There was quite some cheering in the media yesterday. A Belgian electricity provider, Lampiris, performed an “experiment” to see how a blackout could be avoided. Their intention was was to find out if an awareness campaign at a peak moment is working and how big the effect is.
This is how it was done: Lampiris asked its clients who was interested in participating in this trial and was willing to for example shut off the heating, shut off the television, postpone things like ironing/washing/drying and going to the bathroom in the dark. Thursday evening those volunteers got a text message that asked them to use as less electricity as possible between 18:00 and 19:30.
Lampiris claimed it was a real success and it showed that a blackout could be prevented just by asking people to use as less electricity as possible. It saved 100 to 200 MW of energy, half of the production of a small nuclear plant. This conclusion was praised uncritically repeated by the media.
The question I had was: how could they crunch the numbers so fast? Apparently the result was declared in the news of the same day, maybe even in the news of 19:00, before the monitored period was over. How could they have a result so quickly?!?!
I came across a video of the VRT news of 27/11/2014 in which the gain was explained and I made this screenshot:
Total load on November 27, 2014 of the Belgian network as seen on the VRT news
It hasn’t much details, but it made some things clear.
First, the success was indeed declared before the peak demand was over, the shot seemed the measured data from around 19:00.
Second, Lampiris ignored the elephant in the room: sure, there was a slight decrease of load compared to the forecast around peak, but from noon until peak demand there was an even larger decrease. My estimate is about three or four times the decrease that they mentioned as a clear effect that it worked…
Third, at the top left there is a part of the title and I think the complete title is “Total Load”. Which, if true, gives the claim a totally new dimension.
Then I found a screenshot of a wider view (probably from the VTM news). It is indeed the total load page from the Elia website.
Total load on November 27, 2014 of the Belgian network as seen on the VTM news
“Total load” is not exactly “Consumption”. According to the Elia website, “Total Load” means all production injected into the grid, including import of electricity abroad, but excluding import and electricity used for energy storage. This made it very clear why they had their result so quickly. They confused consumption with the power that was put on the grid at that moment.
The amount of electricity put on the grid depends heavily on for example the amount of green energy, changing demands from the industry, the weather and who knows what more. That measured decrease probably is within the normal variation, as being proved by the lots lower demand that afternoon. Another thing that could influence their result is the degree of accuracy of the Elia forecasts.
Actually that graph doesn’t tell us anything about the consumption of electricity at that time. It just showed us the electricity that was put on the grid. Yet the PR guy of Lampiris led us to believe that their action caused this. And the media? Well, they all fell for it big time.
Trust, yet verify 