Draft ERIGRID Laboratory
Chapter 4 - Report
SPEARHEAD Project
Rafael Oliva
July 17, 2018
Date Performed:
Partners:
Director:
Version:
May 2 - 11t, 2018
Jean Alinei
Ronan Blanchard
Luiz Lavado
1.1
Abstract
This document is a draft for the Chapter 4 work in the main Spearhead report
1
Experimental Work
This transnational access focused on the study and characterization of power electronics converters for
renewable energy. A series of experiments took place to test the power converters and the conversion
architecture on various conditions. This section describes the test plan, the procedures, the setups and
the data management of the project.
1.1
Test Plan
The tests were planned to go from more simple to more complex measurements. A summary of the
test setups is provided in table 1.
Table 1: General Summary of all the experimental setups
Experimental
Setup
Exp.
#
Experiment
Description
Setup I
DC/DC
Charge
controller
1
2
Ownwall DC/DC converter as charge controller
Ownwall DC/DC converter as charge controller with batteries, no source
3
Ownwall DC/DC converter as a battery charger (24 V)-
Setup II -
4
Ownwall Basic RTDS Tests
Setup III -
5
Ownwall RTDS + PAS5000 Tests
Setup IV -
6
7
Ownwall MPPT with RTDS
Dual Ownall Tests with RTDS
Setup V -
8
Ownwall Inverter Tests
Setup VI -
9
WindConverter Basic Tests / Software
Setup VII -
10
RTDS WindConverter Lo Level Tests
Setup VIII -
11
Motherboard + Ownwall Comm Tests
Setup IX -
12
Motherboard/ HMI Display + Ownwall Comm Tests
1
2
Tested Units Description
2.1
Ownwall DC/DC Converter
The OwnWall converter is a bidirectional, one-legged power converter capable of buck / boost operation,
which implements isolated measurements and various control algorithm capabilities through the use of
an integrated microcontroller. The firmware can be adjusted for the following functionalities, as stated
in Table 2.
Table 2: OwnWall DC/DC Software functionalities
Function
Nº.
Description
Voltage
control
1
Control through PID+Vref of low-side or high-side voltage
Current
control
2
Control through PID+Vref of low-side or high-side current
MPPT
control
3
Maximum power point tracking through a dedicated algorithm associated
with a voltage or current reference
Bidirectional
current flow
4
Current flow is bi-directional between high-side and low-side
The converter has a simple architecture, as shown in Figures 1 (Power Block) and 2 (Diagram and
implementation photographs). The microcontroller consists of a soldered Arduino Nano module, which
is programmed for greater speed directly in C using free tools (avr-gcc toolchain). Operational parameters can be stored in EEPROM, to configure the converter for different applications. To fully evaluate
the converter operation in test mode, a Backend application written in Python communicates at low
level with the controller through an USB-emulated serial port, and updates a Frontend web application
written in PHP which provides a fast update of the basic internal measurements and status of the
controller. In normal operation, the serial connection is routed through an RS-485 module inserted
under the Nano board, which communicates with the Motherboard at lower speed.
Figure 1: OnWall Converter Power Block Schematic.
2
Figure 2: Ownwall converter – Diagram and implementation.
2.2
WindConverter
Traditionally in Piggot-design turbines a fixed, six pulse rectifier (or Gratz bridge) is used to convert the
three-phase, variable frequency output of the PM generator to DC for battery charging, pump operation
or direct grid-tied inverter supply. Although widely used and extremely reliable, this passive rectification
scheme has a few disadvantages that can be addressed with newer configurations. The issues to be
dealt with in fixed diode bridges are as follows:
1. Need of an additional controller to dump excess power, avoiding overcharge of the battery bank.
2. Lack of control on the output voltage, which is directly linked to the generated voltage.
3. Low efficiency, because of forward voltage drop. Even when using shottky diodes instead of conventional ones, 0.5V voltage drop on each phase @ typical 30A means 45W of overall loss.
4. No aerodynamic optimization can be performed, and no direct possibilities of Maximum Power Point
Tracking (MPPT) conducting to reduction of power losses.
5. Harmonics injection from the generator, creating vibrations, and lowering efficiency.
Many of these issues can be addressed using the principle of active rectification. In recent years a great
number of industrial applications use this principle for 3-phase Power Factor Correction (PFC), with
the converter working with fixed voltage and fixed grid frequency (50/60Hz) levels. Other designs,
such as aircraft electrical generators [Uros, 2014] operating at higher, variable frequencies (350 to 800
Hz) are more similar to the requirements of a small wind PM generator, in which the control needs
to adapt to variable voltage and frequency from the generator. These control schemes are based on
cheap and reliable DSP or DSP-like microcontrollers, which can implement complex calculations and
transformations, and include dedicated PWM output modules to ease implementation.
There are a number of possible architectures for active rectification.
In the case of Windconverter, a three phase boost architecture was selected, as depicted in Figure 3 .
First main constraint of this topology is that DC bus voltage needs to be higher than maximum line to
line voltage of input generator.
This is necessary in order to obtain full control over the rectifier switches. This topology was selected
3
due to its great simplicity. Unless otherwise specified, balanced input voltages will be assumed during
the analysis of this topology.
Figure 3: Three-Phase Boost Rectifier – Basic Diagram [Uros,2014]
To simplify the design of the active rectifier, a series of transformations are applied to the three-phase
acquired signals, supposing a balanced system in direct sequence. These are usually referred to as the
α β-dq transformations, which basically convert a three-phase time-variant signal into a two-dimension
time-varying vector signal (α β-transformation) , and then into a time-invariant signal.
This is achieved through a new reference frame (dq) which rotates together with the same angular
velocity as the vector. The resulting projection of the vector on the new axes is always constant and
therefore the time variance of the AC variables is eliminated.
This transformation is called α β – dq. The implementation of this transformation, in order to obtain
true time invariance of the transformed abc domain variables, requires the introduced dq frame to be
somehow always fixed to the reference vector from the α β domain. Therefore, the dq frame should
rotate at the same speed as the reference vector. This is achieved through the design of a PLL or
Phase-Locked-Loop, which adjusts the frequency/angle to the varying frequency of the generator, thus
keeping the control model fully functional.
The diagram of the three-phase active boost rectifier with the implemented α β-dq transformations,
PLL and EMI reduction filter is shown in Figure 4.
Figure 4: Three-Phase Boost Rectifier with control [Uros,2014]
4
The obtained dq model behaves like a DC circuit and thus in general simplifies analysis and design of
the regulator Also, it provides independent control of active and reactive power flow with d-channel
controlling active, and q-channel reactive power flow. If unity PF is required, it can be obtained simply
by forcing iq to zero through duty cycle dq and the amount of active power transfer is then determined
by id through duty cycle dd .
The Windconverter prototype schematic (Power block) for this system is shown in Figure 5. The
controller used is placed on a separate board and uses a dsPIC 16-bit microcontroller. A 3D rendering
can be seen in Figure 6.
Tested versions of this board (v1.1) used the integrated ADC on the dsPIC controller, while the new
version which was not ready for testing in Greece incorporates two separate MCP3204 ADC units, and
GND isolation for better noise performance.
Figure 5: Windconverter three-phase boost active rectifier - Power Schematic
Figure 6: Windconverter 3D rendering of the designed board
5
The power rating of this prototype is 500W DC, ideally matching a 2.40 m, 24V piggot wind turbine.
The main purpose of the designed prototype is to simplify the development of the control software by
implementing plenty of sensors, and being able to select them by the jumpers present on board.
It is designed to boost up the voltage between 48V and 55V, corresponding to a 48V battery bank.
It incorporates Allegro ACS712 isolated sensors, which can sense bidirectional AC current avoiding
complex shunt measurements.
The switches are implemented with IRF540 MOSFETs, which have a medium RdsON, low losses and
high current flowing capacity. Also since TO-220 packages allow back side heatsinks screwed on board,
the cheapest extruded heatsink solution can be used, as shown in Figure 7.
Figure 7: Windconverter side view of heatsink and power components on designed board
Additionally, parallel DPak Schottky Diodes are implemented because of their better performances
relative to the embedded MOSFET diodes, giving lower forward voltage drop for better efficiency
transient response. RC Snubber filters are added in parallel with the main switches to improve the
transients in MOSFET commutation and improving their lifespan.
The switches are driven by simple and low-cost IR2181 MOSFET drivers (not shown in Figure 5), which
provide 2A driving current required for IRF540 both on high and low side.
Figure 8: Windconverter implemented board tested in Greece
2.3
Motherboard WEMB1
The system proposal of the Modular Power system is shown in Figure 9 . The focus was on a modular
approach, allowing connection of specific modules adapted to a given application, controlled by a
6
common central controller. Each module (1) has its own microcontroller and is adapted to a specific
kind of power conversion (PV, WindPower, etc), while the central controller (3) provides a stabilized
feeder WEFB1 (2), interconnection facilities (5) and a HMI or Human-Machine Interface (6) to interact
with the final user. Items (2), (3), (6) and part of (5) reside on the Power Motherboard WEMB1 which
was designed by J. Alinei, assembled at LyR Ing. and AEA/UNPA during 2017 and tested in Athens
during the International Access 2018.
Figure 9: WEMB1 Components - Block Diagram
The motherboard WEMB1 design can be seen in the 3D render of Figure 10. This board has sections
for implementing the required blocks as indicated in the diagram of Figure 9, (3) removable Main
Arduino Uno /Arduino Mega controller board (left) and expansion shields, (2) removable Main Feeder
WEFB1 board (right), and (5) interconnection facilities including an RS485 interface and Feeder output
connector. The HMI (6) will depend on application but will be resolved with one or more (stacked)
Arduino Shields. A battery sensing and also a LVD (low voltage disconnect) pair of circuits are also
included within WEMB1.
Figure 10: WEMB1 Design 3D Render
7
The schematics of the modified WEMB1 design can be found in Figure 11. The feeder version used
was prepared for 24 V only. The modifications to initial design are marked as A and B. These were
(1) marked as A on schematic, a protection diode Dp1 (1N5819 on Vin) to avoid a back circulation of
power in the not-improbable event of WEMB1 board unpowered and Arduino being updated. And (2)
marked as B on schematic, a pair of jumper switches were necessary because of shared Rx,Tx lines to
communicate via the MAX485/DS3695A and internally to update firmware, in the case of Arduino Uno.
For the systems tested in Athens, an Arduino Mega with more serial ports was used so the switching
was unnecessary.
Figure 11: WEMB1 Schematics - Modified
3
3.1
Test Setup Descriptions
Setup I - Ownwall DC/DC Converter as charge controller, Exp. 1
This first OwnWall converter setup consists of a programmable DC source connected in series with
a resistance as illustrated in Figure 12. This setup was performed with input voltages close to 24V,
measurement of voltages and currents adn connection of the OwnWall converter through the USB/Serial
connection to the Backend application, which allows for extensive visualization and changes in EEPROM
stored parameters for the controller.
8
Figure 12: Basic Setup (I.1) with DC Supply
3.2
Setup I - Ownwall DC/DC Converter as charge controller with batteries,
no source, Exp. 2
This OwnWall converter setup consists of a connection to a 24V battery bank, as in Figure 13, with no
source power supply connection.
Figure 13: Basic Setup (I.2) with Batteries
9
3.3
Setup I - Ownwall DC/DC Converter as battery charger (24 V), Exp. 3
This OwnWall converter setup consists of a connection as a charger for a 24V battery bank, as in Figure
14.
Figure 14: Basic Setup (I.3) as battery charger
3.4
Setup II - Ownwall Basic RTDS Tests, Exp. 4
This OwnWall converter setup consists of a connection to load through the RTDS, as in Figure 15.
Figure 15: OWnwall Setup in connection to RTDS
10
3.5
Setup III - Ownwall RTDS+PAS5000 Tests, Exp. 5
This OwnWall converter setup consists of a connection to load through the RTDS to a PAS5000, as in
Figure 16.
Figure 16: OWnwall Setup in connection to RTDS and PAS5000
3.6
Setup IV - MPPT with Ownwall RTDS+PAS5000 Tests, Exp. 6
This OwnWall converter setup consists of an MPPT control in firmware with PV emulation thru the
RTDS and PAS5000, as in Figure 17.
Figure 17: MPTT and PV curve with OWnwall through RTDS and PAS5000
11
3.7
Setup IV - Dual Ownwall RTDS+PAS5000 Tests, Exp. 7
This Dual OwnWall converter setup consists of tests thru the RTDS and PAS5000 to Resistive Load,
as in Figure 18.
Figure 18: MPTT and PV curve with OWnwall through RTDS and PAS5000
3.8
Setup V - Ownwall Inverter RTDS Tests, Exp. 8
This OwnWall converter setup includes inverter firmware tests thru the RTDS, as in Figure 19.
Figure 19: Inverter OWnwall tests with RTDS
12
3.9
Setup VI - Windconverter Basic Tests, Exp. 9
Figure 20: OWnwall Setup in connection to RTDS
4
Results and Conclusions
The atomic weight of magnesium is concluded to be 24 g mol−1 , as determined by the stoichiometry of
its chemical combination with oxygen. This result is in agreement with the accepted value.
Figure 21: Figure caption.
13
5
Discussion of Experimental Uncertainty
The accepted value (periodic table) is 24.3 g mol−1 ?. The percentage discrepancy between the accepted
value and the result obtained here is 1.3%. Because only a single measurement was made, it is not
possible to calculate an estimated standard deviation.
The most obvious source of experimental uncertainty is the limited precision of the balance. Other
potential sources of experimental uncertainty are: the reaction might not be complete; if not enough
time was allowed for total oxidation, less than complete oxidation of the magnesium might have, in part,
reacted with nitrogen in the air (incorrect reaction); the magnesium oxide might have absorbed water
from the air, and thus weigh “too much.” Because the result obtained is close to the accepted value it
is possible that some of these experimental uncertainties have fortuitously cancelled one another.
6
Answers to Definitions
1. The atomic weight of an element is the relative weight of one of its atoms compared to C-12
with a weight of 12.0000000. . ., hydrogen with a weight of 1.008, to oxygen with a weight of
16.00. Atomic weight is also the average weight of all the atoms of that element as they occur
in nature.
2. The units of atomic weight are two-fold, with an identical numerical value. They are g/mole of
atoms (or just g/mol) or amu/atom.
3. Percentage discrepancy between an accepted (literature) value and an experimental value is
experimental result − accepted result
accepted result
14