It is necessary to capture the in-time and short-term future states of contributing appliances and their state determining underlying parameters, in order to calculate a reliable, local flexibility potential to be communicated. By using this abstracted aggregated value, it is possible to communicate the flexibility potential of the associated appliances, i.e., participating residential and commercial end-users. By applying the aforementioned abstraction and aggregation layer in front of the management applications, a wide spectrum of DR programs can be realized.
Figure 3 shows the generalized approach using end-user inputs for setting the flexibility calculation constraints. The flexibility approach applied to this work focuses on maximum available power over time provided by a general set of (1) deferrable, (2) interruptible, and (3) flexible loads shown in
Figure 3, taking into account the power consumption patterns of the appliances, the current parameter states, and individual, specific end-user inputs for comfort purposes.
3.1. Implementation into Secure Architecture
With the integration of an advanced metering infrastructure (AMI) Germany has implemented a highly secure communication device, called Smart Meter Gateway (SMGw), which enables communication and signal exchange between different entities surrounding the AMI [
32]. A detailed overview of the infrastructure and related requirements, focusing on aspects of modeling and real-time simulation are presented in [
33]. The communication and data exchange is established between the wide area network (WAN), including different external market participants (EMP), the home area network (HAN) with an end-user’s smart and controllable appliances and a visualization interface, and the local metrological network (LMN), consisting of metering devices for electricity. The LMN is also capable of including metering devices for other sectors, such as water or gas. For the communication to smart and controllable devices, so-called controllable local systems (CLSs), the SMGw offers an additional communication channel for a limited and certified set of EMPs. This additional transparent data channel allows the direct communication and data exchange to end-user controllable devices. For energy and power management purposes over a wide variety of different loads or RES, the connection to only one local central management system is beneficial. The information allocated by the local central management system to EMPs is independent from device specific protocols within the HAN devices. Besides this harmonized information allocation, only one device, the management system, has to be configured by the SMGw-administrator as CLS within the SMGw. Other smart and controllable devices are directly connected to the central management device. The proposed implementation is shown in
Figure 4. Only CLS directly connects to the local central management system communicate state and parameter values. In contrast, non-CLS (nCLS) only contributes to the overall power consumption without the adjustment potential. By sending signals to the local central management system, the CLS operation is adjusted to met the pre-defined goals. The key parameter to be determined is the accessible amount of power that can be adjusted in the next hours. The need for a highly secure software and hardware architecture in the energy sector, discussed in [
12,
13], especially when concentrating on the demand side [
16], has only been considered in a few cases [
17]. With the developed methodology and the integration into the presented architecture, this gap can be closed.
3.2. Methodology Description
We consider the time of the day represented by
I intervals where
and
M participating end-users with
. The power consumption
of end-user
m in interval
i can be calculated according to Equation (
2). Whereas
is the non-adjustable power consumption,
can be monitored and controlled for power adjustments through data and signal exchanges between
smart devices and the PMS. Protocols like EEBUS or other protocols can be applied. However, the simple aggregated power consumption of smart devices
is not addressable for the entire value, taking into account the specific appliance’s state and the state determining underlying parameters and their future development.
As an example, we consider a water heater with an attached storage tank. In order to meet the user’s comfort level, the water temperature is between two pre-defined thresholds, the lower and upper storage tank temperature. The modus operandi is such that the water heater will switch to the ON state in order to bring the water temperature to the upper threshold. An external switch OFF signal will lead to a cut in the heating process for only a short time interval as the temperature must be held above the lower temperature threshold for not harming end-user comfort constraints. The developed methodology takes into account not only the current state of the heater, but also the state determining parameters, such as the water and inner room temperature. An additional example of a dish washer can express the problem. The possibility to interrupt the operation cycle of the appliance for a pre-defined couple of minutes introduces
, which sets the maximum duration the operation cycle of the dish washer can be interrupted for. Now, not only this value
has to be considered. The additional value
(see
Figure 3) for the end time of operation needs to be integrated, when calculating the time length of flexibility’s availability. The general value and impact of pausing major household appliances is described in [
34].
Taking into account the aforementioned examples, the controllable part of Equation (
2) can be changed to Equation (
3). The flexibility’s availability is expressed in a step time length of
. Equation (
3) shows the separation of the
D CLS into two bins for accessibility greater (defined as
) and smaller (defined as
)
. CLS being in bin
contributes to flexibility events, as the others belonging to
change their actual status before the possible event can take place.
Assuming that the separation into the two bins changes over time,
can be written as the flexibility matrix with the abstracted information about accessible power, duration, and possible start time of flexibility events for end-user
m, communicated at interval
i. the flexibility matrix
is shown in Equations (
4) and (
5). The row
corresponds to the possible start time of the flexibility event, on the time basis of
. Thus, the first possible allocation of the flexibility appears in
min, taking into account the lack of immediate communication and reaction, and possible trading intervals on energy trading platforms. Column
contains the possible duration of the flexibility event.
The exemplary value
, sent at 7 p.m. for
intervals (see Equation (
6)), shows a accessible flexibility event starting at (
min) 7:45 p.m. Taking into account Equation (
7) the duration
can be calculated.
where
By sending the flexibility matrix repeatedly in a constant time interval of
the accessible system flexibility can be updated for further consideration. The further reduction of rows is possible, when only information about the next possible flexibility event is needed. The information flow is shown in
Figure 5. Beginning with the CLS, the PMS calculates the flexibility potential to fill the flexibility matrices. The here applied dimensions of
are N = 4 and J = 6, which means a maximum forecast of 150 min with an event that starts in n = 4 (60 min) for then
min. As the next step, the flexibility matrices, consisting of information about power, duration, and possible start time of the event, are sent to the abstraction and aggregation layer. By aggregating all connected entities
m in the aggregator system the system flexibility can be transferred to the management layer operated in the aggregator backend. With information about accessible flexibility in the aggregator system, the exact flexibility offered in the markets, such as the intra-day spot market, can be realized. Reacting to power production uncertainty, the communicated flexibility can be used by DSO to ensure and restore a reliable system operation.
To this point, Equations (
2)–(
5) only consider load reduction potential from turn-off appliances as generation substitute. Expanding the described method to appliances with an automated switching functionality, e.g., thermal or electric storage devices shown in
Figure 3 an additional load capacity is added. Equation (
8) shows this additional load potential. Considering the expected time of device
d to stay in the current ON or OFF state, it can either belong to bin
for a turn off flexibility and temporal ON state or to
for turn on flexibility and temporal OFF state. Following the same suggestions as for Equation (
3) only the flexibility, which is still accessible in
and later is relevant for flexibility consideration. These devices belong to
. The formulation of Equation (
4) for a positive or negative flexibility potential of the end-user is introduced. As described above,
shows the load reduction potential, whereas
shows the load increase potential. The system flexibility potential is then
for generation and
for additional power demand. Whereas three device classes and their contributions to flexibility realization are shown in
Figure 3, only interruptible and flexible devices are relevant for further consideration. This is because the flexibility matrices show the maximum possible deviation from the current power consumption.