2.1.2. Profile Generators: LPG and ALPG

To produce the load profiles of the selected households profiles as the result of the Wepro model between the city's main characteristics and the households occupants, we use LPG and ALPG as the load profiles generators. Thus, in this case we optimise a bottom-up approach provided by the generators, scale-up from the household level to the city level based on the down-scale task perform previously in the weighting model, and employ the profile generator's model at the former level.

The main reason of choosing LPG and ALPG because both of them are developed based on behavioral model, which is in line with ClairCity project's goal to model the citizen's behaviour. LPG's model has been selected for use in our model, as it offers a mature model with which to synthesise household energy load profiles based on various occupants' profiles. Pflugrandt has developed the model with a strong focus on modelling the behavioural aspect. The basic elements for modelling a single household in Figure 4 are the desire to do so and expressions of the need to do something. The model specifies weight, threshold and decay time as desired properties [15].

*Weight* is the relative weight of a need compared to all a person's other needs. In selecting for the next action, the minimisation of the deviation requirement is used as a criterion, the weighting acting as a multiplier in this calculation. *Threshold* determines when the person really feels a need, that is, when it is included in the next action selection of the calculation. For example, in reality there is usually no eating after lunch because only 10% of the hunger sensation is evident. Instead, one generally waits until a noticeable feeling of hunger has built up before having dinner. Finally, *Decay Time* describes the half-life, until 50% of the requirement is reached. It has been found that activities at 50% threshold mostly after the two to three times the decay time, depending on the weighting and the other available activities. The decay constant is calculated from the decay time by which the current value of the need is multiplied in each time step [15]. When creating households, it has been found that activities at the 50% threshold are usually executed after two to three times the decay time, depending on the weighting and the other available activities. Furthermore, besides desire, it is also essential to identify the individual's properties (age, gender, sick leave in the year, average duration of illness, needs when

healthy, needs when ill) and load type, which in this case is electricity [14,15]. LPG provides various pre-defined German household profiles.

**Figure 4.** LPG's minimum necessary elements in modelling a decision-making process of a single household [15], where the basic elements are the desire to do so and expressions of the need to do something.

The second profile generator used in our work is ALPG. ALPG employs household occupancy profiles generated by a simple behavioural model, which creates consistent profiles for the devices. The devices' flexibility is specified through four classes: timeshiftables, buffer-timeshiftables, buffers and curtailable. The inflexible electricity profiles are grouped into the following categories: stand-by load, consumer electronics, lighting, inductive devices, fridges, and other. Furthermore, to show annual electricity consumption, the individual profiles are scaled in magnitude, making it easier to alter the profile if there is a change in electricity usage by the external factors. An example of such a change could be the adoption of a new technology, for instance, light-emitting diode (LED) lights. Moreover, the following classes in Figure 5 are implemented in the simulation model: neighbourhood, household, person, device, house, writer and ALPG. Electricity usage in a typical Dutch setting is the focus of ALPG, which is also in line with our work in modelling residential electricity load profiles, with Amsterdam as the case-study city [16,17].

Furthermore, we after applying the capacity allocation into LPG and ALPG the following are closest profiles that reflect the city's proportion of the age groups, gender and labour force mentioned above.

• LPG

The following are the simplified Wepro-based selected pre-defined households profiles in LPG although there could be also several other options that may fulfill the Wepro model composition:

**Couple**, *both of whom work,* with **one child Couple,** *one at work*, *one at home*, with **one child Couple** *both of whom work* **Single** *with work* **Senior** *at home*

The **underlined** entities indicate the age groups, the *blue italic* entities represent the labour force. Moreover, to express the gender shares of each age group, we selected the characters of LPG pre-defined household profiles in Table 3, as follows:

**Table 3.** The selected pre-defined household profiles in LPG based on Wepro model.


Furthermore, we can insert these occupant's list to the Wepro composition in order to validate the model. As illustrated in Figure 6, the selected household profiles can fulfill the Wepro's model composition. Then, we generate these LPG's pre-defined households' load profiles one by one. The LPG can be downloaded free from https://www.loadprofilegenerator.de/. In generating one pre-defined household's load profile, after we download and open the windows program, we can go to "calculation" menu.

**Figure 5.** ALPG's class diagram [17] that shows the cardinality of a class in relation to another. The example of one-to-one (1..1) relationship is depicted between Household and House, where a household lives in a house and a house belongs to a household. The one-to-many(1..\*) relationship is shown between Household and Device, where a household has one or more devices, and each device belongs to a household. Each class from these multiple classes represents a part of the model, which makes the software flexible to be extended in the future work.

Furthermore, we should select some options such as which pre-defined profile to be modelled, geographic location and temperature profile based on temperature dataset that we input before, if the temperature dataset is not provided yet by LPG. That is why we need to pre-process our input data such as temperature dataset in order to be matched with LPG's format. Then we can calculate the household profile one by one which may require a computational processing time and the result is generated in comma-separated values (.CSV) file.

**Figure 6.** The application of the Wepro model's structure for amsterdam's household occupancy profiles in LPG. It consists of the amsterdam's age group share, labour force composition share and gender share of each age group, their capacity of the occupants to be modelled and the selected gender character provided in LPG.

• ALPG

As shown in Table 4, the pre-defined households profiles in ALPG are not as detailed as in LPG, but they simply can fulfill the Wepro model. The pre-defined households class contains seven types of households: Single worker, dual worker, family dual worker, family single parent, dual retired and single retired. Dual profile means a couple. In this case, each type of household corresponds to a category of electricity annual consumption in Kilowatt hour and amount of occupants or persons.


**Table 4.** Pre-defined households configurations in ALPG based [17].

To fulfill the Wepro model and simplify the process, we selected: one single worker, one single retired, two dual worker and one family dual worker. ALPG is open-source code and the code is available at it's github page. Figure 7 shows the snipped code of the households profiles selection, where the ALPG program runs by executing profilegenerator.py.:

Furthermore, the same procedure with LPG, in Figure 8 we inserted these occupant's list to the Wepro composition in order to validate the model.

Accordingly, it indicates a different result in comparison with LPG because in ALPG there is no need to identify the gender characteristics as it has simplified and consistent pre-defined profiles list as provided in Table 4. Moreover, these selected occupancy's list in ALPG may fulfill the Wepro model regardless the gender detail. In consequences, there are five generated households load profiles both in LPG and ALPG. We used the average load profile's value of these generated load profiles in the analysis.

*Energies* **2020**, *13*, 3543

**Figure 7.** The snipped code of the configuration of the selected household profiles in ALPG based on Wepro model.

**Figure 8.** The application of the Wepro model's structure for amsterdam's household occupancy profiles in ALPG. It consists of the amsterdam's age group share, labour force composition share and gender share of each age group, their capacity of the occupants to be modelled and the selected pre-defined profiles provided in LPG.
