*3.2. Validation with Case Study's Measured Data*

Validation can be done by comparing the data generated by Wepro with the city's measured data such as smart-meter data, TUS data or data from the utilities. In this case, we cannot make an in-depth validation as the measured city's data is unavailable. Therefore, a future study would follow to improve our current work when the city's measured data is available.

In practice, we can still compare our model with the standard load profiles for Dutch households published by the Energy Data Services Netherlands (EDSN) to validate whether our hourly average generated load profiles have the same trends the standard Dutch residential load profile. The average normalised standard household load in The Netherlands based on EDSN is provided in Figure 1 of [11]. It is shown that the morning peak starts to increase from 5 am, similar to both our generated models. It then reaches the peak around 10 am, while both our models identify the morning peak around 6 am to 7 am. The EDSN's load remains flat from 10 am to 13 pm, although there is a subtle peak at 12 am during lunchtime. This curve from 10 am to 13 pm is quite similar to the Wepro-ALPG model one. Furthermore, like the Wepro-LPG model, the ESDN's load is decreasing gradually to 4 pm. After that, the curve starts to increase significantly like the curves of both our generated models. The EDSN's model reaches the peak between 19:00 and 19:30 similar to the Wepro-LPG. Furthermore, the load is decreasing quite significantly until 2 am. It remains flat from 2 am to 5 am which is similar to our

generated models. In general, it can be concluded that the generated hourly average share of the Wepro models have similar curve trends as the EDSN's load trend, although the morning peaks in the generated Wepro models have different time characteristics from EDSN's morning peak. Furthermore, both our models and the EDSN's model show a subtle peak during lunchtime. The evening peak occurs after dinner in the Wepro-LPG and EDSN model, while the the evening peak is occurring exactly at the dinner time in Wepro-ALPG.

As an update, the load profile data in EDSN have been moved to de Vereniging Nederlandse Energie Data Uitwisseling's (NEDU) page [33]. The data provided in NEDU's page start from year 2016, therefore data 2016 are used in this initial validation. Smart-meter data is used as a basis for the consumption/production profiles as described in 'Profielenmethodiek elektriciteit', where the documentation is available in Dutch. The raw data are provided in 15 min resolution, which show how much electricity is allocated in that 15 min. The data are obtained from 3,002,450 households type E1A in 2016. The comparison of the standard average Dutch household load profile in 2016, which has similar trends with Figure 1 of [11] for E1A residential type and the hourly average load profile in a year of Wepro-LPG and Wepro-ALPG is plotted in Figure 14.

**Figure 14.** The comparison of hourly average standard residential load profile in The Netherlands based on NEDU's data of 2016, E1A and the hourly average load profile in a year of Wepro-LPG and Wepro-ALPG.

This simple validation is an initial check to see whether our generated load profiles resemble the standard Dutch's household load profile characteristics before going into an in-depth validation with the city's measure data. Moreover, further study in the future is required.
