Empty peak time range error

[JW-Kraft]

I just discovered this issue testing a model with a single appliance for one day.

In the function calc_peak_time_range the variable rand_peak_enlarge is calculated as follows:
rand_peak_enlarge = round(math.fabs(peak_time - random.gauss(mu=peak_time, sigma=peak_enlarge * peak_time)))
Therefore, it can end up being 0 in case peak_time and the result of random.gauss(mu=peak_time, sigma=peak_enlarge * peak_time) are close enough that their difference rounds to 0.

The function then returns an empty array as peak time range (np.arange(peak_time - rand_peak_enlarge, peak_time + rand_peak_enlarge)) which produces an error later on.

For longer timeframes this is just very unlikely to happen.

A simple fix would be to limit rand_peak_enlarge to be at least 1:

rand_peak_enlarge = max(round(math.fabs(peak_time - random.gauss(mu=peak_time, sigma=peak_enlarge * peak_time))), 1)

[FLomb]

Hi, thanks for spotting this issue. I think that the proposed solution makes sense

[Bachibouzouk]

I just need to check that the proposed solution is now slowlier than doing a if rand_peak_enlarge < 1. As this is run only once per simulation I doubt it will have a huge impact on the runtime, but as I already have setup ways to look at execution speed I might as well make use of it :)

[Bachibouzouk]

Closed in #80