Review and Testing of Steffes Electric Thermal Storage Unit with Grid‐Interactive Frequency Regulation Authors(s): Richard Wies, Associate Professor, PI, University of Alaska Fairbanks Nicholas Janssen, Research Assistant, University of Alaska Fairbanks Submitting Organization(s): University of Alaska Fairbanks Institute of Northern Engineering PO Box 755910 Fairbanks, AK 99775‐5910 Prepared for: Intelligent Energy Systems, LLC Dennis Meiners, Principal Under EETF Grant #7310049: “Small Community Self‐Regulating Grid” Subrecipient/Funding Organization: Alaska Energy Authority 813 West Northern Lights Blvd. Anchorage, AK 99503 Last Revised: Wednesday, January 22, 2014
52
Embed
Review and Testing of Steffes Electric Thermal Storage ...acep.uaf.edu/media/127462/Final_Report_ETS_Grid...Jan 22, 2014 · ETS masonry heaters use resistive heating elements to
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.
Transcript
Review and Testing of Steffes Electric Thermal Storage Unit with Grid‐Interactive Frequency Regulation
Authors(s):
Richard Wies, Associate Professor, PI, University of Alaska Fairbanks
Nicholas Janssen, Research Assistant,
University of Alaska Fairbanks
Submitting Organization(s): University of Alaska Fairbanks
Institute of Northern Engineering PO Box 755910
Fairbanks, AK 99775‐5910
Prepared for: Intelligent Energy Systems, LLC
Dennis Meiners, Principal Under
EETF Grant #7310049: “Small Community Self‐Regulating Grid”
Subrecipient/Funding Organization:
Alaska Energy Authority 813 West Northern Lights Blvd.
Anchorage, AK 99503
Last Revised: Wednesday, January 22, 2014
ii
Cover Photo Description and Credits
Cover: Steffes ETS Grid Frequency Regulation Test Setup at UAF Electric Power Lab Photo Credit: Nicholas Janssen, UAF INE Graduate Research Assistant
iii
Table of Contents
Cover Photo Description and Credits ............................................................................................. ii
Table of Contents ........................................................................................................................... iii
List of Figures ................................................................................................................................ iv
List of Tables .................................................................................................................................. v
Acknowledgements ........................................................................................................................ vi
Summary of Findings ................................................................................................................... viii
CHAPTER 1 – INTRODUCTION AND BACKGROUND .......................................................... 9
Problem Statement and Objective ............................................................................................... 9
Scope of Work ............................................................................................................................ 9
Grid Frequency Regulation ....................................................................................................... 10
Electric Thermal Storage (ETS) ............................................................................................... 12
Control Methodologies ......................................................................................................... 13 Thermostatic Control ........................................................................................................ 13 Charge Level Control ........................................................................................................ 13 Electric Load Control ........................................................................................................ 14
Figure 1: Example energy balance of a high-penetration wind-diesel power system. ................. 11 Figure 2: Typical ETS room unit [2]. ........................................................................................... 12 Figure 3: Grid-Interactive ETS (GETS) controller proposed response. ....................................... 14 Figure 4: Laboratory setup circuit diagram. ................................................................................. 16 Figure 5: Steffes ETS laboratory test setup showing the major components. .............................. 17 Figure 6: Steady state GETS system response. (No manual adjustments). .................................. 21 Figure 7: System response to a frequency swell (field perturbed). ............................................... 21 Figure 8: System response to dips in frequency (field perturbed). ............................................... 22 Figure 9: Typical observation with real power peaks shown (no field adjustments). .................. 23 Figure 10: Switching delays between zero and full load (no field adjustments). ......................... 24 Figure 11: Switching delays between 3/4 load and full load (no field adjustments). ................... 24 Figure 12: Visible time delays in the fast switching of small load. .............................................. 25 Figure 13: Visible time delays in the fast switching of medium load. ......................................... 26 Figure 14: Data group “Nov01”. No field adjustments, 23 Ohm load. ........................................ 39 Figure 15: Data group “Nov02”. No field adjustments, 23 Ohm load. ........................................ 40 Figure 16: Data group “Nov03”. Field adjustments made, 23 Ohm load. .................................... 41 Figure 17: Data group “Nov04”. Field adjustments made, 23 Ohm load. .................................... 42 Figure 18: Data group “Nov05”. Field adjustments made, 23 Ohm load. .................................... 43 Figure 19: Data group “Nov06”. No field adjustments, 30 Ohm load. ........................................ 44 Figure 20: Data group “Nov07”. Field adjustments made, 60 Ohm load. .................................... 45 Figure 21: Data group “Nov08”. Field adjustments made, 60 Ohm load. .................................... 46 Figure 22: Data group “Nov09”. No field adjustments, 60 Ohm load. ........................................ 47 Figure 23: Data group “Nov10”. No field adjustments, 60 Ohm load. ........................................ 48 Figure 24: Data group “Nov11”. Field adjustments made, phases ‘b’ and ‘c’ open. ................... 49 Figure 25: Data group “Dec01”. No field adjustments, phase ‘b’ unloaded. ............................... 50 Figure 26: Data group “Dec02”. Field adjustments made, phase ‘b’ unloaded. ........................... 51 Figure 27: Data group “Dec03”. Field adjustments made, phase ‘b’ unloaded. ........................... 52
v
List of Tables
Table 1 - GETS Controller Response Times ................................................................................ 22
vi
Acknowledgements
The testing, measurements, and observations of the Steffes Electric-Thermal Storage unit
with grid-interactive frequency regulation reported in this document were performed by the
Institute of Northern Engineering, University of Alaska Fairbanks under subcontract from
Intelligent Energy Systems, LLC with an Emerging Energy Technology Fund (EETF) Grant
#7310049: “Small Community Self‐Regulating Grid” from the Alaska Energy Authority.
Richard W. Wies, Associate Professor of Electrical and Computer Engineering, University of
Alaska Fairbanks, was the principal investigator. The second contributing author of this report is
Nicholas Janssen, Graduate Research Assistant.
The testing and measurements were performed in the Electric Power Systems Laboratory at
the University of Alaska Fairbanks under the direct supervision of Associate Professor Richard
Wies using available university equipment and supplies.
vii
Abstract
This report documents the findings of a laboratory analysis of a novel grid-interactive
electric-thermal storage (GETS) controller installed in a Steffes 2102 electric-thermal storage
(ETS) unit. The objective of this analysis was to document and describe the frequency response
of an isolated grid driven by a rotating prime mover with an ETS unit connected as a self-
regulating load. The controller is said to respond to changes in grid frequency by activating and
deactivating individual resistive heating elements in order to maintain a balance of real power in
the system. Measurements taken on a laboratory setup confirm that the unit is responding to
changes in grid frequency. The time series data are presented and discussed, though conclusions
about the performance of a network of ETS units in a full-scale grid application are difficult to
make without the development of a simulation model and testing in an Alaska village with a
hybrid wind-diesel system.
viii
Summary of Findings
The ultimate question this project sought to answer was, “Does the grid-interactive electric-
thermal storage (GETS) controller respond to changes in grid frequency in an isolated grid?”
The response of a single ETS unit with the GETS controller to manual frequency changes
was demonstrated in the laboratory using a standalone AC synchronous generator driven by a
DC motor. The results of this testing showed that the controller does respond to deviations from
60 Hz. The collected data provided some insight into the rate and nature of the controller
response.
The GETS controller responded to changes in system frequency f as described below:
1) f < 56.0 Hz: unit shut off with Error 20 due to under frequency condition,
2) 56.0 Hz f 60 Hz: unit stayed active with all resistive elements off,
3) 60 Hz < f < 60.5 Hz: unit cycled on and off resistive elements trying to overcome
generator inertia to correct system frequency to 60 Hz, and
4) f 60.5 Hz: unit remained on with all for resistive elements active.
Analysis of the laboratory results showed that rises in system frequency above 60 Hz were
met by additional load applied by the GETS controller. Similarly, when the unit was loaded and
the frequency dropped below 60 Hz, the controller responded by switching the load back off.
The GETS controller acting on its own was never able to adjust the output to a steady 60 Hz
waveform when the system frequency was between 60 Hz and 60.5 Hz. It was also not clear in
the data that there were four discrete resistive elements turning on and off at a rate proportional
to the frequency shift. The unit tended to alternate between two discrete and disparate loading
levels when attempting to make a frequency correction. Ramp rates and other patterns, such as
obvious time delays, were evident in the response of the ETS heater and are analyzed and
discussed in more detail within the report.
The authors recommended further study into the effects of a large network of self-regulating
ETS units on isolated medium and high-penetration wind-diesel grid stability. Both computer
modelling and in-situ testing would reveal some of the inherent challenges of regulating grid
frequency with multiple self-sensing ETS units with GETS controllers in an isolated grid.
9
CHAPTER 1 – INTRODUCTION AND BACKGROUND
Problem Statement and Objective
Integration of wind turbine generators into electrical grids in Alaska villages could alleviate
the high cost of imported diesel fuel and the associated electric energy prices. However, stability
problems are highly prevalent in systems that approach a medium to high level of wind energy
penetration. Secondary loads are necessary in order to absorb swells in wind energy that result in
grid frequencies above 60 Hz. Ideally, the central power plant, comprised of Diesel Electric
Generation (DEG), would be responsible for grid frequency regulation so long as the total wind
generation does not exceed demand. One proposed method for utilizing excess wind energy for
heating and grid frequency regulation is to power a network of Electric-Thermal Storage (ETS)
devices in village residences with thermostatic heating, charge, and electric load controllers.
ETS masonry heaters use resistive heating elements to charge a “core” of ceramic storage
bricks when the energy is either cheaper or more available. When the home’s thermostat calls for
heat, a blower is activated and air is forced through the ceramic core, and then into the room.
This provides space heat to village residences using excess wind energy that would otherwise
have no application. A grid-interactive ETS (GETS) controller has been added to the unit tested
in this study which was designed to sense grid frequency, and then add resistive electric load to
aid in stabilizing the grid frequency at 60 Hz when increases in wind generation or decreases in
load occur.
The objective of this study was to review, test, and evaluate a Steffes 2102 ETS with a grid-
interactive controller as a secondary load for frequency regulation in medium to high-penetration
wind-diesel grids in Alaska villages. A scope of work for this study is provided in the next
section.
Scope of Work
The scope of work for University of Alaska Fairbanks (UAF) Institute of Northern (INE)
under subcontract with Intelligent Energy Systems (IES) under an Emerging Energy Technology
Fund (EETF) Grant #7310049 “Small Community Self-Regulating Grid,” from the Alaska
Energy Authority (AEA) includes the review and testing of a single Steffes 2102 ETS unit with a
grid-interactive controller. The original grant included a provision to test and verify the operation
10
of the unit with the proposed self-regulating grid controller at the Alaska Center for Energy and
Power (ACEP) Power Systems Integration Laboratory (PSIL)) located on the UAF campus. An
amendment to that provision was proposed and adopted to conduct the testing in the UAF
Electric Power Systems Research laboratory given the small capacity of the single unit and the
need for more than a single day lease for testing.
The steps used to test and verify the operation of the Steffes 2102 ETS unit with the GETS
controller include the following.
1) Procure a Steffes ETS from Steffes through IES with the proposed GETS controller.
2) Power the unit with an isolated variable frequency three-phase AC generator.
3) Take the unit with the GETS controller through a series of positive and negative load
ramps in steps to observe the frequency response to load changes.
4) Record all pertinent data including voltage, current, power, and frequency.
5) Analyze the data to determine if the single Steffes ETS unit with the GETS controller
responds sufficiently to changes in load to regulate the system grid frequency.
Before describing the laboratory test setup, testing procedure, and measurement methods,
some background on grid frequency regulation and ETS control methodologies will be
presented.
Grid Frequency Regulation
Maintaining a constant grid frequency, particularly in an isolated high-penetration wind-
diesel grid, is the challenge of establishing a real time balance between real power generation
and demand. The three main variables are the wind power, the consumer electrical demand, and
the consumer heat demand.
If the amount of real power generation exceeds the demand at any given instant, the resulting
power imbalance will tend to accelerate any rotating masses in the system. This will result in
frequency swings above the nominal 60 Hz. Pseudo-instantaneous increases in wind power
generation and decreases in load are the primary causes.
In order to prevent these frequency swings, a high-penetration wind-diesel hybrid power
system must make use of a secondary (or dump) load (see Figure 1) in order to absorb any excess
11
wind power. A resistive load bank or electric boiler may be used for this purpose. However, if
there is no demand for this heat, the energy is wasted. Some systems may use an electrical
energy storage system, such as a lead-acid battery, but these add complexity and cost. There is
also no way for a fully-charged battery to absorb any additional energy.
Figure 1: Example energy balance of a high-penetration wind-diesel power system.
Some cold-climate communities, such as those located in rural Alaska and Canada, may
benefit from the use of medium to high-penetration wind-diesel systems and the dumped waste
heat. It is common for some remote diesel plants in these regions to operate a waste heat
recovery loop and provide space heating utility to a nearby community building [1]. However,
this solution hinges upon the geographic proximity of the diesel plant to a sizeable heat demand.
A wind-diesel power grid that could simultaneously maximize the use of available wind power
while also satisfying the heating demand of the rest of the community would be highly valuable
to the residents.
12
Electric Thermal Storage (ETS)
Masonry ETS units (see Figure 2) are space heaters consisting of a mass of ceramic bricks
(12) heated by resistive heating elements (14) [2]. When the space demands additional heat, a fan
(16) turns on and circulates room air (10) through the heated bricks. These units are designed to
operate in regions with “off-peak” electricity rates in order to help level load demand during
heating seasons (demand-side management). Since they are capable of delivering heat on
demand, storing heat energy, and absorbing electrical energy, they may be a well-suited
secondary load for isolated high-penetration wind-diesel systems.
Figure 2: Typical ETS room unit [2].
The ETS unit provided for review and testing for this application was the Steffes 2102. The
unit is powered by common 120 VAC with 4 sets of resistive elements for a total absorptive
capacity of 1.3 kW with manufacturer’s data and specifications as shown in Appendix A [2]. In
order to test the ETS unit, the control methodologies for delivering heat, storing heat, and
controlling electric load were reviewed and understood as described in the following sections.
13
Control Methodologies
A central challenge of implementing a network of ETS units in a high-penetration wind-
diesel grid is unit control. Much of the focus with this application of ETS units is on
determination of the optimal number of storage units to satisfy the heat demand while ensuring
enough capacity to consistently store excess wind power [3]. However, for delivering heat on
demand, storing heat energy, and absorbing electrical energy, the ETS units should have
controllers to optimize wind-diesel system performance and use of excess wind energy. The
three types of control related to delivering heat on demand, storing heat energy, and absorbing
electrical energy, respectively, are: 1) thermostatic, 2) charge level, and 3) electric load.
Thermostatic Control
All ETS units are controlled by a thermostat which ensures room temperature is maintained
by turning on a fan to pull air through the heated bricks and force heated air into the interior of
the residence. Since indoor temperatures are directly proportional to outdoor temperatures,
thermostatic control is dependent on the availability of stored heat (charge level control).
Charge Level Control
Charge level control of ETS units involves adjusting the target charge level and charging
rate. Two common methods of charge control used on the ETS units are: 1) charge level
adjustment with outdoor temperature, and 2) full charge level set point with off-peak charging.
In the first method the charge level is adjusted based on the outdoor temperature [4]. Lower
outdoor temperatures indicate a higher demand for space heat. This ensures the unit is adequately
charged in the case of a “cold snap”, when the heat is needed most.
In the second method a full charge level set point is maintained, but charging is limited to
off-peak hours [4]. This ensures there is always enough heating potential to satisfy the next day’s
demand. However, maintaining a high charge level is not necessarily beneficial for the utility. It
is in their best interest that a certain portion of the ETS units remain empty. The challenge is to
determine how many ETS units of a certain power level and storage capacity are needed in order
to provide benefit to both the consumer and the grid.
From the standpoint of ETS charging, the utility needs sufficient storage capacity in order to
absorb anticipated wind events. Maintaining an optimum charge level is imperative to both the
customer and the utility, since the aggregate system storage capacity must be sufficient in times
14
of high wind or low demand [3]. This is where electric load control can play a key role in
absorbing excess electric energy from wind such that grid frequency is maintained.
Electric Load Control
Electric load control of ETS units involves adjusting the number of active resistive heating
elements in response to a system frequency increase above 60 Hz. The additional system load
assists generation controls in maintaining the grid frequency. The increase in system frequency is
normally due to an increase in wind generation or a decrease in system load. Maintaining grid
frequency stability is critical to the operation of a high penetration wind-diesel system as
discussed in Chapter 1. Pursuant to the application of the ETS units in high-penetration wind-
diesel systems for maintaining grid frequency, a grid-interactive controller was developed by the
Steffes Corporation for the purpose of adding resistive elements (electric load) in response to
increases in frequency. This grid-interactive controller is the subject of this review and testing.
Grid-Interactive ETS (GETS)
A grid-interactive ETS (GETS) controller was developed which is said to sense the
frequency of the current powering the ETS unit, and activate resistive elements within the bricks
using controlled relays as the frequency increases above 60 Hz. The controller should
incrementally load or fully unload the unit in response to frequencies above and below 60 Hz,
respectively. The GETS controller was programmed to incrementally activate up to four resistive
elements to increase the electric load on the system in proportion to the frequency increase at a
rate of 25 % power capacity per 0.125 Hz between 60 and 60.5 Hz as shown in Figure 3.
Many of these ETS units with a grid-interactive control would be distributed throughout a
village power grid. In this study only a single ETS unit was tested to determine if the GETS
controller responded and at what rate to changes in frequencies above and below 60 Hz. The
question remains as to how a network of the ETS units would affect the dynamics of an isolated
high-penetration wind-diesel grid when individually responding to frequency increases above 60
Hz at the same time. Answering this question would be best suited for a dynamic simulation
model and an actual installation of these units in an isolated medium to high-penetration wind-
diesel system in an Alaska village.
Chapter 2 provides a detailed description of the laboratory test setup, testing procedure, and
measurement methods used to evaluate the performance of the single ETS unit with the grid-
interactive controller as supplied by the manufacturer. The results of the lab testing will then be
discussed in Chapter 3. Finally, in Chapter 4, recommendations will be provided and conclusions
drawn from our review and testing of the ETS unit with the grid-interactive controller.
16
CHAPTER 2 – TESTING AND MEASUREMENT METHODS
In order to evaluate the response of the GETS controller, a lab-scale three-phase 208-V mini
grid was established. Measurement devices for current and voltage also had to be installed along
with a way of sampling the waveforms at a moderately high frequency. Perturbations were
introduced to the system in order to simulate changes in wind power or electrical load. The
response of the GETS controller was subsequently measured throughout the duration of the
resulting system response. Of primary interest were frequency, voltage, and current.
Laboratory Test Setup
The laboratory setup used for testing the ETS unit is shown in Figures 4 and 5. The
synchronous machine (AC generator) is driven by a 5-kW, 120-V shunt-connected DC motor.
The 5 kW DC motor and synchronous machine are mechanically coupled by a common drive
shaft. Collectively they represent the DEG that would supply power to the village grid.
Figure 4: Steffes ETS laboratory test setup circuit diagram.
17
Figure 5: Steffes ETS laboratory test setup showing the major components.
18
The field currents of the DC motor and AC synchronous generator are independently
controlled by variable resistors. Adjustments in current to the DC motor field winding allow for
manual frequency control through shaft speed. Adjustments in current to the AC synchronous
machine field winding allow for manual voltage control. The loads connected to phases ‘b’ and
‘c’ of the AC synchronous generator are switchable resistive load banks. The Steffes 2102 ETS
unit equipped with the GETS controller is connected between phase ‘a’ and the neutral. Current
and voltage measurement instruments were placed on all three phases of the generator bus.
It should be noted that there are no automatic controls on either the voltage or frequency of
the system. Any changes to the system voltage and frequency had to be made manually by
adjustment of the AC synchronous machine field resistance and the shunt DC motor field
resistance, respectively.
Testing Procedure
The procedure used to test the GETS controller is outlined in the following steps.
1) Motor Startup – First, the shunt field and armature windings of the DC motor were energized.
As current was diverted from the field to the armature by increasing the field resistance, the shaft
speed approached 1200 RPM (synchronous speed for the AC generator). A handheld light-pulse
tachometer was used to confirm the shaft speed.
2) Generator Startup – Once the motor reached 1200 RPM, the synchronous generator field
current was gradually increased by decreasing the generator field resistance to bring the line-line
voltage up to 208 VAC, as measured by fixed instrumentation. Some back and forth adjustments
to both AC synchronous machine field and the shunt DC motor field resistors was needed before
a steady three-phase, 60 Hz, 208 VAC waveform was obtained.
3) Load Activation – The load was applied by closing the breaker and placing 23 ohms on phases
‘b’ and ‘c’ of the generator, with the Steffes 2102 GETS on phase ‘a’ (see Figure 5).
4) Field Adjustment – Having no automatic voltage regulator (AVR) or speed control, manual
adjustments had to be made with every change in load. Application of the initial load would result
in an immediate drop in frequency and voltage.
5) System Perturbation – Three cases were tested:
i) steady state response of the GETS controller (no field adjustments made),
ii) deliberate system perturbations by manual adjustment of the DC motor field, and
iii) deliberate system perturbations by manual removal of the initial load on phases ‘b’ and ‘c’.
19
Measurement of the voltage and current waveforms was performed to evaluate the subsequent
system dynamics throughout the duration of the resulting response.
Measurement Methods
Current and voltage measurements were taken on all three phases of the generator bus (see
Figures 4 and 5). Fluke i410 AC/DC Hall Effect current clamps (see Appendix B) with a 1 mV/A
output were used to measure current in all three phases. Tektronix P5200A high voltage
differential probes with 500X attenuation (see Appendix B) were placed between each phase and
the neutral point for a nominal L-N voltage of 120 Vrms.
Automatic measurements of all six channels were performed by using a National Instruments
PCI-6221 Data Acquisition System (DAQ) and breakout box with a LabView interface. The
outputs of the current and voltage probes were 256-bit 0-5 V signals, which were routed directly
to the breakout box and sampled at 6000 Hz.
Data post-processing was performed on the six channels of data (three phases of voltage and
three phases of current) using MATLAB in order to generate the following time series signals.
RMS Voltages [V] - Calculated in a sliding window of 600 samples (6 cycles).
RMS Currents [A] - Calculated in a sliding window of 600 samples (6 cycles).
Frequency [Hz] - The zero-crossings of the voltage waveforms in each phase were
interpolated in order to find the “instantaneous” value of the frequency.
Real Power [W] – The product of RMS voltage and RMS current was calculated
assuming unity power factor load.
The resulting time series signals for the three cases listed under Testing Procedure: 5) System
Perturbations were plotted as shown in the following chapter and used to make observations
about the response of the GETS controller to system frequency changes.
20
CHAPTER 3 – OBSERVATIONS AND RESULTS
Several observations about the response of the GETS controller were formulated upon
inspection of the results.
1) The GETS controller responded to adjustments in system loading based on measured
frequency with discrete levels of applied or removed load.
2) The controller responded to swells in frequency by adding load until fully loaded.
3) The controller responded to dips in frequency by removing load until fully unloaded.
4) Time delays were present in the controller response.
a) Delays of 1.00 s are observed when the controller switches from no load to full load.
b) Delays of 2.00 s are observed when the controller switches from ¾ load to full load.
5) Fast switching (0.10 s) of small to medium loads was observed.
a) Small (< 1 A) loads switch off for 0.10 s.
b) Medium (≈ 3 A) loads switch on for 0.10 s.
6) The controller cycled the ETS loads on and off at two discrete levels when the system
frequency was between 60 and 60.5 Hz.
Figure 6 shows a portion of the system response with no manual adjustments or interaction of
any kind. The changes in frequency and current were solely a result of the GETS controller
switching load on and off. Phase ‘a’ shows the current drawn by the Steffes heater, which
appears to rise and fall between about 2 A and 10 A. The range of frequency swings was from
about 57 Hz to just over 60.5 Hz.
In order to test system response time, deliberate perturbations in the DC motor field were
performed to adjust shaft speed and thus the output frequency of the AC generator. Initially, the
system frequency in Figure 7 was at 59.95 Hz with no ETS load. The DC motor field current
adjustment to increase shaft speed caused a rise in frequency to 60.25 Hz at t = 9.98 s. No load
was applied until t = 10.53 s: a delay of 0.55 s. By this time, the system frequency had risen to
well above 63 Hz. Also note that the initially applied load was only 5 amps (50% loading). By
the time all elements were activated (t = 11.54 s), a delay of 1.51 s had passed since the increase
in frequency was initiated.
21
Figure 6: Steady state GETS system response. (No manual adjustments).
Figure 7: System response to a frequency swell (field perturbed).
4 6 8 10 12 14
57
58
59
60
61
Time (s)
Fre
quen
cy (
Hz)
Frequency of Voltages
Fan
FbnFcn
4 6 8 10 12 14
2
4
6
8
10
Time (s)
RM
S C
urre
nt (
A)
RMS Line Currents
IaRMS
IbRMSIcRMS
10 10.5 11 11.5 12 12.5
58
60
62
64
66
68
X: 10.03Y: 60.5
Time (s)
Fre
quen
cy (
Hz)
Frequency of Voltages
X: 9.984Y: 60.25
X: 9.809Y: 59.95
Fan
FbnFcn
10 10.5 11 11.5 12 12.5
2
4
6
8
10
12
14
X: 11.54Y: 6.018
Time (s)
RM
S C
urre
nt
RMS Line Currents
X: 10.53Y: 1.745
IaRMS
IbRMSIcRMS
22
Next the controller response to under-frequency was observed. Figure 8 shows the system at
a stable 60.5 Hz with the ETS unit fully loaded. As the field on the DC motor was adjusted to
reduce the shaft speed, the AC generator frequency began to dip. At t = 2.16 s, the frequency had
already dipped to 60.0 Hz. At this point, the ETS unit should have been fully unloaded, but did
not do so until 1.346 s later at t = 3.51 s.
Figure 8: System response to a frequency dip (field perturbed).
In summary, Table 1 shows the controller’s frequency response times and ramp rates for
frequency swells and dips as observed in Figures 7 and 8, respectively. “Time to first response”
was the difference in time between the first point at which the controller could take any action
and the point at which any action was observed. “Time to appropriate response” was the time it
takes the controller to respond with an appropriate amount of loading/unloading according to the
specifications in Figure 3. The implications of these delays are discussed later in this section.
Table 1 - GETS Controller Response Times
Frequency swell Frequency dip Time to first response (s) 0.58 0.49
Time to appropriate response (s) 1.51 1.35 Ramp rate (Hz/s) 8.00 2.67
1 1.5 2 2.5 3 3.5 4 4.5
56
58
60
62
X: 2.16Y: 60
Time (s)
Fre
quen
cy (
Hz)
Frequency of Voltages
Fan
FbnFcn
1 1.5 2 2.5 3 3.5 4 4.50
5
10
15
X: 3.506Y: 8.832
Time (s)
RM
S C
urre
nt
RMS Line Currents
IaRMS
IbRMSIcRMS
23
As seen in Figure 6, the system cycled between at least two discrete levels of loading (2 A
and 10 A). Since the system has four elements, more than just these two discrete levels could be
seen in other scenarios as would be expected. However, it was not obvious in the data that four
discrete, equally-sized resistive elements were switching. In the absence of an AVR, it was better
to view these loading levels in terms of real power (product of current and voltage) rather than as
currents (A).
The following (Figure 9) shows a typical example of one of the eleven recorded scenarios.
Peaks in real power could only be seen at 200 W and 1100 W corresponding to no load (f 60
Hz) and full load (f 60.5 Hz), respectively. Power peaks between 200 W and 1100 W were not
visible here, and were scarcely present in some of the other recorded scenarios (see Appendix C).
Figure 9: Typical observation with real power peaks shown (no field adjustments).
Also present in the frequency response data were what appear to be fixed time delays in the
switching of the loads. When the controller switched from zero to full load, the load was
maintained for a minimum of 1.00 s (see Figure 10). There also appeared to be a minimum
amount of time of about 2.00 s maintained at no load. In addition, when the controller switched
between ¾ and full load (Figure 11), the load was maintained for 2.00 s.
-200 0 200 400 600 800 1000 1200 1400 16000
2
4
6
8x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Pow
er (
W)
Real Power (W)
Pa
Pb
Pc
24
Figure 10: Switching delays between zero and full load (no field adjustments).
Figure 11: Switching delays between 3/4 load and full load (no field adjustments).
4 6 8 10 12 14 16
56
58
60
62
Time (s)
Fre
quen
cy (
Hz)
Frequency of Voltages
Fan
FbnFcn
4 6 8 10 12 14 160
5
10
15
X: 4.338Y: 9.53
Time (s)
RM
S C
urre
nt
RMS Line Currents
X: 3.338Y: 2.044
X: 13.34Y: 9.707
X: 12.34Y: 1.869
X: 7.332Y: 10.29
X: 6.332Y: 1.501
X: 10.34Y: 9.214
X: 9.34Y: 1.642
IaRMS
IbRMSIcRMS
2 4 6 8 10 1255
60
65
70
Time (s)
Fre
quen
cy (
Hz)
Frequency of Voltages
Fan
FbnFcn
2 4 6 8 10 12
2
4
6
8
10
12
X: 10.51Y: 8.538
Time (s)
RM
S C
urre
nt
RMS Line Currents
X: 9.483Y: 9.314
X: 7.483Y: 8.062
X: 6.505Y: 10.02
X: 4.505Y: 8.349
X: 3.483Y: 10.37
X: 1.483Y: 8.393
IaRMS
IbRMSIcRMS
25
Also seen in the data were relatively fast switching transients of small load currents (< 1 A)
that appeared to be the control unit’s load demand (see Figure 12). These switching transients
were likely a characteristic of the control units design with the small load currents OFF for a
predetermined time of 0.10 s. This cycling on and off represented the control unit switching to
attempt to engage load in response to an over-frequency condition between 60.0 Hz and 60.5 Hz.
The control unit was never able to correct the AC synchronous generator frequency to 60 Hz
which was manually set to 60.3 Hz through the adjustment of the shunt field of the DC drive
motor.
Figure 12: Visible time delays in the fast switching of small load.
In addition, there was a slightly larger (≈ 3 A) load that was turned ON at the same rate (see
Figure 13). However, this load was kept ON for exactly 0.10 s. This cycling on and off
represented the control unit switching to attempt to disengage load in response to an under-
frequency condition (f < 60.0 Hz).
2.6 2.7 2.8 2.9 3 3.1 3.2 3.3 3.459
60
61
62
Time (s)
Fre
quen
cy (
Hz)
Frequency of Voltages
Fan
FbnFcn
2.6 2.7 2.8 2.9 3 3.1 3.2 3.3 3.4
8
9
10
11
X: 2.69Y: 9.76
Time (s)
RM
S C
urre
nt
RMS Line Currents
X: 2.79Y: 9.163
X: 2.97Y: 10.08
X: 3.07Y: 9.496
X: 3.251Y: 10.54
X: 3.351Y: 9.883
IaRMS
IbRMSIcRMS
26
Figure 13: Visible time delays in the fast switching of medium load.
A comprehensive set of plots at full test time length for all scenarios tested were produced as
shown in Figures 14 through 27 of Appendix C.
Conclusions and the authors’ recommendations for future work based on results and
observations presented above are discussed in the following chapter.
3.2 3.4 3.6 3.8 4 4.258.5
59
59.5
60
60.5
61
Time (s)
Fre
quen
cy (
Hz)
Frequency of Voltages
Fan
FbnFcn
3.2 3.4 3.6 3.8 4 4.2
1
2
3
4
5
Time (s)
RM
S C
urre
nt
X: 3.926Y: 4.373
X: 3.608Y: 1.574
X: 3.507Y: 4.389
X: 3.407Y: 1.491
X: 3.29Y: 4.421
X: 3.19Y: 1.368
X: 3.708Y: 4.421
X: 3.826Y: 1.698
RMS Line Currents
IaRMS
IbRMSIcRMS
27
CHAPTER 4 – CONCLUSIONS AND RECOMMENDATIONS FOR FUTURE WORK
Conclusions and recommendations for future work based on our review and testing of the
Steffes 2102 ETS with the GETS controller are provided in the following sections.
Conclusions
In this study an individual Steffes 2102 ETS with a GETS controller was reviewed and tested
for response to changes in frequency above and below 60 Hz. The results of the laboratory test
revealed that the GETS controller responded to changes in frequency. In general our findings
suggest that although the Steffes 2102 ETS with the GETS controller responds to frequencies
above 60 Hz by switching on resistive load, we cannot confirm that it responds as indicated with
incremental load based on the magnitude of the grid frequency shift with our test setup.
Limitations in our test setup prevented the assertion of any conclusions about how a network of
these units would respond on a larger grid.
There were three major differences between the test setup and a realistic isolated medium to
high-penetration wind-diesel grid. These differences should be considered when drawing any
conclusions about the data presented in this study.
First, there was no automatic speed controller or voltage regulator on the motor/generator
system. These devices are present on all DEGs and are responsible for controlling grid frequency
and voltage. Without these devices in the laboratory test setup, we could only see how the GETS
controller takes action as a result of certain frequency excursions. We could not make any
conclusions about its ability to assist with grid frequency regulation.
Second, the capacity of motor/generator in proportion to the load capacity of the Steffes 2102
unit was much larger than a village power system with DEGs. By the time a load correction was
made by the controller, the system had already slowed down or sped up well beyond the
allowable frequency limits of a typical grid. This gives the appearance of poor controller
performance (see Figure 6), however, the controller’s response in a situation with more base load
and generation present could improve.
Finally, we were only able to test a single unit on a single phase of the three-phase AC
synchronous generator. This architecture was very different from an actual village electrical
power system with a network of ETS units—each attempting to regulate grid frequency.
28
Understanding these differences was important for interpreting the results of this study.
Visible time delays were present (see Figures 10 and 11) which appeared to have negative
consequences on the laboratory test grid’s system frequency. These time delays could contribute
to grid instability in a larger medium to high-penetration wind-diesel system where it is
important that not all ETS units respond at the same time to pseudo-instantaneous changes in
generation or load. However, the ETS units should respond faster than the DEG. Since the DEG
cannot tell the difference between primary consumer load and secondary ETS load, the ETS
units must unload before the DEG “picks them up” and makes them a part of the primary load.
In conclusion, based on the authors’ review and testing of a single Steffes 2102 ETS unit
with GETS control, while the unit response to frequency shifts was evident, we could not
confirm the incremental load versus frequency response with our test setup. However, based on
our results, recommendations for future testing of ETS units with grid-interactive control were
proposed as discussed in the next section.
Recommendations for Future Work
The concept of the grid-interactive control with multiple ETS units in an isolated wind-diesel
grid should be further investigated through in-situ testing and modelling to determine their effect
on system stability. Two courses of action are recommended in order to further validate the
concept of the ETS with a self-regulating GETS controller.
1) A network of these units needs to be installed on an actual village wind-diesel grid to
collect data and observe the system frequency response. It was not clear from our testing
how the timing of the units in response to frequency changes above and below 60 Hz
would affect system transient stability. System stability could be an issue with the GETS
control competing against the response of the DEG(s) speed control and other ETS units
with GETS control in the network.
2) Secondly, a modelling study should be conducted to improve our understanding of the
effects of time delays, loading levels, and ETS unit cooperation. Such a study would
require an accurate wind-diesel system model and additional knowledge of how the
GETS controller would respond to changes in wind generation from induction machines,
diesel generation from three-phase AC synchronous machines, and system load.
29
In closing, further study of these ETS units with GETS controllers for isolated medium to
high-penetration wind-diesel grids is required to determine their effectiveness in stabilizing grid
frequency.
30
REFERENCES
[1] Isherwood, W., Smith, J. R., Aceves, S. M., Berry, G., Clark, W., Johnson, R., Das, D., Goering, D., & Seifert, R. (2000). Remote power systems with advanced storage technologies for Alaskan villages. Energy: The International Journal, 25(10), 1005–1020.
[2] Steffes 2102 Specification and Data Sheets. Retrieved December 18th, 2013, from the
[3] Hughes, L. (2010). Meeting residential space heating demand with wind-generated
electricity. Renewable Energy: An International Journal, 35(8), 1765–1772. [4] Bedouani, B. Y., Moreau, A., Parent, M. & Labrecque, B. (2001). Central electric
thermal storage (ETS) feasibility for residential applications: Part 1. Numerical and experimental study. International Journal of Energy Research, 25(1), 53–72. doi: 10.1002/1099-114X(200101)25:1<53::AID-ER610>3.0.CO;2-T
[5] Fluke i410 Current Clamp Specification and Data Sheet. Retrieved December 22nd, 2013,
Fluke Company website: http://www.fluke.com/fluke/inen/accessories/Current-clamps/i410.htm?PID=56301
[6] Tektronix 5200A High Voltage Differential Probe Specification and Data Sheet. Retrieved December 22nd, 2013, from the Tektronix Company website: http://www.tek.com/differential-probe-high-voltage
31
APPENDIX A: Steffes 2102 Specification and Data Sheets
Steffes 2102
32
33
34
35
APPENDIX B: Measurement Instrumentation Specification and Data Sheets
Fluke i410 AC/DC Current Clamp
Tektronix P5200A High Voltage Differential Probe
36
Fluke i410 Current Clamp
Specifications
Measurement type Hall sensor
Nominal current range 400 A, AC/DC
Continuous current range 1 A - 400 A AC/DC
Maximum Non-Destructive Current
400 A
Lowest measurable current
0.5 A
Basic Accuracy 3.5% + 0.5 A (% reading + floorspec)
Useable frequency DC - 3 kHz
Output level(s) 1 mV/A
Zero error adjustment Yes
Safety Specifications
Safety CAT III, 600 V
Maximum voltage 600 V
Mechanical & General Specifications
Warranty 1 year
Battery Life 9 V, 60 h
Maximum conductor diameter
30 mm 2 x 25 mm
Output cable length 1.6 m
Shrouded banana plugs Yes
37
Tektronix 5200A High Voltage Differential Probe
P5200A
Attenuation 50X / 500X
Differential Voltage 500X: ±1300 V 50X: ±130 V
Common Mode Voltage ±1300 V
Maximum Input Voltage-to-Earth 1000 V CAT II
Bandwidth 50 MHz
Differential Input Impedance 10 MΩ || 2 pF
Input Impedance between each Input and Ground 5 MΩ || 4 pF
Typical CMRR DC: >80 dB 100 kHz: >60 dB 3.2 MHz: >30 dB 50 MHz: >26 dB
Cable length 1.8 m
1. The differential voltage is the maximum measurable range between the (+) and (-) input leads of the probe. Beyond these limits, the output could be clipped.
2. The maximum common mode voltage and maximum input voltage-to-earth (RMS) are the maximum voltages that each input lead (+/-) can be from ground.
38
APPENDIX C: Additional Scenario Results for ETS GETS Controller Figure 14: Data group “Nov01”. No field adjustments, 23 Ohm load. ...................................39
Figure 15: Data group “Nov02”. No field adjustments, 23 Ohm load. ...................................40
Figure 16: Data group “Nov03”. Field adjustments made, 23 Ohm load. ...............................41
Figure 17: Data group “Nov04”. Field adjustments made, 23 Ohm load. ...............................42
Figure 18: Data group “Nov05”. Field adjustments made, 23 Ohm load. ...............................43
Figure 19: Data group “Nov06”. No field adjustments, 30 Ohm load. ...................................44
Figure 20: Data group “Nov07”. Field adjustments made, 60 Ohm load. ...............................45
Figure 21: Data group “Nov08”. Field adjustments made, 60 Ohm load. ...............................46
Figure 22: Data group “Nov09”. No field adjustments, 60 Ohm load. ...................................47
Figure 23: Data group “Nov10”. No field adjustments, 60 Ohm load. ...................................48
Figure 24: Data group “Nov11”. Field adjustments made, phases ‘b’ and ‘c’ open. ..............49
Figure 25: Data group “Dec01”. No field adjustments, phase ‘b’ unloaded. ..........................50
Figure 26: Data group “Dec02”. Field adjustments made, phase ‘b’ unloaded. ......................51
Figure 27: Data group “Dec03”. Field adjustments made, phase ‘b’ unloaded. ......................52
39
Figure 14: Data group “Nov01”. No field adjustments, 23 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
1
2
3
4
5
6x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2059.5
60
60.5
61
61.5
62
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
40
Figure 15: Data group “Nov02”. No field adjustments, 23 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
2
4
6
8x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2059
60
61
62
63
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
41
Figure 16: Data group “Nov03”. Field adjustments made, 23 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
2
4
6
8x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2059.5
60
60.5
61
61.5
62
62.5
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
42
Figure 17: Data group “Nov04”. Field adjustments made, 23 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
0.5
1
1.5
2
2.5
3x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 1 2 3 4 5 6 7 8 9 1057
58
59
60
61
62
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 1 2 3 4 5 6 7 8 9 100
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
43
Figure 18: Data group “Nov05”. Field adjustments made, 23 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
2
4
6
8x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2056
58
60
62
64
66
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
44
Figure 19: Data group “Nov06”. No field adjustments, 30 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
5
10
15x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 5 10 15 20 25 30 35 4057
58
59
60
61
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fbfc
0 5 10 15 20 25 30 35 400
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
45
Figure 20: Data group “Nov07”. Field adjustments made, 60 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
1
2
3
4
5
6x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2056
58
60
62
64
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
46
Figure 21: Data group “Nov08”. Field adjustments made, 60 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
2
4
6
8x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2058
58.5
59
59.5
60
60.5
61
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
47
Figure 22: Data group “Nov09”. No field adjustments, 60 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
1
2
3
4
5x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2059
60
61
62
63
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
48
Figure 23: Data group “Nov10”. No field adjustments, 60 Ohm load.
-200 0 200 400 600 800 1000 1200 1400 16000
1
2
3
4
5
6x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2058
59
60
61
62
63
64
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
49
Figure 24: Data group “Nov11”. Field adjustments made, phases ‘b’ and ‘c’ open.
-200 0 200 400 600 800 1000 1200 1400 16000
1
2
3
4
5x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2056
58
60
62
64
66
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
50
Figure 25: Data group “Dec01”. No field adjustments, phase ‘b’ unloaded.
-200 0 200 400 600 800 1000 1200 1400 16000
2
4
6
8x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 2 4 6 8 10 12 14 16 18 2056
58
60
62
64
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 2 4 6 8 10 12 14 16 18 200
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
51
Figure 26: Data group “Dec02”. Field adjustments made, phase ‘b’ unloaded.
-200 0 200 400 600 800 1000 1200 1400 16000
2
4
6
8x 10
4
Real Power (W)
Fre
quen
cy
Real Power Histogram
0 5 10 15 20 25 3056
58
60
62
64
66
Time (s)
Fre
quen
cy (
Hz)
Frequency
fa
fb
fc
0 5 10 15 20 25 300
500
1000
1500
Time (s)
Rea
l Pow
er (
W)
Real Power
Pa
Pb
Pc
52
Figure 27: Data group “Dec03”. Field adjustments made, phase ‘b’ unloaded.