Prototype Field Tests. Test Results D6.2 Programme FP7 – Cooperation / Energy Grant agreement number 207643 Project acronym ADDRESS Type (distribution level) Public Date of delivery May 31 st , 2013 Report number D6.2 Status and Version V 1.0 Number of pages 123 WP/Task related WP6 – T6.3 WP/Task responsible KEMA / EDF-SA, Enel Distr., Iberdrola Author(s) Mathieu Caujolle, Luc Glorieux, Philippe Eyrolles, Julien Le Baut, Radouane Irhly, François-Xavier Toledo, Regine Belhomme, Francesco Naso, Olena Morozova, Giovanni Valtorta, Dominic Ectors, Pieter Kropman, Jitske Burger, Jan Maarten van der Valk, Ignacio Delgado, Roberto González Partner(s) Contributing EDF-SA, Enel Distribuzione, VITO, KEMA, Iberdrola Distribución. Document ID ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results_v1.0.pdf proof
126
Embed
Prototype Field Tests. Test Results...Prototype Field Tests. Test Results D6.2 Programme FP7 – Cooperation / Energy Grant agreement number 207643 Project acronym ADDRESS Type (distribution
This document is posted to help you gain knowledge. Please leave a comment to let me know what you think about it! Share it to your friends and learn new things together.
1.1. Scope of the document ................................................................................................ 10 1.2. Structure of the document ........................................................................................... 10 1.3. Notations, abbreviations and acronyms..................................................................... 10 1.4. Acknowledgements ...................................................................................................... 11
2. Prototype Field Tests. ................................................................................ 13
3.1. Description of the test .................................................................................................. 14 3.1.1. Players involved and architecture of the system ....................................................... 14 3.1.2. Equipment installed .................................................................................................... 15 3.1.3. Exchanged messages between actors ...................................................................... 16
3.2. Results ........................................................................................................................... 17 3.2.1. Assessment of the ADDRESS technology ................................................................ 17 3.2.2. Problems encountered ............................................................................................... 38
4.1. Description of the test .................................................................................................. 43 4.1.1. Location ...................................................................................................................... 43 4.1.2. Players involved ......................................................................................................... 44 4.1.3. Architecture of the system ......................................................................................... 45 4.1.4. Equipment installed .................................................................................................... 46 4.1.5. Exchanged messages between players .................................................................... 49
4.2. Results ........................................................................................................................... 49 4.2.1. Assessment of the ADDRESS technology ................................................................ 49
5. France ......................................................................................................... 63
5.1. Description of the test .................................................................................................. 63 5.1.1. Location ...................................................................................................................... 64 5.1.2. Players involved ......................................................................................................... 65 5.1.3. Architecture of the System ......................................................................................... 65 5.1.4. Equipment installed .................................................................................................... 66 5.1.5. Test conditions ........................................................................................................... 74
5.2. Results ........................................................................................................................... 87 5.2.1. Test execution ............................................................................................................ 87 5.2.2. Assessment of the technical performance on the consumer side ............................. 88 5.2.3. Provision of AD services at a cluster level ................................................................. 94 5.2.4. Market simulator....................................................................................................... 106 5.2.5. Complete scenario execution ................................................................................... 116
Figure 10), the peak and off-peak hour identified are:
- Peak-hour: 15 and 22
- Off-peak hour: 6, 13 and 19.
The Aggregator has the prototypes of each cluster; to find a solution to accomplish a request, the
Aggregator needs to know the type of day when the signal is sent; in addition, in the tests two levels of
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 24 of 126
power1 have been considered: 20 kW and 10 kW. The following table present these use cases:
Power 20 kW
signal hour
working day holiday
Peak hour & Decrease power
10 15
22 22
Peak hour & Increase power
10 15
22 22
Off-peak hour & Increase power
6 6
19 13
Off-peak hour & Decrease power
6 6
19 13
Table 4. Spanish field tests - Test Case 1, power signal 20 kW.
Power 10 kW
signal hour
working day holiday
Peak hour & Decrease power
10 10
22 22
Peak hour & Increase power
10 10
22 22
Off-peak hour & Increase power
6 6
19 19
Off-peak hour & Decrease power
6 6
19 19
Table 5. Spanish field test - Test Case 1, power signal 10 kW.
The response from the ATB for a power demand reduction of 20 kW, start time 22h, and duration 1h is
reported in Figure 11.
Cluster 1
Duration signal: 1 h
1 According to the total number of consumers, and in order to be able to achieve the AD requirements, the ATB
has defined two power thresholds in order to increase or reduce the demand from consumers.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 25 of 126
Start time: 22:00
Incentive: If power
consumption in
signal time is less
than 2 kW the user
benefit is 0.01
€/kWh.
Cluster 2
Duration signal: 1 h
Start time: 22:00
Incentive: If power
consumption in
signal time is less
than 2 kW the user
benefit is 0.01
€/kWh.
Cluster 32
Duration signal: 1 h
2 In order to achieve the AD requirement, the ATB sends two signals to different clusters to reduce the demand
and another signal to increase the demand in other moment of the day to try to shift some loads for other period of time.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 26 of 126
Start time: 23:30
Incentive: If power
consumption in
signal time is less
than 2 kW the user
benefit is 0.01
€/kWh.
Figure 11. Spanish field test - Results of test Case 1.
Test Case 2: Functionality of the EBox
The aim of this test case is the analysis of the behavior of the EBox when Aggregator’s signal is
received, how it is interpreted by the EBox and which are the signals that the EBox sends to the smart
plugs and smart devices.
The parameters considered in the test are:
- Power demand: increase / reduction
- Number of slots of the price&volume signal:
• Two slots (one power threshold and incentive)
• Three slots (two power thresholds and incentives)
- Hour of the day: peak / off-peak
- Duration of the signal:
• Two slots: 30m / 1h / 1h30m
• Three slots: 45m
Slots refer to the power levels. If the signal has two slots, there is a unique level power and two
incentives (xx € and 0 €). Depending on the nature of the signal (reducing or increasing the demand),
the power consumption should be above or below the limit. For a signal with 3 slots, there are two
levels power and three incentives (xx €, yy € and 0 €), receiving a different incentive depending on the
level consumption: above, below or between limits.
For the significance of the test, a reasonable number of EBoxes should be studied. The number
depends on the typology of the house and the equipments installed. Additionally, as explained in the
following sections, there is a limitation regarding the communications between EBoxes and
Aggregator. This has constrained the selection of EBoxes: for carrying out the test, the best EBoxes
from communications’ point of view have been selected. For selecting the Eboxes in order to be able
to analyze its behavior according to the information received:
1. different signals were sent in a specific day
2. for this day, information from the Eboxes is received at the end of the day.
After several days, we are able to receive information from all the Eboxes.
Studied cases are the result of the combination of input parameters and signal; they are summarized
in Table 6.
Peak hour Off-peak hour
2 slots 3 slots 2 slots 3 slots
Increase power Duration 30 min
45 min 30 min
45 min 1 h 1 h
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 27 of 126
1 h 30 1 h 30
Decrease power Duration
30 min
45 min
30 min
45 min 1 h 1 h
1 h 30 1 h 30
Table 6. Spanish Field Test. Test case 2.
Figure 12, Figure 13, and Figure 14 depict the response from the smart plugs downstream of EBoxes
which received an AD signal from the Aggregator. In the Figures, the measurements from all the smart
plugs along the day are shown.
Figure 12. Power reduction request at 19:00, working day – Demand from smart plugs.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 28 of 126
Figure 13. Power reduction request at 19:00, holiday – Demand from smart plugs.
Figure 14. Power increase request at 16:00, holiday – Demand from smart plugs.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 29 of 126
The last test carried out in this case is aimed at checking the internal algorithms of the EBox to verify
that the Ebox does not intervene when an Aggregator’s signal is received and the global override is
selected by the user. To check for this, a specific Ebox in override mode was selected and its behavior
was analyzed in comparison with other Eboxes not in override mode that had received the AD signal.
The following image depicts the behaviour from two Eboxes, one of them in override mode and the
other one not in override mode. The figure depicts the daily energy consumption from both Eboxes
(normalized consumption load curves) which have received an AD signal to reduce the demand at
12:00h during one hour. The consumer’s Ebox not in override mode tries to reduce the demand during
the first quarter of hour whereas the consumer’s Ebox in override mode does not work and
consumer’s loads are switched on during this hour.
Figure 15 Ebox comparison - OVERRIDE mode.
Test Case 3: Functionality of the smart devices
The objective of this test case is the analysis of the communications between EBox and smart devices
according to the output s from the EBox algorithm.
To verify the functionality of smart plugs and smart appliances, it is necessary to select an adequate
number of devices and to analyze their functionally in different days. The total number of smart plugs
involved in the tests is 255 divided into 125 to power shiftable loads and 130 to power interruptible
loads. For shiftable devices, the specification is:
- the next start time is scheduled to occur during the user preferred time and according to the AD
signal.
For interruptible devices, the specifications are:
- max duration of the interruption: 2 slots (30 minutes)
- minimum distance between two off-periods: 6 slots (1,5 hours)
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 30 of 126
- max number of off-periods: 8
The following paragraphs illustrate the main results of the tests and all the comments for all these
results will be presented in Deliverable6.3 [12]
Shiftable devices
Out of all the smart plugs involved in the tests, 51% have had their next start time in the preferred time
interval (Figure 16). The rest of the smart plugs did not fulfil the user’s preferences, for the following
reasons:
- Override option is selected in the smart plug the smart plug works correctly (18%)
- No override option selected in the smart plug the smart plugs does not work correctly (31%)
Figure 16. Spanish field test - Results for plugs for shiftable devices.
Interruptible devices
The 86% of the plugs for interruptible devices did fulfil specification during all the days, the 7%
between 80% and 100%, the 1% between 50% and 80% and 6% did do well in less than 50% of the
days (Figure 17).
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 31 of 126
Figure 17. Spanish Field Test – Results for interruptible devices.
For each single specification, the analysis shows that :
- max duration of the interruption (2 slots = 30 minutes): fulfilled by 99.81 % of the smart plugs
(Figure 18);
Figure 18. Spanish field test. Maximum duration.
- min distance between interruptions (6 slots = 1,5 hours): 94.6% of interruptible smart plugs
(Figure 19);
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 32 of 126
Figure 19. Spanish field test - Minimum distance between off-periods.
- max number of off-periods (8): all the smart plugs have presented less or 8 off-periods during
the day (Figure 20);
Figure 20. Spanish filed test - Number of off-periods.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 33 of 126
Smart washing machine
In the Spanish field test there are 25 Smart Washing Machines (SWMs) installed. The behaviour of the
SWM is similar to the one of shiftable devices: its start should be scheduled during the preferred
operation interval.
To evaluate the functionality of the SWM, it was verified that the power consumption occurred within
the operation interval set by the user. The result of the analysis is shown in the Figure 21.
Figure 21. Spanish filed test – Management of smart washing machines
Test Case 4: Global analysis
For these tests, the consumers’ response is analyzed from the Aggregator’s point of view. The focus is
on the behaviour of the consumers regarding the signals received, the day of the week and the hour
and the duration of these signals, for reducing consumption in peak hours and increasing it in off-peak
hour. To this extent:
1. The Aggregator defines the requirements to increase/reduce the demand during a specific day
and hour.
2. The Aggregator runs its internal algorithm and sends AD signals (price&volume signals) to the
most suitable clusters.
3. At the end of the day, the measurements from the selected consumers (all the consumers in
the selected clusters) are assessed in order to verify if the AD requirement from Aggregator’s
has been fulfilled.
Due to the limitation in communications, the number of EBoxes that received the AD signal was lower
than expected; then, it was necessary to extrapolate the data.
Based on the total consumption data received from the EBoxes, the average power consumption was
calculated for the selected cluster(s) and compared to the one of the rest of the ADDRESS
consumers. It is also interesting to represent the normalized load curves, obtained by dividing the load
curve by the maximum value of the load, to check the overall characteristics of the power
consumption.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 34 of 126
Figure 22 and Figure 23 depict the results of the tests for reducing consumption. The difference
between the consumption curves during the price&volume signal period of validity is apparent.
0
20000
40000
60000
80000
100000
120000
140000
160000
180000
0:0
00
:30
1:0
01
:30
2:0
02
:30
3:0
03
:30
4:0
04
:30
5:0
05
:30
6:0
06
:30
7:0
07
:30
8:0
08
:30
9:0
09
:30
10
:00
10
:30
11
:00
11
:30
12
:00
12
:30
13
:00
13
:30
14
:00
14
:30
15
:00
15
:30
16
:00
16
:30
17
:00
17
:30
18
:00
18
:30
19
:00
19
:30
20
:00
20
:30
21
:00
21
:30
22
:00
22
:30
23
:00
23
:30
0:0
0
kWh
Reduce -Load curve
Cluster 3
Without AD Signal
Figure 22. Spanish filed test – AD to reduce consumption by 20 kW at 13:00 to 14:00 the 26th of June (Summer working day) – Cluster 3 selected
0
0,1
0,2
0,3
0,4
0,5
0,6
0,7
0,8
0,9
1
0:0
0
0:3
0
1:0
0
1:3
0
2:0
0
2:3
0
3:0
0
3:3
0
4:0
0
4:3
0
5:0
0
5:3
0
6:0
0
6:3
0
7:0
0
7:3
0
8:0
0
8:3
0
9:0
0
9:3
0
10
:00
10
:30
11
:00
11
:30
12
:00
12
:30
13
:00
13
:30
14
:00
14
:30
15
:00
15
:30
16
:00
16
:30
17
:00
17
:30
18
:00
18
:30
19
:00
19
:30
20
:00
20
:30
21
:00
21
:30
22
:00
22
:30
23
:00
23
:30
0:0
0
Reduce Demand - Power consumption Normalized
Cluster 3
Without signal
Figure 23. Spanish filed test – AD to reduce consumption by 20 kW at 13:00 to 14:00 the 26th of June (Summer working day) – Cluster 3 selected – Normalized curves
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 35 of 126
Figure 24 and Figure 25 depict the results of the tests for increasing consumption. Also in this case,
the difference between the consumption curves during the price&volume signal period of validity is
apparent; in particular, the load curve for the selected cluster (cluster 1) shows increases during the
period of the signal while the load curve of the rest of the consumers decreases.
Figure 24. Spanish filed test – AD to increase consumption by 20 kW at 22:00 the 25th of April (Spring working day) – Cluster 1 selected
Figure 25. Spanish filed test – AD to increase consumption by 20 kW at 22:00 the 25th of April (Spring working day) – Cluster 1 selected – Normalized curves
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 36 of 126
Test Case 5: Consumers’ behavior according to incentives
In this test case, the analysis is focused on the difference of the consumers’ behavior when they
receive a similar price&volume signal but with different incentives. Most of the parameters in these
tests have been kept constant in order to give evidence only to the influence of the incentives; they are
day of the week, hour of the day to start the price&volume signal, power thresholds, duration of the
signal.
The analysis considers the aggregated response from the consumers after they receive the
Aggregator’s signal. These signals were sent manually from the ATB so as to be able to define the
incentives to decrease/increase consumption in a working day/holiday.
To develop this test, many signals were sent to the EBoxes, for each working day and holiday. After
sending these signals, the following weeks the same signals but with different incentives were sent,
with the idea of having two or more signals that only differ in the incentive. Due to communication
problems, the number of signals sent had to be bigger than required.
In the following figures, two examples are shown to ascertain the influence of the incentives. Two daily
load curves for the same day of the week during different weeks and with different incentives are
shown. Depending on the conditions of the day, the power demanded varies greatly from one day to
another; for this reason, these curves have been normalized for an easy comparison.
The first example (Figure 26) refers to a signal sent from the Aggregator to all the consumers (4
clusters) in order to reduce the demand at 19:00 for the same day of the week in different weeks and
compare the behaviour from consumers according to the incentives proposed. It is apparent that the
curve with the highest incentive has a more pronounced power reduction than the other. In the period
when the signal is activated, the energy consumption during the day with more incentive is lower than
the day with less incentive.
Figure 26. Spanish filed test – Daily power consumption during different days with the same power reduction but different incentives.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 37 of 126
In the example of Figure 27, which refers to the case of an increase of consumption, it is possible to
verify that the total energy demanded during the period of the signal has actually increased, and that
the increased consumption has a direct relation with the incentive of the signal.
Figure 27. Spanish filed test – Daily Power Consumption of different days with the same power increase but different incentives
Test Case 6: Consumers’ behavior according to duration
This test case has objectives similar to the test case 5; it is intended to analyse the response of the
consumers when they receive similar price&volume signals but the duration of the signal is different.
As in test case 5, the signals are sent manually from the ATB, this time in order to be able to define
different durations. Most of the parameters have been kept constant in order to give evidence only to
the influence of the incentives; they are day of the week, hour of the day to start the price&volume
signal, power thresholds, incentives. The analysis considers the aggregated response from the
consumers after they receive the Aggregator’s signal. These signals were sent manually from the ATB
so as to be able to define the duration of the signal intended to decrease/increase consumption in a
working day/holiday.
To develop this test, many signals were sent to the EBoxes, for each working day and holiday. After
sending these signals, the following weeks the same signals but with different durations were sent,
with the idea of having two or more signals that only differ in the duration. Due to communication
problems, the number of signals sent had to be bigger than required.
In Figure 28, an example is shown to ascertain the influence of the duration. Load curves are shown
for two different days; for each day, both the power demanded (darker line) and the trend line (lighter
line) are shown; the difference between the days is the duration of the signal. The trend line allows to
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 38 of 126
see the behavior of the curve at the time of the signal. As it can be observed, the longer the duration
of the signal the more difficult to follow the signal. The signal with a duration of 30’ allows Ebox to
follow the AD requirement in an easier way than a signal with a duration of one hour and a half.
This can be justified by the specifications of controllable devices. The specifications of the interruptible
devices don’t allow interruptions with a duration bigger than fifteen minutes; moreover, the minumum
distance between two off-periods must be bigger than 1h 30’. Therefore, it is quite difficul to be able to
reduce consumption during all the period of the signal by disconnecting all devices at same time.
Figure 28. Spanish filed test – Daily power consumption of different days with the same signal but with different duration.
3.2.2. Problems encountered
3.2.2.1 Communications
During the deployment of the test, communications has been one of the main problems. Issues with
the reception and recording of information have caused difficulties into the processes of selection and
analysis of the results. Next, some indicators will be discussed to show the functionality of
communications in the Spanish field test.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 39 of 126
Figure 29. Indicators used for testing communications functionality
The first indicator is the number of EBoxes that send data about what had happened in the house the
day before. The correct operation in this case would be that every EBox sends every day data
packages with information about consumption, interruptions, etc. recorded the day before. However,
information is received with some day delay, or not received at all, as it is shown in Figure 30, where
the number of EBox data received (y axis) per day (x axis) according to the day the information is
checked (see the legend).
Figure 30. Number of packages of data received from E-boxes every day
The second indicator is aimed at checking if communications between Aggregator and EBoxes work
correctly. Figure 31 shows the average value and the percentage of EBoxes that have received
Aggregator’s signal per month versus the total number of EBoxes in the database.
In this figure, the different columns depict the average values of AD signals received by the different
Eboxes based on the daily information registered from these ones and collected in the ATB. It should
be taken into account that the communication from EBoxes to ATB varied every day and additionally
not always the EBoxes were able to communicate correctly with the ATB in order to receive the AD
signal the day before.
E- boxes
Aggregator
ADDRESS_036 ADDRESS_001 ADDRESS_300 …
Smart Plugs Smart washing machine
AACC
Data Packages from E-boxes to Aggregator
Signals from Aggregator to E-boxes
Consumptions from Smart devices to E-boxes
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 40 of 126
0
20
40
60
80
100
120
January February March April
Communications from Aggregator to E-boxes
Av. Signal Received Av. Signal no received
0,00%
10,00%
20,00%
30,00%
40,00%
50,00%
60,00%
70,00%
80,00%
90,00%
100,00%
January February March April
Communications from Aggregator to E-boxes
% Signal Received % Signal no received
Figure 31. EBoxes that have received signal from Aggregator, per month.
Next, the communications between smart household devices and the EBoxes is verified. It is checked
which is the information about the household devices per each EBox in the database. Results about
this indicator are depicted in Figure 32.
0%
20%
40%
60%
80%
100%
January February March April
Smart Devices Communication
Received devices Non received devices
Figure 32. Smart household devices kept into the database vs. theoretical ones.
The last indicator about communications functionality is the performance of communications between
smart devices (plugs, washing machine and AACC) and the EBoxes. It is measured by the number of
slots actually received by the EBox from the smart devices compared with the expected value (Figure
33). The theoretical one is the result of the product between 288 slots per day and the number of
devices installed.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 41 of 126
0%
20%
40%
60%
80%
100%
January February March April
Smart Devices Communication
Received slots Non received slots
Figure 33. Actually received slots vs. theoretical ones.
Table 7 proposes a comparison between the number of slots received and the number of theoretical
slots, taking into account both the number of devices received (1) (see before) and the number of
EBox received (4). In the last case we have obtained the number of theoretical slots (6) taking into
account that every house has five smart devices (5) and each device stores information every five
minutes (288 slots per day).
Month Devices
received (1)
Theoretical
slots (2)
Received
slots (3)
E-box
received (4)
Theoretical
devices (5)
Theoretical
slots (6)
January 7.663 2.206.944 677.922 2.717 13.585 3.912.480
February 10.822 3.116.736 1.012.258 3.168 15.840 4.561.920
March 9.703 2.794.464 905.988 2.983 13.325 3.837.600
Table 7. Device communications.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 42 of 126
0%
20%
40%
60%
80%
100%
January February March April
Smart Devices Communication
Accepted Devices Failed Devices
Figure 34. Actual received devices vs. theoretical ones.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 43 of 126
4. Italy
4.1. Description of the test
The main objective of Italian field test is to validate the DSO algorithms and prototypes developed
within the ADDRESS project to enable and exploit AD products visible on the MV network. In this test,
AD products were emulated by means of a storage system and by modifying load/generation of some
MV customers/producers.
4.1.1. Location
The Italian test site location is Carpinone (Molise Region, Centre of Italy), where the grid has over
capacity which enables tests to be performed without endangering the quality and availability of
supply. At the location a storage system is installed to emulate demand increase of a load area by
charging the battery and demand decrease by discharging the battery.
Figure 35. The test location and the storage system in Italy.
Some of the major characteristics of the Carpinone test location are outlined in the Table 8 and a
graphical outline of the medium voltage network is depicted in the Figure 36.
340 km MV network (300 km overhead lines)
10 MV feeders
157 MV/LV substations
17 MV consumers (total power 13,2 MW)
11 MV DGs (13,25 MW: 5 hydro, 5 PV and 1 biogas)
1 Storage system (1000 kW - 500 kWh)
8110 LV consumers
63 LV DGs (total power 467 kW)
Table 8. Carpinone grid characteristics.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 44 of 126
Figure 36. Carpinone electric system layout.
As depicted in Figure 36, the storage system used to emulate AD in the test is installed on the Pesche
feeder.
4.1.2. Players involved
Some MV consumers and some of the generators connected on the Carpinone HV/MV substation
were involved in the field test. In particular, these generators have provided the generation curves
utilized by the Generation Forecast algorithm and some of them have become available to disconnect
their plants as required by the Test case 3 (section 4.2.1 Assessment of the results). Some of the
major characteristics of the involved generators involved are outlined in the Table 9.
Type of generator P (kW)
Biogas 771
PV 900
Hydro 2500
Hydro 1600
Hydro 2620
Hydro 1150
Hydro 1150
PV 82
Table 9. Involved generators.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 45 of 126
4.1.3. Architecture of the system
The field tests executed in Carpinone by ENEL Distribuzione refer to the left side of Figure 37.
Figure 37. Field Test Overview.
The Italian tests were focused on the so-called upstream of Active Demand: the DSO and TSO part.
The scope of the field tests addressed the functional components, which are to be used by the DSO in
an active grid environment. The DSO algorithms, Medium Voltage Control Centre (MVCC) prototypes
and SCADA functionality (Figure 38) are evaluated, using the requirements and criteria that have
already been defined in the design process and have been tested during the laboratory tests.
Measurements are set-up to capture the output of the said functions. Upon completion of the test the
measurements are compared with the original requirements.
The aim of the test effort is to:
- warrant the ability to maintain a secure & reliable network management
- ensure the usability of (software) functions.
Pla
ye
rs o
r fu
nc
tio
ns
of
the
ele
ctr
icit
y
sy
ste
m
Aggregation platform
EB
Meter or equivalent
Washing machines
AC
Electric heating
Water heaters
Smart plugs
Maybe other
controlled usages
Retailer Other non
controlled usages Balancing
Responsible Party
Producer
Ma
rke
t, c
on
tra
cts
or
dir
ect
bila
tera
l re
latio
nsh
ips
Consumers
DSO
TSO
Italian Field Test
Spanish Field Test
French Field Test
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 46 of 126
Figure 38. ADMS and MVCC architecture.
4.1.4. Equipment installed
The MVCC developed in the ADDRESS project was installed on the AD Server component that runs
on a computer server that resides in the Campobasso Control Centre (see Figure 39). The SCADA
system of Campobasso Control Centre provides I/O data to support the DSO functionality.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 47 of 126
Figure 39. Equipment and measurements in the HV/MV substation.
Also the Remote Terminal Units (RTUs) and active/reactive power measurement devices were
installed in the HV/MV and in some MV/LV substation, in order to provide real measurements for
Distributor State Estimation. The RTU installed in the HV/LV substation transmits measurements to
SCADA via IP network. In the Table 10 devices installed in the MV network and their functions are
described.
SCADA System implementing
MVCC functions
Network Control Centre
(Campobasso)
DSO functions to enable and
exploit AD
TPT2000 (RTU) HV/MV “Carpinone” substation and
“Carovilli” remote MV busbar
Real time network
measurements for DSO
algorithms
RTU, P&Q
measurement
devices
MV/LV substations &
MV producers and consumers
premises
Table 11. Field test devices installed in the MV network.
The installation of the measurement devices in the MV/HV substation (storage included) and in the MV
producers and customer’s premises are illustrated in the Figure 40 and Figure 41.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 48 of 126
Figure 40. Equipment in the MV/LV substations.
Figure 41. Equipment in the MV producers’ and customers’ premises.
Due to the installed devices it was possible to collect the following measurements (in addition to
measures of current and voltage on the top of the MV lines, transmitted to the SCADA through
GSM/GPRS in real time):
- P, Q in 10 MV/LV substations (MV/LV transformer load)
- P, Q in 4 MV Producers premises
- P, Q in 2 MV Consumers premises (MV load)
- P, Q in 1 Storage system
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 49 of 126
4.1.5. Exchanged messages between players
Most of the message exchanged between the MVCC and the external actors involved in the
architecture were simulated. In particular, the interactions with the market and the aggregators were
simulated by manually creating a total supply bid file (with the AD products characteristics offered to
AD market sent to SO for validation) and uploading it in the MVCC interface. The communication with
the producers involved in the tests took place via e-mail when it was necessary to require production
curves or some plants disconnection for the test execution.
4.2. Results
4.2.1. Assessment of the ADDRESS technology
Test Case 1: the DSO as an AD product validator
In this case the DSO receives bids of AD products, which aggregators traded in the market or directly
with other energy market players, to be validated. These bids should be curtailed or accepted by the
DSO by means of the validation algorithms, and the results of the validation (accepted or curtailed
bids) should be published in the MVCC interface. Then, the AD products will be activated by means of
emulation and the real network operation will be observed. The target of the test is to ensure that:
- The DSO algorithms run in a correct sequence without errors;
- The DSO validation algorithms detect and curtail correctly an unfeasible (too large) AD product
and publish the response on the MVCC interface.
The following scenarios were considered:
- Test 1 and Test 1bis: the DSO receives bids for the intraday market that should be curtailed or
accepted
- Test 2: the DSO receives bids for the day-ahead market that should be curtailed or accepted.
Test date/time
Objective Test procedure Expected result Result(OK/FAILED)
29/01/2013 14.00 - 15.00
To ensure that the DSO validation algorithms work properly
1. Run the GF, LF, FT
All the algorithms should run without errors
OK
2. Upload the Total SupplyBid file with SRP and CRP. The AD product 300 kW (250 kW SRP + 50 kW CRP UP + 90 kW CRP DOWN (100% curtaliled)) from 14.00 till 15.00 was offered on the LA DM602023612L001 (the LA where the storage system is connected)
TSB2013_01_29
intraday 300kw LA storage crp down da levare.xml
OK
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 50 of 126
3. Run the OLV algorithm
The OLV correctly accepted the AD products (curtailment factor =0,001)
cim_ieee14_state_rdf
_OLV.xml
OK
4. Run the RTV (because we have also the CRP product in our TSB)
The RTV correctly accepted the CRP product
ValidationResponse_R
TV.xml
OK
5. Download the DSE outputs before the storage activation
The DSE output before the storage activation are realistic (the current and the voltage behaviors were checked and compared with the real measurements in some control points)
outputAdrress
13.41.xml
OK
6. Actuate the AD products with the storage system
The storage system is correctly activated (the P, Q values in the LV measurement xml file on the LA DM602023612L001 were compared with the previous situation, p. 4)
OK
7. Download the DSE outputs after the AD product activation (redo this every 15 min while the AD is activated)
The DSE output during the storage activation (at 14.20) is realistic (the current and the voltage behaviors were checked and compared with the real measurements in some control points)
outputAdrress
14.20.xml
OK
8. Verify the network operation by comparing the DSE outputs before, while and after the AD products
There are no network constraints in the network operation (the DSE outputs after the AD product activation must not exceed the
OK
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 51 of 126
activation limits of the test line (+-10% of nominal voltage)
Table 12. Test1: AD validation for the intraday market.
Notes:
1. The VVC is tested implicitly because it always reacts to achieve the optimal network situation
2. The files related to the tests described above are available on the MVCC interface (for the
indicated date/time).
The first preliminary observations are:
1. All the algorithms run well and give realistic results.
2. Before the AD product activation, the DSO had no network problems to solve and after the AD
product activation by means of the storage system no network constraints appeared (DSE
results).
3. The interpretation of DSE results presents some difficulties due to the fact that they are
presented in .xml format. A user friendly visualization interface should be developed for the
product to be suitable for real applications.
4. The MVCC algorithms behavior was compliant with the expected one.
Test date/time
Objective Test procedure Algorithms behavior Result
(OK/FAILED)
29/01/2013 16.00-17.00
To ensure that the DSO validation algorithms works properly
1. Run the GF, LF, FT All the algorithms should run without errors
OK
2. Upload the TSB file with SRP = 300 kW on the LA DM602023612L001 form 16.00 till 17.00
TSB2013_01_29
intraday 300kw LA storage - scorta II cat.xml
OK
3. Run the OLV algorithm and RTV (because we have also the CRP product in our TSB)
OLV has to accept or to curtail the "big" bids (part of SRP and/or CRP), curtailed CRP/SRP have to be published on the MVCC interface
cim_ieee14_state_rdf
_OLV.xml
OK
4. Download the DSE outputs before storage activation
The DSE output before storage activation are not realistic
outputAdrress
15.20.xml
FAILED
5. Actuate the AD products with the storage system LVMeasurements2013
0129151545.xml
OK
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 52 of 126
The P value on the LA DM602023612L001 is different respect to the P before storage activation
6. Download the DSE outputs after the AD product activation (redo this every 15 min while AD is activated)
The DSE gave an error FAILED
Table 13. Test1Bis: AD validation for the day-ahead market.
The first preliminary observations are:
1. All the algorithms (p.1) runs without errors.
2. Before the AD product activation the DSE gave an unrealistic result, not compliant with the
real network situation. This result could be due to the following reasons: significant differences
between field measurements and historic data or an unexpected reaction of the DSE to the
connection/disconnection of capacitor banks in HV/MV substation. This reliable DSE is
needed for real applications.
Test Date Objective Test procedure Expected result Results
(OK/FAILED)
12/12/2012
To validate the "memory function" of the OLV Algorithm. The memory function should read the results of the last OLV and, if the validated product is performed in the same hours and on the same LAof the new products, the latter will have to be curtailed consequently. This function is requested forthe evaluation of products of the intraday market, considering the results of the OLV of the day ahead market of the previous day.
1. run the GF 2. run the LF
All algorithms should run without problems and in the correct sequence. The algorithms output will be automatically updated to the OLV input folder and will appear in the OLV interface page
OK
The total supply bid has been prepared assuming a SRP load increase of 300 kW from 19:00 to 20:00 on the LA DM602023612L001
Run the OLV. Be sure there is no thick on the Enable button at the bottom of the "input" rectangle
The OLV should not receive any error message in updating the input files including the TSB already prepared in the previous step. The OLV should not curtail the SRP product assuming a typical load (observed from the real measurements), on the LA DM602023612L001, from -60 to -100 kW
TSB2012_12_14 LA storage -300.xml
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 53 of 126
The log file should present the normal sequence of operation: Network model GF and LF input upload TSB upload Iterative calculation steps
The result of the curtailment matrix should be seen both on the interface and on the xml file
No curtailment has been applied
OK
The Total Supply Bid has been prepared assuming SRP load increase of 200 kW from 19:00 to 20:00 on the LA DM602023612L001 and a CRP down (load decrease) of 50 kW
Run the OLV. Be sure there is the thick on the Enable button at the bottom of the "input" rectangle
The OLV should not receive any error message in updating the input files comprising the TSB already prepared in the previous step. The OLV should curtail the SRP product with a curtailment factor value from 0,8 to 1 The OLV should curtail the CRP completely
OK (0,845 CF)
Table 14. AD Validation for the day-head market.
Test Case 2: the DSO as an AD product buyer
In this test case, the DSO has to buy AD products in order to solve some expected network
constraints. A network constraint is simulated by changing the thermal limit of a MV cable in the
network under test. The power flow variation to solve this network problem is calculated preliminarily
and the amount has to be equal to/larger than the AD bid that the DSO has to buy on the market.
Then, the DSO sends a request to the market and receives a response. In case of positive reply, the
AD product is activated by means of the storage system. The observed real network operation has to
be compliant with the DSO forecast (the current through the interested cable has to be lower than the
modified thermal limit).
server.log
cim_ieee14_state_rdf_OLV_20121214182736.xml
TSB2012_12_14 LA storage -200 + crp 50 con memoria.xml
cim_ieee14_state_rdf_OLV_20121214184144.xml
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 54 of 126
Date Scope of the test
Test procedure Expected result Results (OK/FAILED)
19/02/2013
To verify that AD products could be used to solve network problems
1. Run the GF, LF, FT All algorithms run without problems and in the correct sequence
OK
2. A network constraint is created by setting a new (inferior) thermal limit of the MV cable between DM60-4027517 and DM60-2023541 in the MV equipment file. The DSO can solve the problem by means of a load reduction in the adjacent LA. The load reduction is simulated by the storage, which injects the energy in the network.
The changed parameters can be found in the attached file (row 20847-20853 related on the segment DM604027517_DM602023541_0
MVEquipment_20130
215093725m1.xml
OK
3. Upload the Total SupplyBid file with the AD product on Load Area where the storage is installed.
The created TSB contains SRP =50 kW, CRP UP = 50 kW and CRP DOWN = 5 kW products that correspondto a Load reduction = 95 kW on the LA 0001DM602023612L101 from 16.00 till 17.00
TSB2013_02_19
pom2 100kw LA storage crp da levare (1).xml
OK
4.The OLV has to curtail/accept and publish the validation response
The OLV accepted SRP and CRP because they do not exceed the power limit of MV/LV transformer of the LA 0001DM602023612L101 (where the AD product is simulated).
cim_ieee14_state_rdf
_OLV.xml
OK
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 55 of 126
5. Run the RTV (on demand) because the TSB contains the CRP product.
The RTV accepted correctly the SRP and CRP
cim_ieee14_state_rdf
_OLV.xml
OK
6. Download and analyze the DSE output before the AD products activation
The current on the interested cable exceeds the 10 A (in this case it is equal to 16,5 A). It means that the DSO could have a network problem if the current on the interested cable is not reduced
outputAdrress
13.45.xml
OK
7. Activate the validated AD product with the storage (100 kW emission) and analyze the DSE output and LV measurements .xml file
The values of P, Q on the LA 0001DM602023612L101in the LV measurements .xml file has to change with respect to the normal state (before the storage activation)
LVMeasurements2013
0219153826 before storage.xml
LVMeasurements
after storage.xml
OK
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 56 of 126
8. Download the DSE output while (every 15 min) and after the AD product activation. Verify the network operation by comparing the DSE output before and after the AD products activation
There are no network constraints in the real network operation after the AD products activation (the current in the indicated cable is less than 10 A (6,5 A) and the voltage does not exceed the limits of the test line (+-10% of nominal voltage)
outputAdrress
16.30.xml
OK
Table 15. DSO as an AD product buyer.
The observations are:
1. All the algorithms run well and give realistic results.
2. Before the AD product activation the DSO had an issue: the current in the interested cable was higher than 10 A (e.g. 17 A)
3. After the AD product activation by means of the storage system the current on the interested cable was less of 10 A (6,5 A) so the DSO problem
was solved by means of the AD products.
4. The MVCC algorithms behaviour was compliant with the expected one.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 57 of 126
Test Case 3: Localized AD products to limit the power flow towards the transmission system
The target of the test is to limit the power flow rising to the transmission network through the primary
substation using the localized AD product. In particular, the AD contribution of specific load areas will
be programmed within the day-ahead market.
During the test an analysis of historical data of consumption patterns of the primary substation
Carpinone was carried out (green busbar); in particular data from December 2011. Years 2012 and
2011 have similar weather and load conditions.
A significant reduction of power consumption occurs during the central hours of the morning. At the
same hours of Saturdays and Sundays there is also the reversal of power flow to the transmission
network. The maximum reverse flow is about 2,3 MW in the day December 26th (vacation day in
Italy), the reason for this behaviour is that the network has mainly industrial customers, which reduce
their load during the day off.
In this case the DSO is configured as:
• Distribution network manager: the DSO verifies if any network problems and congestions
occur, and verifies the technical possibility to solve the problems with an active demand
product
• Active Demand product buyer: the buys from the market the most suitable active demand
product from an aggregator.
The test steps are:
1. Creation of the total supply bid for the day-ahead market in order to have an AD product
from 11.00 to 12.00 of the day after
2. Launch of the online validation and verification
3. Emulation of the AD product the next day
4. Check the distribution state estimator if "network constraints" occur and if limitation of the
power flow towards the transmission network through the primary substation at Carpinone
does occur
Test 1: 19th December
This consumption profile is foreseen:
• 11:00 to 11:35 308 kW
• 11:35 to 11:45 150 kW
• 11:45 to 12:00 50 kW
19th December
Test procedure Expected result Notes
1. run the GF 2. run the LF
All algorithms can run without problems and in the correct sequence The algorithms output will be automatically updated on the OLV input folder and will appear in the OLV interface page
This test can be executed also for the day ahead
Weather_Isernia_2012_12_19.xml
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 58 of 126
19th December
Test procedure Expected result Notes
The Total Supply Bid prepared assuming an SRP load increase of 250 kW from 11:00 to 12:00 on the following LA: DM602023612L001 DM602023541L001
If we simulate the AD products with the storage system, they have to be simulated in the LA DM602023612L001
Run the OLV. Be sure there is no thick on the Enable button at the bottom of the "input" rectangle
The OLV should not receive any error message in updating the input files comprising the TSB already prepared in the previous step. The OLV should not curtail the SRP product assuming a typical load (observed from the real measurements), on the LA DM602023612L001, from -60 to -100 kW
The log file should present the normal sequence of operation: Network model GF and LF input upload TSB upload Iterative calculation steps
The result of the curtailment matrix can be seen both on the interface and on the xml file
These curtailments were applied: DM602023612L001: 0% DM602023541L001: 77,2%
The curtailment factor on the very low value of the CRP up is 1.00 (totally curtailed); contemporarly the CF on the SRP product is not 0.This error is due to a bug of the iterative calculation process; the error is not fatal and do not compromise the test success
cim_ieee14_state_rdf_OLV.xml
In order to provide the planned AD product it is possible to use only the load area DM602023612L001 (Centro Squadre), where the storage is located. These are the following set points for the storage: 11:00-11:35 308 kW 11:35-11:45 150 kW 11:45-11:57 50 kW
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 59 of 126
19th December
Test procedure Expected result Notes
The OLV should not receive any error message in updating the input files comprising the TSB already prepared in the previous step. The OLV should curtail the CRP completely
DistributionStateEstimator1200.zip
Table 16. Localized AD products to limit the power flow towards the transmission system – Test 1
The observations are:
1. a strong load reduction occurs from 11 to 12 in the morning, but only in some cases an
inversion of the power flow occurs.
2. in case of power flow inversion, the storage cannot always ensure enough power to limit the
flow rising to the transmission network through the primary substation Carpinone.
3. in case of storage activation, the connecting cable is not always able to support the
requested consumption. It becomes necessary to use also other load areas to provide the
requested AD product.
Test 2: 28th January
Tests carried on during December showed that the contribution of the storage only not always is
enough to limit the flow rising to the TSO network through the primary substation at Carpinone. It is
therefore necessary to use other load areas to provide the requested AD product. In particular a
production reduction will be requested to t
he generators on the network; so, the total supply bid must also include the contribution of AD
provided by them.
The total supply bid includes:
Load Area 0001DM602023612L001 W -250000
Load Area 0001DM602023541L001 W -250000
Load Area 0001DM602023503L001 W -150000
Load Area 0001DM602023541L001 W -150000
Load Area 0001DM602027506L001 W -100000
Load Area 0001DM602031519L001 W -100000
Load Area 0001DM602031518L001 W -200000
Load Area 0001DM602031517L001 W -150000
Load Area 0001DM602031514L001 W -100000
It corresponds to the following requirements for the generators on the network under test:
Generator Line Installed power
[kW]
AD foreseen
[kW]
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 60 of 126
Biogas Pesche 700 300
PV Biogas Pesche 950 400
Storage (consumption) Pesche 500 500
TOTAL 1200
In order to provide the AD product, the storage on Load Area “Centro Squadre”
(DM602023612L001) is used.
The foreseen consumption profile is:
• 11:00-11:30 500 kW
• 11:30-12:00 0 kW
28th January
Test procedure Expected result Notes
1. run the GF 2. run the LF
All algorithms can run without problems and in the correct sequence The algorithms output will be automatically updated to the OLV input folder and will appear in the OLV interface page
This test can be executed also for the day ahead
Weather_Isernia_2013_01_28.xml
The Total Supply Bid was prepared assuming an SRP load increase of 1200 kW from 12:00 to 13:00 on the following LA: 0001DM602023612L001 kW250 0001DM602023541L001 kW250 0001DM602023503L001 kW150 0001DM602023541L001 kW150 0001DM602027506L001 kW100 0001DM602031519L001 kW100 0001DM602031518L001 kW200 0001DM602031517L001 kW150 0001DM602031514L001 kW100
Run the OLV. Be sure there is no thick on the Enable button at the bottom of the "input" rectangle
cim_ieee14_state_rdf_OLV_20130128121515.xml
The log file should present the normal sequence of operation: Network model GF and LF input upload TSB upload
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 61 of 126
28th January
Test procedure Expected result Notes
Iterative calculation steps
The result of the curtailment matrix can be seen both on the interface and on the xml file
In order to provide the planned AD product the following generators are involved: Biogas Pesche (kW700) kW300 requested PV Biogas Pesche (kW950) kW400 requested Storage Pesche (kW500) kW500 requested TOTAL 1200 kW
The OLV should not receive any error message in updating the input files including the TSB already prepared in the previous step. The OLV should curtail the CRP completely
DistributionStateEstimator1200.zip
Table 17. Localized AD products to limit the power flow towards the transmission system – Test 2.
The first preliminary analysis is:
1. a strong load reduction occurs in specific hours (11-12 in the morning), but only in some
cases an inversion of the power flow occurs. In case of a power flow inversion the storage
cannot always ensure enough power to limit the flow rising to the transmission network
through the primary substation at Carpinone.
2. in the case of storage utilization, the connecting cable is not always able to support the
requested consumption. It becomes necessary to use also other load areas to provide the
requested AD product
3. an economic analysis is required to quantify costs and benefits of the action taken:
- costs: penalties for the non-production of generators;
- benefits: avoided penalties for the resolution of congestion and / or flow towards the
TSO.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 62 of 126
Test Case 4: MVCC algorithms reaction to network changes
The aim of this test is to verify the algorithms reaction to the relevant network structure modification.
This test was carried out on purpose exploiting a “big” network configuration variation. In fact, this
test was performed while the Carovilli operative centre was disconnected due to scheduled
maintenance works.
Test date/time
Objective Test procedure Algorithms behavior Results
(OK/FAILED)
20/02/2013 16.00-17.00
To verify that the algorithms react correctly after modification of the network structure
1. Run the GF, LF, FT
All the algorithms should run without errors
OK
2. Download the Network model files and the DSE outputs after the Carovilli Operative Centre disconnection
The Carovilli Operative Centre disconnection is detectable in the MV topology .xml file
MVTopology_2013022
0003125.xml
The DSE output is realistic (the current and voltage values were checked in some control points)
output.xml
OK
Table 18. MVCC algorithms reaction to network changes
The observations are:
1. The MVCC correctly receives the measurements and the network model is correctly
updated.
2. The DSE reacts well in case of relevant network structure modification and it is able to give a
realistic evaluation of the new network state.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 63 of 126
5. France
5.1. Description of the test
Extensive tests were performed in EDF laboratory to prepare the French field tests. They were
conducted in conditions very near to those encountered on the field and allowed to assess the
technical feasibility and the performance of the ADDRESS chain components during individual and
interaction tests [6]. They also allowed to identify the limitations and the risks linked to the currently
implemented solutions.
Since these unitary and interactions tests were extensively carried out during the previous testing
stage, they are no longer performed in the field tests. The test cases carried out during the French
field tests are mostly global scenarios to which all the main components of the ADDRESS
architecture contribute.
The objectives of the French field tests were mainly the following:
Using the results of the laboratory tests as a basis, one of the aims was to evaluate the technical
feasibility and the performance of the complete ADDRESS chain.
During the field tests, it was assessed how AD can meet the needs of the electricity system
players using bilateral contracts (SRP and/or CRP). The provision of AD services was based on
the simulation of possible problems or actual needs identified by electricity system players (DSO,
TSO, BRP…) or by other players (retailer, RES producers – PV in our case…). These needs
resulted in appropriate AD requests tested on the field. The AD services such as power reserve
for imbalance management, load shaping (load increase or decrease) for technical or
economical optimizations, voltage control and power control to relieve overloads or network
congestions are considered.
The impact on the network of the obtained AD volumes was examined too. The effects
forecasted using the AD product volumes submitted by the aggregation entity and validated by
the DSO were compared with the effects caused by the load variations observed on the field.
In the field tests, we also studied the response of the consumer portfolio to the incentive signals
defined by the aggregation entities. This study performed on both individual (consumer) and
global (cluster of consumers) levels allowed us to verify if, how and under which conditions the
initial AD need was fulfilled. The forecasted behaviors of the consumer appliances and of the
global cluster were compared to the ones measured on the field.
In addition to the previous test cases, another objective was to test the simulated market
interactions of the aggregation entities and other electricity system participants (e.g. retailer, or
balancing responsible party). Several market scenarios were performed. We aimed at assessing
the impact AD could have on the prices of the simulated electricity market; different market
conditions were considered. The integration of the simulated electricity market inside of the
complete ADDRESS chain of process was also tested.
Finally, consumers’ acceptance and commitment were also assessed but this topic is out of the
scope of this deliverable. In this respect, the studies carried out and the result obtained are
described in ADDRESS Deliverable D5.2 [7].
Several cases were tested in order to cover these objectives of the French field tests. Most of them
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 64 of 126
are about the complete ADDRESS chain of processes providing different service types. Others
concern the electricity market simulator and its interactions with the aggregation entity. All these
cases allow us to verify the capability of the implemented ADDRESS architecture to respond to the
envisioned situations and to test the performance of its modules on real data. Those involving all the
components of the ADDRESS architecture are the following:
Request from the DSO for reducing a local overload constraint,
Request from the TSO for reducing a congestion constraint,
Request from the BRP for managing the imbalance between production and consumption,
Requests in combination with RES for reducing a local voltage constraint anticipated by the DSO
or providing AD services to a RES producer,
Requests from the DSO for limiting the global consumption of the islands over a complete day,
Offers providing AD services to an electricity market.
It is to be noted that the proposed field test cases are generic enough to be illustrative of most of the
services that could be provided by the ADDRESS architecture. The AD requests that have been
considered and tested may indeed correspond to the needs of different players, or in other words
each of the 31 AD services defined in the Deliverable D1.1 [8] may be associated with at least one of
the cases played in the French field tests.
The description of the test cases and detailed test scripts performed in the French field tests can be
found in Chapter 8 and in the Appendix of Deliverable D6.1 [1].
5.1.1. Location
The field tests in France were performed on the two islands of Houat and Hoëdic in the South of the
Brittany region, in the West part of France.
The network considered is composed of 1 MV feeder and 8 MV/LV substations. The island of Houat
is connected to the continent through a 19 km long underwater MV cable, and a second 8.5 km long
underwater MV cable connects the island of Houat to the island of Hoëdic. On Hoëdic, one private
PV plant with a power of 100 kW peak is connected to the LV grid. At certain periods of the year
(e.g. May or June) the PV production may be very high with respect to the consumption of the island.
There are a total of 569 consumers for both islands, of which about 373 are permanent consumers.
Indeed, there is a high rate of secondary houses. Around 30 domestic consumers have taken part in
the tests, leading to a rate of 5 to 8 % of active consumers, which is quite significant for such a
demonstration.
On these islands a lot of the consumers use electricity for their water heaters (water tanks containing
between 150 and 300 litres) taking advantage of the off peak tariff (8 hours per day), and for heating
by means of electric heaters. This represents the major part of the electricity consumption for these
consumers.
The French field tests site is described in more detail in ADDRESS Deliverable D6.1 [1].
As already mentioned, the French field tests are dedicated to the validation of the whole ADDRESS
chain, i.e. from the AD buyers to controllable appliances at the consumers’ premises, which is shown
on Figure 42. The configuration of the chain and of its different components for the carrying out of
the French field tests are described in detail Sections 5.1.3, 5.1.4 and 5.1.5.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 65 of 126
Pla
yers
or
fun
cti
on
s o
f th
e e
lectr
icit
y
syste
m
Aggregation platform
EB
Meter or equivalent
Washing machines
Electric heating
Water heaters
Smart plugs
Retailer Other non controlled usages
Balancing Responsible
Party
Producer
Mark
et, c
ontr
acts
or
direct
bila
tera
l re
lationship
s
Consumers
DSO
TSO
Figure 42. The whole ADDRESS chain tested in France
5.1.2. Players involved
In order to conduct the field test cases described in the previous section, the complete ADDRESS
architecture is considered and tested. Depending on the scenario, different players are involved in
the tests. In every test script, at least the three following players are to be considered: the DSO, the
aggregation entity and the consumers (or more precisely their EBoxes and appliances). The Market
Simulator (MS), simplified representation of the Electricity system players and real local players are
also considered in some of them.
5.1.3. Architecture of the System
As already mentioned, during the laboratory tests, the modules composing the ADDRESS
architecture were validated and most of the limitations and risks they present were identified. Based
on these results, choices were made regarding the ADDRESS architecture implementation for the
French field tests in order to ensure that both the system performance and reliability are guaranteed.
Based on the results of the laboratory tests, a two-part system compliant with the ADDRESS
architecture and fit to the reliability requirements of a field test was retained for the French field tests.
This system is composed of two distinct parts (Errore. L'origine riferimento non è stata trovata.):
the upstream part of the architecture consisting of the DSO Platform, the ATB System, a
simplified modeling of the different electricity system players and the Market Simulator. It is
managed in off-line mode. Since significant risks of failures of the ATB optimization
processes were detected during the laboratory tests, the optimization results could not be
directly applied in the EBox database connected to the internet. The ATB and the DSO
platforms are run with a local copy of the EBox database (renamed ATB database). Using
the database, the whole day-ahead and intra-day sequences are run with the interactions
between the DSO and the ATB platforms being considered. If the sequences prove to come
to their end and succeed, they are approved for application.
the downstream part consisting of the ecosystem of the Energy Boxes. It is always run
connected to the internet in an online mode. The correct operation of the Energy Boxes
(EBox) highly depends on its availability and the robustness of the information present in the
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 66 of 126
EBox database of the aggregation entity (where the EBoxes get the signals generated by
the ATB). In case a problem arises in this part, the consumers could be impacted. Special
care is thus taken so that no erroneous data is input in this database.
The link between both parts is ensured by the synchronization of the two EBox and ATB databases.
This synchronization occurs only when every precaution has first been taken to ensure that no
problem may arise on the field.
The strict separation between both parts was also useful to run some cases for which tweaks were
required. Indeed, some of the algorithm limitations identified during the laboratory tests required
manual interventions in order to by-pass some modules and run the considered test script (e.g. when
testing load increasing signals).
Detailed information on the system implementation deployed for the French field tests and on the
impact of these two separate operation modes on the management of the platforms included in the
ADDRESS system will be specified in the following section.
DSO Platform
Weather
websites
Historical Consumption
server SOAP
webservices
EBox Database
WebServer
ATB Database
2-way Synchronization
EBoxes
EBoxes
EBoxes
EBoxes
Internet
ATB
System
Up
stre
am
ch
ain
Do
wn
stre
am
ch
ain
Simulated Electricity
Market
Figure 43 - Operation mode of the ADDRESS architecture considered in the French field tests
5.1.4. Equipment installed
5.1.4.1 Upstream side of the ADDRESS chain: market, DSO and aggregation function
5.1.4.1.1 DSO platform
The DSO performs different activities depending on the considered test case
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 67 of 126
(
LAC
GenF
FTC
LoadF
OLV RTV
Algorithm managing
engine
Consumption data database
AD flexibility data database
Weather data database
Weather
websites
Historical load data
server
Weather forecast manager
CIM network
converters
Historical
databases AXIS Server ATB
Network description
: Interactions of the DSO platform with external actors
: Interactions within the DSO platform
SOAP
DSO Platform
Figure 44.). All of its activities are run using the DSO platform developed by the EDF team and
where in particular the algorithms developed for the DSO by the project partners have been
integrated. Its specific components and GUIs are used.
The platform is used to anticipate the distribution network constraints and to define the DSO
needs in terms of AD products in regards to these constraints. These flexibility needs are
transmitted to the aggregation entity in order to adjust the local consumption of its
consumers’ portfolio.
The DSO platform operates automatically as the actor responsible for the technical
validation of the Active Demand programs. It communicates the flexibility information to
every aggregation entity, performs the technical validation of the submitted flexibility
programs and returns the curtailed volume to the appropriate aggregation entity.
The DSO platform enables assessing the impact of the real flexibility response of the
consumers on the evolution of the distribution network quantities and its constraints.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 68 of 126
LAC
GenF
FTC
LoadF
OLV RTV
Algorithm managing
engine
Consumption data database
AD flexibility data database
Weather data database
Weather
websites
Historical load data
server
Weather forecast manager
CIM network
converters
Historical
databases AXIS Server ATB
Network description
: Interactions of the DSO platform with external actors
: Interactions within the DSO platform
SOAP
DSO Platform
Figure 44. Main components of the French DSO platform and interactions with external actors.
5.1.4.1.2 Aggregation entity and ATB
The aggregation entity is the central component of the ADDRESS architecture. It interacts with
several actors depending on the considered case.
EBoxes. The main interaction of the aggregation entity with the consumers is through their
EBoxes. The aggregation entity defines and “sends” price and volumes signals to the
consumers’ EBoxes to obtain modifications of their consumption.
Regulated and deregulated players. The flexibility needs expressed by regulated and
deregulated actors are taken into account by the ATB platform algorithms. The signals sent
to the EBoxes depend on the requested flexibility volumes and timeslots.
DSO. The aggregation entity takes into account the local flexibility information defined by the
DSO in order to assess the offer volume it can respond to and to constrain the selection of
the signals that can be sent to the EBoxes. In the same way, the results of the technical
validation carried out by the DSO are integrated by the aggregation entity when choosing the
signals for the EBoxes.
Market. The aggregation entity can offer flexibility volumes on an external electricity market
and take into account the acceptance level of its offer proposal.
The management of its algorithms and their interactions with external actors is done by the two
platforms developed for the French field tests:
(i) the ATB prototype management software, used to generate the content of the ATB
database, and
(ii) the ATB platform, used to manage the day-ahead and intra-day optimization sequences.
Both platforms are part of the system of the aggregation entity, i.e. which will be called here the ATB
system (Figure 44)
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 69 of 126
AXIS Server DSO
: Interactions of the ATB platform with external actors
: Interactions within the ATB platform
SOAP
ATB System
EB Database
WebServer
ATB Database
Consumer Samples
Generation
ATB prototype
management
Response Forecast
Signals Adjustment
Market Price Forecasting
ATB Platform
Optimization Algorithms
2-way Synchronization
EBoxes
EBoxes
EBoxes
EBoxes
Internet
Regulated /
deregulated actors
ESB
Orchestrator Market
Figure 45. Main components of the French ATB system and interactions with external actors.
5.1.4.1.3 Market simulator
The tests performed regarding the Market Simulator (MS) follow several objectives:
Assessing the impact of AD offers generated by the ATB or any aggregation entity on the
wholesale market prices,
Analyzing the market acceptance of these AD offers,
Evaluating the potential benefits for the aggregation entity.
It should be noticed that the market considered for the tests is the (French) SPOT national energy
market where different actors (DSO, TSO, aggregation entities, BRPs…) interact in order to sell or
buy volumes depending on their needs and their resource availability. These volumes are
aggregated in one global offer and one global demand coming from the forecasting tool (Figure 46.).
Other markets could have been considered assuming that they follow similar principles. The SPOT
market was favored since the data required for the studies are public.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 70 of 126
Figure 46. Interaction of the ATB system with the Market Simulator.
During the tests, two MSs have been used. The 1st one is the tool developed in Task 5.3 of WP5 and
the 2nd
one is a tool developed by the EDF team in order to run additional test cases not possible
with the 1st simulator.
1. Initial market simulator
The market simulator developed in Task5.3 is a JAVA program which operates similarly to a real
electricity market. It uses demand and supply bids as inputs and then clears the market. The tool
only accepts hourly and blocks bids for a single market area.
A bid creation tool was also developed in Excel in order to simplify the creation of the bid files for the
test cases.
The integration of the hourly bids is realized as follows:
The supply bids and demand bids are aggregated separately: the aggregated curve is the
sum, for each price, of the volumes from the different bids (Figure 47.). We thus obtain two
curves: the hourly total supply curve and the hourly total demand curve.
The two curves are then compared to find the temporary equilibrium, i.e. the price for which
the demand and supply volumes are equal (or intersect).
This allowed us to understand several wholesale price issues. But several problems were
encountered while integrating the AD offers and motivated the development of an alternative market
simulating tool.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 71 of 126
Figure 47. Aggregation of supply bids
2. Alternative market simulator
The alternative market simulator was developed in Excel by EDF in WP6 in order to run additional
test cases. In this new version, the equilibrium point is found by intersecting linearly the price-volume
curves of the total offer and total demand. This choice was made to simplify the computation
process.
However, when the AD offer of the aggregation entity is partially accepted by the market, the market
simulator will then use the usual starcase model. So, we are sure that the price of the equilibrium is
the price of the AD offer.
5.1.4.2 Downstream part of the ADDRESS chain: Energy Boxes
The equipment installed at each consumer’s premises consists of:
- one EBox composed of PC,
- up to 7 smart wall units to control the electric heaters and the water heater,
- up to 3 temperature sensors,
- 5 smart plugs to control other types of “classical” electric appliances,
- Smart washing machines developed in the project and able to communicate directly with the
EBox installed at 7 consumers.
The configuration is shown in Figure 48.
Bid 1 Bid 2
Total bid
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 72 of 126
Meter EBox PC
ADDRESS
At consumers’
premises
ADDRESS
Control Center
(EDF Clamart)
Electric appliances connected
through smart plugs or wall units
• radiators
• water heater
• white goods…
Smart washing
machine
(7 consumers)
Temperature
sensors
Figure 48. Equipment installed at consumer’s premises.
The EBox is the computing device installed at the consumer’s premises that contains the energy
management software. It manages the controllable devices inside the house (Smart Plugs and Wall
Units, Temperature Sensors, Smart Washing Machine...) and communicates with the aggregation
platform through the so-called EBox database described in Section 5.1.4.1.2 (and
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 73 of 126
AXIS Server DSO
: Interactions of the ATB platform with external actors
: Interactions within the ATB platform
SOAP
ATB System
EB Database
WebServer
ATB Database
Consumer Samples
Generation
ATB prototype
management
Response Forecast
Signals Adjustment
Market Price Forecasting
ATB Platform
Optimization Algorithms
2-way Synchronization
EBoxes
EBoxes
EBoxes
EBoxes
Internet
Regulated /
deregulated actors
ESB
Orchestrator Market
Figure 45.) to get the AD requests (Figure 49.).
The EBox is always operating in an online mode, i.e. it is always connected to the Internet. The
EBox database of the aggregation platform is remotely accessed every 15 minutes with this internet
connection using webservices. This remote access allows in particular to detect if a new incentive
signal has been added in the EBox database so that the optimization algorithm of the EBox is then
launched. During this optimization run the requested load modification operation is taken into
account. It is to be noticed that the incentive signals input in the EBox database are either selected
by the ATB platform algorithms or manually chosen.
During the French field tests, several types of devices have been managed by the EBox. Table 19
summarizes the loads that have been controlled.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Evening peak-shaving Decrease 3 27/02/2013 to 12/03/2013 18:30 15
Morning peak-shaving Decrease 4 14/03/2013 to 03/04/2013 07:45 30
Special evening peak
shaving (sport on TV)
Decrease 1 16/03/2013 21:15 30
DSO request Decrease 1 20/03/2013 09:45 30
DSO request Decrease 1 22/03/2013 11:00 30
DSO request Decrease 1 25/03/2013 10:00 30
DSO request Decrease 2 27/03/2013 to 28/03/2013 12:00 30
DSO request Decrease 2 28/03/2013 to 01/04/2013 17:00 30
Evening peak-shaving
in Spring time
Decrease 7 03/04/2013 to 10/04/2013 20:15 30
DSO request Decrease 8 11/04/2013 to 18/04/2013 11:15 30
Morning peak-shaving
at the end of Off-peak
period
Decrease 7 19/04/2013 to 25/04/2013 05:30 30
Night peak-shaving
(EWH)
Decrease 7 26/04/2013 to 02/05/2013 02:00 30
To shift consumption
during wind turbine
production
Increase 2 04/05/2013 to 09/05/2013 02:30 60
To shift night peak Increase 12 05/05/2013 to 31/05/2013 03:30 60
Night peak-shaving
(EWH)
Increase 5 06/05/2013 to 13/05/2013 03:00 60
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 84 of 126
Scenario Request
Type
Number
of days
Dates (dd/mm/yyyy) Start
time
Duration
(min)
To shift consumption
during wind turbine
production
Increase 1 12/05/2013 04:00 60
To shift consumption
during PV production
Increase 6 17/05/2013 to 22/05/2013 13:30 60
To shift consumption
during PV production
Increase 9 23/05/2013 to 31/05/2013 11:00 60
Table 21. French field tests scenarios
5.1.5.4 Market simulator or MS
5.1.5.4.1 General assumptions
Several assumptions are made in order to test effective impact of AD on an electricity market. We
assume that:
- Each entity develops a consolidated portfolio strategy, has a good knowledge of his costs and is
able to forecast wholesale market prices efficiently. It is thus able to trade-off between his costs
and his potential gains using this information;
- We only focus on a global electricity market: the national exchange market. In fact, the rules for
the national power exchanges are clearly defined. It is also fully representative of the price
determination processes used in the power markets and all the data required for the simulation
are public. We did not consider local markets;
- When assessing AD provider’s costs, Task 5.4 assumes that the AD offers have a marginal
impact on the wholesale market, i.e. they do not affect the wholesale prices [9]. This is the
assumption used by the ATB platform when defining its AD offers. On the contrary, the main
objective of the market simulator is to illustrate the potential impacts of AD on these prices. Both
approaches are thus complementary.
- Each aggregation entity uses its internal market price forecasts, the flexibility available in its
consumers’ portfolio and the consumers’ preferences (focus on comfort, savings…) in order to
identify the optimal time slots for the AD volumes to be proposed to the market. Like in Task 5.1
and Task 5.4, we choose to adopt an “epsilon” margin around the forecasted market prices.
When testing the price sensitivity, several margin values were considered. They ranged from
10% to 40% below the forecasted market prices when offering AD products reducing demand
and from 10% to 40% above the forecasted prices when offering AD products increasing
demand3. Hence, this strategy aims at guaranteeing his AD products will be retained with a high
probability in the merit order when his forecasting are pertinent; in the other cases, losses are
limited thanks to the margins adopted.
- In the scenarios tested, we assume that only one aggregation entity is representing all the offers
from the different aggregators existing in the market. Using a price-volume signal, it is able to
manage the Electric Water Heaters (EWHs). The aggregation entity earns money by taking
3 In order to make the margin significant for market prices around zeros (0€), we have added (subtracted) one
euro (1€) to this margin. This is required since the margin is computed relative to the market price itself.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 85 of 126
advantage of appropriate load shifting on the Spot Market.
5.1.5.4.2 EWH scenarios
In these scenarios, we focus on a specific appliance, the “electric water heater” (EWH). This choice
was made for two reasons:
1. It is a common appliance in France: 9 millions of households are equipped and most of them
with thermal storage (see the description on Deliverable D6.1 [1]);
2. The French field tests showed positive results concerning the shifting of these devices.
The best time slots for EWHs activation are identified while ensuring comfort criteria.
First, we ran the market simulator in order to replay situations that were observed in the actual
French market. These tests were performed with the two market simulators, the one developed by
VTT (TS_MS_Java) and the one developed by EDF (TS_MS_Excel) as an alternative to bypass
some of the other tool limitations. We mainly focused on market situations characterized by unusual
market prices. Two different situations were considered in particular:
One with high market prices (9th February 2012 during the cold period where prices reached
1938€ / MWh in France). The corresponding test cases are TS_MS_Java – 1 to – 3 and
TS_MS_Excel – 1 to – 3 (described in D6.1 [1]).
One with negative market prices (25th December 2012 where the transitory situation caused
by high wind generation made prices drop to -50 €). The corresponding test cases are
TS_MS_Java – 4 to – 6 and TS_MS_Excel – 4 to – 6.
Then, we created an intermediate scenario using the field test flexibility and where the ATB was
bypassed (TS_MS_FRAD – 01). The incentive signals were selected manually on the basis on our
own specifications resulting from market forecast and put in the ATB database to be sent and
applied by the EBoxes. The resulting flexibility response observed in the field was introduced back
into the market simulator (with the appropriate scaling) in order to assess its impact on the market
prices.
The data we use for the EWH scenarios are public. They concern the electrical consumption of the
EWHs in France. The following simplified assumptions have been retained for illustrative purpose. It
is to be noticed that they correspond to a nearly ideal situation from the aggregation perspective.
The unit power of one EWH is around 2 kW and we have about 9 million EWHs controlled in
France. That gives us a global theoretical power of 18 GW. However, we apply a correction
(division) factor of two in order to take into account every constraint that acts on the EWH.
With this assumption, we get a global shiftable load of 9 GW in France.
Since the average heating time of an EWH is of about six hours, the maximum storage
capacity is 54 GWh. It leads us to a consumption of 20 TWh over a year.
In order to be coherent with the volumes treated in the French market and the ones
manipulated by the aggregation entities, the EWH volumes have been scaled (only 15.3%4
of the global volumes were considered for the scenarios)
From public data, we have been able to create a profile that depends on the month, the day
of the week and the hour. An overview of these profiles is given below (Figure 54.).
4 For a given day, MAX EWH power of the day / MAX national demand of the day. This ratio is then multiplied
by the MAX spot power of the day for estimating the EWH potential of the day.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 86 of 126
Figure 54. Illustration of different consumption profiles.
The minimum average sourcing cost can be reached using as much as possible the EWHs
heater at night, when the prices are low. It supposes that the storage is empty at the
beginning of the night. On the contrary, if the storage were at the maximum at night, we
should have to wait until the morning when the consumer begins to use hot water and the
individual prices are higher.
We assume that the aggregation entity has an exact knowledge of the D+1 market prices,
i.e. no error is made on the price forecast. Ideally, the aggregation entity would know the
D+2 and D+1 prices considering the daily profile consumption.
Two options were considered when choosing the reference power, Ptref., i.e. the quantity
used by the aggregation entity to prepare its offers.
o Using the load corresponding to the water consumption. We consider that the relay
controlling the load is always ON. So, the EWH is started as soon as the water
temperature becomes lower than the set point temperature limit.
o Using the load corresponding to the real energy consumption of the EWHs, i.e.
taking into account the control of the starting time thanks to the relay. This reference
is already optimized in order that the consumption of energy will be in the off peak
periods of the day (periods with individual lower prices).
The latter option was chosen because the prices we use for the tests (spot EPEX 2012) are
directly related with the EWH load controlled by the relay. The chart below (Figure 55.)
represents the average daily consumption profile (from 6:00 pm to 5:00 pm in D+1). It takes
into account the national repartition of the off-peak period (not every EWH is managed in the
same way). We will use this similar distribution for every day of the week to simplify our
cases.
Figure 55. Daily EWH consumption profile used for the tests.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 87 of 126
5.1.5.4.3 ATB and MS interaction
In day-ahead, the aggregation platform takes into account the flexibility requests expressed by
bilateral contracts (if any) and proposes the remaining flexibility volumes to the electricity market at
specific prices. The acceptance of the proposed flexibility is then taken into account by the
aggregation platform to define the signals to be sent to each consumer cluster. The optimization
algorithms of the ATB platform also take into account the flexibility constraints imposed by the DSO
for each Load Area in order to ensure the technical feasibility of the chosen signals.
Once the signal selection has ended, the chosen incentive signals are applied on the field.
It is to be noted that up-scaling factors are applied to the offered volumes proposed to the market in
order to make their impact more significant. The inverse scale is then applied to the accepted offer
volumes when the market responds to the aggregation platform. Because of limitations of the current
ATB implementation, the acceptance factor of the volumes corresponding to the payback effect of
the proposed offers cannot be taken into account by the ATB following optimization; only the offered
volume can be modulated by the market acceptance.
5.2. Results
As previously mentioned the field tests in France were mainly focused on assessing the operation
and the performance of the complete ADDRESS chain. The test cases and the detailed test scripts
are given in Chapter 8 and in the Appendix of Deliverable D6.1 [1].
In this Section, the main results of these tests are presented. A detailed description of a complete
test case is also given and the results obtained at each step are illustrated.
5.2.1. Test execution
The table below presents an overview of the tests execution performed during the French field tests
on the complete platform sequence. The column “Successful” indicates whether the successive
steps of the test scripts were carry out successfully or not but does not give any information on
regarding the effective delivery of AD flexibility. The results of the tests in terms of AD delivery are
given in Section Errore. L'origine riferimento non è stata trovata..
NB: For a detailed description of the test scripts, please refer to Deliverable D6.1 [1].
Function Code ID Objective Success
ful
Load decrease
services
TS_CAT3_Loa
d_Dec
1 AD request from the DSO for local overload reduction
OK
2 AD request from the TSO for global congestion reduction
OK
3 AD request from the BRP for imbalance management OK but
ATB
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 88 of 126
Function Code ID Objective Success
ful
bypassed
5
Requests in
combination
with RES
TS_CAT3_Loa
d_Dec
1 AD request to avoid curtailment of a PV plant due to voltage constraint
OK
2 AD request for compensating the peak production of a mainland wind farm
OK
Over-a-day
global load
limitation
TS_CAT3_OA
D_Lim 1
AD requests from the DSO for an over-a-day global load limitation
OK
Market
Simulator
TS_MS_Java
1,3
4,6
Wholesale market prices estimation based on forecasted demand and supply
Evaluation of the MS accuracy given the number of
price/volume couples
OK
2,5 Estimation of the AD impact on the wholesale market prices using the EWH as a shiftable device
Failed6
TS_MS_Excel 1 to 6
Wholesale market prices estimation based on forecasted demand and supply
Estimation of the AD impact on the wholesale market prices using the EWH as a shiftable device
Estimation of the impact of the prices of AD offers
on the earnings / savings
OK
TS_MS_FRAD 1 Test of MS with AD offers related to the load shifting observed in the field tests
OK
Integration of
an electricity
market
TS_CAT3_MA
R_INT 1 Provision of AD services to an electricity market OK
Table 22 - Test execution overview
5.2.2. Assessment of the technical performance on the consumer side
5.2.2.1 Management of the consumption by the Energy Box
As previously mentioned, the management of the electrical heaters defined as Thermal Loads was
not possible due to technical issues: as a consequence, it was decided to declare the electrical
heaters as shiftable loads. Regarding the EBox specifications, it performs the consumption
management of the controlled devices even in the absence of request coming from the aggregation
platform, i.e. it optimizes the consumption of the house without any request:
Interruptions are carried out during peak periods so that the savings are maximum for the
consumers;
5 The ATB platform had to be bypassed because load increase signals were always selected by the ATB algorithm: instead
of using the direct effect of load decrease signals, the payback effect of load increase signals was selected. Load decrease signals were thus selected manually. 6 For several hours, the MS gives unrealistic results significantly different from the ones expected from a
theoretical point of view. The calculation of earnings/savings cannot be done without knowing the part of the aggregator's offer that has been accepted by the market.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 89 of 126
The optimization is done to save a maximum amount of energy, i.e. at the end of the day all
the interruptions allowed by the configuration specified in the EBox for the interruptible loads
have been made;
The distance between two consecutive interruptions is set so that the impact on the
consumer’s comfort limited;
When the priority is given to money savings, Shiftable Loads start during off-peak periods.
Moreover, several issues have been encountered during the field operation, such as:
- A default in a Wall-Unit.
- Stop in the execution of the EB software when a Smart Plug or a Wall-Unit is switched off. In
this situation, the load remained switched off until a remote manual action is undertaken to
restart the EBox software.
- Communication errors between the EBox and the Smart Plugs or the Wall Units. In that case
the EBox doesn't receive power measurements and is not able to send ON/OFF orders.
- The distance between the EBox and the Smart Washing Machine must be short in order to
ensure their communication.
NB: according to the privacy rules applicable in France, we are not allowed to present individual
results recorded in a consumer house, even if these results are anonymized. Consequently, in the
following sections, we will illustrate results obtained at consumers in the French field tests with
similar individual data recorded during the EDF lab tests in the multi-energie house and with
aggregated results obtained on the field.
5.2.2.2 Peak-Shaving using Interruptible Loads
After a signal is received, the EBox runs an optimization to calculate the interruptions of the
controlled appliances. If interruptions are still available (i.e. the total number of interruptions for the
day is not reached) and the economic incentive sent by the aggregation platform is high enough, the
appropriate interruptible loads are switched off at the requested time in order to reduce the
consumption below the required power.
AD request
Start (date & time) Duration (min) Limit (kW)
15-03-2013 07:45:00 30 0.05
The objective of this price-volume signal is to limit the power of the house below 0.05 kW during
30 minutes. The result is presented on Figure 56..
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Table 24. Mean energy consumption of ADDRESS and Reference groups during the period of the French field test (all figures in kWh)
Table 25 presents the variations of the off-peak consumption ratio (LP/Total ratio where
Total=LP+HP) in presence of the ADDRESS systems: the off-peak consumption ratio (raw) is
presented for both groups and years and the corrected ratio for 2013.
Group Raw
LP/Total ratio 2012 Raw
LP/Total ratio 2013 Corrected
LP/Total ratio 2013
ADDRESS 0.407 0.395 0.403
Reference 0.545 0.552 0.558
Table 25. Variation of the off-peak consumption ratio in presence of ADDRESS systems
Both for the ADDRESS group and the Reference group, the mean variations between 2012 and
2013 in Table 24 correspond to energy reductions. So the EBoxes as well as the actions they trigger
when performing standalone optimizations or when responding to flexibility requests do not seem to
have a negative effect on the global energy consumption since the energy reductions observed for
the ADDRESS group is of the same order of magnitude (around 5%) as the one obtained for the
Reference group. They do not seem to have an impact on the share between LP and HP energy
consumptions either. The differences between the two groups are not significant enough to be able
to conclude to a positive effect of the EBoxes on the energy consumption.
NB: it should be recalled that the objective of ADDRESS at the consumer level is not to reduce the
energy consumption but rather to shift the consumption to times that more beneficial for the
electricity systems as a whole [10], [11].
The above results should be considered with great care because:
the energy consumption between the two groups is not equivalent in terms of total energy
consumption even though both categories of consumers have the same types of electric
equipment. The size of the two groups is too small to avoid having a few consumers with a large
influence on the results for the whole group.
Possible changes in the consumers’ individual behavior (holidays…) can also greatly influence
the results considering the size of the groups and the limited historical data available.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 106 of 126
The duration of the field tests is too short to allow to derive more conclusive results on such
topics.
5.2.4. Market simulator
5.2.4.1 EWH scenarios
5.2.4.1.1 Methodology
The whole process applied when running the Market Simulator (MS) in the EWH scenarios is
presented in Figure 72..
In the first step, we run the MS with only global forecasted data (offer and demand). These volumes
originate from exchanges between different actors (DSO, TSO, aggregation entities, BRPs…) that
sell or buy volumes depending on their needs and their resource availability. All of these volumes are
aggregated into one global offer and one global demand.
The purpose of this 1st step is to determine the forecasted market equilibrium without any AD offers.
In real life, the global demand and offer volumes are forecasted based on the historical values.
However, in our scenarios, the bids are read from the public historical curves. Using this information,
we construct a reference (SPOT prices) to compare the results of MS with.
In a second step, load shaping is performed. For the scenarios of December 25th and February 9
th, it
consists in changing the daily EWH reference consumption that has been shown in Figure 55. in
order to optimize the purchasing expenses. This shifting isn’t totally free as it highly depends on the
consumers’ usage. The challenge is to find the best shifting that satisfies the following points:
- The new load positioning has to satisfy comfort constraints;
- The shifting appreciation regarding the spot price is maximum.
It is to be noticed that the initial placement of the French EWH’s consumption is already statically
optimized from a consumer-only point of view. However the aggregation entity, by having accurate
price forecasts, can improve this optimization and be rewarded for providing additional services.
The output of this load shaping step is summarized in Figure 73. and Figure 74. where:
The green curve represents the normal energy consumption for EWH during this day.
The blue one represents the forecasted EPEX spot prices8.
The grey curve represents the usage of hot water;
The red curve represents the optimized placement of energy consumption for the EWH
considering the constraints discussed above.
The difference between the optimized consumption and the reference consumption is considered as
the volume of the AD offer. Negative AD Volumes mean that the aggregator proposes a load
decrease for the concerned period, and positive ones mean that the aggregator/retailer will purchase
this extra volume compared to the reference placement (load increase).
NB: the sum of the optimized consumptions over the day is equal to the one of the reference
consumptions; thus the total of AD volumes is equal to 0.
In the third step, we adopt a 20% margin around the real market prices in order to introduce
uncertainty in our calculations: 20% below the market prices when offering AD products reducing
demand and 20% above the forecasted prices when offering AD products increasing demand. The
impact of the value of this margin on the results is also studied here.
8 The real EPEX Spot prices of the considered day are used for the simulations.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 107 of 126
The MS is then run a second time after integrating the AD offers to obtain the new equilibrium points.
The fourth step, i.e. the exploitation of the results given by the MS, will be presented and discussed
in the Deliverable D6.3 Errore. L'origine riferimento non è stata trovata..
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 108 of 126
Figure 72. Steps for running the Market Simulator with aggregation entity’s offer9
9 For the tests, the AD prices will be calculated using the real SPOT market prices instead of the forecasted ones as shown Errore. L'origine riferimento non è
stata trovata..
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 109 of 126
Figure 73. Illustration of the load shaping of EWH consumption for the first scenario.
Figure 74. Illustration of the load shaping of EWH consumption for the second scenario.
5.2.4.1.2 Results
The following tables (Table 26 and Table 27) summarize the results obtained when running the
whole MS processes on the considered EWH scenarios.
As we can see, the equilibrium prices decrease when the aggregator offers a load decrease and
increase when it consumes more than it is supposed to. The average difference between the two
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 110 of 126
equilibrium prices (Equilibrium Price 2 – Equilibrium Price 1) over the 24 periods is -1,1 €/MWh for
the first scenario and +0.5 €/MWh for the second one. This indicator shows that the AD offers
globally implied a decrease in the market prices. However, the real impact on the purchasing
expenses should rather be weighted by hourly volumes depending on the part of the AD volumes
that has been accepted by the market. This impact will be discussed in the report D6.3 [12].
In order to assess the functioning of the MS, the prices for the equilibrium 1 were compared to the
real SPOT prices that were observed during this day. The purpose is to validate the equilibrium
calculation algorithm that allows obtaining the equilibrium points.
As we can see in Table 28 and Table 29, the results are satisfactory and the relative difference is
lower than 1%. However, for some periods (the 8th and 13th periods in Table 28 for example), the
difference becomes significant. It is related to the sampling of the demand and offer’s data for some
specific periods where these two curves are similar around the SPOT price10
. This shows that the
MS is very sensitive to the input data: for each hour, the more offer and demand points we have
around the forecasted equilibrium the more effective and reliable the results of the equilibrium
calculation algorithm will be.
10
This constraint is explained and illustrated in the methodology paragraph (§5.2.1.3.1.15.2.4.1.1).
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 111 of 126
Table 26. Test results for the scenario of the February 9th
, 2012.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 112 of 126
Table 27. Test results for the scenario of the December 25th
, 2012.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 113 of 126
Table 28. Comparison of the results with the real SPOT prices of the February 9
th , 2012
Table 29. Comparison of the results with the real SPOT prices of the December 25
th , 2012
5.2.4.2 Electric water heaters from the field test
5.2.4.2.1 Methodology
In this case, the methodology is quite the same as the one of the two previous EWH scenarios. The
only change occurs in the load shaping step: since we use the load shifting that has been obtained in
the French field tests rather than the consumption whose placement was optimized, we no longer
need to run the placement optimization tool.
On the field we succeeded to shift the consumption of 10 water heaters (2 kW each): 5 of them were
shifted from May 24th at 11:00pm to May 25th at 4:00 am and 5 were shifted from May 25th at 1:00
am to may 25th at 5:00 am.
Since the real clusters are supposed to be much bigger than the one considered in the French field
tests, the shifted consumption of the EWH was simply multiplied by 500 in order to have more
significant AD offers to propose to the market (10 MWh).
5.2.4.2.2 Results
Compared to the 2 first scenarios, the volumes for AD offers are considerably lower. The impact on
the market prices is thus not as significant as they are in the 2 first scenarios. However we can
notice that, for the 7th period, the 10 MWh load increase has changed the market price by a value of
0.334 €/MWh.
Prototype Field Tests, test results
ADD-WP6-T6.3_Iberdrola_D6.2 – Prototype Field Tests. Test Results
Final v1.0
Copyright ADDRESS project page 114 of 126
Table 30. Test results for the scenario of May 24th and 25th, 2013
5.2.4.3 ATB and MS interaction (simulation day: December 24th 2012)
5.2.4.3.1 Methodology
The volumes of AD offers are calculated in two different situations for the scenario of December 24th:
1. The first one is the load shifting of the French EWH (same global volume and different
consumption hours),
2. The second one is the output of the Aggregation Tool Box (ATB) that has been run with the
French field tests parameters.
In Figure 75, the AD flexibility volumes proposed by the ATB platform for two of the four Load Areas
considered are presented. The reduced volumes available in each cluster (a few kW) explain why
the volumes had to be up-scaled before being sent to the MS.