**5. Results**

This section presents all the gained results from the aggregator standpoint. All the results provided in this section are for surveying the performance of the aggregator model during the ramp period and the DR event itself. Figure 10 shows the consumption reduction profiles after applying DR programs. The results shown in the same figure are with the one-minute time interval between 11:00 where the DR program manager notified the aggregator for the event, until the end of the event (14:00).

**Figure 10.** Results of applying demand response programs by the aggregator.

As it is clear in Figure 10, the aggregator firstly started to apply IDRP since it has the lowest remuneration rate from the aggregator point of view. To do this, the aggregator announced all the IDRP resources one by one and waited for their response as the program is voluntary. Then, all IDRP participants replied with their desired responses (OPT IN or OPT OUT) until 11:30 (program deadline). After that, the aggregator evaluates the use of the IDRP program and as it did not reach the reduction baseline, it decided to apply DLC T1 program. Therefore, the aggregator notifies DLC T1 participants 5 min in advance (11:35) to the starting point of the event (11:40). During this 5 min, the aggregator takes advantage of the available ESSs and start discharging them, so there would be a reduction in the consumption. While all resources in DLC T1 have participated in the event and the reduction rate of DLC T1 has been reached, the aggregator stops discharging the ESSs. The same procedure is also applied for DLC T2, and finally, at 11:50, when the aggregator reached the desired reduction baseline and it is ready to start the event at 12:00.

Moreover, as Figure 10 shows, the starting point of the paid period for each program is the moment that the first participant reduced its consumption, so the aggregator has to pay the contractual remuneration according to the reduced power. In other words, the aggregator receives the remuneration from the DR program manager only for event duration (i.e., in this case, two hours between 12:00 to 14:00). However, the aggregator must start paying the remuneration before the event during ramp period as the DR participants started the consumption reduction. That's why aggregator should pay remuneration to the DR participant with a lower rate than the one that it receives from the DR program manager, so it would be able to manage all the paid periods without a financial downturn.

In order to have a more precise and technical vision to the model, Figure 11 illustrates the experimental results adapted from the real-time simulation model and Resistive Loads Bench as HIL. The results shown Figure 11 are related to the 6 DR participants that own Fan Heater and they are involved in the IDRP (indicated in Table 2). In fact, each consumer load in the Resistive Loads Bench emulates a Fan Heater in each DR participant. The results demonstrated in Figure 11 are adapted from MATLAB™/Simulink and OP5600 in 3600 periods of 0.5 s, which is in total 30 min, between 11:00 to 11:30 while all IDRP resources are announced to participate. In other words, the time step of this model in real-time simulation is set at 0.5 s. This means OP5600 conveys the reference signal (power reference in Figure 11) to the resistive load bench with one-minute time interval, and then, it acquires real-time consumption data with 0.5 s time interval. The actual power measurement curve in Figure 11 shows the real behaviors and reactions of resistive consumer loads, and it is only shown until the IDRP deadline, as after this moment all their consumption was cut.

**Figure 11.** Experimental results adapted from OP5600 and Resistive Loads Bench.

Indeed, employing real-time simulation (OP5600) and laboratory equipment as HIL for emulating consumption profiles have several advantages. One of them is that we validate the actual demand reduction under the technical parameters of the grid, namely voltage variations (as shown in Figure 11). This leads to having a gap between the experimental and simulation results. This gap is clearly

visible in Figure 11 between the red dashed line as Power Reference and the blue line as actual power measurement. Consequently, it is interesting to calculate and compare the remuneration costs of aggregator using both experimental and simulation results. Figure 12 shows the accumulated remuneration costs during the ramp period and the event using simulation profiles.

**Figure 12.** Accumulated remuneration costs of aggregator during the ramp period and the event.

As Figure 12 shows, there are a few remuneration costs for the IDRP program as the remuneration rate and the available capacity were not significant. Also, the costs of DLC T1 has a linear ascending gradient since the available capacity of this program was constant during the event. Finally, as DLC T2 has a fixed remuneration rate per event, it has a constant ratio in the aggregator's remuneration expenses. Table 3 demonstrates the detailed cost calculation for each program. In Table 3, the main focus is given to the first 30 min of the IDRP program as the real-time simulation and HIL methodology have been implemented for this specific program. The actual and simulation profiles are respectively the blue (actual power measurement) and red dashed line (power reference) in Figure 11.


**Table 3.** Remuneration costs for each program paid by the aggregator to DR participants.

Total Cost = 2.2573 EUR (using actual profile); 2.2571 EUR (using simulation profile).

As Table 3 shows, the calculated remuneration cost between 11:00 to 11:29 in IDRP has a difference between the actual and simulation profiles. This cost difference is not significant because in this specific model it is only for six fan heater devices as a part of the IDRP program, which has a little reduction capacity for 30 min. Suppose that the aggregator has a huge number of DR participants, namely 1 million customers with a longer DR event. Therefore, this little difference becomes remarkable in this case as it would mean a huge amount of cost variation between what it is expected and what occurs in actual cases.
