A simple model: the amazing November dip

This post is a follow-up to the previous post, where I looked into the scenario of an unlimited storage device topping off the excess electricity at peaks and filling in the gaps when there is a shortage. I found that a storage capacity of about 2,500 GWh was needed to fill in all the gaps. That number seemed very high to me, so I wanted to check whether other people also found such large numbers.

I quickly found a back-of-the-envelope-calculation by the late David MacKay. He proposes that 33 GW of wind power, delivering on average 10 GW, needs roughly 1,200 GWh backup. This is his calculation:

10 GW × (5 × 24 h) = 1,200 GWh.

He starts from the assumption that it is necessary to bridge five consecutive days of no wind. The difference is that he only considers wind, while I also include another intermittent energy source (solar).

The average delivered power is rather similar in both cases. In my scenario I have ((3,369.05 MW x 0.12) + (3,157.185 MW x 0.24)) x 8.57 = 9,957.95 MW.

Which is a tad below the 10 GW of MacKay is working with. Yet, my result is almost twice as high. Is this the influence of another intermittent power source in the mix? Or just a coincidence? Or did I do something wrong?

I decided to focus on MacKay’s assumption. If that 1,200 GWh is the result of five days without any production, then there needs to be at least a ten day period with hardly any intermittent power production. It could be substantially more, there wasn’t one timeslot that had zero production for both power sources, so the deficit will be smeared out over a longer period.

Looking at the storage state graph, I noticed a large drop from half October until the end of November. During that period, a lot of power was drawn from storage. I displayed that period in an orange frame:

simple energy model (charts007b) - belgium - solar and wind production x8.57 - reference year: 2018 - storage state: november dip

What on earth happened there? It is quite an remarkable drop. Storage went from roughly 2,200 GWh to almost 140 GWh in just 1.5 months. It was not something I expected. I expected that a lot of power would be build up during summer, so when the beginning of the year is quite windy with lots of wind production, then I didn’t expect any problems with an almost empty storage device in the second half of the year. I expected a gradual decrease from September until December.

Looking closer, there are two distinctive drops. The first drop is from the evening of October 15 until October 28. It has periods of shortages (drawing from storage) alternating with periods of some excess power (recharging the storage), but the general trend is down. After a period of leveling off, a second (steep) drop. There are six consecutive days of shortages from November 21 until November 27 and therefor only drawing from storage. After those six days, virtually all the power that was build up during spring and summer was consumed.

A lack of production could not be the (only) cause. There was less production during those two months than the average of all the months, but not thát much less. It is surely a factor, but not the only one.

Other factors are for example the shortening of the days in autumn/beginning of winter (therefor higher demand) and the weather. October 16 was the start of a colder period with rapidly dropping temperatures until the end of the month. November 21 saw minimum temperatures dropping below freezing for the first time. Such moments lead to a higher electricity demand in Belgium.

It was most likely the combination of those factors that led to that drop. There was not such a similar drop in for example 2016, storage gradually decreased from September until December (as I was expecting).

The difference with the MacKay’s calculation is that this scenario doesn’t have five full days with a lack of production. It went slowly over 1.5 months in two periods having a 1,000 GWh drop each. But the storage was not able to recharge after the first drop before the second drop already came along. Therefor the need for almost double the capacity than what was expected from the back-of-the-envelope calculation.

That is also an effect of intermittency. There is no guarantee that storage will be recharged to an appropriate level before the next dip comes along.

4 thoughts on “A simple model: the amazing November dip

  1. rogercaiazza

    The lesson here is that you have to use real data to do the analysis. The next question is how many years do you have to look at before you can determine what the appropriate worst case will be?
    Related question about the load. The shorter days must include increased lighting load. What about holiday lighting? Here in New York there is a bump up in load for residential Christmas lights. I have never seen an estimate of the amount.

    Like

    Reply
    1. trustyetverify Post author

      I probably will do a multi-year scenario somewhere in the future. Also using more realistic assumptions (95% efficiency and drawing 100% from storage are very optimistic). 2018 also seems an exceptional year (at first glance, previous years give higher storage requirements).

      I don’t see a bump in the Belgian data at the end of the year. Demand in the holiday period at the end of the year is lower than in previous and following weeks . If there is more demand due the lighting then I can’t see it. It could be rather hard to estimate the effect (there are many variables, holiday lighting being only one of them).

      Maybe I will see a bump when I look into the New York data next time. Will report back to you then.

      Like

      Reply

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s