AN ABSTRACT OF THE DISSERTATION OF

Transcription

1

2 AN ABSTRACT OF THE DISSERTATION OF Terry Lettenmaier for the degree of Doctor of Philosophy in Electrical and Computer Engineering presented on May 22, 213. Title: Testing of Wave Energy Converters using the Ocean Sentinel Instrumentation Buoy Abstract approved: Annette R. von Jouanne Ocean testing of Wave Energy Converter (WEC) prototypes is necessary to facilitate commercial WEC development. The Ocean Sentinel Instrumentation Buoy, completed in August 212, provides a stand-alone load for WEC prototypes during ocean testing. The first part of this work was to develop the power conversion and data acquisition equipment installed on board the Ocean Sentinel that is used to test WECs. The functionality of this equipment was demonstrated during August-October 212 ocean testing of the half-scale WET-NZ wave energy converter. WET-NZ is an acronym for Wave Energy Technology-New Zealand. The second part of this work was to develop a maximum power point tracking (MPPT) algorithm that can automatically track optimum control settings of WECs as sea conditions change. During the WET-NZ tests, a cycling MPPT algorithm was developed that provided more effective tracking of the optimum WET-NZ control input than was possible with standard perturb and observe MPPT algorithms. The effectiveness of the cycling MPPT algorithm was verified later during more systematic testing of MPPT algorithms using a WEC emulator, developed as part of this project, to simulate WEC operation.

3 Copyright by Terry Lettenmaier May 22, 213 All Rights Reserved

4 Testing of Wave Energy Converters using the Ocean Sentinel Instrumentation Buoy by Terry Lettenmaier A DISSERTATION Submitted to Oregon State University In partial fulfillment of the requirements for the degree of Doctor of Philosophy Presented May 22, 213 Commencement June 213

5 Doctor of Philosophy dissertation of Terry Lettenmaier presented on May 22, 213. APPROVED: Major Professor, representing Electrical and Computer Engineering Director of the School of Electrical Engineering and Computer Science Dean of the Graduate School I understand that my dissertation will become part of the permanent collections of Oregon State University libraries. My signature below authorizes release of my dissertation to any reader upon request. Terry Lettenmaier, Author

6 ACKNOWLEDGEMENTS I would like to thank my committee members Dr. von Jouanne, Dr. Amon, Dr. Brekken, Dr. Raich, and Dr. Zaworski. I d like to give a special thanks to my advisor, Dr. von Jouanne, who has been wonderful to work with. I especially enjoyed working closely with Dr. Amon, Sean Moran, and the rest of the OSU team during the work on the Ocean Sentinel. I also want to thank our partners during the WET-NZ deployment, Callaghan Innovation, Power Projects Limited, and Northwest Energy Innovations, and acknowledge the special assistance that both Alister Gardiner and Lan Le-Ngoc at Callaghan Innovation have given me. Both Alister and Lan were very helpful during the WET-NZ tests, giving valuable insight into operation of the WET-NZ, and Alister also provided a thorough review of my thesis chapter covering that material. I also thank fellow student Bret Bosma for all of his help with wave energy converter modeling that I used for the later parts of my research. I acknowledge support for this work from the US Department of Energy (Award Number DE-FG36-8GO18179) for the Northwest National Marine Renewable Energy Center (NNMREC), and the State of Oregon Capital Funding Program. Finally, and most importantly, I want to thank my wife Laurie for all of the support that she has given me during the last three years.

7 TABLE OF CONTENTS Page 1 INTRODUCTION Background The Northwest National Marine Renewable Energy Center (NNMREC) The Ocean Sentinel Instrumentation Buoy Maximum Power Point Tracking (MPPT) THE OCEAN SENTINEL INSTRUMENTATION BUOY Introduction Instrumentation Power System Onboard Instrumentation and Equipment TRIAXYS Wave Measurement Buoy Umbilical Cable Data Acquisition and Telemetry Systems CompactRIO DAS and Control Watchman 5 DAS and Control Telemetry WEC Electrical Loading and Power Conversion WEC ELECTRICAL LOADING, POWER CONVERSION, AND DATA ACQUISITION EQUIPMENT ON THE OCEAN SENTINEL Introduction Load Banks Switchgear Hardware Converter Hardware CompactRIO Control and Data Acquisition CompactRIO Hardware Data Recording and Display Software Architecture Communication between top level LabVIEW VIs Host Software Real-Time Controller Software Main FPGA Software Slave FPGA Software... 46

8 TABLE OF CONTENTS (Continued) Page 3.6 Power Converter Simulations Bench Testing Commissioning Tests Calibration Check of Power Measurements Thermal Test Characterization of Constant Resistance Control WET-NZ TESTING Introduction Description of the WET-NZ Test Setup Test Sequence Constant Resistance Load Tests Test Method Sea Conditions Data Analysis Results and Discussion Voltage Threshold Switching Tests Test Method Sea Conditions Results for Latching Control Results for Declutching Control MPPT Testing Perturb and Observe Algorithm Cycling Algorithm Conclusions WEC AND POWER CONVERTER EMULATOR Introduction The OSU Autonomous WEC (AWEC) AWEC Model WEC Equations of Motion PTO Model... 1

9 TABLE OF CONTENTS (Continued) Page 5.4 Generation of excitation force time series from spectral records WEC and Power Converter Emulator MATLAB Simulink simulations MATLAB-Simulink Model MATLAB MPPT Routines Characterization of Simulated AWEC with Constant Resistance Loading WEC Characterization with Latching Control Observations and Discussion MATLAB-Simulink Simulation Speed Effect of Ocean Sentinel Power Converter MPPT TESTING USING THE WEC AND POWER CONVERTER EMULATOR Introduction Methods Wave Records used for MPPT Runs Assessment of Optimum R dc Regulation for Comparison of MPPT Results Effect of R dc Transitions on Output Power Perturb & Observe Algorithm R dc control Cycling MPPT algorithm R dc control Effects of T MPPT on Cycling MPPT Algorithm Effects of T mod on Cycling MPPT Algorithm Effects of Percent Deviation on Cycling MPPT Algorithm Effects of Modulation Waveform on Cycling MPPT Algorithm Effects of Gain on Cycling MPPT Algorithm Cycling MPPT Algorithm Simultaneous R dc and T d control Observations and Discussion CONCLUSIONS Conclusions The Ocean Sentinel WET-NZ Testing Maximum Power Point Tracking Recommendations for Further Work

10 TABLE OF CONTENTS (Continued) Page The Ocean Sentinel Maximum Power Point Tracking BIBLIOGRAPHY APPENDICES A. Ocean Sentinel Load Bank Drawings B. Wiring Diagrams and Schematics for Ocean Sentinel Equipment Developed by NMMREC and ESI Motion, Inc C. Ocean Sentinel data formats

11 LIST OF FIGURES Figure Page Figure 1-1 Global wave power distribution; numbers are annual averages in kw/m... 2 Figure 1-2 NNMREC wave energy converter test facilities... 4 Figure 1-3 Location of NNMREC scaled test site... 5 Figure 2-1 WEC testing with the Ocean Sentinel instrumentation buoy... 8 Figure 2-2 Ocean Sentinel instrumentation buoy on station... 9 Figure 2-3 Ocean Sentinel instrumentation power system Figure 2-4 Location of equipment on the Ocean Sentinel Figure 2-5 TRIAXYS wave buoy in and out of the water Figure 2-6 Low power umbilical Figure 2-7 Umbilical junction box Figure 2-8 National Instruments (NI) CompactRIO based DAS Figure 2-9 Watchman 5 DAS Figure 3-1 WEC power conversion, control, and data acquisition system Figure 3-2 Photos of WEC power conversion, control, and data acquisition equipment 23 Figure 3-3 Air cooled load banks: 5 kw (foreground); 2 x 25 kw (background) Figure 3-4 Three-phase ac load bank connections Figure 3-5 Dc connections of 5 kw load bank for use with power converter Figure 3-6 Switchgear schematic Figure 3-7 Photo of switchgear enclosure Figure 3-8 Schematic of converter enclosure Figure 3-9 Photo of converter enclosure Figure 3-1 Schematic of converter cooling system Figure 3-11 Photos of converter coolant system Figure 3-12 CompactRIO master and slave integrated with switchgear and converter.. 37 Figure 3-13 CompactRIO hardware Figure 3-14 Software architecture for CompactRIO system Figure 3-15 Screen shots of the host PC user interface Figure 3-16 Generator control used for 212 WET-NZ tests Figure 3-17 SimPowerSystems model; control at top and power electronics at bottom. 48

12 LIST OF FIGURES (Continued) Figure Page Figure 3-18 Simulated generator phase current at 2 kw output and 33 rpm Figure 3-19 Converter inductor and output currents with 2 kw output... 5 Figure 3-2 Test bench schematic Figure 3-21 Photo of test bench Figure 3-22 Bench test steady state voltage and current waveforms Figure 3-23 Sample current and voltage waveforms recorded during calibration check 55 Figure 3-24 Thermal test results Figure 3-25 Plots showing how applied load resistances were calculated from data Figure 3-26 Measured versus commanded resistance Figure 4-1 The half-scale WET-NZ WEC Figure 4-2 The WET-NZ power-takeoff system Figure 4-3 Test site layout for 212 WET-NZ testing Figure 4-4 Sample time series data for constant load resistance tests... 7 Figure 4-5 Histogram of sample counts for constant load resistance data Figure 4-6 WET-NZ power output versus load resistance, binned by H m and T e Figure 4-7 Examples of two different wave spectra with similar T e Figure 4-8 Average response of WET-NZ to wave energy flux spectra Figure 4-9 Power vs. expected power for all constant resistance load data Figure 4-1 Data analysis example Figure 4-11 Normalized power vs. R dc binned by H m mod & T e mod Figure 4-12 Sample time series data for voltage threshold switching tests Figure 4-13 Histograms of sample counts for latching (left) and declutching (right) Figure 4-14 Normalized output power versus voltage threshold for latching control Figure 4-15 Perturb and observe MPPT algorithm Figure 4-16 Results for perturb and observe MPPT algorithm with T MPPT = 2 min Figure 4-17 Cycling MPPT algorithm Figure 4-18 Results for cycling MPPT algorithm Figure 5-1 Quarter-scale OSU Autonomous WEC Figure 5-2 Two body point observer model used for AWEC simulations... 97

13 LIST OF FIGURES (Continued) Figure Page Figure 5-3 Frequency domain amplitude responses used to model AWEC Figure 5-4 PTO model Figure 5-5 Method used to generate excitation force time series from spectral records 13 Figure 5-6 WEC and power converter emulator and Ocean Sentinel CompactRIO Figure 5-7 Emulator operating with the Ocean Sentinel CompactRIO Figure 5-8 Implementation of the WEC model in the CompactRIO emulator Figure 5-9 Power converter components included in emulator PTO model Figure 5-1 PTO model used for the emulator Figure 5-11 MATLAB-Simulink model of the AWEC Figure 5-12 MATLAB routine for running MPPT algorithms Figure 5-13 Output power versus R dc for regular seas Figure 5-14 Output power vs. R dc for random seas; two-body Simulink results Figure 5-15 Time plot of sample output power data with H m = 1 m and T e = 5 s Figure 5-16 Ocean Sentinel converter latching controls Figure 5-17 Output power versus T d for latching control in regular seas Figure 5-18 Emulator output power versus T d for regular seas Figure 5-19 Emulator waveforms for latching compared to resistance control Figure 6-1 Optimum R dc and P avg for data Figure 6-2 Optimum R dc and P avg for data Figure 6-3 Emulator results showing effect of R dc transitions on output power Figure 6-4 Perturb and observe MPPT results Figure 6-5 Time plots of R dc for P&D algorithm; T MPPT = 1 min and C step = 6.6 Ω Figure 6-6 Cycling MPPT algorithm with sine wave modulation Figure 6-7 R dc regulation for both data sets using cycling algorithm; best parameters. 133 Figure 6-8 Energy capture versus T MPPT for cycling MPPT algorithm Figure 6-9 Energy capture versus T mod for cycling MPPT algorithm Figure 6-1 Energy capture versus percent deviation for cycling MPPT algorithm Figure 6-11 Energy capture vs. percent deviation for sine, square wave modulation Figure 6-12 Energy capture versus gain for cycling MPPT algorithm

14 LIST OF FIGURES (Continued) Figure Page Figure 6-13 Cycling algorithm that simultaneously adjusts both R dc and T d Figure 6-14 Simultaneous R dc and T d control using cycling MPPT; data Figure 6-15 Simultaneous R dc and T d control using cycling MPPT; data

15 LIST OF TABLES Table Page Table 1-1 Phased test program recommended by the European Marine Energy Centre... 2 Table 2-1 Power budget for the Ocean Sentinel instrumentation power system Table 2-2 Conductors for high and low power Ocean Sentinel umbilical cables Table 3-1 Major switchgear components Table 3-2 Major converter components Table 3-3 Major converter cooling system components Table 3-4 CompactRIO master controller components Table 3-5 CompactRIO slave components Table 3-6 CompactRIO data file formats... 4 Table 3-7 Methods used to communicate between different LabVIEW VIs Table 3-8 Half-scale WET-NZ electrical generator specifications Table 3-9 Calibration check data for CompactRIO power measurement Table 4-1 Half-scale WET-NZ device characteristics Table 4-2 Cumulative duration of WET-NZ tests Table 4-3 Settings typically used during voltage threshold load cycling sequences Table 5-1 Full-scale OSU autonomous WEC masses and dimensions Table 5-2 Constants for AWEC model Table 5-3 PTO parameters used for modeling Table 6-1 Cycling MPPT parameters for best performance Table 6-2 Cycling MPPT parameters used for simultaneous R dc and T d control... 14

16 Testing of Wave Energy Converters using the Ocean Sentinel Instrumentation Buoy 1 INTRODUCTION 1.1 Background There is increasing interest in generating electrical power from ocean waves. Ocean waves have several distinct advantages over other renewable energy resources, such as wind and solar. Ocean waves are usually present, they are predictable, and they have high energy density. Ocean wave power density is highest between 3º and 6º latitude and off the west coasts of the world continents, and as shown in Figure 1-1, is approximately 3 kw per meter of wave crest length off the Oregon coast [1], [2]. For the United States (US) it is estimated that 26 TWh per year of electrical energy could be generated from ocean waves, which is approximately 6 percent of the US annual electrical load. This is an amount comparable to the current traditional hydropower contribution [3] Figure 1-1 Global wave power distribution; numbers are annual averages in kw/m

17 2 A number of private companies in the US are now developing wave energy converters (WECs) that will take advantage of the ocean wave resources and convert wave energy to electrical power. Along with the advantages of ocean waves as an energy source, however, come a number of challenges to developing WECs. WECs need to be designed to survive extreme ocean waves. WECs also must be designed for a difficult seawater environment, and both deployment and maintenance is difficult at sea. In order to design WECs that can meet these challenges, developers usually follow a development and test program that initially includes wave tank testing of small scale models, and ultimately leads to prototype testing in the open ocean. For example, the five step, phased test program shown in Table 1-1 is recommended by the European Marine Energy Centre (EMEC) for wave energy converter development [4]. This program includes wave tank testing of both validation and test models during the first two phases, followed by three sea testing phases, initially with a process model at a benign site and eventually with a demonstration unit in the open ocean. This type of phased program mitigates both technical and fiscal risks; as the program progresses through each phase it increases in both complexity and cost. Table 1-1 Phased test program recommended by the European Marine Energy Centre (EMEC) for wave energy converter development (from [4]) Phase Device scaling (by length) Test location Duration (includes analysis) Phase 1 Validation model Phase 2 Design model Phase 3 Process model Phase 4 Prototype Phase 5 Demonstration 1 : : : : 1 2 Full size Wave tank testing 2D flume & 3D basin Wave tank testing 3D basin Sea testing Benign site Sea testing Exposed site Sea testing Open ocean 1 3 months 6 12 months 6 18 months months 1 5 years

18 3 Until recently, ocean test facilities needed for WEC testing during the later phases of development have not existed in the US because of the costly infrastructure and lengthy permitting process required. This has led to the establishment of publicly funded ocean test facilities such as those being developed by the Northwest National Marine Renewable Energy Center (NNMREC). These facilities are leased to commercial wave developers to facilitate WEC development in the US. 1.2 The Northwest National Marine Renewable Energy Center (NNMREC) NNMREC, established in 28, is a partnership between Oregon State University (OSU), the University of Washington (UW), and the National Renewable Energy Lab (NREL). The mission of NNMREC is to facilitate the commercialization of marine renewable energy technologies, inform regulatory and policy decisions, and to close key gaps in scientific understanding with a focus on student growth and development. Center activities include: 1) development of facilities to serve as an integrated, standardized test center for developers of wave and tidal energy; 2) evaluation of potential environmental, ecological and social impacts, focusing on the compatibility of marine energy technologies in areas with sensitive environments and existing users; 3) device and array optimization for effective deployment; 4) improved forecasting; and 5) increased reliability and survivability of marine energy systems. NNMREC wave energy converter test facilities are shown in Figure 1-2. These facilities can be leased by commercial WEC developers. Facilities include a linear test bed at the OSU Wallace Energy Systems and Renewable Facility that can be used for dry testing of WEC drive trains, a 2D wave flume and a 3D wave basin at the O.H. Hinsdale Wave Research Laboratory that can be used to test small scale WEC models, and prepermitted ocean test sites off the coast of Newport, Oregon. NNMREC established a scaled ocean test site in 212 that is used for stand-alone testing of scaled devices without a grid connection through a cable to shore, and is in the process of developing the Pacific Marine Energy Center (PMEC), which will include multiple cables to shore for grid connected testing of full-scale WECs. The PMEC test site will be located further

19 4 offshore and in deeper water than the scaled test site. The scaled site will normally be used to test scaled WEC models during summer months when shorter period waves necessary for those tests occur, while the PMEC site will be used for year around, full scale device testing. OSU Wallace Energy Systems & Renewable Facility 2kW Wave Energy Linear Test Bed OSU Hinsdale Wave Research Facility 2D Wave Flume 14L x 3.7W x 4.6D (m) 3D Wave Basin 48.8L x 26.5W x 2.1D (m) Ocean test sites off of Newport, OR Scaled site for stand-alone testing (Ocean Sentinel Instrumentation Buoy) Pacific Marine Energy Center (future full-scale site with cable to shore) Figure 1-2 NNMREC wave energy converter test facilities The location of NNMRECs scaled test site is shown in Figure 1-3. The site is approximately 2.5 nautical miles offshore from Yaquina Head, north of Newport, Oregon (latitude N, longitude W). Water depth at the site averages 5 m. This test site does not have a cable to shore, due to the high expense of permitting and installing a cable and also because many of the scaled WEC models that will be tested at this site are not sufficiently mature to be connected to the electrical grid for commercial power production. NNMREC has developed the Ocean Sentinel Instrumentation Buoy to facilitate stand-alone WEC testing at this test site. The Ocean Sentinel is a floating platform that can be deployed at the test site along with the device being tested; when

20 5 connected to the device being tested through an umbilical cable the Ocean Sentinel can provide the electrical interface and loading necessary for stand-alone operation. The development of the Ocean Sentinel and its use to test scaled WECs is the focus of this work. Figure 1-3 Location of NNMREC scaled test site 1.3 The Ocean Sentinel Instrumentation Buoy The Ocean Sentinel was developed by NNMREC together with AXYS Technologies beginning in the fall of 211, and was completed in August, 212. The Ocean Sentinel is described in detail in Chapter 2 and also in [5]. The Ocean Sentinel includes electrical loading and power conversion equipment that provides the electrical interface and generator control necessary to test prototype WECs that are not fully developed for grid connection. That electrical loading and power conversion equipment, together with data acquisition equipment that is installed on board the Ocean Sentinel for testing WECS, was developed as part of this work and is described in Chapter 3. Immediately after completion of the Ocean Sentinel in August 212, it was deployed at

21 6 the NNMREC scaled test site and used to test a half-scale WET-NZ wave energy converter. WET-NZ is an acronym for Wave Energy Technology-New Zealand. The Ocean Sentinel and the WET-NZ were moored at NNMREC s scaled test site for a sixweek period from August 22, 212, until October 5, 212, while the testing was performed. The WET-NZ electrical loading was remotely controlled using the Ocean Sentinel equipment during this period. The WET-NZ testing and analysis of the test results were part of this work, and are described in Chapter 4. The WET-NZ testing was concluded in the beginning of October, 212, because the Ocean Sentinel and its moorings were designed for deployment between May-October when conditions at the test site are more benign. Also, short period waves were needed to test the half-scale WET-NZ, and these conditions are less frequent at the NNMREC scaled test site during winter months. 1.4 Maximum Power Point Tracking (MPPT) Maximum power point tracking (MPPT) is a common technique used to adjust the electrical loading applied to solar panels in order to maximize their power output [6]. Different MPPT techniques have also been used to control wind turbines [7]. MPPT control of WECs is more difficult, because WEC output power can have large, random fluctuations in random seas. MPPT control of WECs is possible though; the use of a simple perturb and observe MPPT method to control WECs was demonstrated previously at OSU [8], [9]. During the WET-NZ tests with the Ocean Sentinel, experimentation was done with MPPT control of the WET-NZ. The perturb and observe MPPT method did not work well for controlling the WET-NZ, however, and a new cycling MPPT method was developed during the course of those tests that performed better. These results are described in Chapter 4 along with the other WET-NZ results. Following the WET-NZ tests, an emulator was built that incorporates a WEC model and allows simulated WEC testing to be conducted with the Ocean Sentinel control equipment. This emulator was used to simulate operation of a small, 2 W Autonomous Wave Energy Converter (AWEC) under development as part of a

22 7 concurrent project at OSU. The AWEC was simulated for this project to take advantage of available WEC modeling information and to avoid the use of proprietary information. The emulator is described in Chapter 4. The emulator was used to continue the MPPT work that began during the WET-NZ tests. The performance of both the perturb and observe MPPT algorithm and the cycling MPPT algorithm were characterized in detail. The results showed that the cycling algorithm is easier to set up and is a more reliable method of controlling a WEC than the perturb and observe algorithm. These results are described in Chapter 6.

23 8 2 THE OCEAN SENTINEL INSTRUMENTATION BUOY 2.1 Introduction The Ocean Sentinel instrumentation buoy facilitates open-ocean, stand-alone testing of WECs with no cable to shore. A concept diagram of the Ocean Sentinel testing a WEC is shown in Figure 2-1, and the buoy is shown on station in Figure 2-2. The Ocean Sentinel is a surface buoy, based on the 6-meter NOMAD (Navy Oceanographic Meteorological Automatic Device) buoy design. The Ocean Sentinel can be used to test WECs with average power outputs of up to 1 kw. Deployment to date has been at the NNMREC scaled test site located offshore from Yaquina Head, north of Newport, Oregon as shown in Figure 1-3 [5]. TRIAXYS Wave Measuring Buoy WEC Under Test To shore Approximately 125 m Ocean Sentinel Instrumentation Buoy Load bank Umbilical DAS & Telemetry Power Conversion Figure 2-1 WEC testing with the Ocean Sentinel instrumentation buoy

24 9 Figure 2-2 Ocean Sentinel instrumentation buoy on station WECs being tested are connected to the Ocean Sentinel by an umbilical cable. Three point mooring systems are used for both the WEC and Ocean Sentinel, to maintain small watch circles and to keep the umbilical from twisting. The mooring systems and umbilical are designed for a nominal 125 meter separation between the WEC and Ocean Sentinel. Power generated by the WEC is controlled by switch gear and power conversion equipment located on board the Ocean Sentinel and dissipated in an onboard load bank. WEC output power is measured and recorded on board the Ocean Sentinel, together with wave data transmitted via wireless telemetry from a TRIAXYS buoy moored nearby. Other WEC data measured on board the WEC under test can also be transmitted to the Ocean Sentinel via the umbilical for integration with the wave and power data, if required by the WEC developer. The primary functions of the Ocean Sentinel are as follows: 1. Provide stand-alone electrical loading and power conversion for the WEC under test. 2. Measure and record WEC power output.

25 1 3. Collect and store data transmitted from a wave measuring buoy moored nearby, and also from the WEC under test when required by the developer. 4. Transmit collected data to a shore station via a wireless telemetry system. 5. Conduct environmental monitoring. The Ocean Sentinel was developed by NNMREC and AXYS Technologies between October 211 and August 212. NNMREC developed the umbilical and the WEC electrical loading, power conversion, and the data acquisition that is integrated with that equipment. AXYS provided the hull along with equipment that did not need to be custom designed for WEC testing such as the instrumentation power system, a Watchman 5 data acquisition and control system, meteorological sensors, and cameras. AXYS also provided the TRIAXYS wave buoy. AXYS delivered the Ocean Sentinel to NNMREC in July 212 with the equipment that it supplied installed. NNMREC installed the remaining equipment and tested all systems in July and early August 212. The Ocean Sentinel was then deployed in late August 212 and used for the first time to test the WET-NZ wave energy converter at NNMREC s scaled test site. 2.2 Instrumentation Power System A power system developed by AXYS Technologies supplies power to all equipment installed on-board the Ocean Sentinel and also exports auxiliary power to the WEC under test. A block diagram is shown in Figure 2-3. The features of this system are as follows: 1. The system is capable of continuously providing at least 4 W of power at 24 Vdc and 12 Vac. An estimated power budget is shown in Table 2-1.

26 11 Table 2-1 Power budget for the Ocean Sentinel instrumentation power system Equipment Power Budget (W) Export to WEC under test 15 Power converter 125 NNMREC data acquisition system (DAS) 2 Telemetry/communications 75 On-board instrumentation & other 3 Total 4 2. Power is provided by a large bank of sealed lead acid batteries, which are maintained by a wind generator, solar panels, and a standby diesel generator. 3. The solar and wind generation alone can provide the primary power required with typical Newport, OR summer conditions. 4. The diesel generator is designed for 3-month minimum deployment periods without the need for refueling, though the system is designed to allow refueling at sea. 5. Remote monitoring of the power system status, with battery state of charge, solar and wind turbine performance, generator performance and fuel level, as well as instrumentation loading are all transmitted to shore in real-time. 6. Absorbed glass mat type, sealed lead acid batteries are used to prevent spillage, with smart charge controllers and proper venting to reduce the chance of hydrogen buildup inside the buoy. 7. In addition, the power system has been designed for expansion in the future, to enable NNMREC to install a grid emulator or other power conversion equipment on-board the buoy. Another possible enhancement would allow the use of WEC output power, when available, as a supplemental power source to reduce fuel consumption.

27 12 Figure 2-3 Ocean Sentinel instrumentation power system The location of the solar panels and wind turbine above deck can be seen in Figure 2-2. The rated power of the wind turbine is 1 W, the rated power of each solar panel is 21 W, and the rated power of the diesel generator is 32 W. The diesel generator, diesel tanks, and lead acid batteries are located in the aft three compartments below deck, as shown in Figure 2-4. The diesel tanks have a total capacity of 24 gallons. The lead acid battery bank has a total capacity of 2 A-Hr at 24 V. Umbilical junction box Load banks Umbilical WEC power conversion and data acquisition Diesel generator Diesel tanks Lead acid batteries Figure 2-4 Location of equipment on the Ocean Sentinel

28 Onboard Instrumentation and Equipment The following instrumentation is installed on board the Ocean Sentinel: an anemometer, wind direction sensor, temperature and barometric pressure sensors, and two video cameras with wide-angle fisheye lenses that view the deck and water immediately surrounding the Ocean Sentinel. Wind data from this instrumentation is recorded by the NNMREC DAS for integration with WEC power and wave data. Other equipment installed on board the Ocean Sentinel includes a navigation light, Automatic Identification System (AIS) transponder, Radar reflector, and Global Positioning System (GPS) with satellite communications. In addition, NNMREC anticipates that other environmental sensing equipment will be installed on board the Ocean Sentinel by OSU researchers conducting studies in the future. 2.4 TRIAXYS Wave Measurement Buoy A TRIAXYS wave buoy procured from AXYS Technologies, shown in Figure 2-5, is used for ocean wave and ocean current measurements. This buoy is moored approximately 1 m from the WEC under test, and transmits wave magnitude, period, direction and current data to the Ocean Sentinel via radio telemetry. Accelerometer and rate gyro data is processed on board the TRIAXYS buoy to produce both directional and non-directional wave frequency spectra. An Acoustic Doppler Current Profiler (ADCP) on board the TRIAXYS buoy measures the ocean current profile down to a depth of 4 m. The wave frequency spectra and current profile data are transmitted to the Ocean Sentinel, at configurable intervals. Lead-acid batteries that are charged by eight small solar panels surrounding the top of the buoy provide on-board power.

29 14 Figure 2-5 TRIAXYS wave buoy in and out of the water 2.5 Umbilical Cable The umbilical cable connecting the Ocean Sentinel and the WEC under test is approximately 2 m long. Two different umbilical systems have been designed for the Ocean Sentinel; a high power umbilical with power conductors sized for 125 amps continuous that was developed by 3U Technologies and NNMREC, and a low power umbilical with power conductors sized for 4 amps continuous that was developed by NNMREC alone. The current carrying capabilities are determined by the voltage drop in the power conductors; a voltage drop of 5% or less is desired with a 22 volt output from the WEC generator to reduce errors in WEC power measurement on board the Ocean Sentinel. The power conductor insulation is rated for 1 volts in both umbilicals. The conductors that are included in each umbilical are listed in Table 2-2. The heavier, high power umbilical requires specialized cable machinery on board a large vessel for deployment. The low power umbilical is light enough to be deployed by hand from a small boat, making it the preferred option if it meets the requirements of the specific WEC being tested. Both umbilicals have multiple 6 AWG and 16 AWG conductors that can each be either connected in parallel for use as WEC power conductors, used as auxiliary 12 volt ac power conductors, or used for other purposes depending on the needs for the specific deployment.

30 15 Table 2-2 Conductors for high and low power Ocean Sentinel umbilical cables Conductor WEC output power and ground WEC auxiliary power, neutral sense, or other Fiber optic Number and Size Low Power Umbilical High Power Umbilical (4) 6 AWG (15) 6 AWG (12) 16 AWG (31) 16 AWG None (4) pairs, single mode conductors The low power umbilical was designed and built for the 212 WET-NZ deployment described in Chapter 4. A photo of this umbilical is shown in Figure 2-6; it is made up of two separate multi-conductor 6 AWG and 16 AWG cables that are bundled together. While the high power umbilical includes single mode fiber optic conductors for communications between the WEC and Ocean Sentinel, none are included in the low power umbilical because these were not needed for the 212 deployment. A different low power umbilical will be designed and built that includes fiber optic conductors, if needed for future deployments. Figure 2-6 Low power umbilical

31 16 The routing of the umbilical on board the Ocean Sentinel can be seen in Figure 2-4, where the low power umbilical made up of two separate cables is shown. The umbilical cables enter the Ocean Sentinel over the stern and are routed forward to a junction box, where connections are made to other cables that route down to equipment in the forward compartment. This junction box, with its sealed cover removed, is shown in Figure 2-7 with the low power umbilical connections that were used during the 212 WET-NZ tests. The terminal blocks inside allow reconfiguration of the umbilical conductors so that different numbers of parallel conductors can be connected to the WEC electrical loading equipment, 12 Vac ancillary power, or control circuits in the forward compartment as necessary for different WEC deployments. Cables that run from equipment in the forward compartment to the load banks on deck also route through the junction box but are not connected to the terminal blocks. 6 AWG umbilical conductors 16 AWG umbilical conductors Load bank wiring routed through jbox Terminal blocks for WEC generator power conductors Terminal blocks for ancillary power & signal conductors Cables down to forward compartment Figure 2-7 Umbilical junction box 2.6 Data Acquisition and Telemetry Systems Two independent acquisition systems (DASs) are used on board the Ocean Sentinel: 1) a National Instruments (NI) CompactRIO based system developed by NNMREC that is used to measure and record WEC test data, shown in Figure 2-8, and 2) a Watchman5 DAS procured from AXYS Technologies, shown in Figure 2-9, that is

32 17 used to monitor and control the power system, monitor environmental sensors, and interface with the TRIAXYS wave buoy. The two DAS communicate to shore independently via their own telemetry systems CompactRIO DAS and Control The CompactRIO DAS records the following: 1. The power output of the WEC under test, measured on board the instrumentation buoy. 2. Wave, ocean current, and wind data provided by the Watchman 5 through a serial link. 3. Any data received from a DAS installed on-board the WEC under test. See Figure 2-8 for a diagram of this DAS. This system is integrated with the WEC power conversion and control equipment on the Ocean Sentinel, and is described in detail in Chapter 3 together with that equipment. During the 212 Ocean Sentinel deployment, all data systems on board the WEC being tested were operated independently of the Ocean Sentinel data systems. The Ocean Sentinel CompactRIO DAS, however, is compatible with CompactRIO DAS modules that have been developed by the National Renewable Energy Laboratory (NREL) for installation on board the WEC being tested. NNMREC anticipates integrating the Ocean Sentinel CompactRIO DAS with NREL modules installed on the WEC being tested during future deployments, as indicated by components shown with dashed lines in Figure 2-8.

33 18 crio 925 controller WEC Under Test Fiber optic Ocean Sentinel Instrumentation Buoy Switchgear WEC V, I Power converter crio Slave SWAP Network 82.11b 3G Cellular EtherCAT slave I/O modules TRIAXYS Wave Buoy 9 MHz Watchman 5 Control and DAS Modem Power system Future enhancements Figure 2-8 National Instruments (NI) CompactRIO based DAS Watchman 5 DAS and Control The Watchman5 system, shown in Figure 2-9, has been developed by AXYS Technologies to control power systems and monitor environmental sensors on board numerous meteorological buoys deployed by AXYS. The Watchman 5 has the following functions on board the Ocean Sentinel: 1. Power system control, monitoring, and safety systems. 2. Bilge level monitoring and alarms. 3. Mooring watch circle monitoring and alarm. 4. AIS interface. 5. Interface with the TRIAXYS wave measuring buoy to receive ocean wave and current data. 6. Interface with onboard wind sensors. 7. Interface with onboard video cameras.

34 19 8. Record data from environmental sensors installed onboard the Ocean Sentinel for purposes other than WEC testing. 9. Provide ocean wave, ocean current, and wind data to the CompactRIO DAS through a serial link. Figure 2-9 Watchman 5 DAS Telemetry Independent 3G cellular telemetry systems are used for communications between the CompactRIO and the Watchman 5 DAS and shore. A 9 MHz serial link is used for communication between the TRIAXYS wave measuring buoy and the Watchman 5 DAS on board the Ocean Sentinel. The Watchman 5 DAS also has an independent INMARSAT D+ satellite communication link providing secondary data transmission and system control capabilities when required. Several future enhancements to the Ocean Sentinel telemetry, shown by the dashed lines in Figure 2-8, are planned: 1. Incorporation of a router and redundant telemetry to the CompactRIO using both 3G cellular and the 82.11b Ship-to-Shore Wireless Access Protocol

35 2 (SWAP) network run by the OSU College of Earth, Ocean and Atmospheric Sciences (CEOAS). 2. Routing of Watchman 5 communications through the CompactRIO telemetry system. 3. Addition of communications between the Ocean Sentinel and the WEC under test through optical fiber integrated into the umbilical and Ethernet to fiber converters. During the 212 deployment of the Ocean Sentinel, the SWAP modem and antennas were installed on board the Ocean Sentinel and communications were successfully tested independent of the CompactRIO communications in preparation for the future addition of this system. 2.7 WEC Electrical Loading and Power Conversion The WEC electrical loading and power conversion equipment on board the Ocean Sentinel is described in detail in Chapter 3. This equipment, developed by NNMREC, includes the load banks, switchgear, and power conversion equipment together with the CompactRIO control and data acquisition system.

36 21 3 WEC ELECTRICAL LOADING, POWER CONVERSION, AND DATA ACQUISITION EQUIPMENT ON THE OCEAN SENTINEL 3.1 Introduction This chapter describes the system on board the Ocean Sentinel that provides standalone electrical loading, power conversion, control, and data acquisition for WECs being tested. This system, developed by NNMREC, interfaces WEC generators to the Ocean Sentinel load banks and also records WEC power output and ocean wave data. Generator control is provided for WECs in early stages of development that do not include onboard generator power conversion. See Figure 3-1 for a diagram of this system. CompactRIO Master Junction box Figure 3-1 WEC power conversion, control, and data acquisition system In reference to Figure 3-1, this system consists of the following equipment: 1. Air cooled resistive load banks that dissipate the power generated by the WEC being tested.

37 22 2. Electrical switchgear that includes a disconnect and fuses, contactors for switching the load banks, contactors and soft start circuitry for the power converter, and voltage and current transducers to measure the output power of the WEC being tested. 3. A power electronic converter that can provide a continuously adjustable load to the generator of the WEC being tested. 4. A CompactRIO data acquisition and control system that controls the switchgear contactors and the power electronic converter, and records electrical data from the switchgear and converter, the output power of the WEC being tested, and ocean wave and ocean current data. This system was designed for a high degree of flexibility, because NNMREC expects to test WECs that have different power outputs and generator configurations with the Ocean Sentinel. The system is reconfigurable via terminations in the switchgear enclosure so that each load bank section can be connected for either contactor or converter control, and the load banks can each be reconnected for different voltages and powers. The generator control and data acquisition provided by the CompactRIO can be reconfigured in LabVIEW software as needed for different WECs being tested. In addition, the converter hardware was designed so that power electronics modules can be replaced with different off-the-shelf modules to provide different converter architectures and power ratings depending on the requirements of WECs being tested in the future. The location of this equipment on the Ocean Sentinel is shown in Figure 2-4. Two air cooled load bank enclosures are installed above deck, behind the forward mast. The load banks are mounted on a sliding track that allows them to be slid fore-aft so that the three compartments underneath them can be accessed. The switchgear, power converter, and CompactRIO system are all installed below deck in the forward compartment. See Figure 3-2 for a photograph of the equipment in the forward compartment. The switchgear and converter are housed in separate enclosures. The CompactRIO system consists of both master and slave units. The CompactRIO master is installed in a DAS and telemetry enclosure that also includes the telemetry equipment described in

38 23 Section 2.6.3, and the CompactRIO slave is installed in the converter enclosure. All enclosures are sealed and corrosion resistant, National Electrical Manufacturers Association (NEMA) 4X rated; cables enter the enclosures via sealed connectors or glands. Desiccant is used to maintain low humidity inside the enclosures. DAS & telemetry enclosure Switchgear enclosure Converter enclosure CompactRIO slave CompactRIO master Figure 3-2 Photos of WEC power conversion, control, and data acquisition equipment The switchgear, converter, and CompactRIO data acquisition and control were developed between October 211 and August 212. Switchgear and converter hardware was not completed and ready for testing until early August 212, shortly before deployment of the Ocean Sentinel with the WET-NZ wave energy converter. Early in the development process, the power electronic converter and control was simulated in MATLAB-Simulink. Prototype bench test hardware was built in the spring of 212 that included a complete power converter along with most of the switchgear hardware. This platform was used to test most aspects of the CompactRIO software. Final system testing was then performed in August 212, immediately before deployment of the Ocean Sentinel.

39 Load Banks The air cooled load banks were custom designed and built for NNMREC by Powerohm Resistors, Inc. of Katy, TX. A photo of the two load bank enclosures installed above deck on the Ocean Sentinel is shown in Figure 3-3. PowerOhm drawings for the load banks are included in Appendix A. One enclosure houses a single 5 kw load bank while the other houses two separate 25 kw load banks. All load banks are constructed with stainless steel grid resistors and are cooled by natural convection air flow through the louvered aluminum enclosures. Terminal blocks for external connections are at the bottom of each enclosure. Each load bank consists of three separate grid resistors that each have a center tap. The center tap allows the two halves of each resistor to either be connected for operation in series, parallel, or with one half disconnected. Each load bank can either be configured as a three-phase or dc load by making appropriate connections at the terminal blocks inside the load bank enclosures and also at the terminal blocks inside the switchgear enclosure. Figure 3-3 Air cooled load banks: 5 kw (foreground); 2 x 25 kw (background) When the load banks are configured as three-phase ac loads, several different connections are possible depending on the specific requirements of the WEC being

40 25 tested. All of the possible connections are shown in Figure 3-4. For operation at rated load bank power, the delta-parallel connection with a three-phase voltage of 46 volts must be used. The Ocean Sentinel wiring is sized for these operating currents. In cases where the WEC under test has a lower voltage generator or smaller load steps are needed, however, the other connections can be used; the maximum power for each connection is listed in Figure 3-4. Two different constraints limit the maximum voltage for each load bank configuration: 1) the voltage across each load bank element (half of each grid resistor) can t exceed 23V to avoid overheating the element, and 2) the maximum line current can t exceed the capability of the Ocean Sentinel wiring (31 A or 63 A maximum, respectively, for the 25 kw and 5 kw load banks). These separate limits are highlighted in Figure 3-4. Resistance and maximum voltage, current, and power for different three phase load bank connections 25 or 5 kw Load Bank 25 kw 5 kw 25 kw 5 kw 25 kw 5 kw 25 kw 5 kw Delta series Delta single Delta parallel Wye series Wye single Wye parallel Configuration Equivalent wye per leg resistance Maximum voltage line line (V) Maximum line current (A) Limited by 23V maximum across each load bank element Limited by switchgear wiring capability (31A/63A for 25/5 kw load banks) Figure 3-4 Three-phase ac load bank connections Maximum power (kw) When the power converter is in use, the 5 kw load bank is normally used as the resistive load. Figure 3-5 shows two different dc load bank connections that can be used

41 26 with the power converter and how these connections affect the maximum load that can be applied to the WEC generator. Other connections are also possible. The series-parallel connection provides a resistance R load of 4.2 Ω while the parallel connection provides an R load of 1.5 Ω. Power (kw) Max power versus line voltage Series parallel, Rload = 4.2 Ω Parallel, Rload = 1.5 Ω 2 4 Voltage (Vrms line line) Equivalent ac wye perphsase resistance (Ohms) Min ac resistance vs. line voltage Series parallel, Rload = 4.2 Ω Parallel, Rload = 1.5 Ω 2 4 Voltage (Vrms line line) Figure 3-5 Dc connections of 5 kw load bank for use with power converter The output current of the converter I out must be limited to 11 amps maximum to protect the load bank wiring and converter output inductors. This constrains the maximum power dissipation and also the minimum resistance that can be applied to the WEC generator with respect to generator voltage. Assuming 1% efficiency, the converter can be analyzed as a dc transformer so when I out is limited, constant power loading is provided to the WEC generator. When I out is not limited, maximum load is provided with the converter full on so that converter input current I dc equals I out. In this case the load bank is effectively connected to the WEC generator through the three-phase

42 27 rectifier. The approximate relationships between the voltages and currents on the two sides of the rectifier are shown in equations (3.1) and (3.2) assuming full conduction: V dc 1.35*V LL 1.35* 3*V ln (3.1) I dc *I L (3.2) The resulting relationship between the equivalent ac wye resistance applied to the generator output R wye and the resistance on the dc side of the rectifier R dc (equal to the load bank resistance R load when the converter is full on) is then per equation (3.1). R wye R dc.55*r dc ; R wye V ln ; R I dc V dc (3.1) L I dc Plots of the maximum power and minimum resistance R wye that can be applied to the WEC generator with respect to its output voltage are shown at the bottom of Figure 3-5. The converter operates in current limit so that power is constant for ac voltages V ll above 34 volts and 85 volts for the series-parallel and parallel connections, respectively. At low voltages, below V ll of 17 volts, the parallel connection can provide lower resistance and higher power loading than the series-parallel connection, but at higher voltages the opposite is true. 3.3 Switchgear Hardware A schematic showing the electrical connections for the switchgear is shown in Figure 3-6, and an annotated photo of this equipment installed in the switchgear enclosure is shown in Figure 3-7. A list of the major switchgear components is included in Table 3-1. ESI Motion of Simi Valley, CA designed the mechanical layout of components in the switchgear enclosure and the custom enclosure, under the direction of NNMREC. The enclosure is constructed of anodized aluminum, with welded seams. The cover, not shown in Figure 3-7, is an aluminum plate that screws in place with an o- ring seal. ESI also assembled and wired the switchgear for NNMREC. ESI s detailed wiring schematic for this equipment is included in Appendix B.

43 28 To crio master (DAS/Telemetry) To Converter To Pwr dist. To crio master (DAS/Telemetry) EMI filter Power supplies 126 A/1 AWG (for future 1kW converter) 5 kw load bank 25 kw load bank 25 kw load bank Figure 3-6 Switchgear schematic

44 29 Load bank connectors Generator input cable gland Transient suppressor Disconnect Current sensors Converter output cable gland Converter input cable gland Converter output terminations Ground fault current sensor 5 kw load bank terminations Fuses Soft start relay Soft start resistors Contactor drivers Contactors 25 kw load bank terminations Voltage sensors 24 V & ±15 V Converters Figure 3-7 Photo of switchgear enclosure Table 3-1 Major switchgear components Description Manufacturer Qty Part Number Disconnect Altech 1 VKA316N Contactor 2A Kilovac 6 LEV2 Contactor 1A Kilovac 3 LEV1 Contactor driver Crydom 4 DC6S5 Fuse, 4A, UL type J McMaster-Carr 6 554T22 Fuse, 7A Ferraz Shawmut 3 A5QS7-4 Current sensor LEM 3 IT 4-S Voltage sensor LEM 3 CV 3-5 Converter, 24V to 24V CUI Inc. 1 VHK2W-Q24-S24 Converter, 24V to ±15V CUI Inc. 1 VYB2W-Q24-D15-T Ground fault current sensor Bender 1 W35AB Ground fault module Bender 1 RCMA423-DM1C-1 Soft start relay, 3 ph. Crydom 1 D53TP25D Soft start resistors, 33 Ω TE Connectivity Transient Suppressor Delta 1 LA63

45 3 In reference to the Figure 3-6 schematic, the variable speed, variable frequency electrical generator on board the WEC connects via the umbilical cable to the three-phase disconnect switch in the switchgear. The disconnect switch connects to a transient suppressor that will absorb energy in the case of a lightning strike, and also to three sets of three-phase contactors, through three sets of fuses. The outputs of two sets of contactors are normally connected at terminal blocks to two 46 volts at 25 kw, deltaconnected sections of the load bank. These load bank sections can be reconnected inside the load bank enclosure for 23 volts at 12.5 kw, or they can be reconnected at the terminal blocks inside the switchgear enclosure for wye connections for lower power. A third set of contactors may either be connected through a terminal block to a 46 volts at 5 kw load bank section, or to the input of the power converter. A soft start circuit consisting of a three-phase semiconductor relay and resistors is wired in parallel with the third set of contactors. This circuit limits inrush current to the power converter bus capacitors when the main contactors are closed, and is disabled when the power converter is not in use. The output of the power converter, when in use, can be connected through terminal blocks to parallel load bank elements. All contactors are dc-rated for operation at low frequency, and are designed for frequent switching to provide incremental load control when connected directly to sections of the load bank. When the contactors are connected to the power converter, they are only opened when the WEC output is disabled or fault conditions exist. All contactor coils are controlled by the CompactRIO master, through 5amp solid state relays. Control power to the switchgear is provided by unregulated 24 volt power from the Ocean Sentinel power system, which can vary between 2 volts and 29 volts depending on battery charge. Two power supplies in the switchgear compartment convert the unregulated 24 volts to regulated 24 volts, to power the contactor coils, and to ±15 V, to power the voltage and current transducers. Contactor coils are powered from regulated 24 volts rather than unregulated 24 volts to avoid exceeding the maximum 28 volts coil voltage of the LEV-1 contactors.

46 31 Three voltage transducers and three current transducers are used to measure the three-phase WEC generator voltages and currents. The outputs of these devices are connected to the CompactRIO master, which calculates WEC output power from the voltage and current data. High accuracy transducers are used in order to meet the International Electrotechnical Commission (IEC) WEC power performance standard [1], which requires.5 % of measurement range accuracy for both voltage and current, where the measurement range is determined by the output range of the specific WEC being tested. The three voltage transducers, connected in a wye, are LEM CV3-5 models which have a nominal rating of 35 Vrms and.2% accuracy. These transducers provide.5% accuracy for line-neutral voltages as low as 14 Vrms, or line-line voltages as low as 24 Vrms. They have a frequency bandwidth that ranges from dc to 3 khz. The three current transducers are LEM model IT 4-S models which have a nominal current of 28 Arms and.4% accuracy for frequencies of 6 Hz or less, and.4% accuracy at higher frequencies. These transducers provide.5% accuracy for currents as low as 2.2 amps for -6 Hz and currents as low as 22 amps for frequencies greater than 6 Hz. They have a frequency bandwidth that ranges from dc to 15 khz. These current transducers use a high precision fluxgate technology. The current flows through a magnetic core, and a secondary coil is controlled by closed loop feedback to zero the magnetic flux in the core and measure the current. All wires to the load bank are routed together through a ground fault current sensor that, together with a ground fault module, can detect ground faults in the load banks by measuring common mode currents. A load bank ground fault will occur if one or both load banks becomes partially submerged in seawater. The ground fault module output is connected to the CompactRIO master and integrated with other system fault protection. This method of sensing a ground fault is only effective, however, if the electrical system is ground (seawater) referenced elsewhere in the system, most likely at the WEC generator or the power converter. If the entire electrical system is isolated from chassis ground, an alternate method of detecting ground faults that uses the voltage transducers is possible. In this case, the wye center point of the three voltage transducers is connected

47 32 to ground. Under normal conditions with no ground fault present symmetric three-phase voltages will be measured due to symmetrical parasitic impedances between each phase and ground. When a ground fault occurs, however, an asymmetry will result in the threephase voltages with respect to ground that can be measured. This alternate method of ground fault sensing was used during the 212 Ocean Sentinel deployment with the WET-NZ. 3.4 Converter Hardware A schematic showing the electrical connections in the converter enclosure is shown in Figure 3-8, and an annotated photo of this equipment is shown in Figure 3-9. A list of the major converter components is included in Table 3-2. The mechanical layout of the converter enclosure was designed by ESI Motion under the direction of NNMREC. ESI also designed the enclosure, and installed and wired the equipment inside the enclosure. The design of the converter enclosure is very similar to the design of the switchgear enclosure. ESI s detailed wiring schematic for this equipment is included in Appendix B. To switchgear contactor and soft start 46 V max APS Rectifier Assembly Converter Enclosure APS Converter Assembly Output inductors To master crio 24 V regulated V + unregulated- Gate drive & protection NI CompactRio EtherCAT Slave Crydom DC6S7 + - To coolant pump (24V) Figure 3-8 Schematic of converter enclosure

48 33 The converter consists of a three-phase passive rectifier assembly and a pulse width modulated (PWM), insulated gate bipolar transistor (IGBT) converter assembly. Both are standard products produced by Applied Power Systems (Hicksville, NY), and both have 5 kw power capability at 46 V. The converter assembly includes three half bridge IGBT sections, three output current sensors, a dc bus voltage sensor, and a gate drive and fault protection circuit board that interfaces with the CompactRIO slave. This configuration was selected to make use of off-the-shelf converter equipment designed for three-phase equipment. Three output inductors are mounted outside the enclosure. The converter gate drive is designed for external input of PWM signals, which are provided by the CompactRIO slave, and internally generates the necessary dead time between the upper and lower IGBT in each half bridge. The converter fault protection circuitry monitors output current and cold plate temperature, and shuts off the IGBTs if a fault is detected. When the converter is in use, the rectifier inputs connect to the WEC generator through switchgear contactors. The passive rectifier interfaces with permanent magnet (PM) generators, and provides the dc input to the converter assembly. The three half bridge IGBT sections in the converter assembly, together with three output inductors mounted outside the enclosure, are used in parallel as a dc-dc buck converter to control the load bank.

49 34 Power cables Rectifier modules DC bus current sensor 24 Vdc & pump connectors Coolant quick disconnects RJ 45 EtherCAT connector APS Rectifier assembly CompactRIOslave IGBT modules Dc bus capacitors Dc bus bars Rectifier & converter Cold plates APS Converter assembly Output current sensors Figure 3-9 Photo of converter enclosure NNMREC anticipates that some WEC developers may require a converter to interface with an asynchronous generator. For this purpose, the existing passive rectifier assembly has the same mechanical footprint, wiring connections, and coolant connections as the existing converter assembly, so that a second, identical converter assembly can replace the passive rectifier assembly and be used as an active rectifier to control asynchronous generators. There is also sufficient space in enclosure to allow replacement of the existing 5 kw rectifier and converter assemblies with 1 kw assemblies in the future if necessary. These larger rectifier and converter assemblies are available from Applied Power Systems.

50 35 Table 3-2 Major converter components Description Manufacturer Qty Part Number Converter, 1A/12V Applied Power Systems 1 IAPL1T12-1 Rectifier, 2A/12V Applied Power Systems 1 IAPL-Rect Dc bus current sensor, 1A LEM 1 HAS 1-S Pump relay, solid state Crydom 1 DC6S7 Output inductors, 4 A, 75 μhy* MTE Corp. 3 DCA42 NI CompactRIO EtherCAT slave components described in Section 3.5 * Inductors located outside of converter enclosure The power converter is seawater cooled by a pumped coolant system. The heat producing rectifiers and IGBTs are mounted on cold plates in the rectifier and converter assemblies, and a keel cooler, a heat exchanger that is integrated with the aluminum hull of the buoy, is used to conduct heat to seawater. A schematic of this cooling system is shown in Figure 3-1, and an annotated photo of this equipment is shown in Figure A list of the major cooling system components is included in Table 3-3. This is an effective cooling system because seawater outside the Ocean Sentinel hull is cool (seawater temperature at the NNMREC test site never exceeds 18 ºC), and heat is easily conducted through the aluminum hull. A small, 15 liter per minute, magnetically driven centrifugal pump is used to circulate ethylene glycol coolant. The pump is controlled by the CompactRIO master through a solid state relay; the pump is turned on whenever the converter is enabled. An expansion tank is used to pressurize the system as coolant temperature increases, and an overflow tank catches fluid if the relief pressure of the expansion tank cap is exceeded. A flow meter is included for visual checks of the system, and is not integrated with the DAS. The keel cooler is oversized for the 5 kw rectifier and converter, and can provide cooling for 1 kw or larger converters.

51 36 Figure 3-1 Schematic of converter cooling system Inductors Converter Enclosure Expansion tank Coolant pump Flow meter Overflow tank Keel cooler Figure 3-11 Photos of converter coolant system Table 3-3 Major converter cooling system components Description Manufacturer Qty Part Number Pump, 4 GPM Johnson Pump 1 CM1P7-1, 24V Expansion tank, 5 Sure Marine 1 W2-674 Overflow tank Sure Marine Flow meter Hedland 1 H625-4

52 CompactRIO Control and Data Acquisition The CompactRIO system controls the switchgear and power converter and collects data. See Figure 3-12 for a schematic diagram that shows how the CompactRIO hardware is integrated with the Ocean Sentinel switchgear and power conversion equipment. The CompactRIO master, located in the DAS and telemetry enclosure, controls the switchgear contactors, records WEC output power calculated from voltage and current sensor data, records ocean wave data received from the Watchman 5 via a serial link, and communicates with the host computer on shore via the telemetry system described in Section The CompactRIO slave, housed in the converter enclosure, controls the converter and records converter data. The host computer provides the user interface for the system. Data is both stored in non-volatile memory on board the CompactRIO master, and also sent to the host computer for display. WEC Generator 3Ø Umbilical Disconnect Switchgear Load Banks 3Ø 25 kw 3Ø 25 kw 5 kw V Sense Soft start Converter Host PC TRIAXYS Wave buoy GPS 9 MHz Watchman 5 3G cell modem DAS & Telemetry 3Ø Iac 3Ø Vac Ethernet Serial NI 925 NI 925 NI 9267 NI 9474 crio 9114 FPGA crio 925 controller 3Ø EtherCAT Passive rectifier V, I, T sensors Gate drive NI 9144 EtherCAT slave FPGA To coolant pump Figure 3-12 CompactRIO master and slave integrated with switchgear and power converter

53 CompactRIO Hardware The CompactRIO hardware consists of the host computer on shore and the master and slave units on the Ocean Sentinel. The host is a windows based PC computer. Photos of the CompactRIO master and slave hardware are shown in Figure The master unit is located in the DAS and Telemetry enclosure shown in Figure 3-2. The slave unit is installed in the converter enclosure shown in Figure 3-9. Interface with the switchgear and power conversion equipment is through input/output (I/O) modules that plug into both the master and slave chassis. Each chassis includes a field programmable gate array (FPGA) back-plane. In the case of the master, the FPGA backplane connects directly to the real-time master controller, while in the case of the slave, the FPGA backplane connects to the master controller via Ethernet for control automation technology (EtherCAT) communications over an Ethernet cable. The real-time controller has 512 MB of Random Access Memory (RAM) and 8 GB of nonvolatile storage. The master has a Xilinx Virtex-5 FPGA, while the slave has a smaller 2 M gate FPGA. The master and slave units consist of the major components listed in Table 3-4 and Table 3-5, respectively. Figure 3-13 CompactRIO hardware

54 39 Table 3-4 CompactRIO master controller components Description Manufacturer Qty Part Number Real time controller National Instruments 1 crio-925 FPGA chassis National Instruments 1 crio-9114 Analog input module National Instruments 2 NI 925 GPS module National Instruments 1 NI 9267 Digital output module, 1A National Instruments 1 NI 9474 Table 3-5 CompactRIO slave components Description Manufacturer Qty Part Number EtherCAT extension National chassis Instruments 1 NI 9144 Analog input module National Instruments 1 NI 925 Digital I/O module National Instruments 1 NI 941 Digital output module, 1A National Instruments 1 NI Data Recording and Display The CompactRIO system saves data files to non-volatile memory in the master controller and also displays data on the host computer. Data files saved on the master can be retrieved later from shore via the wireless telemetry using file transfer protocol (FTP). Three separate types of data files are recorded, as summarized in Table 3-6. WEC and converter electrical data, which consists of WEC voltage, current, power, and power converter data, is continuously recorded at a 1 Hz rate in a NI Technical Data Management Streaming (TDMS) format, and is also displayed by the host. High speed voltage and current data, which consists of the three phase to ground voltages and the three-phase currents, is recorded at a 5 khz data rate in one minute duration data files,

55 4 only when commanded through the host interface. This data is not displayed by the host. Wave, wind, current, and power data is continuously recorded for every 2 minute period in a text file, and is also displayed by the host. This data is formatted per a National Marine Electronics Association (NMEA) 183 format. These data file formats are described in detail in Appendix C. Table 3-6 CompactRIO data file formats Data Data Rate Host Display Save to File File Format WEC & converter electrical 1 Hz Yes Continuous TDMS High speed When 5 khz No voltage & current commanded TDMS Wave, wind, Text; 2 min Yes Continuous current, & power NMEA Software Architecture CompactRIO system is programmed in NI LabVIEW software. The architecture of this software is shown in Figure Separate top level programs, or virtual instruments (VIs), run on the host PC, the master real-time controller, the master FPGA, and the slave FPGA. The real time VI is programmed with the LabVIEW Real-Time Module, and the two FPGA VIs are programmed with the LabVIEW FPGA Module. The host PC VI serves as the user interface, sending commands to and receiving data from the real-time VI. The real-time VI together with the master and slave FPGA VIs run independently of the host VI, and continue operating with the last set of commands after host communication is interrupted. The real-time VI communicates with the other VIs and also reads serial port data sent from the Watchman 5, processes commands from the host, and processes and records data. The master FPGA VI interfaces with current sensors, voltage sensors, and contactor drivers in the switchgear enclosure via I/O modules and also with a CompactRIO GPS module that is used for time synching. The slave FPGA VI performs power converter control and interfaces with the converter via I/O modules. The real-time VI loops run at 1 Hz to limit central processing unit (CPU)

56 41 usage to about 35% under normal conditions. The host VI loops also run at 1 Hz. The master and FPGA slave loops run as fast as 1 khz. Figure 3-14 Software architecture for CompactRIO system

57 Communication between top level LabVIEW VIs The different communication methods that are used between the different LabVIEW VIs in the system are listed in Table 3-7. Communications between the host and real-time VI use network streams configured for a one Hertz data rate. The network stream is a NI method of providing communications between two LabVIEW applications over a network. Communications between the real-time controller and the master FPGA use two different methods. One Hertz data and commands are written directly to front panel controls or read from front panel indicators in the FPGA VI. Transfer of high speed, 5 khz voltage and current waveform data uses a direct memory access (DMA) first in, first out (FIFO) method where a DMA engine automatically transfers data from the FPGA RAM to the real-time controller memory. Communications between the realtime controller and the slave FPGA use user-defined variables and a scan engine in the real-time controller; this is the only available method of communicating with the slave FPGA. Table 3-7 Methods used to communicate between different LabVIEW VIs VIs Communicating RT to and from Host RT to master FPGA (5 khz voltage & current data) RT to and from master FPGA RT to and from slave FPGA LabVIEW Communication Method Network streams DMA FIFO FPGA front panel controls and indicators User-defined variables Host Software The host VI provides the following functions within the CompactRIO system: 1. Creates network stream end points for communications with real-time VI. 2. Provides the front panel user display for one Hertz electrical data.

58 43 3. Decodes NMEA 183 ocean wave, wind, and ocean current data messages and provides a front panel user display. 4. Provides front panel controls for user input of contactor and power converter commands. 5. Provides optional automatic control of power converter commands with either 2 minute cycling or maximum power point tracking (MPPT). The host VI is built around the front panel user interface, which consists of multiple pages that can be selected by tab control. Screen shots of two different front panel pages are shown in Figure 3-15; one that includes control inputs for the contactors and power converter and also displays electrical data, and another that displays wave statistics data. Other pages display wave spectral data, ocean current data, meteorological data, further electrical data, provide file recording control, and MPPT power converter control. Electrical data is updated at 1 second intervals with the previous 5 minutes of data displayed; wave, wind, and ocean current data is updated every 2 minutes with the previous one week of data displayed. The data file recording page includes control of 5 khz voltage and current data recording, which is saved in 1 min intervals when prompted, and includes the option to turn individual electrical or wind, wave, and current data channels off to conserve real-time controller non-volatile memory.

59 44 Figure 3-15 Screen shots of the host PC user interface Real-Time Controller Software The real-time VI provides the following functions within the CompactRIO system: 1. Interfaces with the host VI, main FPGA VI, and slave FPGA VI.

60 45 2. Receives three-phase voltage and current data at a 5 khz rate from the master FPGA via DMA FIFO transfer, and creates 1 Hz data of the waveform data type. One second average WEC output power data is calculated from the voltage and current waveform data. The power calculation includes compensation for umbilical cable losses based on the measured current. 3. Reads 2 m wave, wind, and current data in NMEA 183 format from the serial port. 4. Averages 1s power data over each 2 m wave, wind, and current data measurement period, and combines the results with the wind, wave, and current data. 5. Records 1 Hz electrical, 5 khz voltage and current, and 2 m wind, wave, current, and power data to non-volatile memory. 6. Synchronizes the real-time controller clock to GPS time, in order to maintain synchronism with other data systems. 7. Determines converter primary and soft-start contactor states from host commands using a state machine. These functions are provided by multiple, parallel loops in the VI, as shown in Figure Single-process shared variables are used to communicate between the different loops. The variables are configured with real-time FIFO buffering in cases where deterministic data transfer is required Main FPGA Software The main FPGA VI is the simplest top-level VI, and provides the following functions within the CompactRIO system: 1. Interfaces with analog input modules that record three-phase voltage and current transducer inputs, and sends 5 khz waveform data to the real-time VI using DMA FIFO data transfer.

61 46 2. Interfaces a GPS module and a contactor driver module with front panel indicators and controls. The three-phase voltage and current transducer inputs are sampled at 4 khz by the first three inputs, respectively, of two separate NI 925 analog input modules. The NI 925 was selected for its high accuracy, and while each module does not have simultaneous sampling between different inputs, the same inputs in the two different modules do have simultaneous sampling to within 25 ns. All voltages and currents are oversampled at 4 khz, then each 8 samples are averaged to provide 5 khz data. The 5 khz data is transferred to the real-time VI using a DMA FIFO method that provides high speed, buffered data transfer to the real-time VI Slave FPGA Software The slave FPGA VI provides the following functions within the CompactRIO system to control the power converter: 1. Calculates converter output current command from a resistance command and provides proportional-integral current loops to regulate power converter output current. 2. Generates PWM gate drive signals for the power converter. 3. Monitors voltage, current, and fault signals from the power converter via an analog input module; shuts down power converter in case of fault and sends data to real-time VI via user-defined variables. The converter control functionality provided by the slave FPGA VI is shown in Figure The control input is the generator load resistance command that is input by the user through the host interface. The output (load bank) current commands I * 1, I * 2, and I * 3 are calculated from the resistance command and V dc, the measured bus voltage, per equation (3-1). 3 (3-1)

62 47 This equation assumes 1% converter efficiency, so some error is expected between the measured load resistance and the command; see Section for test results that quantify this error. The output currents I 1, I 2, and I 3 are regulated using three proportional-integral (PI) control loops; output current signals provided by the converter assembly are sampled with an analog input module. The calculation of the output current command and the PI control is implemented in one loop in the VI; this loop operates at approximately 15 khz. The output of the PI control loops is input to a second loop in the VI that generates the PWM gating signals. Communications between the two loops are through local variables. Permanent magnet generator (WEC) 3Ф Rectifier Module I dc 33 μf IGBT Converter Output inductors I 1 I 2 I 3 I out R loadbank V dc Gate drive Voltage threshold control V dc V thresh R dc * LowV C R loadbank x Sqrt V dc x 1/3 I1* I2* x x PI PI D1 D2 PWM PWM R dc * HighV Constant Rdc control R dc * I3* + - PI D3 x Output current loops Implemented in CompactRIO slave FPGA PWM Figure 3-16 Generator control used for 212 WET-NZ tests In the PWM loop, three, 3 khz triangle waves that are each staggered by 12º are generated. The PI outputs are compared to these triangle waves to produce three PWM patterns that have duty cycles proportional to the PI outputs. These PWM patterns are used to drive the upper IGBTs in the converter, through a digital output module; these three signals are inverted to drive the lower three converter IGBTs. The dead time between the upper and lower IGBTs is provided by the gate drive circuitry in the converter. The three output currents I 1, I 2, and I 3 are sampled at a 9 khz, and this

63 48 sampling is performed in synchronism with the 3 khz triangle waves used to generate the PWM signals in order to avoid instability due to the output current fluctuation that occurs over each PWM cycle. 3.6 Power Converter Simulations Simulations of the power converter and control were performed early in the design process using the PowerSimSystems module in MATLAB-Simulink. The simulation model, shown in Figure 3-17, included passive rectification of a permanent magnet generator and the complete PWM switching operation of the six IGBTs in the power converter. The power converter control that is implemented in the CompactRIO slave FPGA, described in Section 3.5.8, was also included. Vdc 2 5-7v 1 u Iout 1 I1 I1error PI(s) > Iout command K- Iout*/3 I3 I2 I2error I3error PI(s) PI(s) Triangle Delay 1/3 cycle < > < > Gating signals 1 Delay 2/3 cycle < A B C PM Generator A Vabc Iabc B a b C c VIac Vgen 4 Igen 5 Diode Bridge A B C Gating signals u 66u 8k + - IGBT half bridge 1 C1 G1 E1 / C2 G2 E2 v 2 Vdc IGBT half bridge 2 C1 G1 E1 / C2 G2 E2 -C- Dead time IGBT half bridge 3 C1 G1 E1 / C2 G2 E2 Vabc A Iabc B a b C c VIdc Vout 3 Iout 1 L1 75u L2 75u L3 75u Rload Figure 3-17 SimPowerSystems model; control at top and power electronics at bottom The primary purposes of simulating the power converter were to 1) predict the passive rectified, three-phase generator current waveforms for different loading, and 2) to

64 49 verify functionality of the control and current sharing between the three output inductors. The generator from the half-scale WET-NZ wave energy converter that was tested later in 212 was modeled for these simulations. These generator parameters are listed in Table 3-8. Table 3-8 Half-scale WET-NZ electrical generator specifications Manufacturer UQM Technologies Model 38 Type 3 phase, permanent magnet synchronous generator Back emf Sinusoidal Pole pairs 9 Inductance 225 μhy/phase Voltage constant 55 Vrms line-neutral/1 rpm Phase connection Wye connected with neutral isolated Samples of the simulation results are shown in Figure 3-18 and Figure 3-19 for a 2 kw generator output and a generator speed of 33 rpm. Figure 3-18 shows one of the generator phase currents. The non-sinusoidal waveform results from the passive rectification. The generator frequency is 5 Hz. Figure 3-19 shows the three inductor currents I 1, I 2, and I 3 and the total output current I out, which is the sum of I 1, I 2, and I 3. Each inductor current has a significant 3 khz switching current ripple. The current ripple in I out is much lower, however, and is at 9 khz due to the staggered phasing of the IGBT switching patters for the three outputs. Igen (A) Time (s) Figure 3-18 Simulated generator phase current at 2 kw output and a speed of 33 rpm

65 5 I 1 (A) I 2 (A) I 3 (A) I out (A) Time (s) Figure 3-19 Converter inductor and output currents with 2 kw output 3.7 Bench Testing A test bench of the switchgear and converter system was built to operate with the CompactRIO and test software early in the design process, before final hardware was complete. A schematic of the test bench is shown in Figure 3-2, and a photo is shown in Figure The test bench included a complete power converter assembly and the primary switchgear components. Most components are identical to those listed in Table 3-1 and Table 3-2 for the switchgear and converter, respectively, and were later available as spares for the Ocean Sentinel. Lower cost three-phase voltage and current sensors were substituted in the test bench, however, that did not have as high of accuracy as the devices used in the final hardware. The converter was either operated at low power or for very short times at high power, so that liquid cooling was not necessary. Testing was performed with the input connected directly to a voltage source rather than a permanent magnet generator. The test bench was operated both with the output short circuited and also with several different resistive loads to simulate different connections of the Ocean Sentinel load bank.

66 51 Vsense Vsense Vsense Driver Driver NI 925 NI 925 NI 9474 Driver Figure 3-2 Test bench schematic NI 925 NI 941 Converter Inductors Switchgear crio slave crio master Figure 3-21 Photo of test bench Sample test bench output voltage and current waveforms recorded with an oscilloscope are shown in Figure These waveforms were recorded while the test bench was operating with a 17 Vdc bus voltage and the output (load bank) short

67 52 circuited. This allowed operation at high output current with a low power input. The output current command was set to 1 A while these recordings were made. The converter was operating at very low duty cycle due to the low output voltage caused by the shorted output. The waveforms are similar to the simulated waveforms shown in Figure 3-19 in that the three currents that flow through the inductors had a 3 khz current ripple while the total output current had a 9 khz current ripple due to the staggered phasing of the PWM gate signals driving the IGBTs. Voltage (2 V/div) Current (A) V1 V2 V3.1.2 Time (s) I1 I2 I Time (s) Current (A) Iout = I1 + I2 + I3.1.2 Time (s) Figure 3-22 Bench test steady state voltage and current waveforms with V dc =17 V and short circuit output

68 Commissioning Tests Commissioning tests were performed to verify proper operation of the complete WEC electrical loading and data acquisition system after all equipment was installed on the Ocean Sentinel in August 212. Most of the systems were tested at the Port of Toledo Boat Yard before the Ocean Sentinel was launched, however, some tests could only be performed dock-side after launching and others after the Ocean Sentinel was operating at sea with the WET-NZ wave energy converter. Three important aspects of the design that were verified during commission tests were the accuracy of the threephase power measurements, the effectiveness of the liquid cooling for the converter, and the relationship between the commanded and measured resistive loading applied to the WEC generator Calibration Check of Power Measurements Because the bench test described in Section did not use the same current and voltage transducers as the final switchgear hardware, the first opportunity to verify the accuracy of the CompactRIO power recordings was during commissioning tests at the Port of Toledo Boat Yard. To test power measurements, the switchgear and converter system was operated with a 6 Hz, three-phase, 23 volt, 5 amp ac source connected in place of the WEC generator. Loads were applied directly by one 25 kw load bank via switchgear contactors, and also with power converter control of the 5 kw load bank. A secondary power measurement was made with an AEMC 3945 power quality analyzer, which has an accuracy of 1%, to check the CompactRIO recordings. The results are shown in Table 3-9. The error between the CompactRIO and AEMC 3945 power measurements was less than 1% at all load levels tested. The accuracy of the Ocean Sentinel power measurement is expected to be much better than 1% because the current and voltage transducers used for the measurement were selected for.5 class accuracy, to meet IEC specifications for WEC power performance assessment [1]. The results shown in Table 3-9 are consistent with this expectation, considering the 1% accuracy of the AEMC 3945 used as a reference.

69 54 Loading Method Table 3-9 Calibration check data for CompactRIO power measurement Voltage Average 3Ø Vll rms (V) Current Average 3Ø I rms (A) Power (W) AEMC 3945 Power (W) CompactRIO Power Measurement Error 25 kw load bank % Converter % Converter % Converter % Converter % Converter % Voltage and current waveforms recorded during the calibration checks are shown for both contactor switched loading of the 25 kw load bank and power converter loading in Figure When the load was provided by the power converter, the current was highly distorted because no inductance was added in series with the three-phase source. Sinusoidal currents existed with contactor switched loading. The results in Table 3-9 show that the power measurement error between the CompactRIO and AEMC 3945 was significantly higher when loading was applied by the power converter, apparently due to higher current harmonics, than when the load was directly applied by the resistive load bank. What portion of this error was due to inaccuracy of the Ocean Sentinel measurements versus inaccuracy of the AEMC 3945 measurements is unknown.

70 55 Voltage (V) Voltage (V) Phase A line voltage and current 25 kw load bank load Va Ia I rms = 27 A V ln rms = 112 V Time (s) Phase A line voltage and current converter load Va 1 Ia I rms = 3 A V ln rms = 112 V Time (s) Current (A) Current (A) Figure 3-23 Sample current and voltage waveforms recorded during calibration check The calibration check of the power measurement described above was made without the umbilical cable in the system. It is important to note that the voltage drop in the umbilical cable is another source of error in the Ocean Sentinel power measurements. The Ocean Sentinel measurements of WEC output power include compensation for the umbilical cable voltage drop based on the measured current and the estimated umbilical resistance, however, variations in cable resistance with temperature are not compensated. Since the umbilical is designed to keep voltage drops to less than 5% of rated WEC output voltage at full power and the umbilical temperature is uniformly cooled by seawater, it is expected that the additional error from this source will be less than 1% Thermal Test Final testing of the liquid cooling system for the Ocean Sentinel converter, described in Section 3.4, was performed dock-side at the OSU Ship Operations facility in Newport, OR. This was the first opportunity to test the cooling system with seawater outside the Ocean Sentinel hull; cooling is provided through the hull by the keel cooler. The power converter was operated with a 6 Hz, three-phase, 28 volt, 5 amp ac source connected in place of the WEC generator. The cold plate and seawater temperature that

71 56 was recorded for over an hour while the power converter was operated with a 5 amp input is shown in Figure Cold plate temperature was recorded by the CompactRIO; seawater temperature recordings from the National Oceanographic and Atmospheric Administration (NOAA) South Beach tide station that is adjacent to the OSU Ship Operations dock were used. Temperature (Deg C) Power (W) or Current (A) 3 28 Converter Cold Plate Temperature Seawater Temperature 22 2 Cold plate temperature :26 13:4 13:55 14:9 14:24 Time (hours:seconds) 14:38 14:52 15:7 15: Ac line current (avg 3 Ø) Output currrent (avg 3 outputs) Dc bus current Power V ll rms = 28 V Converter Power and Currents 13:26 13:4 13:55 14:9 14:24 14:38 14:52 15:7 15:21 Time (hours:seconds) Figure 3-24 Thermal test results The Figure 3-24 results show that at the end of the test period, cold plate temperature approached 29 ºC, approximately 15 ºC above the 14 ºC seawater. The converter is rated for a maximum cold plate temperature of 8 ºC when operating at full power, and seawater temperature at the NNMREC ocean test site can be as high as 2 ºC, so a seawater to cold plate temperature rise of up to 6 ºC is acceptable under worst case conditions. A larger power source was not available to operate the converter at its full 5 kw, 46 volt, 63 amp capability during the commissioning test. Because converter power dissipation is roughly proportional to both operating current and voltage, the power dissipation can be expected to be roughly three times higher under full power conditions than for the 28 volt, 5 amp test condition. Because the cold plate to seawater temperature rise of 15 ºC measured during the test is only one-fourth of the 6

72 57 ºC acceptable under worst case conditions, however, the test results confirm that the cooling system does have sufficient capability for full power operation of the converter. The coolant flow rate was less than one gallon per minute while these tests were conducted. A higher coolant rate would reduce the cold plate to seawater temperature rise. If a larger power converter is installed on board the Ocean Sentinel in the future, some simple modifications to the existing coolant system are possible that will increase the coolant flow rate. The pump, rated for 4 gallons per minute at low pressure, can be replaced with a larger model, also, quick release disconnects that are used to connect coolant hoses to the power converter restrict coolant flow and these can be replaced to improve performance Characterization of Constant Resistance Control During testing of the half-scale WET-NZ wave energy converter, resistive loads were applied to the electrical generator using the Ocean Sentinel power converter. The loading applied was per the commanded resistance R dc from the CompactRIO host interface, using the regulation system described in Section implemented in the CompactRIO slave FPGA. Final measurements of the applied versus commanded resistance were not be made until the converter was operated with the WEC generator at sea. Previous testing had been performed with 6 Hz, three-phase ac sources in place of the generator. The WET-NZ generator operates at higher frequencies, as high as 4 Hz, and includes generator reactance that smoothes input currents. Voltage, current, and power data collected during initial operation of the WET-NZ with constant resistance commands was analyzed using the method showed in Figure 3-25 to estimate applied resistances. In the left plot, one second data collected with a R dc command of 32 Ω is plotted with converter dc bus voltage on the vertical axis and dc current on the horizontal axis. The slope of the trend line gives the actual dc resistance applied. In the right plot, data collected with a 32 Ω R dc command is plotted with WEC output power on the vertical axis and the sum of squares of the three rms phase currents

73 58 on the horizontal axis. In this case, the slope of the trend line gives the average per phase equivalent three-phase wye resistance that is applied. 2 R dc command = 32 Ω 2 R dc command = 32 Ω Dc Voltage (V) 15 1 Power (W) R dc measured 5 R ac measured = trendline slope = trendline slope = 25.7 Ω = 13.5 Ω Dc Current (A) I I Ib I (A 2 a_rms2 + I b_rms2 + I c_rms2 (A 2 ) ) Figure 3-25 Plots showing how applied load resistances were calculated from data The method shown in Figure 3-25 was used to determine the applied dc and ac wye resistances for five different commanded resistance values between 4 Ω and 64 Ω. A plot of measured versus commanded resistance is shown in Figure As expected, the results show that applied dc resistance is less than commanded dc resistance, because the regulation system described in Section assumes 1% converter efficiency while in reality, there are converter losses that need to be supplied by the WEC generator. The difference between applied and commanded resistance is greater at larger resistances, when the converter operates at lower load and lower efficiency is expected. The ac wye resistance is always lower than the dc resistance R dc ; the ratio between the two is very close to.55, which is predicted by equation (3.1).

74 59 Measured Resistance (Ohms) Measured versus Commanded Resistance Rdc measured (Ohms) Rac wye measured (Ohms) y =.42x x y =.32x x Commanded Resistance Rdc (Ohms) Figure 3-26 Measured versus commanded resistance

75 6 4 WET-NZ TESTING 4.1 Introduction The Ocean Sentinel was deployed for the first time in August 212 to test an experimental half-scale version of the WET-NZ wave energy converter. The Ocean Sentinel and WET-NZ were moored at NNMREC s open-ocean test site north of Newport, OR for a six-week period from August 22, 212 until October 5, 212 while the testing was performed. The WET-NZ tests had a number of objectives, as follows: 1. To deploy in open ocean for the first time in US waters, demonstrate and characterize the performance of the WET-NZ at half scale. 2. To demonstrate operation of the Ocean Sentinel, and to gain experience testing a WEC with the Ocean Sentinel. 3. To gain experience deploying both the Ocean Sentinel and a WEC in the ocean. 4. To perform environmental monitoring during ocean testing of a WEC. The results of the WET-NZ tests are presented here, focusing on the first two objectives. The operation of both the WET-NZ and the Ocean Sentinel were successfully demonstrated during the deployment, and the performance of the half-scale WET-NZ was characterized under a wide range of loading and sea conditions. The Ocean Sentinel power converter together with the CompactRIO control and data acquisition system were used to control the load applied to the WET-NZ power-takeoff (PTO) and to collect WEC power and ocean data throughout the deployment period. This allowed experimentation with different adaptive control methods during the deployment. Following the deployment, data collected by the Ocean Sentinel was analyzed in order to characterize WET-NZ performance when operated with the different control methods. Although the Ocean Sentinel was operated by NNMREC staff throughout the WET-NZ deployment, all WEC data collected on board the Ocean Sentinel during this period is proprietary to the WET-NZ developers. NNMREC has been given permission to publish this data here. Data was also collected on board the WET-NZ using an

76 61 independent data acquisition system; that data is not presented here, although reference is made to analysis of this data to support interpretation of the NNMREC results. 4.2 Description of the WET-NZ A sketch of the half-scale WET-NZ and a photo of it at sea with the Ocean Sentinel is shown in Figure 4-1. Characteristics of the device are listed in Table 4-1. The WET-NZ is the product of a research consortium between Callaghan Innovation, a New Zealand Crown Entity, and Power Projects Limited (PPL), a Wellington, New Zealand private company. The Oregon deployment was project managed by Northwest Energy Innovations (NWEI), a Portland, OR firm. The device tested in 212 was half-scale by length; output power scaling per the Froude similitude criteria was 1/11 relative to a nominal full scale device [4]. The WET-NZ consists of a long submerged hull, with a power pod mounted on top that includes a cylindrical float and the power take-off system. The hull of the half-scale WET-NZ tested with the Ocean Sentinel was fabricated at Oregon Iron Works in Portland, OR, and the power pod was fabricated and assembled in New Zealand.

77 62 Power pod Float Hull Figure 4-1 The half-scale WET-NZ WEC Table 4-1 Half-scale WET-NZ device characteristics Length scaling ratio* ½ Power scaling ratio (Froude)* 1/11 Peak power 2 kw Draft 15 m Spar natural period 15 s Float natural period 3.5 s *Relative to full-scale device The float of the WET-NZ is coupled through its shaft to the PTO system and rotates up and down in the waves to generate power. The WET-NZ is designed to be a slack-moored and self-reacting design; the hull is flooded with seawater to give it a large inertia for the float to react against [11]. The natural period of the half-scale WET-NZ spar, which consists of all of the device other than the float, is 15 seconds, and the natural period of the half-scale float is 3.5 seconds. Due to these natural periods, Callaghan Innovation simulations predicted that the half-scale device would not generate significant

78 63 power for portions of the wave spectra with periods longer than approximately 9 seconds. A full-scale device, however, is expected to have longer natural periods and to produce power from longer period waves present in the open ocean wave spectra. The power-takeoff system for the WET-NZ is shown in Figure 4-2. A crankshaft that connects to the shaft of the float extends and retracts hydraulic cylinders. The hydraulic cylinders provide pressure to a hydraulic system that includes a hydraulic motor and a small accumulator. The hydraulic motor drives the permanent magnet generator that was controlled by the Ocean Sentinel power converter during the test. The specifications for this generator are listed in Table 3-8. The hydraulic drive is configured such that the generator only rotates in one direction. An accumulator provides a selected amount of energy storage within the hydraulic transmission system, so that the generator speed and torque do not necessarily decrease to zero when the shaft of the float reverses direction twice per ocean wave cycle. This causes a non-linear relationship between the speed and force of the float with respect to the speed and torque of the generator. Hydraulic cylinder Float center of rotation Crankshaft Figure 4-2 The WET-NZ power-takeoff system

79 64 It should be noted that the WET-NZ device contained on-board batteries, generator, load bank and control system as backup in the event that the umbilical connection to the Ocean Sentinel was lost. This charging system created a substantial continuous load on the WEC PTO, irrespective of the power supply provided from the umbilical. The Ocean Sentinel instrumentation was unable to measure this load. The backup systems were not required at any stage during the deployment. 4.3 Test Setup The WET-NZ, the Ocean Sentinel, and the TRIAXYS wave buoy were deployed at the NNMREC scaled test site with the layout shown in Figure 4-3. Refer to Figure 1-3 for a map showing the location of the test site. Three-point mooring systems were used for both the WET-NZ and Ocean Sentinel. The TRIAXYS wave buoy was moored to the north side of the test site where it was more protected from pleasure and fishing vessel traffic. The four corners of the 35 meter by 25 meter test area were marked by corner buoys, per US Coast Guard requirements. Ocean swell is typically from the westnorthwest at the test site in August and September, with winds typically from the northwest.

80 65 Figure 4-3 Test site layout for 212 WET-NZ testing The low power umbilical described in Section 2.5 was used to connect the WET- NZ to the Ocean Sentinel. This umbilical provided the following electrical connections during the test: 1. The three-phase WET-NZ generator output to the Ocean Sentinel switchgear. 2. System grounds of the WET-NZ and Ocean Sentinel volt ac Ocean Sentinel power to instrumentation on board the WET-NZ. 4. The coils of WET-NZ generator output contactors to the Ocean Sentinel safety system. This umbilical did not provide communications between the WET-NZ and Ocean Sentinel, so data systems located on board the WET-NZ recorded data independently of the Ocean Sentinel and used their own independent telemetry system. All data systems were time synchronized by GPS.

81 66 The WET-NZ generator was controlled by the Ocean Sentinel power conversion equipment and the CompactRIO data acquisition and control system that is described in detail in Chapter 3 throughout the test. The specific configuration used for the WET-NZ tests were as follows: The Ocean Sentinel power converter was used alone to provide the WET-NZ generator load. The two additional 25 kw contactor controlled load banks that are also available on the Ocean Sentinel were not used. The load for the Ocean Sentinel power converter was provided by the 5 kw load bank, with the individual elements wired in series-parallel for a load bank resistance of 4.2 Ω. 4.4 Test Sequence The following tests were performed during the WET-NZ deployment: 1. Constant resistance load characterization 2. Voltage threshold switched resistance load characterization for two modes: Latching (lower resistance below a voltage threshold) Declutching (higher resistance below a voltage threshold) 3. Maximum Power Point Tracking (MPPT) using two algorithms: Perturb and observe Cycling 4. Survivability tests with no load applied. The cumulative operating hours spent performing each of these tests during the six-week deployment period are listed in Table 4-2. In order to conduct each test under a wide range of sea conditions, each test was performed in short segments that were interspersed throughout the deployment period. The greatest portion of the deployment period was spent collecting constant resistance load data. This was the default operating condition when other tests were not being conducted. Much smaller portions of the deployment period were dedicated to the voltage threshold switching and the MPPT tests.

82 67 Some constant load resistance data was also collected when operating with MPPT algorithms that held the resistance setting constant for 2 minute intervals. Table 4-2 Cumulative duration of WET-NZ tests Test Cumulative Duration Total deployment period 156 Hours Device loaded by Ocean Sentinel 785 Hours Constant resistance loading 559 Hours Voltage threshold load switching : latching 1 Hours Voltage threshold load switching : declutching 43 Hours Maximum Power Point Tracking (MPPT)* 144 Hours Other (load applied) 83 Hours Survivability tests (no load) 3 Hours Not operating 268 Hours * Constant resistance load data was collected during some MPPT tests Conditions with high seas were selected for the no-load survivability tests. Three separate, one hour long tests were performed when no load was applied to the WET-NZ generator but the generator output contactors were left closed, so that the generator was still connected to the Ocean Sentinel switchgear through the umbilical. This allowed the Ocean Sentinel DAS to record no-load generator voltage data during these periods. No issues were encountered during these survivability tests; detailed data are not presented here. There were also periods of time when the WET-NZ was not operated; both during initial at-sea commissioning and during maintenance periods. Data was not recorded during most of this time. 4.5 Constant Resistance Load Tests During these tests, the WET-NZ generator was loaded by the Ocean Sentinel power converter so that generator output current was proportional to generator output voltage to give a constant resistance load. This method provides an approximation to constant damping control of the WEC float where the force applied to the float is kept proportional to the speed of the float, assuming that float force is proportional to generator torque and float speed is proportional to generator speed. Generator torque is

83 68 proportional to current, and generator speed is proportional to voltage. In the case of the WET-NZ, however, the PTO hydraulics and the rotating float create non-linearities between float and generator speed, and also between the float force and generator torque, so that fixed resistance control only approximates constant damping control. Testing was performed with constant resistance control because it was simple to implement. Although PTO non-linearities were expected to have some effect, results were still expected to be consistent with analysis of the WEC that assumed constant damping. The constant resistance control of the Ocean Sentinel power converter was implemented in the CompactRIO, as described in Section and shown in Figure The dc bus current in the converter was controlled to be proportional to the dc bus voltage, rather than directly regulating the ac output of the generator. To avoid confusion, all results and analysis are presented in terms of the control quantity R dc, which is the commanded resistance at the converter dc bus shown in Figure 3-16; R dc is referred to as the load resistance. Due to the three-phase rectification, the effective three-phase wye resistance applied to the generator is lower than R dc by a factor of approximately.55 at high load. The relationship is non-linear, however, due to higher relative losses in the converter at low load (high resistance); see Figure 3-26.

84 Test Method During initial testing, a load cycling method was developed where the Ocean Sentinel CompactRIO host was programmed to cycle repeatedly through a sequence of different R dc settings every 2 minutes, in order to collect data with different fixed loads applied under similar sea conditions. The 2 minute period for each load step was synchronized with the 2 minute measurement period of the TRIAXYS wave buoy. This load cycling method was used for the remainder of the test, and is illustrated in Figure 4-4, which shows plots of time series data recorded during a short segment of the test. In this example, the R dc sequence Ω was used. The upper plots show the average power recorded for each 2 minute interval together with the R dc setting for that interval. The lower plots show the significant wave height (H m ) and energy periods (T e ) recorded by the TRIAXYS wave buoy for each 2 minute period. Similar data plots were used to assess performance during the course of the test. In this example output power is usually higher at the intermediate R dc values than at 8 Ω or 128 Ω. Most testing was performed with the R dc sequence Ω although during some periods a 4 Ω step was added. In addition to data collected using the load cycling technique, a small amount of constant resistance load data was also collected during the MPPT tests.

85 7 Resistance (Ohms) Hm (m) Rdc % Power Rdc cycled in 2 minute intervals between Rdc = Ohms 9/14 : 9/14 2:24 9/14 4:48 9/14 7:12 9/14 9:36 9/14 12: 9/14 14:24 9/14 16:48 Date Time (UTC) Hm Te /14 : 9/14 2:24 9/14 4:48 9/14 7:12 9/14 9:36 9/14 12: 9/14 14:24 9/14 16:48 Date Time (UTC) Figure 4-4 Sample time series data for constant load resistance tests 25% 2% 15% 1% 5% % Power (% of deployment maximum) Te (s) Sea Conditions By the end of the deployment period, a large set of 2 minute data samples had been collected for constant resistance operation with a wide range of R dc values. A histogram of the 2 minute sample counts for each one-half meter wide H m and one second wide T e bin is shown in Figure 4-5 that includes data for all R dc settings used. H m and T e are calculated from zero and first negative moments m and m -1 of the wave spectra recorded by the TRIAXYS wave buoy, per equations (4-1) and (4-2). H m 4 m (4-1) T e m 1 m (4-2)

86 71 Sample Count (2 min sample periods) Aug 27 - Oct 5, constant Rdc operating data Hm Bin Center (m) Te Bin Center (s) Figure 4-5 Histogram of sample counts for constant load resistance data; all R dc settings included The half-scale WET-NZ device was not expected to produce significant power from portions of the wave spectra with periods greater than approximately 9 seconds. More test time was therefore desired in conditions with lower T e, especially in the 6 to 8 second range, than occurred during the deployment. Much of the testing was carried out in longer period seas, where the WET-NZ responded more to the shorter wind waves in the spectra than to the ocean swell Data Analysis After the deployment period, data collected by the Ocean Sentinel was analyzed using MATLAB to plot characterization curves showing the power output of the device

87 72 with respect to the R dc settings under different sea conditions. Data included average power, R dc, and both ocean wave statistics and spectra for each 2 minute sample period. The simplest method of presenting power output with respect to R dc under different sea conditions is shown in Figure 4-6, where the data is binned by H m and T e and individual plots are presented for each bin. WET-NZ output power is shown as the percent of maximum WET-NZ output power measured during the deployment period to protect IRL proprietary data. For the most part the Figure 4-6 plots do not show distinct trends in output power with respect to R dc. This is partly due to the large variation in wave spectra that occurred within the same H m -T e bins and the variation in WEC output power that resulted. This effect is illustrated in Figure 4-7, where the wave spectra for two different 2 minute data periods are plotted that both have the same T e (9.1 seconds), but significantly different spectral shapes. The upper spectrum has a single peak at.11 Hz (9.1 seconds), while the lower spectrum has two peaks, one at.1 Hz (1 seconds) and another at.17 Hz (5.9 seconds). The second peak in the lower spectra is due to the presence of locally generated, shorter period wind waves in addition to longer period ocean swell. The half-scale WET-NZ was not expected to produce significant power for portions of the wave spectra with periods greater than 9 seconds (frequencies below.11 Hz). The device was therefore expected to produce higher power from seas with the lower spectrum in Figure 4-7, which has more energy content above.11 Hz, than from the lower spectrum when the same R dc settings were used, even though both data samples are included in the same T e bin in Figure 4-6.

88 73 Hm 2.75m Te 7.5s Hm 2.75m Te 8.5s Hm 2.75m Te 9.5s Hm 2.75m Te 1.5s Hm 2.75m Te 11.5s Output Power (% of deployment maximum) 5 1 Hm 2.25m Te 7.5s Hm 1.75m Te 7.5s Hm 1.25m Te 7.5s Hm 2.25m Te 8.5s Hm 1.75m Te 8.5s Hm 1.25m Te 8.5s Hm 2.25m Te 9.5s Hm 1.75m Te 9.5s Hm 1.25m Te 9.5s Hm 2.25m Te 1.5s Hm 1.75m Te 1.5s Hm 1.25m Te 1.5s Hm 2.25m Te 11.5s Hm 1.75m Te 11.5s Hm 1.25m Te 11.5s Rdc Figure 4-6 WET-NZ power output versus load resistance, binned by H m and T e

89 74 S (m 2 /Hz) S (m 2 /Hz) Frequency (Hz) Sept 28, 8:4 UTC: H m = 1.4 m T e = 9.1 s Oct 1, 17:2 UTC: H m = 1.6 m T e = 9.1 s Frequency (Hz) Figure 4-7 Examples of two different wave spectra with similar T e To remove the effect of spectral variation from the WEC output power versus R dc curves, it was necessary to normalize WEC output power data with respect to the expected power using the average response of the device to the energy flux spectra. The average response of the WET-NZ was estimated from the output power and spectral data using equation (4-3) and a least squares method: P 1 P J 1 f 1 J 1 f 2... J 1 f m l f 1 2 J 2 f 1 J 2 f 2... J 2 f m l f 2 * (4-3) P n J n f 1 J n f 2... J n f m l f m where [P 1 P 2 P n ] T is a vector made up of all 2 minute average power measurements made while a constant resistance load (all R dc values) was applied to the WET-NZ; J n (f 1 ) J n (f 2 ) J n (f m ) are the frequency components of energy flux spectra associated with the n th power measurement P n ; and [l(f 1 ) l(f 2 ) l(f m )] T is the average response of the WET- NZ to the energy flux spectra that was solved for. The average response represents the frequency distribution of what is commonly referred to as the capture length of a WEC [1]. The spectral components for energy flux, or power per meter crest length of the seas, were calculated from the wave spectra using equation (4-4):

90 75 J f i ρ g C g f i S f i (4-4) where J(f i ) denotes the energy flux in W/m at frequency f i, ρ is seawater density (125 kg/m 2 ), g is the acceleration of gravity (9.8 m/s 2 ), C G (f i ) is the group velocity at frequency f i, and S(f i ) is the wave spectral component at frequency f i measured by the TRIAXYS wave buoy. The group velocity at frequency f i, was calculated using equation (4-5): C G f i 1 2 g tanh k k i h 1 2k ih (4-5) i sinh 2k i h where k i is the wavenumber at spectral frequency f i and h is the depth (5 m at the test site). The wavenumber was calculated using the dispersion relationship, equation (4-6), which requires a recursive solution: 2π f i 2 g k i tanh k i h (4-6) Note that equation (4-5) for the group velocity is often simplified by a deep water approximation when the depth is greater than half the wavelength (kh greater than π), however, at 5 m depth this approximation is only valid for wave periods less than 8 seconds and was not used for this analysis. The average response of the WET-NZ to the energy flux spectra, estimated using equation (4-3), is shown in Figure 4-8. The vertical axis scale is not included to protect WET-NZ proprietary data. 1675, twenty minute data samples were used in the calculation; data with all values of R dc were included. To improve the fit to the data, the solution was held constant within frequency intervals corresponding to integral wave periods. As expected, the response of the WET-NZ was negligible for frequencies less than.11 Hz (9 second period). The response was greatest for frequencies between.17 Hz and.2 Hz (5 second to 6 second periods). The results for frequencies higher than about.2 Hz are less accurate because the spectral energy was usually low above that frequency. The average response was used to calculate the expected output power of the WET-NZ based on the measured wave spectra, per equation (4.7): J f 1 J f 2... J f m l f 1 l f 2... l f m (4.7)

91 76 where J(f i ) are the energy flux components at frequency f i calculated per equation (4-4) and l(f i ) are the average response components at frequency f i shown in Figure 4-8. To check this method, the WET-NZ output power data is plotted against expected power, calculated per equation (4.7), in Figure 4-9. The output power is well correlated with the expected power, considering that some variation in output power is expected due to wide range of R dc settings included in the data sets. Average response (m) 1 9s 1 8s 1 7s 1 6s Average response frequency (Hz) 1 5s 1 4s Figure 4-8 Average response of WET-NZ to wave energy flux spectra 7 Power (% of deployment maximum) Expected Power (% of deployment maximum) Figure 4-9 Power vs. expected power for all constant resistance load data

92 77 To best present the power versus R dc characterization data, it was desirable to bin the data using a method that better segregates it per the expected response of the WET- NZ than simply binning per T e and H m as done in Figure 4-6. This was done by calculating modified versions of H m and T e that take into account the response of the WET-NZ. This method is shown by example in Figure 4-1. In the top plot of Figure 4-1, the average response of the WET-NZ is plotted along with a normalized version of this response that has a maximum of one. A sample wave spectrum is shown in the center plot. A modified version of this spectrum is shown at the bottom of Figure 4-1 that takes the response of the WET-NZ into account. The modified spectrum is the product of the normalized response in the top plot and the spectrum in the center plot and is the portion of spectrum that the WET-NZ responded to. The modified significant wave height H m mod and modified energy period T e mod were calculated from moments of the modified spectrum using equations (4-1) and (4-2), and were used for data binning.

93 78 Average WET NZ response Normalized response Average response Normalized response Frequency (Hz).5 Average response (m) Example spectrum: Oct 1, 17:2 UTC: Hm = 1.6 m Te = 9.1 s Results for example: H m mod =.9 m T e mod = 5.6 s Modified S (m 2 /Hz) S (m 2 /Hz) 2 1 Wave spectrum Frequency (Hz) 1.8 Modified spectrum Frequency (Hz) Figure 4-1 Data analysis example Results and Discussion Characterization curves for WET-NZ output power versus R dc are shown in Figure 4-11, where the analysis methods described above are used to normalize and bin the data. Separate plots are shown in each data bin, where the binning is done by H m mod and T e mod. Power is normalized to the expected power calculated per equation (4.7). While significant scatter still exists in these plots, clear trends in the data can be seen for the data bins with larger H m mod and T e mod, with a maximum in output power occurring somewhere in the range of 1 Ω to 5 Ω.

94 79 Hm mod m Te mod 5.25 s 2 Hm mod m Te mod 5.75 s 2 Hm mod m Te mod 6.25 s Normalized Power = P measured /P expected Hm mod.875 m Te mod 5.25 s Hm mod.625 m Te mod 5.25 s Hm mod.875 m Te mod 5.75 s 2 1 Hm mod.625 m Te mod 5.75 s Hm mod.875 m Te mod 6.25 s 2 Hm mod.625 m Te mod 6.25 s Rdc Figure 4-11 Normalized power vs. R dc binned by H m mod & T e mod WET-NZ simulations of the half-scale device predicted a more pronounced optimum power at a lower value of R dc than is seen in Figure WET-NZ has subsequently analyzed PTO data collected on board the WEC and has shown that significant power losses and other inefficiencies existed within the half-scale implementation of the PTO throughout the test period. Causes included the continuous load from the charger system and overly conservative settings on the hydraulic overload protection systems. These results are proprietary to WET-NZ and are not presented here. The effect was additional loading on the WET-NZ float throughout the deployment, so that the device was not operating at an optimum load regardless of the resistance applied to the generator. Accordingly, only a small portion of the power that was extracted from the waves during the tests reached the WEC electrical generator for measurement by the Ocean Sentinel. These effects broadened the power response of the device so that R dc only had a modest influence on output power during the tests, and are consistent with the

95 8 results shown in Figure WET-NZ has been able to identify and quantify the different PTO loss paths, further analyze device operation during the test, and determine improvements to the scaled PTO design that will correct these problems for future deployments. 4.6 Voltage Threshold Switching Tests During these tests, the WET-NZ generator was controlled by the Ocean Sentinel power converter so that two different values of R dc were applied, the first below and the second above a generator output voltage threshold V thresh. Assuming that generator voltage is proportional to float speed and generator current is proportional to float force, this is a simple method of applying different values of damping to the float above and below a speed threshold, in order to implement what is often referred to as either latching or declutching control in the literature [13]: 1. Under latching control, the float is locked into position twice per wave cycle at the moments when float velocity decreases to zero, then released after a set time delay and operated with constant damping until the velocity decreases to zero again; 2. Under declutching control, the float is released to move without resistance twice per wave cycle at the moments when float velocity decreases to zero; after a set time delay a constant damping is applied to the float. Ideally that level of damping is applied until the float velocity decreases to zero again. Latching control was implemented by setting R dc to a low value below V thresh and to a higher value above V thresh ; the opposite was used for declutching control. Both latching and declutching are reactive methods of control that attempt to introduce a phase shift between the force applied to the float and the velocity of the float; latching causes the velocity to lag the force and declutching causes the velocity to lead the force. The control provided by the implementations of these methods used for the WET-NZ tests differed somewhat from that described in the literature, due to the non-linear relationship between float speed and generator speed caused by the hydraulic accumulator in the

96 81 WET-NZ PTO. Due to the PTO accumulator, the WET-NZ generator speed does not necessarily decrease to zero when the float speed goes to zero each wave cycle. As a result, when R dc was switched at V thresh, the change in loading did not necessarily occur every wave cycle for low V thresh settings, also, variations in the phase shift probably existed from one wave cycle to the next when the load switching did occur. Although not ideal, the voltage switching method was relatively simple to implement in the Ocean Sentinel power converter control and allowed the latching and declutching tests to be performed without implementing more sophisticated hydraulic control in the WET-NZ PTO. The voltage threshold control of the Ocean Sentinel power converter was implemented in the CompactRIO, as described in Section and shown in Figure The voltage of the dc bus is compared to V thresh to determine whether to apply a high or low voltage resistance setting. A hysteresis of 1 volts, not adjustable during testing, is included in the voltage comparison to eliminate jitter. The V thresh, high voltage R dc setting, and low voltage R dc setting are inputs from the CompactRIO host interface. When operating at either the high or low R dc setting, the control regulates the dc bus current in the converter so that it is proportional to the dc bus voltage using the same method used for constant R dc control Test Method A similar 2 minute cycling method was used during the voltage threshold switching tests as was used for the constant resistance tests (see Section 4.5), except for during these tests V thresh as well as the high voltage R dc setting could be cycled to different values for each 2 minute load step. The most common settings used during the voltage threshold tests are listed in Table 4-3 for both the latching and declutching cases. In the case of latching control, the low voltage R dc was always set as low as possible, to 4.2 Ω, which is the resistance of the Ocean Sentinel load bank as configured for the WET-NZ tests. High R dc settings of either 32 Ω or 64 Ω were used, which were intended to be in the optimum range for constant resistance loading. Voltage thresholds in the

97 82 range of V to 6V were used for the latching tests; V thresh of V gives a constant R dc load equal to the high voltage R dc setting and was included as a baseline for comparison. These thresholds were very small relative to the peak no-load voltages observed during the tests. In the case of the declutching tests, a low voltage R dc setting of 128 Ω was used together with either 32 Ω or 64 Ω above V thresh ; voltage thresholds were between V and 2V. A larger voltage switching hysteresis than 1 volts would have been better for the declutching tests to delay switching back to high R dc until later in the wave cycle, however, a 1 volt hysteresis was fixed in the control and could not be adjusted during the test. More data was collected for both the latching and declutching tests with the high voltage R dc set to 32 Ω than 64 Ω. Experimentation was also done with various other voltage threshold and resistance settings for short periods during the deployment. Table 4-3 Settings typically used during voltage threshold load cycling sequences Low Voltage R dc High Voltage R dc V thresh Sequence Latching control 4 Ω 32 Ω V - 1V - 2V - 4V 6V 4 Ω 64 Ω or V - 2V 4V 6V Declutching control 128 Ω 32 Ω V - 5V 1V 128 Ω 64 Ω V - 1V 2V Time series plots of data recorded during a short segment of the test are shown in Figure 4-12 to illustrate the load cycling method used. In this example, where the latching control is being tested, a V thresh cycling sequence of volts was used, while R dc was switched between 4 Ω below V thresh and either 32 Ω or 64 Ω above V thresh. The load settings were changed every 2 minutes, with the 2 minute load step intervals synchronized with the TRIAXYS wave buoy measurement periods. Load sequences used during other test intervals often only included the 32 Ω high voltage R dc setting.

98 83 Resistance (Ohms) Hm (m) /25 9:36 9/25 12: 9/25 14:24 9/25 16:48 9/25 19:12 9/25 21:36 9/26 : 9/26 2:24 Date Time (UTC) Vthresh RhighV % Power Hm Te Vthresh cycled in 2 minute intervals between Vthresh = V RlowV = 4 Ω, RhighV = 32 or 64 Ω 9/25 9:36 9/25 12: 9/25 14:24 9/25 16:48 9/25 19:12 9/25 21:36 9/26 : 9/26 2:24 Date Time (UTC) Figure 4-12 Sample time series data for voltage threshold switching tests 5% 45% 4% 35% 3% 25% 2% 15% 1% 5% % Power (% of deployment maximum) Te (s) Sea Conditions Histograms that show the number of 2 minute sample counts collected in each one-half meter wide H m and one second wide T e bin are shown in Figure 4-13 for both the latching case (left) and declutching case (right). Data collected with all high voltage R dc settings are included.

99 84 Sample Count (2 min sample periods) Aug 27 - Oct 5, latching control Sample Count (2 min sample periods) Aug 27 - Oct 5, declutching control Hm Bin Center (m) Hm Bin Center (m) Te Bin Center (s) Te Bin Center (s) Figure 4-13 Histograms of sample counts for latching (left) and declutching (right)

100 Results for Latching Control Data collected while the WET-NZ was under latching control with R dc at 32 Ω above V thresh and 4 Ω below V thresh are shown in Figure These were the most common R dc settings used. Plots of normalized power versus V thresh are shown binned by H m and T e ; power is normalized to expected power calculated from the measured wave spectra per equation (4.7). Data with V thresh at zero indicates a constant R dc load and is the baseline case for comparison. In the 2.25 meter H m and 1.5 second T e bin, the WET-NZ output power clearly increased under the voltage threshold control relative to constant resistance control; it appears that the optimum V thresh is near 2 V. Trends are less clear in other data bins due to scatter in the data and the limited amount of data available, however, it appears that the voltage threshold control was generally more effective at higher T e that are beyond the normal response range of the half-scale WET- NZ, shown in Figure 4-8, which cuts off above a period of about 9 seconds.

101 86 Hm 2.25m Te 8.5s 2 Hm 2.25m Te 9.5s 2 Hm 2.25m Te 1.5s 2 Hm 2.25m Te 11.5s 2 Hm 2.25m Te 12.5s Normalized Power = P measured /P expected Hm 1.75m Te 8.5s Hm 1.25m Te 8.5s Hm.75m Te 8.5s Hm 1.75m Te 9.5s Hm 1.25m Te 9.5s Hm.75m Te 9.5s Hm 1.75m Te 1.5s Hm 1.25m Te 1.5s Hm.75m Te 1.5s Hm 1.75m Te 11.5s Hm 1.25m Te 11.5s Hm.75m Te 11.5s Hm 1.75m Te 12.5s Hm 1.25m Te 12.5s Hm.75m Te 12.5s Voltage Threshold (V) Figure 4-14 Normalized output power versus voltage threshold for latching control with 32 Ω > V thresh, binned by H m and T e Results for Declutching Control Based on observations of time series data made during the test, output power always decreased under declutching control compared to the baseline case with constant R dc control, regardless of the V thresh or R dc values used. Because only limited declutching data was collected for each high voltage R dc setting with similar sea conditions, plots of this data are not shown. 4.7 MPPT Testing MPPT is a control technique that samples the output power of a device and adjusts a control parameter to obtain maximum power for any given operating condition. The technique is most commonly used to adjust the load impedance applied to solar panels,

102 87 however, previous work at OSU [8],[9] has shown that this technique can also be used for WEC control. During the course of the WET-NZ deployment, NNMREC experimented with MPPT control of the WET-NZ using two different MPPT algorithms that were implemented in the CompactRIO host during the course of the deployment: 1) a perturb and observe algorithm, and 2) a cycling algorithm. The MPPT algorithms were used to control either R dc while operating under constant resistance loading or V thresh while operating under latching voltage threshold control. Optimum control of V thresh was not successful using either algorithm because, as can be seen in the Figure 4-14 plots, voltage threshold control of the WET-NZ was only effective in a limited range of sea conditions. Optimum control of R dc was not achieved with the perturb and observe algorithm, but optimum control of R dc was achieved using the cycling algorithm even though, as shown in Figure 4-11, the optimum R dc was not well pronounced in the case of the half-scale WET-NZ due to high PTO losses Perturb and Observe Algorithm A block diagram of the perturb and observe MPPT algorithm that was implemented in the CompactRIO host is shown in Figure This is the same algorithm described in [8] for controlling the load applied to a WEC, and is commonly used to control solar panels [6]. Regulation of R dc is illustrated in Figure 4-15; the algorithm can also be used to regulate V thresh. Starting with some load resistance setting R dc k-1 for the k-1 interval, the WEC is controlled with this resistance value for a fixed MPPT time period T MPPT, and at the end of the interval the average power P avg k-1 is calculated. Assume that R dc is then increased by a fixed control step C step to give the resistance R dc k used for the k th interval. At the end of the k th interval, if the average power P avg k is greater than P avg k-1, R dc is again changed in the same direction by C step and R dc k+1 is higher than R dc k, but if power decreases R dc is changed in the opposite direction. The algorithm continues to search for an optimum R dc setting in this manner, incrementing or decrementing R dc by C step indefinitely depending on whether power increases or decreases from one interval to the next. R dc is never increased beyond a

103 88 maximum setting or decreased below a minimum setting; the algorithm steps R dc in the opposite direction if necessary to avoid this. The selection of the two parameters T MPPT and C step have a significant effect on the operation of the algorithm. It is possible to implement the algorithm so that R dc is stepped either arithmetically or geometrically; in one case C step is either added to or subtracted from the previous value, while in the other case the previous value is either multiplied or divided by C step. T MPPT T MPPT P P avg k-1 P avg k R dc R dc k-1 R dc k R dc k+1 C step No Increase R dc By C step Decrease R dc by C step No Yes R dc k = maximum? R dc k = minimum? Yes Yes Is P avg k > P avg k-1? No No Is P avg k > P avg k-1? Yes Figure 4-15 Perturb and observe MPPT algorithm Optimum regulation of R dc using the perturb and observe algorithm was not successful for the half-scale WET-NZ, because output power changed more from one MPPT interval to the next due to variations in sea conditions than due to changes in the R dc setting. This can be seen in Figure 4-16, which shows the results of a trial run using this algorithm with T MPPT set to 2 minutes, geometric incrementing and decrementing by a C step of 1.41, and minimum and maximum R dc limits of 8 Ω and 128 Ω, respectively. R dc and output power are shown together in the top plots, and H m and T e are shown together in the bottom plots. Based on the results shown in Figure 4-11, under optimum

104 89 regulation R dc should range between about 1 Ω and 5 Ω, but during this trial run R dc ranges between 23 Ω and 128 Ω. Output power is affected by both H m and T e, having a positive correlation with H m and a negative correlation with T e. R dc has less effect on output power than either H m or T e. When power increases due to changing H m or T e, R dc swings widely, and when power decreases due to changing H m or T e R dc cycles back and forth without regulating to the optimum range. While it might have been possible to correct these problems by using a shorter interval T MPPT and the same C step, this was expected to cause a large amount of dithering in the control. Resistance (Ohms) Wave height (m) Rdc Power /13 4:48 9/13 7:12 9/13 9:36 9/13 12: 9/13 14:24 9/13 16:48 9/13 19:12 9/13 21: Date Time (UTC) 1 7 9/13 4:48 9/13 7:12 9/13 9:36 9/13 12: 9/13 14:24 9/13 16:48 9/13 19:12 9/13 21:36 Date Time (UTC) Hm Te Power (W) Wave period (s) Figure 4-16 Results for perturb and observe MPPT algorithm with T MPPT = 2 min

105 Cycling Algorithm A block diagram of the cycling MPPT algorithm that was implemented in the CompactRIO host is shown in Figure This algorithm was developed during the course of the WET-NZ tests to overcome the difficulties encountered with the perturb and observe algorithm. When using this algorithm to regulate R dc, it is cycled alternately between a nominal value R dc nom plus and minus a deviation R dc dev throughout each MPPT period T MPPT. At the end of each period the difference between the average power calculated for the portions of time that the WEC was operating at the high and low R dc settings, P and P1 respectively, is used to calculate R dc nom for the next MPPT period. The R dc cycling time T interval was typically set to 1 minute during the WET-NZ tests and the time period T MPPT was typically set to 2 or 6 minutes. R dc dev is always a fixed percentage of R dc nom, and R dc nom is changed in proportion to the difference P-P1 each period based on a gain parameter. By cycling repeatedly between alternate R dc settings each MPPT interval, the effect that R dc has on WEC output power can be measured even if output power changes substantially over the MPPT period due to changes in sea state. The selection of T interval, T MPPT, the percent deviation, and the gain have a significant effect on the operation of the algorithm.

106 91 P P1 mean( P, P1) Figure 4-17 Cycling MPPT algorithm The cycling algorithm was successfully used to regulate R dc to an optimum range during the WET-NZ deployment. See Figure 4-18 for time data recorded during a 45 hour run using this algorithm. In this case, T interval was set to 1 minute, T MPPT was initially 2 minutes during the first half of the run then changed to 6 minutes for the second half of the run, the percent deviation was 1%, and the gain was set to 3. Based on the results shown in Figure 4-11, WET-NZ power output was optimized with R dc between approximately 1 Ω and 5 Ω. R dc was set to 128 Ω at the beginning of the run, and under MPPT control R dc slowly decreased until after 6 hours it was less than 5 Ω. For the next 19 hours R dc generally stayed between 2 Ω and 4 Ω while output power varied significantly. At time 1/2 :, R dc was perturbed to 8 Ω, and under MPPT control it then slowly increases to around 2 Ω by the end of the run. The ability of the MPPT algorithm to maintain R dc within the optimum 1 Ω to 5 Ω range while output

107 92 power varied significantly demonstrated successful regulation. Experimentation during several other, shorter test runs also indicted successful regulation. Resistance (Ohms) Hm (m) /3 19:12 1/1 : 1/1 4:48 1/1 9:36 1/1 14:24 1/1 19:12 1/2 : 1/2 4:48 1/2 9:36 1/2 14:24 1/2 19:12 Date Time (UTC) Cycling algorithm regulating Rdc, T interval = 1 min Rdc nominal 9/3 19:12 1/1 : 1/1 4:48 1/1 9:36 1/1 14:24 1/1 19:12 1/2 : 1/2 4:48 1/2 9:36 1/2 14:24 1/2 19:12 Date Time (UTC) % P % P1 Hm Te Figure 4-18 Results for cycling MPPT algorithm 6% 5% 4% 3% 2% 1% % Power (% of deployment maximum) Te (s) 4.8 Conclusions The Ocean Sentinel performed superbly throughout its first deployment, and the ability of the power converter together with the CompactRIO control and data acquisition to control the WET-NZ generator load and collect data proved to be very effective for evaluating WEC performance. The method of using the CompactRIO host to cycle between alternate control settings in synchronism with the TRIAXYS measurement period that was developed early in the test was particularly useful for collecting the halfscale WET-NZ characterization data. Due to Froude scaling, open ocean seas generally had longer periods than desired for the half-scale model testing. In addition, the device performance was affected by a number of loss mechanisms in the half-scale PTO that caused additional loading on the float and caused the device to operate off-optimum during the deployment. These loss

108 93 mechanisms have since been identified and quantified by WET-NZ. Although the nonideal sea conditions and high PTO losses made data difficult to interpret, several conclusions were made concerning the half-scale WET-NZ: Controlled external resistance loading via the Ocean Sentinel allowed essential on-board data to be collected and the PTO operation characterized in detail under a wide range of real sea conditions. The low response of the WEC to wave periods greater than 9 seconds that was predicted by Callaghan Innovation design analysis was verified. An optimum in the device output with respect to the constant resistance load applied to the generator was observed and characterized An improvement in the response of the device was seen when latching control was used in long period seas, while no improvement was seen with declutching control. The latching control, which introduces a phase lag in the velocity of the float with respect to the force on the float, apparently provided better impedance matching under these conditions. While not reported here, internal PTO characteristics that prevented optimum resistive loading conditions to be achieved during this deployment have been identified and addressed. A perturb and observe MPPT algorithm was not able to regulate generator load resistance R dc while output power fluctuated due to changing sea conditions. A new cycling MPPT algorithm that was developed during the deployment did successfully regulate R dc under similar conditions..

109 94 5 WEC AND POWER CONVERTER EMULATOR 5.1 Introduction This chapter describes a WEC and power converter emulator that simulates operation of 1) a WEC being tested, 2) the Ocean Sentinel power electronics, and 3) the TRIAXYS instrumentation buoy together using desktop hardware. This emulator, developed after completion of the WET-NZ tests in the fall of 212, interfaces with the Ocean Sentinel CompactRIO control and data acquisition system described in Section 3.5. The emulator was developed for two reasons: 1) to perform further testing of MPPT controls after completion of the WET-NZ tests, and 2) to allow future testing of Ocean Sentinel CompactRIO software when a WEC is not being tested. The WEC and power converter emulator was built with a second set of NI CompactRIO hardware and was programmed in LabVIEW. A set of cables connects the emulator to the Ocean Sentinel CompactRIO hardware. The emulator incorporates a model of the OSU autonomous WEC (AWEC) that is being developed as part of a concurrent project at OSU, and simulates operation of that device using ocean wave time data derived from National Data Buoy Center (NDBC) wave spectra recordings. The AWEC model was used to take advantage of work already being done at OSU and because the design is not proprietary. The emulator performs real-time simulations in order to operate together with the Ocean Sentinel CompactRIO hardware. A MATLAB-Simulink model was also developed to simulate AWEC operation, and these simulations could be performed at much faster rates. The MATLAB-Simulink model does not include a detailed power electronics converter model, uses longer simulation time steps to maximize simulation speed, and models AWEC operation with ideal constant damping (constant generator load resistance) control. For simulations performed with constant damping, which covered most simulations performed for this work, the emulator and MATLAB-Simulink results were similar. This allowed MATLAB-Simulink simulations to quickly extend the emulator results to include a wide range of WEC operating conditions and control inputs.

110 The OSU Autonomous WEC (AWEC) A solid model rendering of a quarter-scale version of the AWEC is shown in Figure 5-1. The AWEC is being developed as part of a concurrent project at OSU; a quarter-scale device was built during the winter of 213 and will be tested at the O.H. Hinsdale Wave Research Laboratory in the spring of 213 [14], [15], [16]. This is a point absorber device that produces power from the heave motion of the waves using a cylindrical float that slides up and down over a tubular spar. The PTO system consists of a lead screw that is connected to the spar and a roller screw that is assembled with the rotor of a permanent magnet generator. The lead screw and generator are mounted on a generator plate that moves up and down with the float, so that generator speed is proportional to the relative speed between the float and spar. PTO system Lead screw Generator plate Roller screw & generator Float Mooring attachments Spar Spar flotation Float Plate Figure 5-1 Quarter-scale OSU Autonomous WEC While a quarter-scale AWEC was built for 213 wave tank testing, a model of the full scale AWEC design was used for this project s simulations to take advantage of