn't still be the correct analysis year; by your definition anything in the first Q is the previous year.

Show an example that you think is a problem...

ADDENDUM

Or, of course, you could alternatively define the analysis year to be the same as the actual year of the Q1 dates in which case the remainder is in the next analysis year and then one would write that

yrFin=yr;                % initialize to existing year
ix=(mo>3);               % logical index to Jan-Mar
yrFin(ix)=yr(ix)+1;

It's totally arbitrary as to which you choose.