Ice Prevention or Removal on the Veteran's
Glass City Skyway Cables
Prepared by: Douglas K. Nims, Victor J. Hunt, Arthhur J. Helmicki, Tsun-Ming T. Ng
Prepared for: The Ohio Department of Transportation, Office of Statewide Planning & Research
State Job Number 134489
August 2014
Final Report
1 Technical Report Documentation Page
1. Report No. 2. Government A cc ess ion No. 3. Rec ipient's Catalog No. FHWA/OH-2014/11
4. Title and Subtitle 5. Report Date (Month and Year) August 2014 Ice Prevention or Removal on the Veterans Glass City Skyway Cables 6. Performing Organization Code
7. Author(s) 8. Performing Organization Report No.
Douglas Nims, Victor Hunt, Arthur Helmicki, Tsun-Ming Ng
9. Performing Organization Name and Address 10. Work Unit No. (TRAIS)
University of Toledo 2801 W. Bancroft St. 11. Contract or Grant No. Toledo, OH 43606 SJN 134489
12. Sponsoring Agency Name and Address 13. Type of Report and Period Covered Ohio Department of Transportation Final Report Research Section 1980 West Broad St., MS 3280 14. Sponsoring Agency Code Columbus, OH 43223
15. Supplementary Notes
16. Abstract
The Veteran’s Glass City Skyway is a cable - stayed bridge in Toledo, Ohio owned by the Ohio DOT. Five times in the seven winters the VGCS has been in service, ice has formed on the stay cables. Ice up to 3/4” thick and conforming to the cylindrical shape of the stay has formed. As the stays warm, ice sheds in curved sheets that fall and can be blown across the bridge. The falling ice sheets pose a potential hazard and may require lane or bridge closure. Because of the specialized knowledge required, this problem required a team including experts in icing, the VGCS construction, the structural measurement system on the bridge, and green technology. The VGCS stay sheaths are made of stainless steel, have a brushed finish, lack the usual helical spiral and have a large diameter. No existing ice anti/deicing technology was found to be practical. Therefore, ODOT elected to manage icing administratively. A real-time ice monitoring system for local weather conditions on the VGCS and the stays was designed. The system collects data from sensors on the bridge and in the region. The study of the past weather and icing events lead to quantitative guidelines about when icing accretion and shedding were likely. The monitoring system tracked the icing conditions on the bridge with a straightforward interface so information on the icing of the bridge is available to the bridge operators. If the conditions favorable to icing occurred, the monitoring system notified the research team and appropriate ODOT officials. If ice has formed, the monitor tracks the conditions that might lead to ice fall.
17. Key Words 18. Distribution Statement
No restrictions. This document is available to the public through the Ice, Bridges, Cable-stayed, Hazard Mitigation, Ice Removal, Ice Prevention National Technical Information Service, Springfield, Virginia 22161
19. Security Classif. (of this report) 20. Security Classif. (of this page) 21. No. of Pages 22. Price Unclassified Unclassified 316 $ 652,894.58
Form DOT F 1700.7 (8-72) Reproduction of completed pages authorized
2 Ice Prevention or Removal on the Veteran's Glass City Skyway Cables
Prepared by: Douglas K. Nims Victor J. Hunt University of Toledo University of Cincinnati
Arthur J. Helmicki Tsun-Ming T. Ng, Ph.D., P.E. University of Cincinnati University of Toledo
August 2014
Prepared in cooperation with the Ohio Department of Transportation and the U.S. Department of Transportation, Federal Highway Administration
The contents of this report reflect the views of the author(s) who is (are) responsible for the facts and the accuracy of the data presented herein. The contents do not necessarily reflect the official views or policies of the Ohio Department of Transportation or the Federal Highway Administration. This report does not constitute a standard, specification, or regulation.
3 Acknowledgments
The authors would like to acknowledge the University of Toledo graduate students Mr. Ali Arbabzadegan, Mr. Joshua Belknap, Mr. Nutthavit Likitkumchorn, and Mr. Clinton and University of Cincinnati Infrastructure Institute graduate students, Mr. Shekhar Agrawal, Mr. Biswarup Deb, Mr. Jason Kumpf and Ms. Chandrasekar Venkatesh, who played a significant role in the research and the writing of this report. Chapter 1, Introduction, was primarily written by Mr. Ali Arbabzadegan with contributions from Mr. Clinton Mirto. Chapter 3, Phase I Research, was primarily written by Mr. Arbabzadegan with contributions from Mr. Joshua Belknap and Mr. Clinton Mirto. Chapter 4, Weather Background, Modeling, and Analysis, was written by Mr. Belknap with contributions from Mr. Arbabzadegan, and Mr. Mirto. Chapter 5, Development of the VGCS Dashboard and Initial Dashboard Results was written by students from UCII. Chapter 6, New Local Weather Sensor Testing, was written by students from UCII. Chapter 7, Experimental Studies on the Sheath Specimens, primarily written by Mr. Arbabzadegan, Mr. Likitkumchorn and students from UCII. Chapter 8, Deployment of New Sensors and Upgrade of Dashboard, was written by Mr. Arbabzadegan and students from UCII. Chapter 9, Sensor Development, was written by Mr. Likitkumchorn. Mr. Mirto assisted in the editing of this report. University of Cincinnati graduate students Mr. Biswarup Deb, Mr. Chandrasekar Venkatesh, Mr. Nithyakumaran Gnanasekar, and Ms. Monisha Baskaran were instrumental in maintaining and updating the Dashboard for this past winter and delivering the standalone computer system to ODOT Dr. Sridhar Viamajala graciously allowed the Scott Park Icing Experiment Station to be built on an unoccupied portion of the concrete pad used for his sustainable energy research. This project was performed under the aegis of the University of Toledo – University Transportation Center. The continuous support of Director Richard Martinko and Associate Director Christine Lonsway has made this project possible. The authors thank Ms. Kathleen Jones and Dr. Charles Ryerson of the U.S. Army Cold Regions Research and Engineering Laboratory for the frequent discussions about the project and extensive analysis, support in developing the criteria for the ice fall dashboard, and their help in the editing of this report. This project was sponsored and supported by the Ohio Department of Transportation. The authors gratefully acknowledge their financial support. Mr. Mike Gramza, P.E. and Mr. Tim Keller, P.E. were the technical liaisons and the authors appreciate their support and input throughput the project. The author would also like to thank Mr. Mike Madry, Mr. Dave Kanavel and Mr. Matt Harvey from ODOT for access to the bridge and assistance in observing the icing events and Mr. Jeff Baker, P.E. (now retired from ODOT) for his assistance in defining criteria for the ice fall dashboard and reviewing the User Manual
4 Dedication
This report is dedicated to the late Professor K. Cyril ‘Cy’ Masiulaniec of Mechanical, Industrial, and Manufacturing Engineering of the University of Toledo. Cy was one of the initial investigators on this project and he was active until the week before his passing when he developed the final details and directions for mounting the thermistors. He will be remembered for his attention to detail and patient thorough explanations of the thermal science that made a significant contribution to this project. Cy was always willing to step up and help our students, his department and the college in many ways. UT students consistently recognized him as an outstanding teacher and he received the UT College of Engineering Award for Teaching Excellence.
5 Abstract
The Veteran’s Glass City Skyway (VGCS) is a large cable - stayed bridge in Toledo, Ohio owned by the Ohio Department of Transportation (ODOT). The VGCS carries I- 280 over the Maumee River. Five times in the seven winters the VGCS has been in service, ice has formed on the stay cable sheaths. Ice accumulations have been up to approximately 3/4” thick and the ice conforms to the cylindrical shape of the stay sheath. As the stays warm, they shed the ice in curved sheets that fall up to two hundred and fifty feet to the roadway and the pieces of ice can be blown across several lanes of traffic on the bridge deck. The falling ice sheets require lane closures or even closure of the entire bridge and could present a potential hazard to the traveling public. Because of the unique nature of the problem, the need for a quick response and the specialized nature of the icing knowledge required, this problem has been addressed with an expert team. The team includes experts in icing from the U.S. Army Cold Regions Research and Engineering Laboratory and the NASA Glenn Icing Branch, the ODOT project managers from the bridge construction, the engineers who designed and implemented the existing structural strain measurement system on the bridge, and experts in green technology. The stay sheaths of the VGCS are unique: they are made of stainless steel, have a brushed finish, lack the usual helical spiral and have a large diameter. No existing ice anti/deicing technology was found to be practical for the VGCS. Therefore, ODOT elected to manage icing administratively. To do this, the research team designed a real-time monitoring system for local weather conditions on the VGCS and the stays as well as the surrounding area. The monitoring system collects a comprehensive set of data from local sensors on the bridge as well as other sensors in the Toledo region. The study of the past weather and icing events lead to quantitative guidelines about the weather conditions that made icing accretion and shedding likely. These guidelines form the core of the algorithms in the ice monitoring system implemented on the bridge. The monitoring system tracked the icing conditions on the bridge with a straightforward interface so information on the icing of the bridge is readily available to the bridge operators. If the conditions favorable to icing occurred, the monitoring system notified the research team and appropriate ODOT officials. If ice forms, the monitor tracks the conditions that might lead to ice fall. The benefits of completing this project include observations of an icing event, review of historical icing events, a building a local weather station on the bridge and stays to collect real-time data on icing and developing the monitoring system. Because no commercial sensor for directly measuring the presence or state of ice on the sheath exists, an electrical resistance based sensor has been developed.
6 Table of Contents
Cover Sheet ...... 12 Technical Report Documentation Page ...... 2 Disclaimer ...... 3 Acknowledgments ...... 4 Dedication ...... 5 Abstract ...... 6 Table of Contents ...... 7 List of Figures ...... 12 List of Tables ...... 19 Chapter 1: Introduction ...... 21 Section 1.1: Bridge Background ...... 21 Section 1.2: Summary of Goals and Objectives ...... 24 Section 1.3: Summary of Results...... 25 Section 1.4: Organization of this Report ...... 27 Chapter 2: Goals, Objectives, Research Approach and Benefits ...... 29 Section 2.1: Overview of Chapter ...... 29 Section 2.2: Goal ...... 29 Section 2.3: Objectives ...... 29 Section 2.4: Expert Team Approach to the Research ...... 33 Section 2.5: Benefits ...... 36 Section 2.6: Chapter Summary ...... 37 Chapter 3: Phase I Research ...... 39 Section 3.1: VGCS Sheaths ...... 39 Section 3.2: Literature Review ...... 40 Section 3.2.1 Known Icing Problems on Other Bridges ...... 40 Section 3.2.2 Anti-Icing/Deicing Technologies found in literature ...... 41 Section 3.3: Technology Matrix ...... 48 Section 3.4: Sensors on the VGCS...... 50 Section 3.4.1: Sensors on the VGCS prior to the 2012 – 2013 Winter ...... 50 Section 3.4.2: Sensors added in 2012 – 2013 ...... 51 Section 3.4.3: Sensors added in 2013 – 2014 ...... 51 Section 3.5: Chapter Summary ...... 53 Chapter 4: Weather History, Modeling and Analysis ...... 55
7 Section 4.1: Introduction ...... 55 Section 4.2: Description of the basic weather that gives rise to an ice storm ...... 55 Section 4.3: VGCS Weather History ...... 56 Section 4.4: Lessons Learned from Previous Icing Events ...... 73 Section 4.5: Analysis ...... 74 Section 4.6: Chapter Summary ...... 76 Chapter 5: Development of the VGCS Dashboard and Initial Dashboard Results ...... 77 Section 5.1: Introduction ...... 77 Section 5.2: Weather Data ...... 80 Section 5.2.1: Introduction ...... 80 Section 5.2.2: Data Sources ...... 80 Section 5.2.3: Data Classification ...... 83 Section 5.2.4: Data Collection and Storage ...... 86 Section 5.3: Ice Accumulation Determination Algorithm ...... 87 Section 5.3.1: Data Update Time ...... 88 Section 5.3.2: Ice Accumulation Algorithm ...... 88 Section 5.3.3: Station Individual Weights ...... 89 Section 5.3.4: Threshold weights ...... 90 Section 5.3.5: Ice Shedding ...... 91 Section 5.4: Ice Persistence Algorithm ...... 91 Section 5.4.1: Ice States ...... 91 Section 5.4.2: Ice Accumulation Persistence Algorithm ...... 92 Section 5.4.3: Ice Presence Confirmation ...... 95 Section 5.4.4: Ice Shedding Persistence Algorithm...... 96 Section 5.5: Monitor Website ...... 99 Section 5.5.1: Dashboard Main Panel ...... 100 Section 5.5.2: Weather Map ...... 101 Section 5.5.3: History ...... 103 Section 5.5.4: Implementation Tools ...... 104 Section 5.6: Performance Testing ...... 104 Section 5.6.1: System Reliability Test ...... 104 Section 5.6.2: Ground Truth ...... 106 Section 5.7: Conclusions ...... 125 Chapter 6: New Local Weather Sensor Testing ...... 126
8 Section 6.1: Introduction ...... 126 Section 6.1.1: Geokon Thermistors ...... 126 Section 6.1.2: Dielectric Wetness Sensor ...... 127 Section 6.1.3: Solar Radiation or Sunshine Sensor ...... 127 Section 6.1.4: Rain Tipping Bucket ...... 128 Section 6.1.5: Goodrich Ice Detector ...... 128 Section 6.2: Geokon Thermistor 3800-2-2 ...... 129 Section 6.2.1: Laboratory experiment on temperature measurement using Geokon Thermistors ...... 130 Section 6.2.2: Installation of Geokon Thermistors 3800-2-2 at the VGCS on Stays 8 & 20 ...... 135 Section 6.3: LWS-L Dielectric Leaf Wetness Sensor ...... 145 Section 6.3.1: Laboratory experiment on measurement of output voltage using LWS-L Leaf Wetness Sensor...... 145 Section 6.4: Sunshine Sensor BF5 ...... 149 Section 6.4.1: Laboratory experiment on measurement of solar radiation using Sunshine Sensor BF5...... 150 Section 6.5: Met One Rain Tipping Bucket ...... 153 Section 6.5.1: Laboratory experiment on measurement of precipitation using Tipping Bucket ...... 154 Section 6.6: Goodrich Ice Detector ...... 156 Section 6.6.1: Laboratory experiment on measurement of ice presence/thickness using Goodrich Ice Detector 0872F1 ...... 157 Section 6.7: Conclusions ...... 161 Chapter 7: Field Study of Temperature Effect on Stay Sheaths ...... 162 Section 7.1: Introduction ...... 162 Section 7.2: Design of Icing Experiment Station ...... 162 Section 7.3: Design of the UT Icing Tunnel and Design ...... 164 Section 7.4: Icing Accretion and shedding Experiments at Scott Park...... 168 Section 7.5: Thermal Experiments at Scott Park ...... 172 Section 7.6: Anti/de-icing Fluid Experiments at Scott Park ...... 175 Section 7.7: Coating Experiments at Scott Park ...... 176 Section 7.8: Coating Experiments using Icing UT Tunnel ...... 178 Section 7.8.1: Testing Procedure ...... 178 Section 7.8.2: Experiments – Icing Progression ...... 179 Section 7.8.3: Result Summery of Icing Tunnel Coating Tests ...... 199
9 Section 7.9: Field Experiment Trips ...... 200 Section 7.10: Conclusions ...... 211 Chapter 8: Deployment of New Sensors and Upgrade of the Dashboard ...... 213 Section 8.1: Introduction ...... 213 Section 8.2: Self Supporting Instrumentation Tower Design ...... 213 Section 8.2.1: Tower Design ...... 213 Section 8.2.2: Anchorage System Design ...... 214 Section 8.3: VGCS Ice Sensors Bridge Installation trip (May 16-17, 2013) ...... 215 Section 8.4: Changes to the Ice Accumulation Algorithm ...... 220 Section 8.5: Changes to the Ice Shedding Algorithm...... 225 Section 8.6: Changes to the Dashboard ...... 227 Section 8.6.1: Dashboard Main Panel ...... 228 Section 8.6.2: Map (Weather Data by location) ...... 229 Section 8.6.3: New Sensors Plotting ...... 231 Section 8.7: Insights Gained from the Operation of the Upgraded Dashboard ...... 235 Section 8.7.1: Ice Events (Winter 2013/2014) ...... 236 Section 8.7.2: Sensor Performance ...... 244 Section 8.7.3: Issues and Observations from Winter Performance ...... 249 Section 8.8: Conclusions ...... 250 Chapter 9: Ice Presence and State Sensor Development ...... 252 Section 9.1: Introduction ...... 252 Section 9.2: Ice Presence and State Sensor Laboratory Testing ...... 252 Section 9.2.1: Sensors and Data Acquisition System ...... 252 Section 9.2.2: Design of Experiments ...... 254 Section 9.2.3: Laboratory Test Results ...... 257 Section 9.3: UT Icing Sensor in Full Scale Experiments ...... 265 Section 9.3.1: Specimens and Data Acquisition System Setup ...... 266 Section 9.3.2: Full Scale Outdoor Tests ...... 270 Section 9.3.3 Full Scale Experiments Result ...... 272 Section 9.4: Conclusion and Next Steps ...... 275 Chapter 10: Transition and Maintenance ...... 276 Section 10.1: Introduction ...... 276 Section 10.2: Standalone Computer System ...... 276 Section 10.3: Maintenance ...... 276
10 Chapter 11: Conclusion, Benefits, Implementation and Future Work ...... 279 Section 11.1: Summary of Goals and Objectives ...... 279 Section 11.2: Results ...... 280 Section 11.3 Benefits ...... 284 Section 11.4: Implementation ...... 285 Section 11.5: Transition and Long Term Maintenance ...... 286 Section 11.6: Archiving of Supporting Documents ...... 286 Section 11.7: Recommendations for Future Work ...... 286 Bibliography ...... 289 Appendix A: Technology Matrix ...... 306
11 List of Figures Figure 1: Veteran’s Glass City Skyway (photo credit will be provided) ...... 21 Figure 2: Veteran’ Glass City Skyway’s Illuminated Glass Pylon (ODOT, 2010) ...... 22 Figure 3: Ice Accumulation on the East Side of VGCS (Baker, 2007) ...... 23 Figure 4: Ice on the Pylon and the VGCS Glass ...... 23 Figure 5: Large Piece of Ice Almost Hitting a Car ...... 24 Figure 6: Application of Superhydrophobic Coating on the Surface (Ryerson, 2008) ... 42 Figure 7: DC Bias Deicing where Electrolysis forms Bubbles (Ryerson, 2008) ...... 43 Figure 8: Pulse Electro Thermal Deicing (PETD) (Ryerson, 2008) ...... 44 Figure 9: Ice Being Released using Ice Dielectric Heating (Ryerson, 2008) ...... 44 Figure 10: Navy Vertical Launch Systems with Electrically Heated Door Edges (Ryerson, 2008) ...... 45 Figure 11: Infrared Heaters above the CRREL Entrance (Ryerson, 2008) ...... 46 Figure 12: Aviation Facility using Infrared Radiant System (Ryerson, 2008) ...... 46 Figure 13: Photonic Deicer for Deicing of Power Lines (Couture, 2011) ...... 48 Figure 14: Damaging ice storm footprint map, 1946-2014 in the lower 48 states and portions of the lower tier of Canada...... 56 Figure 15: Dashboard readout for February 21, 2011 ...... 59 Figure 16: Overview of ice accreting on stay at 10:29 PM Sunday evening ...... 60 Figure 17: Close up of ice accreting on stay at 10:29 PM Sunday evening ...... 60 Figure 18: Stay cable diagram with ice accumulation...... 61 Figure 19: Ice Accumulation up east side of stay February 22, 2011 ...... 63 Figure 20: Frozen Rivulets and bare metal on the west side of stays February 22, 2011 ...... 63 Figure 21: Thermocouple reading between ice and stay February 23, 2011 ...... 64 Figure 22: Thermocouple reading between the ice and stay February 24, 2011 ...... 65 Figure 23: Cracking in ice from chipping away ice, February 23, 2011 ...... 66 Figure 24: Section where ice was chipped away to take temperature readings February 23, 2011 ...... 67 Figure 25: Ice thickness measurements on back stay 19 February 23, 2011 ...... 67 Figure 26: Ice thickness measurements on back stay 19 February 23, 2011 ...... 68 Figure 27: Ice accumulation on pylon glazing February 24, 2011 ...... 69 Figure 28: Ice on bridge deck after 80-90% had shed, February 24, 2011 ...... 69 Figure 29: Weather Summary for the week of February 20, 2011 (Weather Underground, 2011) ...... 71 Figure 30: Solar radiation counts February 22, 2011 ...... 72 Figure 31: Solar radiation counts February 23, 2011 ...... 72 Figure 32: Solar radiation counts February 24, 2011 ...... 73 Figure 33: Process Flow Diagram ...... 79 Figure 34: Map Showing Distances of Weather Stations from VGCS ...... 83 Figure 35: Ice Determination Algorithm ...... 89 Figure 36: Dashboard Speedometer ...... 91 Figure 37: Ice Accumulation Flowchart ...... 93 Figure 38: Sample Ice Accumulation Message Alert ...... 94 Figure 39: Dashboard with Ice Accumulation Alert ...... 95 Figure 40: Ice Presence Flowchart ...... 95
12 Figure 41: Ice Shedding Flowchart ...... 97 Figure 42: Sample Ice Shedding Message Alert ...... 98 Figure 43: Dashboard with Ice Shedding Alert ...... 99 Figure 44: State Transitions possible from Red Level 3 ...... 99 Figure 45: Dashboard Main Panel ...... 101 Figure 46: Dashboard History Panel ...... 103 Figure 47: Weather Summary on Feb 20, 2011 ...... 107 Figure 48: Screenshot Showing Ice Accretion on VGCS ...... 108 Figure 49: Weather Summary on Feb 21, 2011 ...... 109 Figure 50: Weather Summary on Feb 22, 2011 ...... 110 Figure 51: Ice Accumulation on Stays on Feb 22, 2011 ...... 110 Figure 52: Weather Summary on Feb 24, 2011 ...... 112 Figure 53: Example of Ice Shedding Alert ...... 112 Figure 54: Ice Falling from VGCS on Feb 24, 2011 ...... 113 Figure 55: Weather Summary on Feb 25, 2011 ...... 114 Figure 56: Feb 24, 2011 Algorithm Performance Graph ...... 115 Figure 57 : Contribution of the Icing Criteria and Weather Stations ...... 116 Figure 58: Solar Radiation Variation on Feb 24, 2011 ...... 117 Figure 59: Features of the Past Icing Events ...... 119 Figure 60: Dec 12, 2007 Algorithm Performance Graph ...... 120 Figure 61: Mar 28, 2008 Algorithm Performance Graph ...... 122 Figure 62: Dec 17, 2008 Algorithm Performance Graph ...... 123 Figure 63: Jan 03, 2009 Algorithm Performance Graph ...... 124 Figure 64: Geokon 3800-2-2 Thermistor ...... 130 Figure 65: Naked Thermistor Bead (photo credits, John Flynn, Geokon Inc.) ...... 130 Figure 66: Canary Systems Multilogger Software ...... 131 Figure 67: Measurement trend of eight thermistors ...... 132 Figure 68: Thermistors kept in freezer ...... 133 Figure 69: Thermistors immersed in water left to freeze ...... 133 Figure 70: Readings simultaneously noted by handheld GK 404 ...... 133 Figure 71: Standard thermometer immersed in setup to record temperature ...... 133 Figure 72: Thermistor Characteristics at Freezing ...... 134 Figure 73: Side View of Gage Locations at VGCS ...... 135 Figure 74: Custon Thermistor Mount Fabricated for Installing on Stay Surface ...... 136 Figure 75: Thermistor Placed on East Side of Stay ...... 136 Figure 76: Thermistors Placed on Upper Side of Stay ...... 136 Figure 77: Far View of Thermistor Installation of Stay ...... 137 Figure 78: Thermistor Cables Being Routed to Multiplexer Inside White Box ...... 137 Figure 79: Stay Sheath Cross Section Showing Thermistor Positions ...... 138 Figure 80: Stay 20 Thermistors Temperature Trend ...... 139 Figure 81: Stay 8 Thermistors Temperature Trend ...... 140 Figure 82: Characteristics for Stay 20 Thermistors on March 15 ...... 143 Figure 83: Characteristics for Stay 20 thermistors on March 9 & 10 ...... 144 Figure 84: Leaf Wetness Sensor functional diagram ...... 145 Figure 85: Experimental setup of data logger CR1000 with LWS-L Leaf Wetness Sensor ...... 146
13 Figure 86: Droplets of Water Sprinkled on Leaf ...... 146 Figure 87: LWS-L partially immersed in cup of water ...... 147 Figure 88: LWS-L immersed in cup left to freeze ...... 147 Figure 89: LWS Wetness Test ...... 148 Figure 90: LWS Freezing Temperature Test ...... 148 Figure 91: Sunshine Sensor BF5 (side view) and Detailed Construction ...... 149 Figure 92: Sunshine Sensor BF5 Set Up on a Deck for Unobstructed Exposure to Solar Radiation ...... 150 Figure 93: Solar radiation characteristics over an extended period of 16 days ...... 151 Figure 94: A typical partly cloudy day chosen to see the daily solar radiation characteristics ...... 152 Figure 95: A typical clear sunny day taken as example to see the daily solar radiation characteristics ...... 153 Figure 96: Rain Tipping Bucket (from top left clockwise) distant view, top view and inside view ...... 154 Figure 97: Rain Bucket lab ...... 155 Figure 98: Gessler Buret ...... 155 Figure 99: Rain Bucket accuracy experiment (actual vs tipping volume) ...... 156 Figure 100: The Goodrich Ice Detector (external and function diagrams) ...... 157 Figure 101: Ice Detector Mounted for Experiment ...... 159 Figure 102: Microcare Anti-Stat Freezing Spray ...... 159 Figure 103: Probe before Spraying ...... 159 Figure 104: Probe After Spraying ...... 159 Figure 105: Frequency/Ice thickness characteristics of 0872F1 during freezing spray experiment ...... 160 Figure 106: Thickness measurement using calipers ...... 160 Figure 107: Google Earth Screenshot of Scott Park ...... 162 Figure 108: Experimental Setup ...... 163 Figure 109: Sensors on South-faced Specimen ...... 164 Figure 110: Data Acquisition System ...... 164 Figure 111: SolidWorks Design for the UT Icing Tunnel ...... 165 Figure 112: UT Icing Tunnel ...... 165 Figure 113: Testing Section of the UT Icing Tunnel ...... 166 Figure 114: Misting System in the Testing Section ...... 167 Figure 115: Panasonic HX_A100D Camera (Panasonic 2013) ...... 167 Figure 116: Mounting System of Testing Section ...... 168 Figure 117: Spraying a Mist of Water on North-faced Specimen ...... 168 Figure 118: Pattern of Ice Accumulation on Outdoor Tests ...... 169 Figure 119: Water beneath the Ice Layer before Shedding ...... 169 Figure 120: Ice Shedding Steps ...... 170 Figure 121: Stay’s Behavior in Icing Test – 2/15 to 2/18 ...... 171 Figure 122: Stay’s Behavior in Icing Test – 2/20 to 2/22 ...... 171 Figure 123: Thermal Experiment Setup ...... 173 Figure 124: Strands Configuration in Thermal Tests ...... 173 Figure 125: Deicing Pattern in Thermal Test ...... 174 Figure 126: Accumulated Ice in Anti-icing Thermal Test ...... 174
14 Figure 127: Formation of Ice in Chemical Anti-icing Test ...... 175 Figure 128: Drip Tube System used in Chemical Deicing Test ...... 176 Figure 129: Hydrobead Sprayed on Half of the Specimen ...... 177 Figure 130: Water Droplets due to Hydrobead ...... 177 Figure 131: Specimen’s Behavior in Coating Test ...... 178 Figure 132: Uncoated - 40 Micron - 0:00 min ...... 179 Figure 133: Uncoated - 40 Micron - 0:15 min ...... 179 Figure 134: Uncoated - 40 Micron - 0:30 min ...... 180 Figure 135: Uncoated - 40 Micron - 0:45 min ...... 180 Figure 136: Uncoated - 40 Micron – 1:00 min ...... 180 Figure 137: Uncoated - 40 Micron – 1:30 min ...... 181 Figure 138: Uncoated - 40 Micron – 2:00 min ...... 181 Figure 139: Uncoated - 40 Micron – 4:00 min ...... 181 Figure 140: Uncoated - 40 Micron – 6:00 min ...... 182 Figure 141: Uncoated - 40 Micron – 8:00 min ...... 182 Figure 142: Uncoated - 40 Micron – 10:00 min ...... 182 Figure 143: Uncoated - 40 Micron – After Test ...... 183 Figure 144: None Coating - 40 Micron – Shed Ice Sheet ...... 183 Figure 145: Hydrobead-Coated Specimen ...... 184 Figure 146: Hydrobead – 40 Micron – 0:00 min ...... 184 Figure 147: Hydrobead – 40 Micron – 0:15 min ...... 185 Figure 148: Hydrobead – 40 Micron – 0:30 min ...... 185 Figure 149: Hydrobead – 40 Micron – 0:45 min ...... 185 Figure 150: Hydrobead – 40 Micron – 1:00 min ...... 186 Figure 151: Hydrobead – 40 Micron – 1:30 min ...... 186 Figure 152: Hydrobead – 40 Micron – 2:00 min ...... 186 Figure 153: Hydrobead – 40 Micron – 4:00 min ...... 187 Figure 154: Hydrobead – 40 Micron – 6:00 min ...... 187 Figure 155: Hydrobead – 40 Micron – 8:00 min ...... 187 Figure 156: Hydrobead – 40 Micron – 10:00 min ...... 188 Figure 157: Hydrobead – 40 Micron – After Test ...... 188 Figure 158: Hydrobead – 40 Micron – Shed Ice Sheet...... 189 Figure 159: PhaseBreak TP – 40 Micron – 0:00 min ...... 189 Figure 160: PhaseBreak TP – 40 Micron – 0:15 min ...... 190 Figure 161: PhaseBreak TP – 40 Micron – 0:30 min ...... 190 Figure 162: PhaseBreak TP – 40 Micron – 0:45 min ...... 190 Figure 163: PhaseBreak TP – 40 Micron – 1:00 min ...... 191 Figure 164: PhaseBreak TP – 40 Micron – 1:30 min ...... 191 Figure 165: PhaseBreak TP – 40 Micron – 2:00 ...... 191 Figure 166: PhaseBreak TP – 40 Micron – 4:00 ...... 192 Figure 167: PhaseBreak TP – 40 Micron – 6:00 ...... 192 Figure 168: PhaseBreak TP – 40 Micron – 8:00 ...... 192 Figure 169: PhaseBreak TP – 40 Micron – 10:00 ...... 193 Figure 170: PhaseBreak TP – 40 Micron – After Test ...... 193 Figure 171: PhaseBreak TP – 40 Micron – Shed Ice Sheet ...... 194 Figure 172: WeatherTITE – 40 Micron – 0:00 min ...... 194
15 Figure 173: WeatherTITE – 40 Micron – 0:15 min ...... 195 Figure 174: WeatherTITE – 40 Micron – 0:30 min ...... 195 Figure 175: WeatherTITE – 40 Micron – 0:45 min ...... 195 Figure 176: WeatherTITE – 40 Micron – 1:00 min ...... 196 Figure 177: WeatherTITE – 40 Micron – 1:30 min ...... 196 Figure 178: WeatherTITE – 40 Micron – 2:00 min ...... 196 Figure 179: WeatherTITE – 40 Micron – 3:00 min ...... 197 Figure 180: WeatherTITE – 40 Micron – 4:00 min ...... 197 Figure 181: WeatherTITE – 40 Micron – 6:00 min ...... 197 Figure 182: WeatherTITE – 40 Micron – 8:00 min ...... 198 Figure 183: WeatherTITE – 40 Micron – 10:00 min ...... 198 Figure 184: WeatherTITE – 40 Micron – After Test ...... 198 Figure 185: WeatherTITE – 40 Micron – Shed Ice Sheet ...... 199 Figure 186: Stay Specimens at Different Angles and Orientations ...... 201 Figure 187: Data-logging System Setup ...... 201 Figure 188: Sunshine Sensor Setup...... 201 Figure 189: Ice Detector Placed Right Beside Stay ...... 201 Figure 190: Stay Thermistors Zip-tied on Sheath ...... 201 Figure 191: Leaf Wetness Sensor Taped on top of Specimen ...... 201 Figure 192: Ice Detector at Various Times Throughout the February 16 Experiment . 202 Figure 193: Leaf Wetness Sensor at Various Times Throughout the February 16 Experiment ...... 203 Figure 194: Ice Detector Characteristics (Toledo experiments on February 16) ...... 204 Figure 195: Characteristics of stay thermistors (Toledo, February 16) ...... 205 Figure 196: Leaf Wetness Sensor ice melting characteristics ...... 205 Figure 197: LWS-LS with Different Slants ...... 207 Figure 198: Top & Side Thermistors Setup ...... 207 Figure 199: Ice Detector Setup ...... 207 Figure 200: First Spray Shower ...... 207 Figure 201: Garden Hose mount on ladder (left) & hand held (right) for experiment on ice detector & leaf sensors ...... 208 Figure 202: Ice Detector at Various Times during Experiment (Left and Middle during ice accretion; right during deicing) ...... 208 Figure 203: Stay thermistor characteristics (Toledo experiments February 20 – 21) .. 209 Figure 204: Leaf Wetness Sensor Characteristics (Toledo, February 20 – 21) ...... 210 Figure 205: Ice Detector characteristics (Toledo, February 20 – 21) ...... 211 Figure 206: Tower Anchorage System ...... 214 Figure 207: Rohn’s Weather Tower Drawing ...... 215 Figure 208: Tower mounted near stay 19 ...... 215 Figure 209: Initial Plan by UT Research Team for Tower Mounting ...... 216 Figure 210: Leaf Wetness Sensor Zip-tied to Cross-arm ...... 217 Figure 211: Rain Bucket mounted on cross-arm using leveling bracket ...... 218 Figure 212: Sunshine Sensor attached to cross-arm with steel U-bolts ...... 218 Figure 213: Ice Detector Mounted using Steal Worm Band Clamps ...... 219 Figure 214: Ice Detector Mounted Close Up ...... 219 Figure 215: Sensor Cable Conduit ...... 219
16 Figure 216: CR1000 Datalogger Setup Insider Tower Cabinet ...... 219 Figure 217: Close up of Weather Tower ...... 220 Figure 218: Completed New Weather Station Near Stay 19 ...... 220 Figure 219: Flowchart of existing Ice Accumulation Algorithm (Agrawal, 2011) ...... 222 Figure 220: Flowchart for revised ice accumulation algorithm ...... 224 Figure 221: Flowchart of existing Ice Shedding Algorithm (Agrawal, 2011) ...... 226 Figure 222: Flowchart for revised ice shedding algorithm ...... 227 Figure 223: Dashboard Main Panel ...... 228 Figure 224: Example Snapshot of Weather Map, with Pop-up for Ice Detector ...... 230 Figure 225: Last 48 hour report of Solar Sensor (Global Radiation) ...... 231 Figure 226: Last 48 hour report of Leaf Wetness Sensor ...... 231 Figure 227: Stay 20 Thermistors plot (January 1 – July 1) ...... 232 Figure 228: Stay 8 Thermistors plot (January 1 – July 1) ...... 232 Figure 229: Ice Detector plot (June 1 – July 1) ...... 233 Figure 230: Leaf Wetness Sensor plot (June 1 – July 1) ...... 233 Figure 231: Rain Tipping Bucket plot (June 1 – July 1) ...... 234 Figure 232: Sunshine Sensor plot (June 1 – July 1) ...... 234 Figure 233: Ice Detector & LWS Characteristics during Ice Event, December 9, 2013237 Figure 234: VGCS Icing camera view before noon ...... 239 Figure 235: Ice Detector & Leaf Wetness Sensor characteristics on February 20 ...... 240 Figure 236: Ice detector & Leaf wetness Sensor characteristics on April 3 ...... 242 Figure 237: Rain Tipping Bucket & Leaf Wetness Sensor characteristics on April 3 ... 242 Figure 238: Solar Radiation & Stay Thermistor 8X08TWS characteristics on April 3 .. 243 Figure 239: Leaf Wetness Sensor characteristics winter 2013/14 ...... 245 Figure 240: Stay Thermistor characteristics winter 2013/14 ...... 245 Figure 241: Sheath thermistors warming faster than outer (March 4, 2014) ...... 246 Figure 242: Rain Tipping Bucket characteristics winter 2013/14 ...... 247 Figure 243: Ice Detector characteristics winter 2013/14 ...... 248 Figure 244: Solar radiation Sensor characteristics winter 2013/14 ...... 249 Figure 245: Relative distribution of alarms triggered by new sensors ...... 249 Figure 246: UT Icing Sensor Circuit ...... 253 Figure 247: Electro Spacing Area of the UT Icing Sensor ...... 253 Figure 248: UT Icing Sensor Connected to Data Acquisition System ...... 254 Figure 249: Dashboard of UT Icing Sensor ...... 254 Figure 250: 1-mm Electro Spacing UT Icing Sensor ...... 255 Figure 251: 7-mm Electro Spacing UT Icing Sensor ...... 255 Figure 252: Water Measurement ...... 256 Figure 253: Ice Measurement ...... 256 Figure 254: 75% Slush Measurement ...... 256 Figure 255: 50% Slush Measurement ...... 256 Figure 256: 25% Slush Measurement ...... 256 Figure 257: Ice Measurement at 6 mm thickness ...... 257 Figure 258: Ice Measurement at 13 mm thickness ...... 257 Figure 259: Ice Measurement at 19 mm thickness ...... 257 Figure 260: Resistance of Ice for 1-mm Electro Spacing Sensor ...... 258 Figure 261: Dashboard Screenshot of Ice Measurement ...... 258
17 Figure 262: Resistance of 75% Slush for 1-mm Electro Spacing Sensor ...... 259 Figure 263: Dashboard Screenshot of 75% Slush Measurement ...... 259 Figure 264: Resistance of 50% Slush for 1-mm Electro Spacing Sensor ...... 260 Figure 265: Dashboard Screenshot of 50% Slush Measurement ...... 260 Figure 266: Resistance of 25% Slush for 1-mm Electrode Spacing Sensor ...... 261 Figure 267: Dashboard Screenshot of 25% Slush Measurement ...... 261 Figure 268: Resistance of Water for 1-mm Electro Spacing Sensor ...... 262 Figure 269: Dashboard Screenshot of Water Measurement ...... 262 Figure 270: Resistance of Ice for 7-mm Electro Spacing Sensor ...... 263 Figure 271: Resistance of 75% Slush for 7-mm Electro Spacing Sensor ...... 263 Figure 272: Resistance of 50% Slush for 7-mm Electro Spacing Sensor ...... 264 Figure 273: Resistance of 25% Slush for 7-mm Electro Spacing Sensor ...... 264 Figure 274: Resistance of Water for 7-mm Electro Spacing Sensor ...... 264 Figure 275: Resistances for 6-mm Thickness and 7-mm Electro Spacing Sensor ...... 265 Figure 276: VGCS Stainless Steel Specimens ...... 266 Figure 277: HDPE Specimen and Frame Structure ...... 266 Figure 278: North Facing Specimen with 120 Stands Inside ...... 267 Figure 279: Sensors Setup on VGCS Specimen ...... 268 Figure 280: Sensors Setup on HDPE Specimen ...... 268 Figure 281: Cross Section and Sensor Setup Orientation of both Specimens ...... 268 Figure 282: UT Icing Sensor on HDPE Specimen ...... 268 Figure 283: MicroStrain V-Link ...... 269 Figure 284: MicroStrain TC-Link ...... 269 Figure 285: MicroStrain WSDA-Base (Signal Receiver) ...... 270 Figure 286: V-Link and UT Icing Sensor ...... 271 Figure 287: Ice Testing ...... 271 Figure 288: Slush Testing ...... 271 Figure 289: Water Testing ...... 271 Figure 290: UT Icing Sensor Initial Test ...... 272 Figure 291: Misting Water on VGCS Specimen ...... 273 Figure 292: Ice Accumulation on VGCS Specimen ...... 273 Figure 293: Stay Behavior in Icing Experiment ...... 274 Figure 294: Flowchart for Stand Alone System ...... 278
18 List of Tables Table 1: Viable Technologies ...... 31 Table 2: Information Required to Revolve Uncertainties ...... 31 Table 3: Team Members Roles and Expertise ...... 35 Table 4: Sheath Roughness Test Data ...... 40 Table 5: Most Viable Solutions for the VGCS ...... 50 Table 6: Uncertainties that Needed Resolved and Corresponding Sensors...... 52 Table 7: Ice Accumulation Weather Conditions ...... 56 Table 8: Ice Falling Weather Conditions ...... 57 Table 9 Weather Conditions for February 20, 2011 (Kumpf et. al, Weather Underground, 2011) ...... 61 Table 10: Interstice Temperature February 23 ...... 65 Table 11: Weather conditions for February 24, 2011 (Kumpf et. al, Weather Underground , 2011) ...... 68 Table 12: Ice Accumulation Criteria ...... 74 Table 13: Ice Fall Criteria ...... 74 Table 14: Sensor System at RWIS Stations ...... 81 Table 15: Airport Information ...... 82 Table 16: Distances of Weather Stations from VGCS ...... 83 Table 17: METAR and RWIS Precipitation Measurements for Ice Accumulation ...... 84 Table 18: Ice Accumulation Criteria ...... 85 Table 19: METAR and RWIS Precipitation Measurements for Ice Shedding ...... 85 Table 20: Ice Shedding Criteria ...... 86 Table 21: Final Ice Accumulation/Shedding Criteria ...... 86 Table 22: Weather Station Weights ...... 89 Table 23: Dial States Explanation ...... 92 Table 24: Tools Used To Design Dashboard ...... 104 Table 25: Dates for Past Ice Events that were Tested ...... 105 Table 26:Weather Statistics for December 12, 2007 Ice Event ...... 105 Table 27: Summary of Events when Ice Accumulation occurred in 2011 ...... 106 Table 28: Interstice Temperature on February 23, 2011 ...... 111 Table 29: Station Comparison for the 2011 Winter ...... 116 Table 30: Overall Performance of Dashboard on Past Icing Events ...... 124 Table 31: Comparison of readings taken by all 3 methods ...... 133 Table 32: New Stay Thermistors List ...... 138 Table 33: Sky Cover and Precipitation During the Period ...... 141 Table 34: Weather Report on March 15 ...... 142 Table 35: Wetness Test ...... 146 Table 36: Impurity Test ...... 147 Table 37: Impurity Test ...... 147 Table 38: Rain Bucket Lab Experiment 1 with 5 Minute Sampling Rate ...... 155 Table 39: Rain Bucket Lab Experiment 2 with 30 Minute Sampling Rate...... 156 Table 40: Caliper Test ...... 160 Table 41: Icing Sensors Initial Observations ...... 161 Table 42: Approximated ice thickness comparison of coatings and droplet sizes ...... 199 Table 43: Event History (February 16, 2013) ...... 200
19 Table 44: Event History (February 20-21, 2013) ...... 206 Table 45: Summary of VGCS Sensor Installation Trip ...... 217 Table 46: Ice Accumulation Station Functions ...... 223 Table 47: Ice Fall Station Functions in algorithm ...... 226 Table 48: Chronology of winter 2013/2014 icing event triggers ...... 236 Table 49: Web Report Tool: Sample Icing Events and Comments, December 2013 .. 243
20 Chapter 1: Introduction Section 1.1: Bridge Background The Veteran’s Glass City Skyway (VGCS), formerly known as the Maumee River Crossing is a large cable-stayed bridge on Interstate 280 that crosses over the Maumee River in Toledo, Ohio. The VGCS is owned by the Ohio Department of Transportation and is considered as the most expensive project ever undertaken by ODOT (Wikipedia, 2013). The construction began in 2001 and the bridge was opened for service in July 2007. The entire project consists of 8,800 feet of approaches and main span. The main span is approximately 1,225-feet in length, consists of a single pylon that rises 216 feet above the bridge deck, and has a single plane of stays, seen in Figure 1 below. The VGCS, which is an important connector for multimodal transpoortation and economic development, carries three lanes of traffic and has thousands of vehicles crossing daily.
Figure 1: Veteran’s Glass City Skyway (photo credit wiill be provided) The VGCS has several novel features: the cradle system for the stays, the stainless steel stay sheathes, and the illuminated glass in thee pylon. The VGCS is one of two installations of a new cradle system (Figg, 2005). This particular system eliminates the need for cable anchorage in the pylon by carrying stays from one side of the bridge deck to the other. This allows the tower to be more slender than what is possible with a conventional anchorage arrangement. The pylon is illuminated with internal LED lighting that is infinitely variable, an example of one lighting schemes can be seen in Figure 2 below. This makes the pylon visible for miles at night and the pylon can be lit to reflecct local events or the time of the year.
21
Figure 2: Veteran’ Glass City Skyway’s Illuminated Glass Pylon (ODOT, 2010) Under some winter conditions, ice forms on the stay cables of the VGCS. Ice accumulation can exceed a 1/2 inch and may persist for several days on the stays. Ice then sheds in semi-cylindrical sheets from the cable sheaths. Shedding of an individual stay can occur in less than a minute. Ice shedding is triggered by a combination of rising temperatures and solar radiation. The sheets may fall over two hundred and fifty feet to the roadway. Due to their aerodynamic shape, they can glide or be blown across several lanes of traffic. In some instances, large ice sheets have crossed all the lanes of traffic and fallen in the river. The potential of falling sheets typically requires lane or bridge closure for the duration of the ice persistence. Lane closures result in the inconvenience to the traveling public as well as loss to economic activities. Falling ice is a safety hazard to motorists and determining ice presence remotely is problematic. Currently, ice presence is determined manually, putting ODOT personnel in harm’s way. Figures 3 and 4 show ice accumulation on the stays and pylon of VGCS in the 2011 icing event.
22
Figure 3: Ice Accumulation on the East Side of VGCS (Baker, 2007)
Figure 4: Ice on the Pylon and the VGCS Glass Figure 5, which was captured during 2011 ice fall event, shows a large piece of ice, circled in red, falling into the third lane of traffic while vehicles are still travelling over the bridge (Belknap, 2011).
23
Figure 5: Large Piece of Ice Almost Hitting a Car Section 1.2: Summary of Goals and Objectives After four icing events in the first two winter seasons of the VGCCS being open, reseaarch was undertaken to assist ODOT in implementing an icing management procedure for the VGCS. The research followed a phased approach. The first phase focused on review of available technologies, selection of potential technologies for the VGCS and costing of the potential technologies whereas the second phase focused on the development and implementation of a monitoring system and sensorss. The first phase objectives were:
Identify available technologies and procedures that coulld potentially solve the icing problem.
Assess the state of the art via literature review and consultation with icing experts.
Examine the advantages, disadvantages, and applicability of each identified technology on the VGCS.
For each viable solution, develop a detailed description of the implementation, define required validation testing, perform a bbenefit/cost analysis, develop a budget for implementation and define a time frame for implementation.
Develop a real-time icing condition monitor. This objective was added by the research team in response to a request by ODOT to make the research immediately actionable by the bridge operators. As the project progressed, as phase II was undertaken. Phase II built on the background developed in the first phase. The objectives for Phase II altered to account for the knowledge gained concerning the state of the art and practice in anti/deicing and to address the need to better understand the microclimate on the bridge. The Final Phase II objectives were:
24 Collect data to resolve uncertainties in the bridge microclimate and the conditions on the stays. To understand the icing behavior it was necessary to gain knowledge about how and when ice was forming on the stays, stay sheath temperatures and the local conditions on the bridge,
Make a recommendation on two to four viable active solutions. This required experiments on anti/deicing techniques as well as literature review and discussion with experts.
Improve the user friendliness, algorithms and error handling of the icing monitor.
Develop of an ice presence and state sensor. No such commercial senor exists and data about the ice persistence and water flow beneath the ice is essential to understanding shedding. Through experimentation, no practical active or passive anti/deicing solution was ever identified, as discussed in Chapter 7 of this report. This ultimately led to a new overall objective, which was to improve the monitoring of icing events in order to provide ODOT with the best information to manage their response to an icing event. The goals, objectives, and uncertainties will be provided in more detail in the following chapter. Section 1.3: Summary of Results Past icing events were reviewed, the mechanisms for icing where explored, and the basic conditions that are favorable to icing accretion and shedding were ascertained. Historically, roughly two icing events occur each year. Icing on the VGCS occurs when there is general icing in the area. There have been five major icing events on the VGCS. The last of which was in February 2011. Conditions are favorable for ice accretion when one of the following conditions occurs: i. Precipitation with air temperature at the bridge below 32o F, or ii. Fog with air temperature at the bridge below 32o F, or iii. Snow with air temperature at the bridge above 32o F. The ice accretion rate is generally slow because during an ice storm precipitation rates are low and much of the water runs off the stays. Once the ice accretes on the stays and pylon, it may persist until shedding conditions occur. Temperatures above 32o F and/or solar radiation cause ice fall. Water flowing beneath the ice layer was observed prior to the ice fall in 2011 and is thought to be a precursor to ice fall. If there is ice on the stay, the weather conditions that cause ice fall are: i. Air temperature above 32o F (warm air), or ii. Clear sky during daylight (solar radiation). Given the unique features of the VGCS, the paucity of literature directly on point, and the urgency of addressing the problem, an expert team was selected to address this problem. The research team that had expertise in icing, icing instrumentation, icing test facilities, the VGCS construction and VGCS instrumentation was formed to address the
25 issues of ice prevention and mitigation on the VGCS. A comprehensive review all anti/deicing technologies that could be identified regardless of their technology readiness level was performed. A matrix of over 70 potential technologies was developed. The matrix describes the advantages and disadvantages of each technology. To simulate icing events and use a test bed for experiments an icing field station was designed and built. It had three full scale sheath specimens ten feet long. One of these specimens included strand. The station had a local weather station and a wireless data acquisition. The initial set of experiments verified that ice accretion and shedding similar to that which occurs on the bridge could be replicated. The icing station was then used for experiments on anti/decing chemicals, anti-icing coating, heat for anti-icing and deicing, and tests of instruments. The technologies that were the most viable were identified. They were: i. Deicing/anti-icing chemicals which would not present a biohazard when leached into the river such a sodium chloride; agricultural products, such as beet based deicers, and calcium chloride ii. Anti-icing coatings iii. Heat. The VGCS stays are mostly hollow so there is a potential to internally heat the stays. Experiments to evaluate the efficacy of each viable technology were carried out. The anti-icing chemical experiments showed that on the stainless steel surface of the sheath the chemicals tested did not persist. The deicing experiments showed that the chemical tested was not viscous enough to sheet across the sheath surface. These results are consistent with the results in the literature. In addition, to not performing the desired anti/deicing functions, chemicals would require a distribution system so they were deemed impractical. Several anti-icing coating were tested in the icing wind tunnel and at the icing experiment station. The coatings did not significantly delay the onset of ice, which stuck to the stay specimens and most did not change the shape of ice that shed. The coating that was outdoors for an extended duration of time became opaque and gummy, therefore, it would alter the appearance of the stays. These results are consistent with the results in the literature. Additionally, coating would be difficult to apply so they were deemed impractical. Introductory heating experiments were carried out at the icing experiment station. The heating was effective at deicing and partially effective at anti-icing. The requirement to heat each stay would require an expensive heating system. At that point, heating was deemed impractical so no advanced experiments or thermal analyses were conducted. Thus, no active or passive system was identified which had sufficient level of promise to justify detailed estimates of installation, operation or maintenance costs. When it was judged that the regional weather information and the RWIS did not provide enough information to assess the microclimate and icing behavior, a local weather station was installed on the bridge. The combination of the existing sensors and the
26 local weather station gives a good picture of the conditions on the bridge. Prior to deployment in the field, experiments on the sheathing specimens at the field station and in the laboratory coupled with the literature review lead to the conclusion that the proposed sensors functioned as desired and they were recommended for installation. To make the research immediately actionable by ODOT operations, a real-time icing condition monitor was developed. The research team designed a real-time monitoring system to track icing conditions on the bridge with a straightforward interface so information on the icing of the bridge was readily available to the bridge operators. This monitoring system is referred to as the “icing dashboard” or simply “the dashboard” because the information necessary to support ODOT operations is presented on one simple visual display. When conditions favorable to icing occur the dashboard alerted the research team. If the conditions favorable to icing persisted, ODOT was notified and, as required, requests for verification of ice accretion were made. The basis of this monitoring system is the smart mix of the automated algorithm and the visual observations, which helped aid in training the system for more optimal performance. The system uses an intelligent decision making process based upon initial criteria from past weather data analysis with parameter adjustments made after visual observations. Dashboard has done well in detecting ice accumulation each time, but the analysis done on the algorithm results and onsite observations from research team members and ODOT have been used to refine the algorithm as well as the interface. The dashboard has proven to be a valuable resource for the bridge operators as well as a valuable tool for reviewing weather events. The automated ice detection and monitoring dashboard for the VGCS was developed, implemented, successfully tested, and has been transferred to ODOT. No suitable sensor to detect the continued presence of ice or the transition from ice to water exists. Therefore, development and field testing of a suitable sensor were undertaken. The resistance based sensor detects the presence of ice and can differentiate between ice and liquid water. The sensor is designed to be mounted on the sheath and can detect the layer of water which forms beneath the ice just prior to shedding. The sensor has been tested in the laboratory and at the icing experiment station. The transition of the dashboard to District Two has concluded. A local standalone computer with the dashboard on it has been provided to the District. The standalone version maintains the basic functionality of the dashboard algorithms and alert system and provides links to the icing weather instrumentation on the bridge. A person at the computer can monitor the conditions on the bridge and determine the causes of alerts. Section 1.4: Organization of this Report Chapter one described background information regarding the VGCS, introduced the problem statement of helping ODOT operations with icing problems on the VGCS, and gave summaries of goals and objectives as well as results.
27 Chapter two discusses goals, objectives and benefits as well as introduces the expert team. Chapter three describes phase I research, which involved investigating the VGCS stay sheaths, performing a literature review regarding icing events on other structures as well as potential anti/deicing technologies, constructing a technology matrix to narrow down the numerous technologies to a few viable ones was constructed, and providing the history of sensor presence on the VGCS. Chapter four looks into the basic weather that gives rise to ice storms, the VGCS’s weather history, lessons learned from previous icing events, and accretion and shedding algorithms. Chapter five thoroughly discusses the development and testing of the icing dashboard as well as its initial results. Chapter six looks into each of new sensors implemented onto the bridge as well as describes both the laboratory and field tests performed on the new sensors. Chapter seven discusses the experimental studies performed on the sheath specimens at the outdoor icing experiment station located at the University of Toledo’s Scott Park Campus. This chapter gives detailed analysis and discussion regarding the potential technologies tested as well as the new sensors that were eventually implemented. Chapter eight describes the design and implementation of the local weather tower on the VGCS. Chapter nine discusses the development of the University of Toledo ice presence and state sensor. Chapter ten looks into the transition as well as the near-term and long-term maintenance of the icing dashboard. Chapter eleven provides a conclusion and recommendations for future work.
28 Chapter 2: Goals, Objectives, Research Approach and Benefits Section 2.1: Overview of Chapter This chapter describes the overall goals of the project, the objectives that were achieved to reach those goals, the approach that was taken to reach the objectives and the benefits that accrued to ODOT from this project achieving its goals. Section 2.2: Goal Under some winter conditions, ice forms on the cables stays of the VGCS. Ice accumulations have been observed at a thickness of 3/4”. The ice accumulation depends on the temperature, precipitation and duration of the storm. The accreted ice conforms to the cylindrical shape of the stay sheath. Thus, as the stays warm, the ice sheds in curved sheets. These curved sheets of ice then fall up to two hundred and fifty feet to the roadway below and may be blown across several lanes of the bridge deck depending on wind conditions and/or ice sheet aerodynamics. The falling ice sheets require lane closures and could present a potential hazard to the traveling public. The overall goal of this research was to assist ODOT in implementing an icing management procedure for the VGCS. This procedure may be active, passive or administrative. Active procedures involve anti/deicing measures that are typically powered and activated only when needed. Passive procedures operate without power and are continuously available, and include coatings or other technologies that are permanently in place. Administrative procedures focus on obtaining information about the condition of ice on the stays and pylon and managing the response to icing incidents with or without taking anti/deicing measures. Section 2.3: Objectives The research followed a phased approach. The first phase focused on review of available technologies, selection of potential technologies for the VGCS and costing of the potential technologies. The second phase focused on the development and implementation of a monitoring system and sensors. The original objectives of this study included the conceptual design and rough costing of three to five reasonable options, which included active or passive anti-icing or deicing approaches applicable to the VGCS, for ODOT. Investigation of a wide range of technologies was completed. No practical anti/deicing technology was identified. Therefore, the objective shifted to the monitoring of icing events in order to provide ODOT with the best information to manage their response to an icing event. The original objectives as well as the modification of objectives will be described below. The initial overall objectives of this study were to present three to five reasonable options to ODOT for ice protection on the VGCS as mentioned above. The highest priority was to identify cost effective methods to prevent the formation of ice on the stays. If suitable methods for ice prevention were not identified, the secondary objective was to identify methods to safely and efficiently remove ice from the stays without damaging the structure or causing additional safety concerns and delays to the public.
29 The first phase objectives were as follows: 1) Identify available technologies and procedures that could be used to solve the icing problem. Sixteen potential technologies were identified. Fourteen ice protection technology categories are acknowledged for anti-icing, deicing, and ice detection in the work by Ryerson (Ryerson 2009). There are many technologies from Ryerson’s work that are potentially applicable to the VGCS cables, which include: chemicals; icephobic coatings; structure design; expulsive techniques; heat; high-volume water, air, and steam; infrared energy; piezoelectric methods; pneumatic boots; vibration and appropriate ice detection methods. Proprietary methods such as pulse electro-thermal de-icing (PETD), a technique incorporating nano-fibers and a piezoelectric system proposed for aircraft will also be considered (Petrenko 2009; Prybyla 2009, and Near 2009, respectively).
2) Assess the state of the art through a literature review and consultation with the icing experts. Given the unique features of the VGCS, the paucity of literature directly on point, and the urgency of addressing the problem, an expert team is a superior way to quickly gain familiarity with the state of the art as well as define testing procedures and identify available facilities.
3) Examine the advantages, disadvantages, and potential applicability of each identified technology on the VGCS.
4) Identify the most viable solutions. It is expected that the most practical solutions will be novel adaptions or combinations of existing solutions.
5) For each viable solution, develop a detailed description of the implementation, define required validation testing, (either in situ or offsite), perform a benefit/cost analysis, develop a budget for implementation and define a time frame for implementation. Because we expect that the solutions will be novel, it is anticipated that some validation testing will be required.
6) Issue an interim report providing a summary of the findings from steps 1 through 4 and the recommendations and economic analysis from step 5 (Nims, 2011).
The research from Phase I resulted in the identification of several viable technologies, which can be seen in Table 1. These technologies fell into three separate categories, which were chemical distribution, chemicals, and internal heating. The technologies deemed viable for chemical distribution included the use of drip tubes or cable climbers with supply hoses or tanks. The chemicals that were further investigated were sodium chloride, calcium chloride, and agricultural products. As for internal heating, forced air, air with piccolo tube, steam heating elements and electrical heating elements were considered.
30 Table 1: Viable Technologies Category Specific Technology Chemical Drip Tube Distribution Cable climber with supply hose or tank Sodium Chloride, Calcium Chloride, Agricultural-based deicing Chemicals products Potential options to be explored are: forced air, air with piccolo tube, Internal Heating steam heating element and electrical heating elements
As part of Phase I, any proposed implementation was investigated in such a way that the implementation would be as “green” as possible. If any of the potentially viable solutions identified above in 5) required the use of a local power source, then cleaner alternative forms of energy, such as solar power, was investigated and utilized if possible. If the recommendation involved the application of chemicals, then the potential environmental consequences were considered and avoided if possible. At an icing team meeting during Phase I work (the meeting notes are in the interim report (add cite)), it was identified that there was insufficient information concerning the ice accumulation conditions, the ice shedding conditions, the microclimate of the bridge and the effectiveness of the viable technologies to reasonably cost alternatives. Thus, the team and ODOT decided that the uncertainties listed in Table 2 needed to be resolved.
Table 2: Information Required to Revolve Uncertainties Required Information Uncertainties to be resolved Presence of ice and/or It is difficult to be certain when ice is forming on the stay, how liquid water on stay fast it is accumulating and if it is persisting. Stay Sheath The temperature of the stays during an icing event is unknown. Temperature It is considered as one of the reasons for shedding. Solar radiation may contribute to ice shed. Solar radiation Sky Solar Radiation raises the stay temperature and the temperature between the ice sheet and the sheath. The bridge has its own microclimate: precipitation amount and Local Weather type, droplet size, wind speed, wind direction, visibility needs to Conditions be determined on the bridge. Characteristics of the distribution of the heat along the stay from Heat flow along stay air flow and through the stay cross section from a local source, and across a stay and the VGCS specific constants for thermal analysis, need to section be determined. The efficacy of the chemicals, the effect of the chemicals on the Efficacy of anti/deicing brushed surface of sheaths, and a practical method for applying chemicals the chemicals are unknown. Visual record of Observation of the unquantifiable aspects of icing on the VGCS. conditions Aerodynamic effects of A drip tube is a possible chemical distribution system. How the drip tube drip tube effects the aerodynamics of the stays.
31 In response to a request by ODOT at a project progress meeting to make the research immediately actionable by ODOT operations, a real-time icing condition monitor was developed. This monitoring system is referred to as the “icing dashboard” or simply “the dashboard” because the information necessary to support ODOT operations is presented on one simple visual display. The need to resolve the uncertainties in Table 2 and build on the capabilities of the dashboard led to a modification of Phase II research, which was initially focused on the implementation of viable technologies. Final Phase II objectives were as follows: 1) Collect data to resolve uncertainties in Table 2. Some of the data may come from existing sensors while some of the data required new sensors (discussed later in this report), laboratory experiments and on-site observation. The collected information should be sufficient to allow accurate costing, resolve uncertainties to reduce the risk of deploying an icing strategy that does not work, and be useful for improving and updating the icing dashboard. The uncertainties to be resolved and the reason for resolving the uncertainty is listed in Table 2 above.
2) Make a recommendation on two to four viable active solutions. To make a decision on the viability on an active system, it is necessary to have a reasonable estimate of the cost and the practical implementation strategy.
3) Improve the icing dashboard. The dashboard tracks the icing events in a format that is easy to understand, is useful for managing icing incidents and archives data. Local condition data that is collected from the bridge will be used to increase algorithm intelligence and error handling. The improvements focused on the enhancement of the visual display, refinement of the accretion and shedding algorithms and incorporation of data for a local weather station on the bridge.
4) Development of an ice presence and state sensor. No suitable sensor exists. Therefore, development and field testing were undertaken.
5) Transition the dashboard board and local weather station to ODOT District 2 so that the functionality of the dashboard and the information from the icing sensors is available to the operators of the VGCS.
As with Phase I, any proposed implementation was to be as “green” as possible. If the recommended solution involved the application of chemicals, then the potential environmental consequences of the chemical waste stream were addressed and “green” alternatives for conventional chemicals were investigated and utilized. The experimentation of the viable technologies will be thoroughly discussed in chapter 7 of this report. Through experimentation, no practical active or passive anti/deicing solution was ever identified. This ultimately led to a new overall objective, which was to improve the monitoring of icing events in order to provide ODOT with the best
32 information to manage their response to an icing event. Section 2.4: Expert Team Approach to the Research Because of the unique nature of the problem, the need for a quick response and the specialized nature of the icing knowledge required, the VGCS icing problem has been attacked with an expert team. The primary requirement was a team of researchers who are experts in ice and professionals familiar with the bridge. These are supplemented by team members who are expert in instrumentation, “green” energy and “green” chemistry. The team includes national expertise in icing from the U.S. Army Cold Regions Research and Engineering Laboratory and the NASA Glenn Icing Branch, expertise on the VGCS design and instrumentation, and experts in green technology. This team will address the unique features of the VGCS stays and provide recommendations to the Ohio Department of Transportation for the most practical and cost effective ice sensing, anti- icing and deicing systems for the VGCS. An expert team was the best way to rapidly assess the state of the art. This approach allowed the research team to confirm that a practical solution for ice anti/deicing for the VGCS does not currently exists. The icing experts have identified the information that must be collected and understood to design an effective anti/deicing solution. The research team consists of the following members: Jeff Baker, P.E., Independent consultant who was formerly the construction manager for VGCS, familiar with all aspects of VGCS construction and operation; experience with VGCS icing incidents. Nabil Grace, Ph.D., College of Engineering Dean, University Distinguished Professor, Lawrence Technological University; Director, Center of Innovative Materials Research; director of the LTU Comprehensive Environmental Test Chamber which has large scale icing test capacity. Michael Gramza, P.E., ODOT lead, District Construction Engineer for District 2, and former construction project manager of the VGCS. Cyndee Gruden, P.E., Ph.D., Associate Professor of Civil Engineering, University of Toledo; environmental engineer with expertise in management of deicing waste streams. Art Helmicki, Ph.D., Professor, Department of Electrical and Computer Engineering, University of Cincinnati; Director, Applied Systems Research Lab, a designer of the data collection system for the VGCS; expertise in sensor and signal processing. Victor Hunt, Ph.D., Research Associate Professor, Department of Electrical and Computer Engineering, University of Cincinnati; expertise in bridge instrumentation, a designer of the existing VGCS instrumentation system. Kathleen Jones, U.S. Army Cold Regions Research and Engineering Laboratory, Expertise; expertise in static and dynamic loads on structures due to atmospheric
33 icing; leader of freezing rain survey team; wrote ice load section for ASCE7 Standard, Minimum Design Loads for Buildings and Other Structures. Richard Martinko, P.E., Director UT-University Transportation Center and Intermodal Transportation Institute; former deputy director of ODOT District 2, former assistant director of ODOT, and former ODOT project principal of all phases of the VCGS project. Cyril Masiulaniec, Ph.D., Late Associate Professor, University of Toledo, Department of Mechanical, Industrial and Manufacturing Engineering; expertise in icing and thermodynamics. Douglas Nims, Ph.D., P.E., PI of this project, Associate Professor of Civil Engineering, University of Toledo; instrumentation and structural study of the VGCS; management of engineering consulting and academic teams. Tsun-Ming “Terry” Ng, Ph.D., Professor, University of Toledo, Department of Mechanical, Industrial and Manufacturing Engineering; expertise in icing and sensor. Currently, working on a study of icing on wind turbine blades.. Andrew Reehorst, NASA Glenn Icing branch; expertise in icing sensors; experience with ice accumulation and icing test facilities. Charles Ryerson, Ph.D., U.S. Army Cold Regions Research and Engineering Laboratory, Manager of CRREL’s Icing Program, Deep; deep and broad experience with aircraft and structural icing. Familiar; familiar with icing test facilities. His 2009 study on off-shore facilities is similar to this VGCS study. Thomas Stuart, Ph.D., Professor of Electrical Engineering University of Toledo; expert in power, PI of an ODOT funded research study of a solar installation near to provide power to the VGCS site. Mario Vargas, Ph.D., NASA Glenn Icing Branch, lead. NASA Glenn has an icing wind tunnel and the researchers are familiar with the capabilities of icing test facilities. Ted Zoli, S.E., Vice President of HNTB, expertise in icing; an extensive history of working with icing issues including testing structures on Mount Washington. Currently, he is engaged on two other cable stayed bridges with icing issues.
34 Table 3: Team Members Roles and Expertise Icing Local Green Team member Brief Description of Primary Activity/Expertise Expert Knowledge Expert
Former construction manager for VGCS, familiar with all aspects of VGCS construction and operation, experience with Jeff Baker X icing incidents.
Lawrence Technological University (LTU). Director of a unique low velocity wind/ freezing/icing/rain/load testing Nabil Grace X facility.
Mike Gramza X ODOT lead, former project manager of VGCS, able to provide input on ODOT operation needs.
Cyndee Gruden X University of Toledo. Expertise in management of de‐icing chemicals
Kathleen Jones X CRREL, national icing expert, leader in icing risk, member and former chair of ASCE‐7 committee on icing
University of Cincinnati. Instrumented VGCS, expertise in instrumentation and testing, support for implementation Art Helmicki X and testing costing
University of Cincinnati. Instrumented VGCS, expertise in instrumentation and testing, support for implementation Victor Hunt X and testing costing
Rich Martinko X University of Toledo. Understanding of ODOT operations, administrative support
Cy Masiulaniec X Late of the University of Toledo. Icing expertise, lead in performing thermal analyses and experiments.
University of Toledo. Lead in administrative support. Instrumented VGCS, lead in developing background Doug Nims X information for alternative, support for thermal calculations, lead in report writing and costing.
Terry Ng X University of Toledo. Icing expertise, lead in sensor development and experiments.
Andy Reehorst X NASA Glenn, icing sensor expert
CRREL, national icing expert, recently completed oil platform study which is parallel to the present VGCS study, Charles Ryerson X familiar with other test facilities nationally
Tom Stuart X University of Toledo. A lead in the design of the VCGS solar array, expertise in power management
NASA Glenn lead, aircraft icing expert, intimately familiar with test facilities at NASA Glenn and familiar with other Mario Vargas X test facilities nationally,
HNTB. National icing expert, consultant on VGCS design and construction, experience with icing problems on existing Ted Zoli X X bridges
35 Section 2.5: Benefits The benefits accruing to the traveling public, operators of the VGCS, District 2 and ODOT in general include: benefits accruing to ODOT and D02 from the dashboard and database include. 1) Comprehensive review of existing active and passive technologies: With the support of team member Charles Ryerson and drawing extensively, on his studies of icing technology, all of the known anti/de-icing technologies were investigated. The included over 70 technologies and is described in the technology matrix summarized in this report and presented in detail in Belknap, 2011.
2) Comprehensive review of past weather events; Team member Kathy Jones reviewed the icing events in northwest Ohio for the past twenty years including the first four icing events on the bridge. Vehicles were damaged in at least two of the first four icing events. A summary of this work is presented in chapter 4.
3) Detailed study of the 2011 major icing event: This was the fifth major icing event on the bridge since its opening. The icing team was onsite from the initial rainfall through the icefall for the February 2011 major icing event. Pieces of ice several feet long and up to three-quarters of an inch thick fell. The bridge was closed for several hours. The team was able to capture video and images of the ice shedding that lead to increased understanding of the icing behavior. A summary of the study of the 2011 icing event is presented in chapter 4.
4) Ice accretion and shedding algorithms: The study of the past weather and icing events lead to quantitative guidelines concerning the weather conditions that made icing accretion and shedding likely. These guidelines form the core of the algorithms in the ice monitoring system implemented on the bridge.
5) Development and implementation of the dashboard: In response to ODOT’s request for a way to make the results of the research easily actionable by the operators of the bridge, a real-time monitoring system was implemented. Initially, the information from existing sensors on the bridge and in the surrounding region was feed into the ice accretion and shedding algorithms and the results displayed on a graphical user interface. This interface was design so it displayed information about the icing status of the bridge in a simple on screen format, much as the dashboard of a car is designed to put the information essential to the operation of the vehicle in a visually compact format. At present, the dashboard reflects information from the initial sensors as well as a bridge mounted weather station and camera as well as temperature sensors on the stay sheaths.
6) Design, installation and use of a local weather station on the bridge. When it was identified that the existing sensor system on the bridge and in the surrounding
36 regional area was not adequate to monitor the microclimate on the VGCS and icing conditions of the stays, it was decided that a suite of local sensors was required. The research team identified an array of commercially available sensors that could provide most of the required information. The team then procured the sensors. A weather tower with local sensors and a camera as well as stay as mounting brackets to attach thermistors directly to the sheath were designed and installed. The sensors were made operational and their data was incorporated into the dashboard.
7) Field and laboratory studies of anti-icing and de-icing technologies: Experiments on anti/deicing chemicals, anti-icing coating and anti/deicing application of a heating system were carried out on full scale sheath specimens at the icing experiment station and in the icing wind tunnel at the University of Toledo. These studies coupled with the literature review demonstrated that no existing technology was appropriate for anti/de-icing on the VGCS.
8) Development of an ice presence and state sensor: No commercial sensor for directly measuring the presence or state of ice on the sheath exists. An electrical resistance based sensor has been developed. The sensor detects the presence of ice and can detect the layer of water, which is a precursor to ice shedding between the ice and the sheath. This sensor has been tested in the icing wind tunnel and at the icing experiment station. It is ready for deployment.
9) Database: The dashboard collects a comprehensive set of from the regional and local sensors on the bridge. It records all the icing and shedding alerts, serves as a log for all the observations and has the capability of exporting and plotting the data. This provides a database than can be used for study of the icing behavior of the bridge.
10) Archive: In addition to the weather data, the dashboard serves as a repository of all references, reports, presentations and other documentation of this project. This allows convenient access to the information for ODOT and researchers.
Through this project, the safety of those crossing the bridge has improved, the understanding of icing events has advanced and a sensor array capable of ascertaining the state of icing on bridge has been installed. Section 2.6: Chapter Summary The overall goal of this research was to assist ODOT in implementing an icing management procedure for the VGCS.
The objectives to support this goal were to Evaluate the state of the art in anti/deicing technologies through literature review and experimentation
37 Install sensors to understand the microclimate and icing on the bridge Design a real-time monitoring system to track icing conditions on the bridge with a straightforward interface so information on the icing of the bridge was readily available to the bridge operators.
An expert team approach was followed. A team with local expertise in the VGCS and expertise in anti/decing was formed and carried out the tasks to achieve the objectives. The objectives to support this goal were to Evaluate the state of the art in anti/deicing technologies through literature review and experimentation Install sensors to understand the microclimate and icing on the bridge Design a real-time monitoring system to track icing conditions on the bridge with a straightforward interface so information on the icing of the bridge was readily available to the bridge operators.
The overall benefit is increased safety for the traveling public. The benefits of completing this project were: Comprehensive review of existing active and passive technologies. Identification that no existing technology was suitable for anti/deicing the VGCS. Comprehensive review of past weather events. Detailed study of the 2011 major icing event. Ice accretion and shedding algorithms. Making real time icing information about the bridge available to the bridge operators. Development and implementation of the dashboard. Design and installation of a local weather station on the bridge. Field and laboratory studies of anti-icing and de-icing technologies. Development of an ice presence and state sensor. Creation of an icing database. Creation of an information archive.
38 Chapter 3: Phase I Research The research performed as well as the findings for Phase I of the VGCS project will be presented in this chapter. Phase I research included the VGCS’s stay sheath analysis, literary review, the completion of a technology matrix, and the identification of uncertainties as well as sensors that will resolve them. The research performed in Phase I allowed for better understanding of icing events that have occurred on the VGCS as well as viable ice protection technologies. During Phase I research, a number of uncertainties pertaining to the microclimate of the VGCS were identified. These uncertainties must be resolved in order to provide a practical anti/deicing technology and/or better monitoring of icing events, thus, several sensors were proposed. The Phase II research is discussed in chapters 5 through 9. Section 3.1: VGCS Sheaths The stay sheaths of the VGCS are unique. Typical stay sheaths are high-density polyethylene (HDPE). However, for the VCGS, stainless steel sheaths were chosen over HDPE due to their low life cycle cost. The brushed stainless steel surface also is an aesthetic enhancement for this signature bridge. The reflection of the light off the stays provides a unique appearance and enhances the effect of the illuminated pylon. Some characteristics of the stays sheaths, such as being made of a 1/8 inch thick 316L stainless steel, a brushed finish, and having a larger than typical diameter, may contribute to the icing problem the VGCS is experiencing. The stainless steel has a brushed finish. A Bendix Profilometer Peak Counter was used to determine the surface roughness on a sample piece of the VGCS sheath. The surface roughness could be a factor for ice clinging to the stays. After a team visit to the NASA Glen Icing Branch, it was determined that the sheath was roughly comparable to the smoothness of an aircraft, and therefore, did not facilitate icing. The test was run by The University of Toledo machine shop supervisor John Jaegly in the Material Science Lab room 1061 North Engineering. The machine uses a carbide tip moved across a surface. Table 4 shows readings taken across the grain of the brush finish and with the grain. Multiple readings were taken on multiple areas. Readings are in mirco-meters.
39
Table 4: Sheath Roughness Test Data Across the Grain With the Grain Test Area Test Area Test Area Test Area Test Area Test Area
1 2 3 1 2 3 1.269 1.212* 1.372* 0.228* 0.353* 0.470* 1.282 1.185* 1.347* 0.229 0.397* 0.485* 1.289 1.157 1.315* 0.231 0.405 0.507 1.146 1.345 0.218 0.409 0.503 1.150 1.374 0.206* 0.427 0.503 1.173* 1.367 0.381* 0.506* 1.171* 1.347* 0.390* 0.499* 1.373* 0.401*
Average Average Average Average Average Average 1.280 1.151 1.362 0.226 0.413 0.504
Avg. of 3 1.264 Avg. of 3 0.381 Note: * measurements are not included in the calculation of the average because the apparatus was being moved during the readings.
Section 3.2: Literature Review Icing is a worldwide problem for large bridges and other industrial facilities in cold climates; therefore, a broad literature regarding both structures that have been affected and anti-icing/deicing technologies were reviewed. This section will first discuss known icing events that have been found in literature, then anti-icing/deicing technologies found in literature, and finally the technology matrix for the Veterans Glass City Skyway. Section 3.2.1 Known Icing Problems on Other Bridges Leonard P. Zakim Bunker Hill Bridge: This particular bridge is an A-type cable stayed bridge that crosses the Charles River in Boston, Massachusetts. In March of 2005, the Boston area experienced winter conditions that caused the cables sheaths of the Leonard P. Zakim Bunker Hill Bridge to ice. The ice then fell off of the stays in large sheets and onto the roadway below. Officials and design engineer considered this weather to be a “fluke,” thus, no technology was investigated (Daniel, 2005) Penobscot Narrows Bridge: The Penobscot Narrows Bridge is an I-type cable stayed bridge that allows traffic to cross over the Penobscot River between Verona Island, ME and Prospect, ME (Penobscot Narrows Bridge and Observatory, 2014). The bridge was completed and opened in 2006 and experienced weather that caused icing for the first time in 2014. Due to the irregular occurrence of icing, the state DOT has taken an observation approach, thus, no technology is currently being investigated or deployed
40 (Gluckman, 2014) Port Mann Bridge: The Port Mann Bridge is an A-type cable stayed bridge that allows traffic to cross the Fraser River in Vancouver, B.C.. The bridge was opened to traffic in 2012 and experienced winter conditions that resulted in the accretion and shedding of “wet” snow in December 2012. Several technologies have been investigated, which includes: heating of the stays, the use of water, the use of a helicopter, cable collars, coatings, chemicals, sensors, etc. Currently, cable collars have been deployed on several stays (Meiszner, 2013). They have been relatively successful, but at times they get stuck on their way down the stays. In addition to the cable collars, a dashboard has been set up in order to provide real-time conditions on the bridge. Ravenel Bridge: The Ravenel Bridge is an A-type cable stayed bridge that connects Mt Pleasant, SC and peninsular downtown Charleston SC. In late January 2014, the bridge experienced weather that caused the cable stays to ice. Once the stays warmed up, ice began to shed causing damage to numerous vehicles passing below. There are reports of ice sheets as large as 8 to 10 feet falling from the cable sheaths (ABCNews4 WCIV- TV., 2014). Currently, there is not a technology deployed on this bridge. Severn Bridge: consisting of the Aust Viaduct, Severn Bridge, Beachley Viaduct, and Wye Bridge; stretches from England to Wales. The bridge was closed for ice falling off the stay cable on two occasions, February 6, 2009 and December 22, 2009 (Severn Bridge, 2011). Svinesund Bridge: The Svinesund Bridge connecting Norway and Sweden has an arc section of 188 meters (~617 feet). The single arc superstructure supports two-lane bridge decks on either side. To prevent ice formation on the arc during the winter, a temperature sensor controlled electric cable system was installed in the top section (Net Resources International, 2011). Southern Quebec, western New Brunswick, and eastern Ontario were covered with thick ice in 1998 due to a significant ice storm. Bridges and tunnels were closed because of the weight concerns as well as falling from superstructures (Countryman Electric). Uddevalla Bridge: The Uddevalla Bridge is an A-type cable stayed bridge located in Uddevalla, Sweden that crosses the Sunninge sound (Uddevalla Bridge in Sweden, 2012) .This bridge was opened to traffic in 2000 and has been experiencing frequent icing problems ever since (Bowers, 2014). Pulse electro-thermal de-icing (PETD) has been deployed on one cable and one pylon of the bridge for testing. Field testing, though successful, revealed a mechanical design flaw (Petrenko, 2011). This technology has been proven to be successful in icing conditions. In addition to PETD technology, the bridge consists of a large array of sensors that relay data back to a dashboard, which gives an early warning based on the microclimate of the bridge.
Section 3.2.2 Anti-Icing/Deicing Technologies found in literature Chemicals are considered as both deicing and anti-icing technologies. They are widely available commercially and are used in several industries such as aviation, off- shore oil industry, transportation departments, and marine structures. Chemicals have both dry
41 and wet applications. In anti-icing, chemicals are either used to reduce the adhesion strength between ice layer and the substrate or prevent the formation of ice. In deicing technology, chemicals are usually used to melt ice layers during or after ice storms. The most widely used chemicals are calcium chloride, magnesium chloride, potassium chloride, calcium magnesium acetate, urea, and agricultural based chemiccals. The major concerns regarding chemicals are environmental and corrosion issues as well as application persistence. Coating is another technology that is an active subject of development and testing. Coatings are a passive anti-icing technology, which means theey are applied to surfaces to reduce the adhesion strength of ice to the surfaces and to prevent the formation of ice on the surfaces (Ryerson, 2008). In Kulinich’s work, ice repellency of hydrophobic coating in different materials with different surface topographies was evaluated (Kulinich, 2011). A new approach of this technology would be the development of super hydrophobic material which causes water to bead into small drrooplets. Menini has developed a new coating with ice phobic characteristics on aluminum alloys which is widely used for several industries such as transmission lines, aircraft wings, and fuselage (Menini, 2011). Figure 6 shows water droplets on a surface which is covered by a superhydrophobic coating.
Figure 6: Application of Superhydrophobic Coating on the Surface (Ryerson, 2008) Concerns with coatings are efficacy in preventing of ice formation on the surfaces and their persistence through the winter. New designs, materials, and details can be considerred for new cable-stayed bridges in cold climates to prevent ice accumulation on their stays. Recently, innovative deicing technologies have been developed which use electricity as a technique to melt ice. These techniques cause ice to de-bond at the ice/substrate interface and then use external forces, such as gravity or wind drag, to remove the ice from the substrate. Electrical techniques have developed in three fundameental subcategories. These techniques include the following: 1) the application of a DC bias voltage to the ice/substrate interface, 2) pulse electro-thermal deicing, and 3) ice dielectric heating (Ryerson, 2008).
42 In the first technique, a small DC current is applied to the ice/substrate interface through conductors. The DC current, through electrolysis, ablates the ice into cavities filled with an oxygen and hydrogen gas mixture. The electrolysis-driven aablation reduces the contact area between the ice and substrate, and consequently reduces the adhesion strength. Figure 7 shows the presence of bubbles in the interface which helps in removing the ice. Experiments have demonstrated proof of concept, but the conductors are typically destroyed due to the high currents.
Figure 7: DC Bias Deicing where Electrolysis forms Bubbles (Ryerson, 2008) In pulse electro thermal deicing (PETD), a thin film conductor is used in the interface which melts the thinnest ice layer and external energy, wind or gravity, is used to remove the ice. Instead of heating the substrate continuously as done by most electrothermal deicing systems, the PETD rapidly heats the ice-subtrate interface for a few seconds with a high current, low voltage pulse. The rapid rise of temperature to only a few degrees warmer than freezing melts a thin layer of ice and causes ice debonding. Thhe short electrical pulse is the source of the energy savings. The best reference of that technology is a paper which was written by Petrenko (Petrenko, 2011). In this paper, Petrenko presents the PETD method, its theory, results of computer simulations, and extensive data from laboratory tests as well as several large-scale implementations. One of the promising features of this method would be the low energgy for deicing. Figure 8 shows a thin metal-foil heater which is used in the ice/substrate interface to melt the ice layer.
43
Figure 8: Pulse Electro Thermal Deicingg (PETD) (Ryerson, 2008) One of the implementations of PETD is the Uddevalla Bridge in Sweden (Petrenko, 2011). A system in current usage was installed on one cable, which is over 200m in length and 25cm in diameter, and one pylon. The system is no longer used on the Uddevalla Bridge. The third electrical technique is ice dielectric heating. That metthod uses high frequency excitation from 60 kHz to 200 kHz to melt an ice layer. Figure 9 shows an ice layer releasing from electrical transmission cable using the ice dielectric heating method.
Figure 9: Ice Being Released using Ice Dielectric Heating (Ryerson, 2008)
44 The next technology, which is considered as new teechnology, 20 years old, is electro- expulsive deicing systems (EEDS). EEDS uses a variety of technologies to create small amplitude and short duration mechanical pulses to remove the ice from the substrate. The most applicable design for cable stayed bridges is an electrically actuated system which was designed and developed by NASA Ames (Ryerson, 2008). There are several technologies which use heat as a method to deice or anti-ice structures. The most applicable ones are electrothermal, hot air, and water deicing. Electrothermal heating is using electrical resistance as a sourcce in either deicing or anti- icing. In electrothermal heating, heating of substrate occurs as a result electrical resistance in wires such as nichrome wires or carbon layers. Though this method is considered as an efficient heating system considering all of the energy conducted through the wires is converted to heat, its use of electrical energy is costly. An example of this method is the heating of automobile rear windows. Figure 10 shows an application of heating cables to prevent icing of hatches and bulk- head doors which is used by the Navy.
Figure 10: Navy Vertical Launch Systems with Electrically Heated Door Edges (Ryerson, 20008) Another source of heat for deicing and anti-icing is hot air. The example of this technology is the automobile windshield defroster. Hot air is used widely in the aviatioon industry. The U.S. Air Force has used jet engines mounted onto trucks as a source of energy to blow warm air across the wings of iced aircraft (Ryerson, 2008). Heated pressure washers and steam nozzles use hiigh pressure jets to remove ice from surfaces. The Navy considers this technology to be viable and less expensive for deiicing of ships (Ryerson, 2008). The efficiency of a hydraulic system is dependent to variety of factors such as: nozzle size, flow rate, wind, and distance. Though viable, steam is less available today than when steam power was common.
45 Another technology, which is considered unique, is iinfrared deiicing. This method is a kind of remote technology which heats the objects through absorption of infrared energy. There are still additional experimental and analytical needs to understand the deicing process using infrared heaters (Koeing, 2011). Figure 11 shows the use of electric infrared heaters to deice the door entrance at the Cold Regions Research Laboratory (CRREL) facility. Figure 12 shows the application of infrared deicing in the aviation industry.
Figure 11: Infrared Heaters above the CRREL Entrance (Ryerson, 2008)
Figure 12: Aviation Facility using Infrared Radiant System (Ryerson, 2008) Millimeter wave technology is another technology for deicing and detecting the presence of ice on surfaces. The ice naturally absorbs microwave energy and heats. That technology is applicable to any situations where water is available to absorb microwave energy and heat the surfaces (Ryerson, 2008). The topic of atmospheric icing on cables has been studied inteensively in the past. However, most of these studies were conducted for icing on to transmission and power lines, which behave differently than cable stay sheatths. Atmospheric icing is a general description for various phenomena which include: rimme, fog, freezing rain, and wet snow.
46 An enormous amount of research has been conducted to understand the nature of the problem and to predict the icing load on the structure. In order to accomplish this goal, numerous ice and snow models have been developed in order to predict the thickness of ice and/or snow as well as the weight associated to the ice and/or snow. By being able to predict the aforementioned values, it becomes possible to examine the effect of this load on man-made structures. Some scholars such as Makkonen (2010), Admirat (2008), Sakamoto (2000), Finstad (1988), and Nygaard (2013) have developed their own icing and snow models. Although they have developed different models, the significant parameters in the icing process are nearly the same for all of them. Additionally, all of the models follow the ISO standard equation for icing of structures (ISO12494 2001): Where,
α : Collision efficiency - the ratio of the droplets that hit the cable to the total number of droplets in the windward side (Dubach et al., 2005);
α : sticking efficiency - the ratio of the droplets that stick on the cable to the total number of droplets that hit the cable (Dobesch et al., 2005);
α : Accretion efficiency - a representation of the amount of the droplets that will freeze to the total number of droplets that hit the surface;
A : Cross-sectional area perpendicular to object;
V : Particle impact speed perpendicular to object;
w : Water content [mass concentration of the ice particles]. The major difference between each model lies in the parameters of collision, sticking, and accretion efficiency; all of which depend on the event itself. Therefore, given the above equation, the factors that affect the formation of ice include: wind speed, precipitation type, precipitation amount or visibility (visibility can be substituted if the amount is unknown), and the size of the cable. In addition, some other factors that can be used for detecting the right conditions for an icing event are temperature and humidity. In the future, the existing analytical studies could be advanced to address the torsional rigidity and geometry of the stays versus power lines. The stay specific models would be used primarily for forecasting ice accretion. The forecast weather data would be input in to the models and total accretion predicted. This forecasting of the ice accretion could be used to reduce the false alarms of the monitoring system. Several anti/deicing technologies have been tested as passive or active solutions such as deicing or anti-icing. These solutions prevent or diminish the accumulation of ice on the surface of a cable, or in this case a stay sheath. Two examples of these technologies
47 include internal heating systems and the super hydrophobic coatings, which were tested in the indoor icing tunnel and the outdoor experimentation station at the University of Toledo. These experiments will be discussed in more detail later in this report. Another method that was recently introduced by Couture (2011) utilized a photonic deicer to melt accumulated ice on power lines; this is displayed in Figure 13 below.
Figure 13: Photonic Deicer for Deicing of Power Lines (Couture, 2011) Cable collars are another technology that may be used in order to remove material such as snow and ice from cable stay sheaths. Cable collars work by using gravity to force them down the stay, ultimately removing the material, as seen in the video in the article by Meiszner (Meiszner, 2013). They are typically in the shape of a ring that has a significant amount of weight. This active technology has been used on the Port Mann Bridge to remove “wet” snow and has been relatively successful at doing so. Common issues associated with this technology include getting stuck in the process of removing the material from the stay, sliding over ice rather than removing ice, and the potential for damaging the sheath. Section 3.3: Technology Matrix The objective of literature review was to assemble a comprehensive list of all potential solutions for VGCS. To reach that goal, a technology matrix which has a description of the technology, a discussion of its advantages and disadvantages, a rough estimate of the cost of each technology, and the status was developed (Belknap, 2011, Ali 2013)). The technology matrix lists 75 potential technologies in 13 different categories. The categories of technologies include: 1- Chemical and chemical distribution: Anti- icing/deicing chemicals such as salt or agricultural base products and practical systems to distribute them.
48 2- Coatings: A layer applied to the surface of the sheath which prevents ice accretion on surfaces.
3- Design: Changing the shape of stay’s sheath to prevent ice accumulation.
4- Electrical deicing systems (electro-expulsive): Using repelling forces between conductors to produce an explosive force that ejects ice from the sheath.
5- Pneumatic expulsive deicing systems: That is the system which an inflatable boot covers the stays. When the boot is inflated, the ice cracks and falls down.
6- Heat: Use of thermal systems to prevent formation of ice or remove accumulated ice.
7- Infrared radiant heat: Use of radiant infrared heating to warm up the stays to prevent ice accumulation or remove accumulated ice.
8- Heating the ice-substrate interface: Applying heat directly to the interface between the ice and sheath. This reduces energy demand.
9- High-velocity water, air, or steam: Use of a high velocity stream of fluid to force the ice to fall off from stays.
10- Manual deicing methods: Chip or scrape ice off the stays.
11- Piezoelectric: Attach a piezoelectric actuator to the sheath surface to break the bond of ice to the stay causing shedding to occur.
12- Vibration or covers: Using vibration to break the ice-surface bond or covering the stays to prevent ice formation.
13- Ice detection: Sensors can monitor ice accumulation or detect the presence of ice.
Solving the icing problem of the VGCS is considered as applied research instead of basic research. Due to this fact, a technology selection meeting was held in June 2010. The notes for advantages and disadvantages of the technology matrix are available in a report which was submitted to ODOT, Innovation, Research and Implementation Section (Nims, 2010). Table 5 summarizes the most viable technologies which seem to apply for the icing problem of VGCS.
49 Table 5: Most Viable Solutions for the VGCS
Category Specific Technology Sodium Chloride Agricultural Products Chemicals Beet Heat Calcium Chloride Hydrobead Coating Internal Heating
Heat - Forced air - Air with piccolo tube
Section 3.4: Sensors on the VGCS A key advance from phase I to phase II was the addition of a local weather station on the bridge. Prior to the 2012 – 2013 icing season, the VGCS had several existing sources of which data could be collected to better understand the local weather. The existing sources included several local airports as well as an ODOT Road/Runway Weather Information System (RWIS). In the research team meeting at the end of phase I, it was agreed that the existing sensor array was not adequate to capture the microclimate of the bridge, which is needed knowledge for understanding the VGCS icing events. To improve data collection, thermistors were added at several location on the bridge in the fall of 2012 and a weather tower was added in the summer of 2013. The detail of the new sensors is in chapter 6 and a discussion of the weather history from the sensors is in chapter 4. Section 3.4.1: Sensors on the VGCS prior to the 2012 – 2013 Winter The main existing local source was the ODOT RWIS site 142, which was installed on the VGCS. The RWIS station on the VGCS included the following sensors which were used to understand the weather conditions prior to the 2012-2013 icing season: 1. Linux RPU – RWIS Elite weather station platform. A full-feature weather station capable of sensing a variety of road weather conditions, gathering traffic data and activating roadside devices. This platform supports a full range of atmospheric sensors as well as pavement temperature and condition sensors.
2. Wireless Pavement Sensor X6 – Collects traffic and weather information. A self-contained, in-pavement sensor that utilizes Vehicle Magnetic Imaging
50 (VMI) technology to detect vehicle count, speed and classification. In addition, the sensor measures pavement temperature and condition.
3. RM Young Ultrasonic Wind Sensor – Measures wind speed and direction.
4. RM Young Air Temp/Dew Point Sensor – Measures humidity and temperature. Has a special plate to block direct and reflected solar radiation while allowing air passage.
5. Weather Identifier and Visibility Sensor (WIVIS) – Precipitation Identifier/Classifier with Visibility. The Weather Information and Visibility Sensor (WIVIS) determine the type, intensity and rate of the precipitation that is occurring, as well as the visibility.
Additional weather information was also collected from Toledo Express Airport and the Toledo Executive Airport. Information from these sources included temperature, dewpoint, wind speed and direction, cloud cover and heights, visibility, barometric pressure, precipitation amount, lightning can be collected from airport stations (Agrawal, 2011). Section 3.4.2: Sensors added in 2012 – 2013 In October of 2012, an array of thermistors were installed on several of the VGCS stays. The installed thermistors allow a local measurement of stay temperature before, during, and after icing events to be obtained, which is vital to the understanding of ice accretion and shedding. The thermistors locations, uses, etc. is further explained in Chapter 6 of this report. Section 3.4.3: Sensors added in 2013 – 2014 In the past, the information on icing events was limited to direct observation on the bridge and/or restricted to data regarding weather conditions at the RWIS and airport stations. In order to help ODOT anticipate icing events, take the necessary action to inform the public in order to keep them safe, and to improve the performance of the dashboard for managing icing events, weather data pertaining to the microclimate of the VGCS was required. Table 6 summarizes the required information for managing upcoming icing events and the sensors that correspond to resolving the uncertainties.
51 Table 6: Uncertainties that Needed Resolved and Corresponding Sensors Required Uncertainties that need to be resolved Sensor Information It is difficult to be certain when ice accumulates on the stays except field observation. Ice Goodrich Ice Presence of Ice thickness also triggers the criteria in falling Detector conditions The temperature of the VGCS stays during icing Stay events is unknown. This temperature is Thermistors Temperature considered as one of the reasons for falling conditions
Solar radiation can cause the sheath surface Sky Solar temperatures to go above freezing even if the Sunshine Radiation ambient temperature is below freezing. This Sensor can trigger shedding of ice off the stays
The VGCS has its own climate. Type and Local Weather LWS/ Rain amount of precipitation, wind speed and Conditions Tipping Bucket direction need to be determined
Visual records of Observation of the stays condition during icing Camera icing conditions events can be valuable
Taking the above information into consideration, several sensors were requested by the research team in order to reach the aforementioned objectives. The sensors requested and brief descriptions are as follows: 1. Goodrich Ice Detector: The Goodrich ice detector detects ice accumulation on an ultrasonic axially vibrating tube. It also measures precipitation transitions between liquid and solid condition (Goodrich, 2009). One of the unique features of this sensor is that it differentiates rain from freezing rain.
2. Leaf Wetness Sensor (LWS): Leaf wetness is a parameter which is used to describe the amount of dew or precipitation left on the surface. The LWS has a potential to detect if water is liquid or frozen.
3. Sunshine Sensor: It has been observed that solar radiation on the VGCS stays is a condition that can trigger ice shedding. This particular sensor measures both global and diffuse radiation as well as sunshine duration. The
52 sunshine sensor uses photodiodes with a computer generated shading pattern for measuring solar radiation (Delta-T Devices, 2002).
4. Electrically Heated Rain and Snow Sensor – R. M. Young’s Model 52202 (Campbell Scientific): It is an electrically heated precipitation gage which provides year-round measurement of rain and snow. It has been observed that ice accumulation and ice persistence depend on the rate and amount of precipitation. This sensor also uses a wind screen to minimize the effect of wind on the rain measurement.
5. Weatherproof Camera: The goal of having active weatherproof camera on the self-supporting instrumentation tower is to observe the unquantifiable aspect of icing events and track the performance of the dashboard. Reviewable visual record of icing events gives valuable information before, during, and after storms.
The aforementioned sensors were installed in the summer of 2013 with the weather tower. They provide critical information pertaining to the microclimate of the VGCS, thus, aiding in the understanding of icing events. A more detailed description may be found in Chapter 6 of this report. Section 3.5: Chapter Summary This chapter thoroughly described phase I research. During phase I research, the VGCS stay sheaths were investigated, literature review regarding icing events on other structures as well as potential anti/deicing technologies was performed, a technology matrix to narrow down the numerous technologies to a few viable ones was constructed, and the history of sensor presence was given. The VGCS stay sheaths are very unique when compared to the typical HDPE sheath. The unique features of the stay sheaths include being made of a 1/8 inch thick 316L stainless steel, a brushed surface, and a larger than typical diameter. The stay characteristics were chosen for aesthetic reasons and due to the low life cycle costs of stainless steel. Initially, the brushed surface of the sheaths was thought to potentially increase ice adhesion. After a team visit to the NASA Glen Icing Branch, it was determined that the sheath was roughly comparable to the smoothness of an aircraft, and therefore, did not facilitate icing. A roughness test was performed, which confirmed this belief. Phase I literature review included icing events on structures, mostly bridges, from all around the world as well as potential anti/deicing technologies. The literature review showed that this is a relatively common problem for cable stayed bridges that are located in areas that experiences weather that gives rise to icing storms; this is described in detail in the following chapter. Additionally, it allowed for all potential technologies to be investigated and discussed in order to determine a viable technology to be used. However, it was determined that no such technology existed. Technologies that were investigated are as follows: chemicals, coatings, the application of a DC bias voltage to
53 the ice/substrate interface, pulse electro thermal deicing, ice dielectric heating, electro- expulsive deicing systems, electro thermal heat, hot air, water deicing, infrared deicing, millimeter wave technology, photonic deicer, and cable collars. A technology matrix was then assembled in to gather, organize, and describe all of the technologies. This technology matrix included 75 potential technologies that were separated into 13 different categories. These technologies were then narrowed down to the most viable technologies at a technology selection meeting in June, 2010, which can be seen in Table 5 above. The entire technology matrix can be seen in Appendix X. Additionally, at the technology selection meeting, it was determined that there was insufficient information concerning the ice accumulation conditions, the shedding conditions, the microclimate of the bridge and the effectiveness of the viable technologies. This ultimately led to the addition of sensors to the existing sensor array, which had been deemed inadequate. The existing array included the following sensors: Linux RPU – RWIS Elite weather station platform, Wireless Pavement Sensor X6, RM Young Ultrasonic Wind Sensor, RM Young Air Temp/Dew Point Sensor, Weather Identifier and Visibility Sensor, and a Tipping Bucket Rain Gauge. Thermistors were added prior to the 2012 – 2013 icing season and a Goodrich Ice Detector, Leaf Wetness Sensor, Sunshine Sensor, Electrically Heated Rain and Snow Sensor – RM Young’s Model 52202 (Campbell Scientific), and a weather proofing camera was added prior to the 2013-2014 icing season as part of the weather tower installation.
54 Chapter 4: Weather History, Modeling and Analysis Section 4.1: Introduction A brief look into the causes of ice storms as well as the history of icing events that have occurred on the VGCS will be presented in this chapter in order to provide a better understanding of icing events, nature of ice accretion and ice shedding, and what happens during ice storms. Since the VGCS was opened for service in July of 2007, there have been five major icing events that have occurred. Jones’ report describes the first four icing events and the weather conditions which preceded them (Jones, 2010). The last icing event, which occurred in February 2011, was directly observed by the research team. Since February 2011, there have been minor icing events that the Dashboard has collected data on. These minor icing events will also be discussed. The weather history and the observed icing behavior described in this chapter serves as the basis for the ice accretion and shedding algorithms in the Dashboard. Section 4.2: Description of the basic weather that gives rise to an ice storm Freezing rain has been the cause of four of the five icing events on VGCS. A prolonged freezing rain event requires a layer of cold (below-freezing) air at ground level, warmer air aloft, a high pressure system to hold the cold air in the place, and precipitation. The duration of a freezing rain event depends on how long the high pressure stays in the place and variations in the thickness of the cold air layer at that location. For a major icing event in Toledo, typically, the warm air aloft comes from the Gulf of Mexico and the cold air originates in Canada. Liquid precipitation in the warm air layer (that may have fallen from the clouds above as snow) supercools as it falls through the cold air layer at the surface. If that layer is not too thick the supercooled drops remain liquid. These freezing rain drops are likely to freeze when they impact a cold surface, primarily because of convective and evaporative cooling, with some contribution from supercooling. If the surface cold air layer is thicker the drops might freeze as they fall, forming ice pellets. These particles of ice accumulate on the ground, but are likely to bounce or slide off the bridge stays. On the cold side of a freezing rain event the precipitation is likely to be falling as snow, and on the warm side the precipitation is plain rain. As an event like this evolves, the precipitation type at one location may change between rain, freezing rain, ice pellets, and snow, sometimes with different types of precipitation occurring at the same time.
55
Figure 14: Damaging ice storm footprint map, 1946-2014 in the lower 48 states and portions of the lower tier of Canada. Section 4.3: VGCS Weather History Five icing events have occurred on the VGCS. Kathleen Jones, CRREL expert and research team member, has prepared a report that describes the first four icing events as well as the conditions which caused them (Jones, 2010). The last event, which occurred in February 2011, was documented by thee University of Toledo graduate students. The icing events and basic features are listed in Table 7 and Table 8.
Table 7: Ice Accumulation Weather Conditions
Ice Event Precipitation EEvent December 2007 Freezingg rain and Fog March 2008 Snow, rain, and fog December 2008 Snow and fog; freezing rain and fog January 2009 Freezing rain and fog February 2011 Freezing rain, cleaar
56 Table 8: Ice Falling Weather Conditions
Ice Event Ice Fall Weather
December 2007 Rain with temperature above freezing
March 2008 Sun with temperature above freezing
December 2008 Rain, gusty winds and temperatures above freezing
January 2009 Gusty winds, temperature above freezing
February 2011 Light wind, overcast, and temperature above freezing
It is also possible for wet snow to accumulate on the stays and cause an ice event. This wet snow, that includes both snowflakes and liquid water can accumulate on the stays. That means that an icing event can begin with the air temperature above freezing. Icing on the cables of VGCS may also occur in supercooled clouds or fog. The Liquid water content of the fog is inversely proportional to visibility. Fog droplets, with typical diameters of a few 10s of microns, have essentially zero terminal velocity, so move only when they are carried by the wind. Therefore, in cloud icing is likely to be significant only in a thick fog and high winds. To have a better understanding of icing events, nature of ice accretion and shedding, as well as what happens during ice storms, a brief summary and lessons learned from past icing events on the VGCS is presented below. Jones report describes the first four icing events and weather conditions that preceded them. The last icing event, February 2011, was captured and documented by the University of Toledo graduate students. December 2007: The data from Toledo Express Airport and Metcalf Field indicated freezing rain and fog occurred on December 9 - 10, which is believed to have caused ice accretion on the stays. Rainfall with temperatures above freezing triggered the ice shedding from stays, which took place on December 12. Ice shedding resulted in the closure of two out of three lanes of traffic as well as damaged vehicles. (Jones, 2010) March 2008: Weather data revealed that a snow and rain mixture with temperatures falling below freezing, concurrent with a fog, caused ice formation on the stays on the evening of March 27. The shedding of ice occurred in the afternoon of March 28. Clear
57 skies and air temperatures above freezing on March 28 were considered to be the shedding triggers. During the ice fall event, the center and left lanes in both directions had to be closed and at least one vehicle was damaged (Jones, 2010) December 2008: On December 17, ice was first observed on the stays. Data gathered from Toledo Express Airport, Metcalf Field, and Toledo Blade indicated that freezing rain, snow, and fog were the conditions that caused ice accretion on the stays. On December 24, ice shedding occurred with temperatures above freezing and gusty winds. It should be noted that ice persisted on the stays for 7 days. Throughout this event, the left and center lanes were closed for 5 days, starting on December 19. January 2009: Ice first was observed on January 3 and shedding occurred on January 13. The data from airport weather stations showed that freezing rain accompanied by fog caused ice accretion on the stays. Temperature rising above freezing and gusty winds on January 13 triggered the ice shedding from the stays. Note, the left lanes were closed until January 21, 2009. This could be due persistence of ice on the stays after the initial shedding event, which then melted over the next week (Jones, 2010). February 2011: This icing event was observed and recorded from the time of ice accretion, which started on the evening of February 20, to the time in which ice shedding occurred, which began on the morning of February 24. This event will be given as a day by day account in order to provide a clear and better understanding of icing on the VGCS. Researchers were on the bridge regularly throughout the event to capture photographs and video as well as properly document the behaviors between the accreted ice and the VGCS stay sheath. A detailed description of this event is in Belknap, 2011. Photos and video from this event are permanently archived by ODOT Research. The forecast for the night of February 20 was freezing rain, followed by a drop in temperature. On a local television station’s weather website (Storm Tracker 11, 2011), the forecasters predicted snow changing to freezing rain. The update for the overnight was scattered rain or freezing rain with additional ice accumulation. With low temperatures and precipitation, the conditions were conducive to ice accumulation. The ice dashboard was monitored that night by the research team. RWIS Station 582016, located on the bridge, was tracked watching both the readouts from the dashboard and the cameras on the stays. At 9 pm, the dashboard ticker was reporting Y3, indicating that the conditions had been conducive to ice accumulation for 6 hour and visual observation should be made to ascertain if ice had accumulated. At 10:29 PM, a researcher and the ODOT shift supervisor visited the bridge and confirmed the presence of ice. With this confirmation, the dashboard was put on alert status. Figure 15 shows a screenshot of the ice dashboard on February 21. The “Record of the last 48 hours” at the bottom of the dashboard shows the progression of conditions from “clear” to “alert”.
58
Figure 15: Dashboard readout for February 21, 2011 Also, at 9 pm, ODOT personnel and researchers obbserved ice on the stays. At 10:30pm, the ODOT shift supervisor and the researchers returrned to the bridge to observe the stays. At that time, the air temperature was 30° F and the precipitation was cold blowing rain and the prevailing wind was from the east. Rain droplets were of average size annd liquid water was clinging to or blowing off the stays. On the stay checked, the ice was 1/4 inch thick on top of the stay, about ½ inch thick on the east side of the stay, there were small icicles and the ice was about 1/2 in thick on the bottom of the stay, and on the west side the stay was bare except for occasional frozen rivulets. Clear ice was accumulaating. Clear ice with few air bubbles deposited by freezing rain or freezing drizzle is referred to as glaze ice. This is opposed to white rime ice with more air bubbles which results from super-cooled fog or cloud drops carried by the wind (Ryerson, 2011). There was no sign of melting - rain was running down and blowing off the icicles. The deck outside the parapet was covered with slushy ice. The parapet had crusty wet ice. Inside of the parapets on the median was an accumulation of wet ice that was 1/2 to 1 inch thick. The ice covered the eastern face of the cable. There was little to no ice on the western face. The conditions on the stays appeared uniform as far up the stay as visible. Figure 16 shows an overview of the ice accumulation and Figure 17 shows a close up of the ice accumulating on the east face of the stay. Figure 18 is a schematic of the distribution of the accumulated ice.
59
Figure 16: Overview of ice accreting on stay at 10:29 PM Sunday evening
Figure 17: Close up of ice accreting on stay at 10:29 PM Sunday evening
60
Figure 18: Stay cable diagram with ice accumulation Table 9 compares the I-280 bridge RWIS station weather readiing with the ice dashboard and Toledo Metcalf Field weather reading.
Table 9 Weather Conditions for February 20, 2011 (Kumpf et. al, Weather Underground, 2011) 20/Feb/2011 5-6 pm 6-7 pm 7-8 pm 8-9 pm 9-10 pm 10-11 pm RWIS Temp 30° F 30° F 30° F 30° F 30° F 30° F Precip None Rain Rain Rain Raain Rain Surface Wet Wet Wet Wet Wet Wet Surface Temp 30° F 30° F 30° F 30° F 30° F 30° F Dashboard Icing Conditions Clear Y1 Y2 Y3 Alert Alert Toledo Metcalf Temp 27° F 28° F 28° F 28° F 26° F 28° F
On February 21, 2011, Storm Tracker 11 (2011) recapped that 1/4” to 1/2” of ice accumulated Sunday. They called for an additional mix of snow, freezing rain, and sleet for the morning and 3-5 inches of snow in the evening. The wintery mix and snow fell throughout the day and accumulated on the stays. A layer of snow was between the ice already on the stay and the new accumulation in somme areas. ODOT placed barrels out at the inside shoulder. This allows the barrels to be qquickly reconfigured to close lanes. The research team stopped a total of four times, three times on main span and once on the back span. They chipped a hole in the ice to meeasure thickness at the main span near stay 6, which was inspected from the median. The ice on the east side was roughly ½ inch thick with closely spaced icicles on the bottom. Then the team looked for any variation in conditions along the length of the bridgee, however, they appeared to be consistent when viewed near stay 14 from the north bound side. Near stay 10 visual inspection from inside the truck on from the south bound side it was seen that below the
61 damper collar the ice appeared very thin or even possibly bare spots and above the damper collar ice appeared thicker with pronounced frozen rivulets. The team stopped one time on the back span near stay 8 where conditions were seen that roughly matched what was observed near stay 10. The wind was from the east as it has been throughout the storm. Generally, on the east side, the ice appeared to be thicker than on the west side. The ODOT supervisor felt the coating on the east side was thicker than he had seen before. On the west side, there were some spots that appeared to have a very thin coat below the damper collar. Above the damper collar, ice was thicker and the frozen rivulets appeared more pronounced than on the east side. The ice above the collar appeared uniform as high as it could be seen on both the east and west sides of the stay. However, it is impossible to discern anything more than gross icing further up than about mid-height. On February 22, 2011, the temperature started at 15° F and climbed to 21° F by mid- afternoon then dropped to the teens again at sundown (Weather Underground, 2011). Although the air temperature was below 32° F, the sun was out and the solar radiation was 575 Watts/m^2 at 1:35 pm. At this time, the researchers observed liquid water under the ice on the stays and ice fall due to solar radiation seemed imminent. However, the ice did not fall when due to the solar radiation and the liquid water refroze at the end of the day. The research team noted that water was dripping from the icicles on the bottom of the stays. There was ice covered snow in many spots along the stays. The ice cover was uniform from the bottom up to the pylon, as observed with binoculars. Cracking in the ice was observed on stay 4 and there was significant cracking of the ice at bottom of the sleeve. There was ice on the east face of the pylon glass, but there was not any ice on the west face of the pylon glass. ODOT moved the orange barrels out closing the inside lane. Figures 19 and 20 portray the ice accumulation on the stays for February 22. The eastern face was coated and the western face was bare except for frozen rivulets.
62
Figure 19: Ice Accumulation up east side of stay February 22, 2011
Figure 20: Frozen Rivulets and bare metal on the west side of stays February 22, 2011 On February 23, 2001, the Weather Underground (2011) archived the outside temperature starting at 19°F in the early morning and rising thrroough the day to 28° F in the late evening. No precipitation occurred and there was an overcast sky the entire day. Wind speed reached 9 mph mostly South to Southwest (Weather Underground, 2011).
63 The research team continued to monitor the conditions on the bbridge. Temperature readings and pictures were taken at intervals on the stays. Liquid water was seen under the ice again even though there was an overcast skky and the ambient temperatures in the mid-20’s. As the team checked later in the day, the ice was breaking off easily and liquid water was observed under the ice. The temperature in the interstice, gap between the ice accumulation and stay sheath, was measured throughout the da. A Fluke 561 Infrared Thermometer was used to measure surface temperatures and the temperatures in the interstice. The interstice temperature was measured with a type K contact thermocouple. The lead of the contact thermocouple is approximately 0.1 inch in diameter and during the day it would slide easily into the interstice. Sometimes the lead could slide in as far as 6-10 inches with little resistance. Figure 21 shows a measurement being taken. In Figure 22, the indicated temperature of the stay surface, measured with the infrared thermometer, is 27.0 F. Because of the high emissivity of the brushed aluminum surface, it is likely that the temperature of the stay surface is overestimated. In laboratory trials with reflective surfaces, the infrared thermometer consistently read higher than the actual temperature. The temperature indicated in the interstice, measured by the contact thermocouple, was 32 F. The lead for the contact thermocouple can be seen inseerted into the interstice. Table 10 shows the variation in the temperature in thhe throughout the day.
Figure 21: Thermocouple reading between ice and stay February 23, 2011
64
Figure 22: Thermocouple reading between the ice and stay February 24, 2011
Table 10: Interstice Temperature February 23 Time Interstice Air StayNote Temperature Temperature 8:15 am 24 F 21 F 20B No visible liquid water 8:50 am 24 F 21.9 F N/A No visible liquid water 9:20 am 24 F 21.9 F 20B 9:30 am 28 F 21.9 F 15B 9:45 am 26 F 21.9 F 11B Liquid water under ice 12:15 30 F 24.1 F 19B Liquid water under ice. Large pm pieces of ice broke free easily. 1:00 pm 31 F 24.1 F 20B 1:00 pm 32 F 24.1 F 19B 1:15 pm 35 F 24.1 F 18B Liquid water under ice. Sheets break free easily. 2:55 pm 32 F 25 F 20B Liquid water that had bled from under the ice refroze on the stay. 3:57 pm 32 F 27 F 20B 5:23 pm 31 F 27 F 19B Liquid water under ice. Sheets break free easily.
Even though the day was overcast, the UV radiation penetrated the clouds and passed through the clear glaze ice. It then warmed the surface of the sheath under the ice. The resulting infrared radiation could not escape the interstice, which resulted in a significant increase in the temperature within the interstice. Due to this dramatic temperature
65 increase in the interstice, a patch of ice was able to be removed easily. Figure 23 portrays the fragility of the ice when the interstice temperature is significantly increased. Note the liquid water on the surface of the stay. The contact thermocouple indicated temperature in the interstice was 36 F. This seems a bit unreasonable because the water in the gap was in close contact with the overlaying ice. However, water moving under the ice was sometimes visible and would run out from under the ice in small streams.
Figure 23: Cracking in ice from chipping away ice, February 23, 2011 At 5:23 pm stay 19B was examined closely and it was observeed that there was liquid water under most of the ice. Figure 24 shows the extent of the layer of water. Only the small section of ice at the bottom of the stay was frozen solidlyy to the sheath. Ice this precariously attached could easily be dislodged. All the research observations were made at ground level, so it is possible that the ice layer further up the stay was materially different. Figures 25 and 26 are schematics of the ice buildup and liquid water observed at different stays and times.
66
Figure 24: Section where ice was chipped away to take temperaturee readings February 23, 2011
Figure 25: Ice thickness measurements on back stay 19 February 23, 2011
67
Figure 26: Ice thickness measurements on back stay 19 February 23, 2011 On February 24, 2011 at 5am, a senior researcher was on the bridge making observations. The research team met on the bridge around 8am. Cameras were given to two junior researchers to capture the ice fall event. At 8:40am, the ice fall event began. The ice fell in large chunks, at random intervals and from random stays. Whether conditions during the shedding event were temperatures above freezing (32° F), overcast sky, and light wind. Traffic was stopped at 9:30am because pieces were falling close to vehicles in the outer lane as well as off the bridge and into the Maumee River. The video footage showing the actual lane closure is MOV04201 with the last semi-truck driving over the bridge at 5:28, film time. The ice continued to fall until 11am. By this time, 80-90% of the ice had already fallen. Some ice remained on the pylon glass as well as on the stays close to the pylon. Temperature at the time of the ice fall was above freezing (32° F), overcast, and light wind. The south bound lanes were opened before 11:30 am; the north bound lanes remained closed for several hours because of the ice on the pylon. Table 11 compares the I-280 bridge RWIS with the iice dashboard and Toledo Metcalf Field weather readings. Figure 27 show ice accumulation on the pylon glazing at the very top. Figure 28 show the bridge deck after a majority of the ice had fallen off.
Table 11: Weather conditions for February 24, 2011 (Kumpf et. al, Weather Underground , 2011) 24/Feb/2011 6-7 am 7-8 am 8-9 am 9-10 am 10-11 am RWIS 280 Bridge Temp 33° F 33° F 34° F 34° F 34° F Precip None None None None None Surface Wet Wet Wet Wet Wet Surface Temp 33° F 33° F 34° F 34° F 34° F Dashboard Icing Conditions R1 R2 R3 R3 R3 Toledo Metcalf Temp 33° F 33° F 34° F 34° F 34° F
68
Figure 27: Ice accumulation on pylon glazing Februaary 24, 2011
Figure 28: Ice on bridge deck after 80-90% had shed, February 24, 2011
69 In summary, ice shed from the stay cables was confirmed. Videos from the ice fall event documented the size and fall pattern of the ice. Large rectangular pieces of ice from stays 20B-16B and 20A-16A, roughly, had pieces making it to the north bound outer lane of traffic and even completely off the bridge. When the ice started accumulating on the stays Sunday night, barrels were set out one lane of traffic. As the days went by and the ice was predicted to fall, the barrels were moved out so that only one of three lanes of traffic could cross the bridge in the northbound direction and leaving two of the three lanes of traffic open in the southbound direction. This was decided because the ice had accreted most heavily on the eastern side of the stays and the typical wind direction throughout the persistence and shedding period was coming from the west. On Thursday February 24th, after about a half hour of ice falling, ODOT closed the lane behind stay 20B on the north bound side. Vehicles were routed off the bridge that could turn around. Once 80-90% of the ice was down, the vehicles remaining on the bridge were instructed to cross with caution. Once all the vehicles were off the bridge was closed until all the ice had fallen. The last ice to fall fell from the pylon glass in the early afternoon. Figure 29 shows the weather for the week of February 20, 2011. Figures 30, 31, and 32 graph the solar radiation for February 22, 23, and 24, 2011, respectively.
70
Figure 29: Weather Summary for the week of February 20, 2011 (Weather Underground, 2011)
71 Solar Radiation Feb. 22, 2011
700 600 500 400 Series1 300
Watts/m^2 200 100 0 1:29:00 AM 3:00:00 AM 4:10:00 AM 5:42:00 AM 7:31:00 AM 9:13:00 AM 1:22:00 PM 2:46:00 PM 4:00:00 PM 5:33:00 PM 6:57:00 PM 8:10:00 PM 9:34:00 PM 12:00:00 AM 10:25:00 AM 11:48:00 AM 10:54:00 PM Time
Figure 30: Solar radiation counts February 22, 2011
Solar Radiation Feb. 23, 2011
250
200
150 Series1 100 Watts/m^2 50
0 1:34:00 AM 2:52:00 AM 4:01:00 AM 5:30:00 AM 6:51:00 AM 8:08:00 AM 9:28:00 AM 1:31:00 PM 2:44:00 PM 4:05:00 PM 5:28:00 PM 6:38:00 PM 8:01:00 PM 9:13:00 PM 12:00:00 AM 10:32:00 AM 12:02:00 PM Time
Figure 31: Solar radiation counts February 23, 2011
72 Solar Radiation feb. 24, 2011
350 300
^ 250 200 Series1 150
Watts/ m 100 50 0 6:01:00 AM 6:44:00 AM 7:28:00 AM 8:10:00 AM 8:56:00 AM 9:48:00 AM 1:14:00 PM 1:50:00 PM 2:36:00 PM 3:16:00 PM 3:59:00 PM 4:55:00 PM 5:52:00 PM 6:26:00 PM 10:35:00 AM 11:17:00 AM 12:04:00 PM 12:37:00 PM Time
Figure 32: Solar radiation counts February 24, 2011 Section 4.4: Lessons Learned from Previous Icing Events There are several lessons to be learned from the five previous icing events. The first is how ice is accretes onto the stay sheaths. In all five events, accretion occurred in conditions where freezing rain and/or snow were present. These events are typically, followed by a sharp temperature drop and accompanied by fog. Additionally, it has been noted that in some events there is minimal precipitation, yet significant amounts of ice still accumulates, therefore it is possible that some of the ice that is accreted onto the stay sheaths comes from supercooled drizzle or cloud droplets (Jones, 2010). Another lesson to be learned is how the ice sheds off of the stay sheaths. In four of the five events, shedding which cleared the ice off the stays occurred when the air temperature warmed to above freezing and was accompanied by gusty winds, clear skies or sunshine, and rain. The exception was the January 2009 event. Then ice shed intermittently and partially. Although, this was true for the last major icing event, which was the only to be directly observed and documented, it was learned that the temperature doesn’t necessarily need to be above freezing in order for shedding to occur. This is due to a greenhouse effect in the interstice between the ice layer and sheath surface. It was observed that considerable water flowed under the ice layer at temperatures in the mid-20’s. This effect caused ice to be removed easily, thus, displaying the possibility that shedding could potentially occur at temperatures below freezing (32 F). The ice detector has proven to be sensitive and accurate for ice accretion. It may be possible to substitute
73 Section 4.5: Analysis The common weather conditions prior to the previous ice events on the bridge led to the development of criteria to use when checking for icing event conditions. Weather conditions (for at least 6 hours) that would likely cause Ice Accumulation: 1. Precipitation with air temperature at the bridge below 32o F, or 2. Fog with air temperature at the bridge below 32o F, or 3. Snow with air temperature at the bridge above 32o F. Weather conditions that would likely cause Ice Fall: 1. Air temperature above 32o F (warm air), or 2. Clear sky during daylight (solar radiation). In order to automate the process of predicting ice fall events, an algorithm was developed based upon the above criteria to evaluate weather data. The weather data collected consists of RWIS measurements and METAR data from the local airports. Taking these criteria and the available data into account, a specific set of criteria was developed for Ice Accumulation and Ice Fall (Tables 12 and 13).
Table 12: Ice Accumulation Criteria Source Condition Description METAR or Freezing (Air Temp. <= 32o F & Precipitation type RWIS Rain is Rain) OR (Precipitation type is Freezing Rain) METAR Freezing Air Temp. <= 32o F & Precipitation type Fog is Fog METAR or Wet Snow Air Temp. > 32o F & Precipitation type RWIS is Snow
Table 13: Ice Fall Criteria
Source Condition Description METAR Warm Air Air Temp. >= 32o F METAR or Clear Sky Condition type is RWIS Clear
The data sources used, including the secondary sources used for redundancy, are: METAR Data Sources 1. http://www.wunderground.com/history/airport/KTOL/2011/05/13/DailyHistory.html ?format=1 2. http://www.wunderground.com/history/airport/KTDZ/2011/05/13/DailyHistory.html ?format=1
74 3. http://weather.noaa.gov/weather/current/KTOL.html 4. http://weather.noaa.gov/weather/current/KTDZ.html RWIS Data Source
1. ftp://ftp.dot.state.oh.us/pub/doit/ssi_rwis/ 2. http://www.buckeyetraffic.org/reporting/RWIS/results.aspx The RWIS stations report only four precipitation types (Rain, Snow, Fog, None/Other) while the METAR stations report more than 30 types, many of which are similar and could be grouped into the four RWIS types. A similar grouping is applied to the METAR data for the Ice Fall criteria. The only available metric for sky cover is from the METAR data. This metric had several values, four of which are used to classify the sky cover as “Clear”, with all other values grouped as “Not Clear”. Several assumptions guided the design of the dashboard. The assumptions below are based on Kathleen Jones’ report (Jones 2010a), discussions with research team members Kathleen Jones and Jeff Baker and Mike Madry, ODOT Northwood Outpost, concerning the icing events on the VGCS. The assumptions are rough guidelines. There will be exceptions to the assumptions. Assumptions: Ice accumulates in a discrete time period and does not fall during that period. The threshold of concern is radial ice accumulation of ¼” It takes a long event (roughly 12 hours or more) for ice to accumulate. If the sky is overcast and the temperature is less then 32°F, the stay sheath temperature after the icing event remains below freezing. (Note: On Tuesday February 23, 2011 this assumption was revealed to be flawed.) Ice can accumulate on the stays from fog or precipitation other than freezing rain, e.g., wet snow accumulations can lead to ice accumulation on the stays. Air temperature can be used as a reasonable approximation for stay sheath temperature. Wind by itself does not trigger an ice fall. Previously accumulated ice falls when either of these conditions occur o The stay sheath temperature rises above freezing. o If the sky is clear, sunlight could trigger the ice fall at temperatures below freezing.
75
Section 4.6: Chapter Summary
This chapter described the weather that gives rise to ice storms, the VGCS’s weather history including previous icing events, lessons learned from those previous icing events, and accretion and shedding algorithms.
The weather system most often associated with major icing is warm air from the Gulf of Mexico overriding cold air from Canada. This leads to liquid water falling on a cold surface. However, other conditions and weather systems have also lead to ice accretion. Historically, roughly two icing events occur each year. The last major icing event on the VGCS was in 2011. Once the ice accretes, it persists until shedding conditions occur. Temperatures above 32 and/or solar radiation cause ice fall. The ice fall in four of the five previous events was accompanied by temperatures rising above 32. If the solar radiation level is high enough, ice can shed. In February of 2011, copious amounts of water flowing beneath the ice were observed when the outside temperature was several degrees below freezing. This layer of water below the ice is a precursor to ice shedding. There is a greenhouse effect that occurs when the solar radiation passes through the ice and heat is trapped between the sheath and the ice. Ice accretion and shedding do not occur simultaneously. The findings from the weather study and observation include: The VGCS is not a special icing structure which accretes ice at a rate different from the surrounding structures. The hazard arises because the aerodynamic ice sheets from the stays can fall on vehicles crossing the bridge. Ice can come directly from precipitation or supercooled drizzle or fog. The development of ice accretion and shedding rules
76 Chapter 5: Development of the VGCS Dashboard and Initial Dashboard Results Section 5.1: Introduction When no existing ice prevention or removal technology appeared to be practical and/or economical for the VGCS, ODOT elected to proceed with a monitoring system to assist them in managing icing incidents. Literature review and contact with experts revealed no existing monitoring technology. Therefore, a novel monitoring system was designed, implemented and executed which is the first step in finding a long-term solution. The goal was to integrate a complex set of data and produce a concise graphic interface that put actionable information at the bridge operator’s fingers tips. Much as a dashboard puts key information about automobile operation in the visual field of a driver. A dashboard was developed that helps in monitoring ice events and other related parameters at VGCS for continuous flow of weather data. It also helps in getting actionable information in the hands of those who must anticipate and respond to an icing event. The working of the initial dashboard was based on a set of newly designed algorithms, with the following key features:
Dashboard is only using existing sensors, thus, there is no additional cost of instrumentation or installation. Many are maintained by others. An existing suite of local weather stations is utilized to form a virtual 2-D area or network of weather data in order to detect approaching weather conditions or patterns conducive to ice accretion and/or shedding. The dashboard works on a set of algorithms, which is the result of an extensive study of the causes and patterns of icing. The solution used in the dashboard is very flexible and can be modified as per user’s requirements. The solution can be used for any location/site and not just restricted to VGCS. This chapter explains the working, algorithm, user interface and performance of the dashboard in detail. It can also be seen that since designing and implementing a practical solution will take several winters, the current solution needs the regular attention of the user (i.e., Ohio Department of Transportation) at several parts of the solution for a successful implementation. The primary objective of this aspect of the larger project was to leverage existing weather data from sources available on the web in order to develop a virtual instrument. This virtual instrument allows weather researchers, infrastructure researchers, and transportation personnel all to monitor for potential icing events from any Internet connected device. Listed below is the list of tasks for this phase of the project:
Add weather data to existing VGCS web interface and database for possible use in algorithms below. Develop a “check engine” light (e.g., green (no ice), yellow (ice, but no shedding), red (ice with possibility of shedding) that responds to the algorithm and sends out corresponding alerts to a list of ODOT’s choosing.
77 Develop a reporting function that will allow ODOT to: o Verify that alerts are responded to o Declare an icing event o Capture time stamp and observation notes/comments Develop database of ground truth field data collected during actual icing events to compare against Dashboard performance. Develop export function for historical data archived on the VGCS weather website. Run calibration studies based on historical/archived/ground truth data and characterize probabilities of false alarms and missed detections (i.e., false positives and false negatives). To accomplish this, a dashboard was developed which included the virtual instrument to deliver the right information based on the task list mentioned above. The dashboard also provides a rich toolset for more detailed monitoring and assessment based on regular collection and storage of weather data from multiple sources. In addition the dashboard provides a way to interact with all data collected by location on a map and plotting the different types of measurements over time. Icing is a complex problem, and no solution has been designed to remedy it, therefore, to design an online monitoring system of continuous weather data is especially helpful, a divide and conquer approach was followed. This approach is explained in Figure 33 by a process flow diagram. Following are the major steps followed: 1. Icing Experts: First step is to get necessary information from icing experts that includes researchers in Cold Region Research and Engineering laboratory (CRREL) and ODOT. The information is then analyzed to determine the icing criteria for all the three stages in icing namely ice accumulation, ice persistence, and ice shedding. 2. Data Collection: This is a crucial phase and involves three steps: Realizing reliable weather sources Choosing appropriate weather parameters that need to be considered in the algorithm design, since different sensor system measure different weather parameters. Collecting weather data from different sources and make it available to use 3. Data Processing: This phase involves the main analysis and design; it is explained through a flowchart below. The main steps are: Ice accumulation checks for the last hour, get the results and store in the database Ice accumulation check for the last few hours to determine ice persistence In case ice is detected, manual check should be done followed by reporting to dashboard Ice shedding checks for the last one-hour, get the results and store in the database Ice shedding check for the last few hours to determine ice persistence 4. User Interface: The final step is the design of a simple user interface that lets the user to check icing status at just one-click. It also provides lot of useful data to researchers pertaining to ice events.
78
Figure 33: Process Flow Diagram Overview of this chapter: Section 2 explains different criteria for ice accumulation and ice shedding. It also explains about different types of weather stations and the parameters that are important for ice event determination. Section 3 explains the algorithm used to determine ice accumulation and ice shedding for one-hour time. Section 4 explains the algorithm for calculating likelihood of iciing using the results from Section 3. Section 5 explains the features and design of the dashboard. Section 6 deals with the performance testing of dashboard for the past and recent icing events.
79 Section 5.2: Weather Data Section 5.2.1: Introduction To accomplish the objectives mentioned in the Section above, the first step is to gather information regarding weather conditions pertinent to icing. Based on statistical analysis of meteorological data [Savadjiev], there are various meteorological variables that affect the process of ice accretion and ice shedding. Extensive analysis was done on 57 icing events, which occurred in the period between February 1998 and January 2000 at the Mont Bélair in Quebec, Canada. Savadjiev also declared that freezing rain is the most important reason behind ice accretion. K.F. Jones performed a pervasive analysis of the icing events’ that occurred in Toledo between 2000 and 2010, and found some of the common properties that were persistent during each ice events. Some of the results taken from Jones’ work are listed below:
Ice accumulation occurred in both freezing rain and snow, both accompanied by fog. Ice shedding occurs when the air temperature warms to above freezing, which may be accompanied by rain, sunshine, or gusty winds. Freezing rain was associated with ice accumulation on the Skyway stays in three of the four ice events. The results from Savadjiev and Jones led to the development of criteria to use when evaluating for ice events conditions: Criteria that would likely cause Ice Accumulation: 1. Freezing Rain: Precipitation with air temperature below 32oF. 2. Freezing Fog: Fog with air temperature below 32oF. 3. Wet Snow: Snow with air temperature above 32oF. Criteria that would likely cause ice fall are as follows: 1. Warm Air: Air Temperature above 32oF. 2. Solar Radiation: Clear sky during daylight. Section 5.2.2: Data Sources To capture the icing criterions mentioned above, meteorological data are taken from different weather sources. They are: (1) Road Weather Information System (RWIS): RWIS can be defined as a combination of technologies that uses historic and current climatologically data to develop road and weather information (for example, now casts and forecasts) to aid in roadway-related decision making. The three main elements of RWIS are: Environmental sensor system (ESS) technology to collect data. Models and other advanced processing systems to develop forecasts and tailor the information into an easily understood format.
80 Dissemination platforms on which to display the tailored information. A typical RWIS contain data for the air temperature, dew point temperature, surface temperature, relative humidity, wind speed and direction, and precipitation type. The reading is taken from buckeye traffic website and it shows one of the readings for the RWIS station at Veterans Bridge. The station name, temperature and precipitation readings are labeled. Here’s listed the links to the RWIS data including the second sources for redundancy and reliability: ftp://ftp.dot.state.oh.us/pub/doit/ssi_rwis/ http://www.buckeyetraffic.org/reporting/RWIS/results.aspx To determine the icing events at Veterans’ Glass Skyway Bridge, weather data from four RWIS stations are considered. The sensor system of the four RWIS stations are tabulated below:
Table 14: Sensor System at RWIS Stations Site # 140 141 142 150 ID 582013 582014 582016 582024 Site I-475 @ US-23 I-75 @ I-475 Split I-280 @ VGCS I-280 @ Libbey Description Split - Lucas co. - SLM 4.9 Lucas Road NLF ID SLUCIR00475**C SLUCIR00075**C SLUCIR00280**C SWOOSR00420**C Latitude 41.68768° 41.67463° 41.65845° 41.52236° Longitude -83.69355° -83.57298° -83.51022° -83.46285° Atmospheric WIVIS Hawkeye WIVIS Generic Precip Sensor Wind Sensor RM Young RM Young RM Young RM Young R/H Temp Theis Theis Theis Theis Sensor Pavement 9 GH / 2 2 FP2000/ 2 GH / 6 GH 1 FP2000 Sensor Repeaters 2 Repeaters
(2) Meteorological Terminal Aviation Routine (METAR): METAR is a format for reporting weather information. A METAR weather report is predominantly used by pilots in fulfillment of a part of a pre-flight weather briefing, and by meteorologists, who use aggregated METAR information to assist in weather forecasting. METAR typically come from airports or permanent weather observation stations. Reports are generated once an hour, but if conditions change significantly, a report known as a SPECI may be issued several times in an hour. A typical METAR contain data for the temperature, dew point, wind speed and direction, cloud cover and heights, visibility, barometric pressure, precipitation amount, lightning, and other information. The sensor data is taken from wunderground website. The links to the METAR data including the second sources for redundancy and reliability are:
81 http://weather.noaa.gov/index.html http://www.wunderground.com/history/airport To determine the icing events at Veterans’ Glass Skyway Bridge, weather data from two Airports are considered. The sensor system of the two airports are tabulated below:
Table 15: Airport Information Site # Toledo Express Airport Metcalf Field Airport ID KTOL KTDZ Latitude Degree 41.5886° 41.5631° Longitude Degree -83.8014° -83.4764° Observing LAND SURFACE COOP LAND SURFACE ASOS Program AB ASOS ASOS-NWS ASOS-FAA
Source of the above data is: http://www.faa.gov/air_traffic/weather/asos/?state=OH The observing system in both the airports are of the type Automated Surface Observing Systems (ASOS). It is a joint effort of the National Weather Service (NWS), the Federal Aviation Administration (FAA), and the Department of Defense (DOD). The ASOS system serves as the nation's primary surface weather observing network and is designed to support weather forecast activities and aviation operations and, at the same time, support the needs of the meteorological, hydrological, and climatological research communities. The basic weather elements measured by ASOS observing systems are: Sky condition: cloud height and amount (clear, scattered, broken, overcast) up to 12,000 feet Visibility (to at least 10 statute miles) Basic present weather information: type and intensity for rain, snow, and freezing rain Obstructions to vision: fog, haze Pressure: sea-level pressure, altimeter setting Ambient temperature, dew point temperature Wind: direction, speed and character (gusts, squalls) Precipitation accumulation Source of the above data is: http://www.weather.gov/ost/asostech.html As explained above the implemented ice monitoring system for the Veterans’ Bridge uses a total of six weather stations which is tabulated below:
82 Table 16: Distances of Weather Stations from VGCS Weather Station Station Typee Arial Distance from Veteraans 140-IR 475 @ US 23 Split RWIS 6.4 miles 141-IR 75 @ SLM 4.9 475 Split RWIS 11.2 miles 142-I-280 @ VGCS RWIS 0 miles 150-I-280 @ Libbey Road RWIS 10.2 miles Toledo Express Airport Aiirport 12.2 miles Metcalf Field Airport Aiirport 10.3 miles
There are a couple of reasons behind choosing the above six weather stations for pooling weather data for the ice monitoring system, some of them are listed below:
Figure 34: Map Showing Distances of Weather Stations from VGCS (1) Distance from the site: The distances of the weather stations from the Veterans’ are shown in the table above. It shows that all the choseen stations are within 12 miles radius of the experimental site. Figure 34 is a map showing the weather stations. (2) Reliability: The local weather stations are operated by various sources and data is obtained via Weather Underground, which would have varying levels of reliability. The airports are run by the National Weather Service (NWS), which is the moost dependable weather service in the country. Section 5.2.3: Data Classification The weather data collected consisted of RWIS measurements and METAR data as explained above. Since the two weather stations measure a variety of weather
83 parameters, it’s important to filter out the pertinent ones that matches the ice criterion need. As we know that Ice events can be classified in two main stages namely: Ice Accumulation: The criterion for ice accumulation was given in above sections. Criteria 1: - Freezing Rain: Precipitation with air temperature below 32oF. Criteria 2: - Freezing Fog: Fog with air temperature below 32oF. Criteria 3: - Wet Snow: Snow with air temperature above 32oF. We can see in all three criterions that temperature is a common parameter, whose absolute value can be received by any weather station but the precipitation type can have different values. Table 5.4 lists all the precipitation types values for RWIS and METAR measurements. Then we will try to classify the precipitation types that can be used for the ice accumulation determination. It can be seen that there are 30+ precipitation types measured by METAR data and couple of them can be used to determine criteria 1, 2 and 3 as shown in Table 17. On the contrary, RWIS measures only ‘Rain’ and ‘Snow’ for criterions 1 and 3 respectively. Taking these criteria and the data being collected into account, a specific set of criteria was developed for Ice Accumulation shown in Table 18.
Table 17: METAR and RWIS Precipitation Measurements for Ice Accumulation METAR Precipitation Types RWIS precipitation Types Used For CRITERIA 1 Mist Rain, Light Rain, Heavy Rain Light Freezing Rain, Light freezing Drizzle Used for CRITERIA 1 Used for CRITERIA 2 Rain Fog, Light Freezing Fog
Common to CRITERIA 1 and 2 Ice Pellets Light Ice Pellets Used for CRITERIA 3 Snow, Light Snow, Heavy Snow, Blowing Snow Ice Pellets, Light Ice Pellets Used for CRITERIA 3 Snow Light Freezing Fog, Light Freezing Rain Light Freezing Drizzle Unused types Unused types Clear, Haze, Partly Cloudy Other Scattered Clouds Unknown Overcast Mostly Cloudy
84 Table 18: Ice Accumulation Criteria Source Condition Description RWIS Freezing Atmospheric Temp. <= 32o F & Precipitation type is Rain Rain RWIS Wet Snow Atmospheric Temp. > 32o F & Precipitation type is Snow METAR Freezing (Atmospheric Temp. <= 32o F & Precipitation type is Rain) Rain OR (All precipitation types listed under criteria 1 in the above table) METAR Freezing (Atmospheric Temp. <= 32o F & Precipitation type is Fog) Fog OR (All precipitation types listed under criteria 2 in the above table) METAR Wet Snow (Atmospheric Temp. > 32o F & Precipitation type is Snow) OR (All precipitation types listed under criteria 3 in the above table)
Similar to the ice accumulation, data classification can be done for ice shedding. Ice Shedding: The criterion for ice shedding was given in above sections. Criteria 1: - Warm Air: Air Temperature above 32oF. Criteria 2: - Solar Radiation: Clear sky during daylight. We can see that one of the criteria needs temperature as a parameter, whose absolute value can be received by any weather station but the second criteria require precipitation type giving information for the sky cover, which can have different values. Let’s first go through the complete list of the precipitation types that METAR and RWIS measure and then try to classify the precipitation types that can be used for the ice shedding determination.
Table 19: METAR and RWIS Precipitation Measurements for Ice Shedding RWIS precipitation METAR Precipitation Types Types Fog, Light Freezing Fog Mist Rain, Light Rain, Heavy Rain, Light Freezing Rain Rain Thunderstorm Heavy Thunderstorms Light Thunderstorms, Thunderstorms and Rain Ice Pellets, Light Ice Pellets, Light Freezing Drizzle Snow, Light Snow, Heavy Snow, Blowing Snow Snow Unknown, Overcast Other Mostly Cloudy None Used for CRITERIA 2 Clear Haze Partly Cloudy Scattered Clouds
85 For the purposes of the dashboard, the sky is considered clear if one of the following cases are true: 1.) the sky is clear or 2.) the obstruction of the solar radiation is small. The precipitation types for the dashboard algorithms include: clear, haze, partly cloudy or scattered clouds. Based on the two ice shedding criteria previously mentioned and the available precipitation types reported, the only metric for sky cover available was from the METAR data. This metric had several values, four of which are used to classify the sky cover as clear whereas all other values are evaluated as sky cover not clear. Taking these criteria and the data being collected into account, a specific set of criteria was developed for Ice shedding shown in Table 20.
Table 20: Ice Shedding Criteria
Source Condition Description RWIS Warm Air Atmospheric Temp. >= 32o F METAR Warm Air Atmospheric Temp. >= 32o F METAR Clear (Sky Condition type is Clear) OR (Any precipitation types listed under criteria 2 in the above table)
Table 21 summarizes different checks need to be done according to the algorithm and checks actually being doing in the dashboard.
Table 21: Final Ice Accumulation/Shedding Criteria Type of Ice accumulation check Ice shedding check station RWIS ☒ Temperature less than 32°F and precipitation type: ☒ Temperature greater rain than or equal to 32°F ☒ Wet snow with temperature greater than 32°F ☐ Clear sky ☐ Fog with the temperature less than 32°F Airports ☒ Temperature less than 32°F and precipitation type: ☒ Temperature greater rain than or equal to 32°F ☒ Wet snow with temperature greater than 32°F ☒ Clear sky / Scattered ☒ Fog with the temperature less than 32°F Clouds / Partly Cloudy during day time (8am to 6pm) Legends ☒ - Conditions checked in dashboard ☐ - Conditions not checked in dashboard
Section 5.2.4: Data Collection and Storage Once the relevant weather data from RWIS and METAR measurements have been identified, we collect them in the local database. Since METAR records are updated every 1-hour, the automated program runs every 1 hour for the data collection. RWIS measurements are updated every 10-minutes, so the automated program runs every 10 minute for the data collection. The automated program is written in the language
86 Python. Data being collected is stored in MySQL database in the UCII server. The tables in the database used in data storage are listed below: (a) METAR: Store METAR data. (b) RWISatmos: Store RWIS atmospheric measurements. (c) RWISsurface: Store RWIS surface measurements. (d) RWIStraffic: Store RWIS traffic measurements.
Table (a) METAR Data: The fields in this table are as follows: “Unixtime” – Time of record (in Unix time) “Temperature” – Atmospheric temperature reading (in o F) “Events” – Precipitation type/Sky cover in detail “Conditions” – Precipitation type/Sky cover in detail “Airport” – Airport KTOL or KTDZ There are few other fields recorded in this table, which are not used in the algorithm. They are: “Dewpoint”, “Humidity”, “Pressure”, “Visibility”, “wind_dir”, “wind_speed”, “gust_speed”, and “precipitation”.
Table (b) RWIS Atmospheric Measurements: The fields in this table are as follows: “unixtime” – Time of the record (in Unix time) “Sysid” – System-id 1 for the RWIS station. For the station 582014 Sysid is 582. “Rpuid” – System-id 2 for the station. For the station 582014, Rpuid is 14. “ApAir_T” – Atmospheric temperature (in o F) “Pc_Type” – Precipitation Type (1 – Rain, 2 – Snow, etc.) There are other fields recorded in this table, which are not used in the algorithm. They are: “RcdType”, “ApAir_Dewpoint”, “ApAir_RH”, “ApW_SpdAvg”, “ApW_SpdGust”, “ApW_DirAvg”, “ApW_DirMax”, “ApPrs_Barometric”, “Pc_Intens”, “Pc_Rate” “Pc_Accum”, “Vis_Distance”. *Tables (c) and (d) are stored only for future use and not used in the current algorithm. Section 5.3: Ice Accumulation Determination Algorithm Once the data sources and the criteria are decided, we need to use them to determine the potential for Ice Accumulation or Ice Shedding occurrences. The determination of ice conditions at each of the weather stations can be done and used to further evaluate the likelihood of an icing event.
87 Section 5.3.1: Data Update Time It must be noted that each of the weather stations has its own data collection time. This has a considerable significance on the time between which the algorithm is run. Since METAR data is important in both the Ice accumulation and Ice Fall determination and its update time is 1 hour, the algorithm cannot run for less than one-hour time difference to avoid checking same records for consecutive algorithm run. That’s why the least count between two runs in this algorithm is one hour. Section 5.3.2: Ice Accumulation Algorithm Sensors in any environment can occasionally misread the actual measurement so for each of the six weather stations, we evaluate all the records for the last hour. If at least 80% of the total records from the last hour meet any (or combination) of the three ice accumulation criteria, then the station has met the icing criteria as a whole for the last hour and is given a Boolean value ‘1’. If this condition is not satisfied by a weather station, the respective station is provided a Boolean value ‘0’. This is then used to find the likelihood of ice accumulation by multiplying the condition of each weather station (0 for not met, 1 for met) by the station weight and summing each result. If the total weight calculated, as above, is greater than a set threshold, we consider that potential icing conditions have been met for the last hour. The pictorial representation of the algorithm for determination of Ice accumulation is shown in Figure 35. The mathematical equation representing the algorithm is:
s ∗w WL, if WL TH; Icing possible
Here, s = Station condition (0 or 1), value is 1 if 80% of the records meet ice accumulation conditions in the last hour, otherwise 0. w = Station weight n = 6 stations WL = Weather Likelihood TH = Threshold, 0.3 in this case
88 Figure 35: Ice Determination Algorithm Section 5.3.3: Station Individual Weights
As seen in the algorithm, w is the weight/vote provided to each weather station. This plays a very crucial role in determining the ice accretion or ice shedding likelihood.
Table 22: Weather Station Weights
Weather Station Station Type Weight
140-IR 475 @ US 23 Split RWIS 0.1
141-IR 75 @ SLM 4.9 475 Split RWIS 0.1
142-I-280 @ VGCS RWIS 0.3
150-I-280 @ Libbey Road RWIS 0.1
Toledo Express Airport Airport 0.3
Metcalf Field Airport Airport 0.3
Threshold weight: 0.3
89 These station weights are based on two factors listed below: (1) Station’s geographic location relative to the VGCS Bridge – Since the ice conditions at VGCS Bridge is to be determined; it is obvious to provide stations near to VGCS high weights. As seen in the Figure 34, the weather stations distances from the VGCS Bridge influences weight, so the weather station on the bridge has the highest weight. (2) Amount of useful weather information that could be retrieved – As it can be seen in section 5.2.3, the weather data from METAR measurements offers more precipitation type parameters for ice accumulation determination and all the three ice accumulation criterions can be determined by METAR measurements, airports are given high weights in spite of their larger distance from VGCS Bridge. Section 5.3.4: Threshold weights Once the hourly weather data are utilized to determine whether ice accretion criterions is/are met for individual weather stations, the results from each station can be used to determine ice accretion as a whole. In this algorithm, [0.3] has been taken as the threshold weight. In future if we want to disregard one of the stations, we just need to make its weight zero without having any change in the algorithm. If we want to add a new weather station, we just need to add one more element in the weight array, which will suffice our need. Here, we want to briefly discuss the origin of the somewhat arbitrary selection of the voting weight and matching threshold of 0.3 in the dashboard algorithm. This decision goes back to the very start of the dashboard and work done at UC on the initial dashboard. At the time, we anticipated using multiple weather sources as voters each with individual weights all voting together. An alert would be set if the total votes exceeded a predetermined threshold. This is the basic scheme we use today. At the time, we did not know how many voters would be included or what the threshold would be set to. We adopted a "metric" system of sorts giving voters weights which were multiples of 0.1 figuring this would give us enough latitude to make adjustments. As time went on, it turned out we never had more than 3 voters for any given rule. In some rules we only had 1 voter. Thus, the weights ended up at 0.1, 0.2, or 0.3 depending on the reliability and proximity of each voter, and the threshold ended up at 0.3.
90 Section 5.3.5: Ice Shedding Hazardous ice shedding requires that a minimum thickness of ice accretion to have occurred first. During the design of the initial dashboard, there was not any equipment installed on the VGCS to measure the degree of ice accumulation, so physical observations were required for this determination. Once ice has accumulated to a degree of concern, the ice shedding potential is theen evaluated from the data collected at each weather station. The same check for 80% of the last hour of Weather data from RWIS and METAR is used to determine the station’s overall condition. Since the ice shedding check is significantly more critical and occurs at a faster rate, any indication of clear sky or sunlight has a high weight. This makes the weight of the airpport’s METAR sky cover measure as high as the temperature reading of the weather station on the bridge. From here, just like the ice accumulation, the assessmment result (0 or 1) for each weather station in table 22 is multiplied by the station weights and each result is summed. Again, if the total is greater than the threshold, ice shedding is considered possible during at this point in time.
Section 5.4: Ice Persistence Algorithm The determination of possible ice conditions (“yes”=1 or “no” = 0) happens each data collection cycle as discussed in the previous section. There is a duration of time, historically as long as days, that the ice accumulation has persisted. Therefore, a set of states and corresponding conditions for transition between states have been developed in order to assess the ice persistence. The resulting states were also driven by the need to have a dial or speedometer style gauge on the dashboard where each state is a position on it with the current state marked by a pointer as in a car. Transitioning between states is limited by the maximum measurement rate of the available date sources to get all pertinent measurements necessary for decision making. Specifically in this case the METAR data from the airport is upddated once per hour, which limits the rate of determination and state transition to once per hour. Section 5.4.1: Ice States To depict the level of ice accretion or ice shedding, eight different states are introduced in a speedometer style gauge shown below.
Clear No Ice present Y’s Ice accumulation likely (Three levels are introduced for different likelihoods) Alert Ice presence confirmed R’s Ice Shedding likely (Three levels are introduced for different likelihoods) Figure 36: Dashboard Speedometer
91 Table 23 provides an explanation of the eight states. Section 5.4.2: Ice Accumulation Persistence Algorithm An algorithm is designed to monitor icing on the stays, which uses five different states during ice accumulation process. They are: ‘Clear’, ‘Y1’, ‘Y2’, ‘Y3’, and ‘A’. Figure 37 shows the flow chart of the persistence algorithm for ice accumulation.
Table 23: Dial States Explanation
Nomenclature Color Significance
Clear Green No ice present on stays
Y1 Yellow Ice accretion possible. Icing conditions met for at least 1 hour, monitoring continuing
Y2 Yellow Ice accretion likely. Icing conditions met for past 8 hours, monitoring continuing.
Y3 Yellow Ice accretion very likely. Icing conditions met for past 10 hours, visual verification required.
Alert Orange Ice presence on the stays confirmed
R1 Red Ice shedding possible. Ice shedding conditions met for past 1 hour, monitoring continuing
R2 Red Ice shedding likely. Ice shedding conditions met for past 2 hours, monitoring continuing.
R3 Red Ice shedding very likely. Ice shedding conditions met for past 3 hours, visual verification required.
92
Figure 37: Ice Accumulation Flowchart The pictorial representation of each state transition in the ice accumulation process is provided in the user manual. Section 5.4.2.1: Transitions from state ‘Clear’ At start, the state is at ‘C’, clear (no ice). From ‘C’, the states may go either to ‘A’ (if ODOT reports presence of ice) or Y1 (if the icing criteria are satisfied for the last hour). These two actions are done priority wise, the priority of ‘A’ being higher than that of ‘Y1’. ‘Y1’ signifies icing is possible but the time for which ice accummulation criteria are met is one hour or less. The state transition diagram is shown in the Figure 36 above.
93 Section 5.4.2.2: Transitions from state ‘Yellow 1’ From ‘Y1’, the states may go either to ‘A’ (if ODOT reports presence of ice), Y2 (if the icing criteria are satisfied for a minimum of 6 out of last 8 hours) or ‘C’ (if none of the above two conditions are satisfied). ‘Y2’ signifies iciing is likely to happen. Section 5.4.2.3: Transitions from state ‘Yellow 2’ From ‘Y2’, the states may go either to ‘A’ (if ODOT reports presence of ice) or ‘Y3’ (if icing criteria are satisfied for at least 8 of last 10 hours) or ‘Y1’ (if none of the above two conditions are satisfied and also icing criteria did not satisfy for Y2 state). ‘Y3’ means icing is very likely to happen and now a visual check must be done to confirm the presence of ice. Once Y3 state has been reached, the algorithm will take the following actions:
No further transition until ODOT response is received. Send mail and text message to icing officials. Will keep sending alerts (mails/text) to officials until a response is received. Officials are needed to sign onto the dashboaard and provide response on actual icing condition at the bridge. A sample mail/text message is pasted below, which is used to notifying the icing officials about the ice accretion on the stays.
Figure 38: Sample Ice Accumulation Message Alert Also at state Y3, dashboard will display blinking Yellow signal to catch the attention of users.
94
Figure 39: Dashboard with Ice Accumulation Alert After the Y3 state is reached, the algorithm will wait for a response for ice presence. This led the algorithm to move to the next stage of the icing event process.
Section 5.4.3: Ice Presence Confirmation Different states used in the Ice presence process are ‘C’, ‘Y1’, and ‘A’. Figure 40 below shows the flow chart of the algorithm explained above.
Figure 40: Ice Presence Flowchart At Y3, further states are determined using the received response. If the response says ‘there is ice present on the stays’, the state will move to ‘A’, which means ice is confirmed. If the response says ‘there is no ice present on the stays’, the state will move to ‘Y1’.
95 There are two reasons why upon receiving a negative response, the state will move to Y1 and not C.
Ice may be present on the stays but it might not be enough to cause an ice fall hazard. In that case, the response received may be negative but a more insightful move is to assign Y1 as the new state in action to the response. In case of a mistake in submitting the response, assigning Y1 will take lesser time than ‘C’ to go to Y3 state, which is desirable. From here, we move to the last stage of the ice events, i.e. state transition algorithm for the Ice Fall process. Section 5.4.4: Ice Shedding Persistence Algorithm Different states used in the Ice Fall process are ‘R1’, ‘R2’, ‘R3’, ‘A’, and ‘C’. Figure 41 below shows the flow chart of the algorithm explained above.
96
Figure 41: Ice Shedding Flowchart The pictorial representation of each state transition in the ice shedding process is provided in the user manual.
Section 5.4.4.1: Transitions from state ‘Alert’ From ‘A’, the states may go either to ‘C’ (if ODOT reports that there is no ice on the stays) or ‘R1’ (if the ice fall criteria are satisfied for the last hour). These two actions are done priority wise, the priority of ‘C’ being higher than that of ‘R1’. ‘R1’ means ice fall is possible. Since the ice shedding process is faster and more critical than the ice accumulation process, an alerting system is implemented at every state change. Therefore, if the ice fall criteria are met during the last hour, the state will be changed
97 from A to R1, and also e-mails and text messages are sent to the ODOT officials notifying about the weather and the states. Section 5.4.4.2: Transitions from state ‘Red 1’ From ‘R1’, the states may go either to ‘C’ (if ODOT reports no presence of ice) or ‘R2’ (ice shedding criteria are satisfied for the last hour) or ‘A’ (if none of the above two conditions are satisfied). ‘R2’ means ice fall is likely to happen. Also mail/text message is sent to icing officials notifying them about the change. Section 5.4.4.3: Transitions from state ‘Red 2’ From ‘R2’, the states may go either to ‘C’ (if ODOT reports no presence of ice) or ‘R3’ (ice shedding criteria are satisfied for the last hour) or ‘R1’ (none of the above two conditions are satisfied). ‘R3’ means ice shedding is very likely to happen. Also mail/text message is sent to icing officials notifying them about the change. Once R3 state has been reached, the algorithm willl take the following actions:
No further transition until ODOT response is received. Send mail and text message to icing officials. Will keep sending alerts (mails/text) to officials until a response is received. Officials are needed to sign onto the dashboaard and provide response on actual icing condition at the bridge. A sample mail/text message is pasted below, which is used to notifying the icing officials about the ice accretion on the stays. Also at state R3, dashboard will display blinking Red signal to catch the attention of users as shown in Figure 43. After the R3 state is reached, the algorithm will wait for a response. If the response received is “no ice present on the stays”, the state will change to ‘C’ and if the response received is “ice present on the stays”, the state will change to ‘A’. For example, a transition for R3 to ‘A” would be caused by nightfall when the temperature drops and the solar radiation diminishes. This would make it likely the ice would persist through the night. A possible scenario causing the transition to C would be ice fall.
Figure 42: Sample Ice Shedding Message Alert
98
Figure 43: Dashboard with Ice Shedding Alert The process flow for ice accumulation, ice presence and ice fall combined together can be seen in the process state diagram shown in Figure 44 below.
Email & Text Msg
Figure 44: State Transitions possible from Red Level 3 Section 5.5: Monitor Website The algorithm design and working is explained in above Sections. This Section explains how the algorithm is implemented for the front end user. The toool through which the algorithm results are made accessible to users is the ‘Dashboard’. The implementation of the algorithm and the dashboard was divided into three parts in order to integrate data from and manage connectivity between separate systems. The first part was to automate the collection and data warehousing of weather measurements of the RWIS
99 stations and airport METAR data. These processes needed to be stable and robust so we had a reliable set of weather measurements from all the necessary sources. This system was not required for development of the next part, but having the necessary data stored locally inherently increases performance, reliability and robustness. From the developed algorithm and all the necessary conditions, software was developed to carry out the automated evaluation process and scheduled to run once per hour as previously mentioned. Finally a set of web pages were developed which is called the dashboard. As defined above the data collection and the results of the algorithm is stored in the database running in UCII server. This information is made available to the user through dashboard. Below listed are few of the dashboard’s important features:
Provide user-friendly speedo-style display of current icing status Collect and maintain database of weather conditions Automatically process incoming weather data and update icing status Generate alerts during icing events User-friendly display and navigation of weather data User-friendly display and navigation of icing event history
Section 5.5.1: Dashboard Main Panel The main panel of the dashboard contains the icing speedometer showing all the states including {G, Y1, Y2, Y3, O, R1, R2, and R3}. The main panel also includes the reporting function for ODOT, which can be used to report icing status after visual inspection. The ticker on the main panel shows icing conditions of the last 48 hours. The main panel includes the links to all other pages of the dashboard. In Figure 45, a screen shot of the web site is illustrated showing the dashboard main panel. As shown, the dashboard main panel provides a user-friendly dial that shows the icing status. The dashboard main panel is changed when icing reaches the level when an alert is necessary (e.g., see Figures 39 and 43).
100
Figure 45: Dashboard Main Panel Section 5.5.2: Weather Map Dashboard includes an interactive map of the weather stations where current sensor readings are shown and where historical readings can be plotted on a timeline. There are also cameras installed on the bridge that can be seen from this interactive map. For this purpose Google Maps API is used. Google Maps is a free web mapping service provided by Google, which offers street level maps for pedestrians, cars, and public transportation. It has an extendable API (Application Programming Interface) that can be used to develop custom Google Maps based applications. The Google map on the dashboard is used as an interface to provide the details about the various sites being monitored for determining the icing conditions at VGCS. It also contains the location of various sites, their past/current weather conditions along with a few reference links. Basically the features included in the weather map aare:
Locating weather stations Showing the live camera installed on the site Weather readings for all stations Site information of all the stations This section is a walkthrough of the working of the Google map on the dashboard. Once the accordion labeled “Map (Weather Data by Location)” on thhe dashboard is clicked, a Google map will open which has several markers on it (e.g., see Figure 34). There are two green markers written ‘A’ on it. These are the two Airports namely KTOL (Toledo Express Airport) and KTDZ (Metcalf Field Airport) whose weather data are being monitored by the dashboard.
101 There are four red markers written ‘R’ on it. These are the four RWIS stations namely Site 140-IR 475 @ US 23 Split, Site 141-IR 75 @ SLM 4.9 475 Split, Site 142-I-280 @ Veterans Glass City Skyway, and Site 150-I-280 @ Libbey Rd whose weather data are being monitored by the dashboard. There are three yellow markers written ‘L’ on it. These are the local weather stations near to the VGCS Bridge. They are: East Toledo, Oregon. These are not considered in the algorithm. The pink marker is the link to the live cameras installed on VGCS. RWIS Stations Red markers on the map represent RWIS stations. On clicking, an information box will be popped up that contains the weather station information, station id, current weather conditions and last 48 hours graphs to atmospheric and surface weather readings. Clicking the “Buckeye Traffic Readings” link on the information window opens up a new window having the weather reading from http://www.buckeyetraffic.org/. Clicking all other links on the information window opens up a new window having the last 48 hours weather readings. Below is an example for last 48 hours plot for the parameter ‘Air Temperature’. The ‘Export’ button will export the Air Temperature data for the last 48 hours into the excel sheet. Airports Green markers on the map represent nearby airports. On clicking them, an information box will be popped up that contains the current METAR report, plots of last 24 hours METAR reports, last 4 hours PIREP readings, plots of PIREP Icing information since Nov 13 2010, and WIKI references to METAR & PIREP. Clicking the “METAR” link on the information window opens up a new window having the current METAR reading from: http://english.wunderground.com/history/airport/ Clicking the “PIREP” link on the information window opens up a new window having the latest PIREP readings from: http://aviationweather.gov/adds/pireps/ Local Weather Stations Yellow markers on the map represent nearby weather stations. On clicking them, an information box will be popped up that contains the weather information at the local station. Attached below is the screenshot for the local weather station: East Toledo. Clicking the “Rapid Fire Panel” link opens up a new window having the weather information from: http://www.wunderground.com/weatherstation Clicking the “More information” link opens up a new window having the weather information from: http://www.wunderground.com/weatherstation
102 Live Camera Pink markers on the map represent link to the four live cameras installed at the Veteran Skyway Bridge. Clicking on any will get the live view with the images getting refreshed every five seconds. Section 5.5.3: History The History section of the Dashboard provides many additional features that are useful for the users for analyzing and reporting purposes. This section provides a detailed explanation of the same. Figure 46 below shows the main panel of the history section of the dashboard.
Figure 46: Dashboard History Panel The main functionalities of this section of the dashboard include: (a) List of events (states other than green) betweenn dates selected by the user: User can select two dates between which the states need to be displayed. As shown in the Figure above, two dates are selected i.e. 03/11/2011 and 05/28/2011, and a click on update button will display all the events occurred between these dates. Also clicking on any event will provide the cause of the event. (b) List of responses received by dashboard: All the responses received by the dashboard over the time are displayed in this section. This also includes the name of the person who posted the comment. Given below iis a screenshot. (c) Summary: This section provides statistics of the weather parameters for all the six weather stations. User can select two dates between which the statistics needs to be produced and which is then showed up upon clicking the update button.
103 Section 5.5.4: Implementation Tools To implement and design dashboard functionalities, several languages/tools/software are used. The list for the same is provided below:
Table 24: Tools Used To Design Dashboard Category Tool Used Purpose Programming Language Python Data Pulling Main Algorithm Scripting Language PHP Website design Database MySQL Data Storage Graphing Tools JPGraph Weather charts Matplotlib Test results Map Google Maps API Weather Map
Visuals Microsoft Visio, JavaScript Flowchart
Section 5.6: Performance Testing The dashboard became fully functional in the month of January 2011 (Jan 15 2011, 18:05:05) and since then it has been running on the University of Cincinnati Infrastructure Institute server, monitoring the weather conditions at Veterans’ Glass City Skyway. To test the performance of the dashboard, rigorous testing methods were implemented, which is the main agenda in this Section. The various tests done on the dashboard can be divided into two main sections: 1. System reliability test a. Weather station dependability 2. Ground truth a. Study of Feb 2011 ice events b. Study of past ice events Section 5.6.1: System Reliability Test Checking the weather data obtained from various RWIS stations constitutes the reliability test. For this purpose, all RWIS stations within 10 miles radius of the VGCS site are considered for the study. This could also be used as a cross check for the choice of weather stations and provides us a better insight on whether the stations needs to be changed/added for future ice analysis. A complete list of RWIS and Airports stations used for the study is given in the table below: Section 5.6.1.1: Weather stations dependability Weather readings are checked for the above-mentioned stations during the period when ice events occurred from 2006 through 2009. A table is provided for the date/time for which weather readings are studied.
104 Table 25: Dates for Past Ice Events that were Tested Icing Events during last 3 years Dates for which the data is studied 12 Dec 2007 2 Dec 2007 3:00PM - 23 Dec 2007 3:00PM 28 Mar 2008 18 Mar 2008 3:00PM – 8 Apr 2008 3:00PM 17 Dec 2008 7 Dec 2008 3:00PM – 28 Dec 2008 3:00PM 3 Mar 2009 24 Dec 2008 3:00PM – 14 Jan 2009 3:00PM
The following parameters are checked for stations fidelity:
Null readings: Number of readings which doesn’t have any value Bad readings: Checked +/- 3 standard deviation for outliers These checks are done for all the four icing events and the results are tabulated below:
Table 26:Weather Statistics for December 12, 2007 Ice Event
Site Site Site Site Site Site Site Site Airport Airport Dec 12 2007 137 139 140 141 142 146 147 150 KTOL KTDZ
Total number of records 5777 5892 5264 5834 5778 5566 5911 5868 841 828 Max 54 54 54 54 54 54 54 54 53.6 54 Temperature Temperature Min (deg F) 6 10 11 14 13 7 8 6 8.1 8.1 Temperature Null count 0 0 1 0 0 0 11 0 0 0 'None' 4761 4817 4396 4621 4871 4270 4769 4771 0 0 'Yes' 1010 1073 0 1206 0 0 1012 1073 0 0 Precipitation 'Rain' 0 0 422 0 322 444 0 0 108 143 (Occurences) 'Snow' 0 0 377 0 361 422 0 0 107 84 'Fog' 0 0 0 0 0 0 0 0 18 13 Null count 6 2 69 7 224 430 130 24 608 588
RWIS sites 140, 141, 142, 150 are used in the algorithm and it can be seen from the table that the precipitation types reading from sites 141 and 150 do not read anything except ‘Yes’ and ‘None’. Whereas the RWIS sites 146 shows a good dependability in terms of weather readings. A similar analysis was done for the other icing events. All this analysis leads to the following conclusions:
RWIS sites 137, 139, 141, 147, 150 – have just “Yes” or “None” as types of precipitation, so they cannot be used for ice determination
105 RWIS sites 141, 146, 150 categorizes all the precipitation types so they can be used for ice determination
Null or bad readings for all the RWIS sites are 82 out of 182,222 records in total i.e. 0.045%, which is a very small number and thus sites 141, 146 and 150 can be used in the ice determination algorithm without having too many outliers Section 5.6.2: Ground Truth After the tests performed on data reliability, the next phase is the testing of the actual algorithm. Since the inception of dashboard on Jan 15, 2011, there have been several occurrences of icing precipitation on the VGCS site. Section 5.6.2.1: Study of Feb 2011 ice events Ice Accretion detected by the Dashboard Table below provides a statistics on the algorithm results within the period between January 15 and June 15, 2011.
Table 27: Summary of Events when Ice Accumulation occurred in 2011 Maximum Ice Events Level Reason Comments Dates reached Freezing Feb 02 2011 Y1 No action taken since ice accretion was minimal Rain Freezing Feb 06 2011 Y1 No action taken since ice accretion was minimal Rain Freezing Feb 07 2011 Y1 No action taken since ice accretion was minimal Rain Feb 20-25 G-Y3-O- Multiple Described in detail 2011 R3 Due to bad precipitation data from VGCS RWIS Feb 25-27 Freezing Y3 station, the dial was stuck at Y3. So, it was manually 2011 Rain reset Mar 5-6 Wet Ice accretion occurred but it was not enough to cause Y2 2011 Snow Ice fall hazard. No action taken Mar 10-11 Wet Ice accretion occurred but it was not enough to cause Y2 2011 Snow Ice fall hazard. No action taken Wet Mar 30 2011 Y1 No action taken since ice accretion was minimal Snow Wet Apr 18 2011 Y1 No action taken since ice accretion was minimal Snow
106 As it can be seen that in the period from Jan 15 to June 15 2011, there have been several occurrences when the ice accretion was detected by the dashboard out of which only one event had sufficient ice to pose ice shedding hazard. The event from Feb 20-25 is described in detail bellow: Feb 20: The ice accretion process started at 15:00 because the RWIS station at the VGCS met freezing rain criteria and the dial in the dashboard became Y1. After then, the criteria for freezing rain constantly met for 8 hours in one or more stations including the two airports and the RWIS at VGCS. This caused the dashboard to display Y3 at 21:00. The ice accumulation was confirmed by the visual inspection done by UT at 22:00, thus the dial in the dashboard moved to Orannge, i.e. alert. Now the algorithmm is waiting for the ice shedding conditions to come. Local TV forecast: The forecast for that night was freezing rain, then a drrop in temperature. On a local television station’s weather website (Storm Tracker 11, 2011), the forecasters predicted snow changing to freezing rain. The update for the overnight was scattered rain or freezing rain with additional ice accumulation. With low temperatures and precipitation, the conditions were conducive to ice accumulation.
Figure 47: Weather Summary on Feb 20, 2011 Figure 48 shows the close up view of ice accretion oon Feb 20, 2011.
107
Figure 48: Screenshot Showing Ice Accretion on VGCS Feb 21: For the entire day, the weather conditions were such that none of the ice fall criteria met, and the dashboard remained at orange, i.e. Alert. As reported by the research team, University of Toleedo: The ice remained on the stays throughout the day Monday. The wintery mix and snow fell, accumulating on the stays. A layer of snow was between the ice already on the stay and the new accumulation in some areas. ODOT placed barrels out at the inside shoulder. This allows the barrels to be quickly reconfigured to close lanes. The research team stopped a total of four times, three times on main span and once on the back span. Main span near stay 6, inspected the east side from the median, they chipped a hole in the ice to measure the thickness. The ice on the east side was roughly 1⁄2 inch thick with closely spaced icicles on the bottom. Viewed near stay 14 from northbound side, were specifically looking for a variation along the length oof the bridge, conditions appeared the same as near stay 6. Near stay 10 from inside the truck on from the south bound side, below the damper collar the ice appeared very thinin or there may have been bare spots, above the damper collar ice appeared ththicker with pronounced frozen rivulets. Stopped one time on the back span, viewed near stay 8 from the south bound side, conditions roughly the same as observed when viewing near stay 10 from the south bound side. The wind was from the east as it has been throughout the storm. Generally, on the east side, the ice appeared to be thicker than on the west side. The ODOT supervisor felt the coating on the east side was thicker than he had seen before. On the west side, there were some spots that appeared to have a very thin coat below the damper collar. Above the damper collar, ice was thicker and the frozen rivulets appeared more pronounced than on the east side.
108 Both the east and the west sides, the ice above thee collar appeared uniform as high as it could be seen. However, it is impossible to discern anything more than gross icing further up than about mid-height.
Figure 49: Weather Summary on Feb 21, 2011 Feb 22: For the entire day, the weather conditions were such that none of the ice fall criteria met, and the dashboard remained at orange, i.e. Alert.
109
Figure 50: Weather Summary on Feb 22, 2011 Figure 51 shows the ice accumulation on the stays on Feb 22, 2011.
Figure 51: Ice Accumulation on Stays on Feb 22, 2011
110 Feb 23: For the entire day, the weather conditions were such that none of the ice fall criteria met, and the dashboard remained at orange, i.e. Alert. But there is one thing that caught the attention. The research team from The University of Toledo measured the temperature between stay and ice with a K contact thermocouple. One noticing observation was that temperature between the ice and stays (interstice) and atmospheric temperature were different. It was observed that interstice temperature was greater than air temperature. At 1:00PM interstice temperature was recorded 32°F whereas air temperature was 27°F. This may be due to greenhouse effect occurring in between ice and stays. Below provided are the results from The University of Toledo.
Table 28: Interstice Temperature on February 23, 2011
Interstice Time Stay Note Temperature 8:15 AM 24 °F 20B No visible liquid water 8:50 AM 24 °F N/A No visible liquid water 9:20 AM 24 °F 20B 9:30 AM 28 °F 15B 9:45 AM 26 °F 11B Liquid water under ice 12:15 PM 30 °F 19B Liquid water under ice. Large pieces 1:00 PM 31 °F 20B 1:00 PM 32 °F 19B 1:15 PM 35 °F 18B Liquid water under ice. Sheets break 2:55 PM 32 °F 20B Liquid water that had bled from 3:57 PM 32 °F 20B 5:23 PM 31 °F 19B Liquid water under ice. Sheets break
Feb 24: At 7:00 am, the Airport at Metcalf field reported temperature above 32F, thus for the first time the dashboard showed R1 to indicate possible ice shedding. It continued for two more hours and the dashboard moved to R3 at 9:00 am. The entire contribution goes to Metcalf Field Airport. Dashboard then generated alert signals and requested visual inspection. At 16:00, dashboard received a response from UT that little ice is still remaining, so the dashboard moved to orange. The comment for the same, taken from the dashboard, is pasted below: “As per discussion with ODOT and Dr. Nims, most of the ice has shed due to rise in temperature but a little ice is still remaining on the stays.” Again at 18:00, due to high temperature at each station, dashboard moved to R1 and in two more hours, it moved to R3. As programmed, it sent alert signals for visual inspection at 20:00. Now dashboard is at R3 and waiting for a reply.
111
Figure 52: Weather Summary on Feb 24, 2011
Figure 53: Example of Ice Shedding Alerrt Feb 25: At 00:59 am, a response was received that temperature is again dropping and Ice Accumulation is again possible. Comment from the dashboard is pasted below. “Temps dropping with ice persisting. Second round of possible icing followed by shedding.” With this, dashboard moved to orange again. Finally at 7:59 am, a response received that ice has shed and the possibility of second round of icing has diluted. The response received at 7:59 am is pasted below: “No ice apparent. Additional accumulation expected last evening did not materialize. ODOT has reopened all lanes.”
112 The dashboard was reset to Green at 8:59 am.
Figure 54: Ice Falling from VGCS on Feb 24, 2011 Figure 54 shows that 80-90% of ice has fallen. Conclusion: The ice event on February 20th reached Y3 due to persistent freezing rain, which was confirmed by visual inspection, andd a response was submitted resulting in a state transition to alert. The ice remained on the stays unttil February 24th and was monitored by officials for possible ice shedding. The ice, eventually, from this event did fall into traffic lanes and in pieces large enough that could havve damaged vehicles. However, well before this point ODOT officials closed the bridge until the ice had sublimated and it was safe for traffic to return. The dashboard reset to Green (all clear) after everything became normal. Thus, the performance of the overall system during the winter months of 2011 proved to be successful especially in determining ice accumulation and persistence.
113
Figure 55: Weather Summary on Feb 25, 2011 The overall algorithm can be simulated on a graph as shown below. The graph shows the actual weather data (temperature and precipitation types) oof the six weather stations and the algorithm response on those data. The simulation is done for the time period Fab 21 to Mar 06 2011. Parameters shown in the graph: 1 Temperature versus time (for six weather station) 2 Average temperature versus time (aggregated for six weatther station) 3 Ice accretion criteria satisfied 4 Ice shedding criteria satisfied Figure 56 is a two-axis graph, in which x-axis is the time in dates (Feb 21 to Mar 06 2011), left y-axis is the temperature in F, and right y-axis is the % of snow/rain per hour. The graph shows the temperature versus time graph for all the weather stations as well as the average temperature versus time. The rain occurrences (in % of record per hour) are shown in blue crosses whereas snow occurrence (in % of record per hour) is shown by red crosses. The yellow triangles are the hours when ice accumulation criteria are met (as per the algorithm described in above Sections). The red squares are the hours when ice-shedding criteria are met (as per the algorithm described in above Sections).
114
Figure 56: Feb 24, 2011 Algorithm Performance Graph Weather Station Performance Evaluation The analysis of algorithm for one icing event has been done in the above section. This section explains the contribution of each icing criteria towards those icing events. Also, it explains the contribution of all the six weather stattions towards the icing events. This study is done as it provides a deeper understanding about the role of each icing criteria and weather station so that in future the focus could be laid on the most important factors affecting the icing and de-icing. The table below summmarizes the overall weather conditions at all the six weather stations during the period between Jan 15 2011 to May 27 2011 (date of creation of this document). This report is generated from the “History” tab of the dashboard. Table 29 shows temperature and precipitation readings for the six weather stations. It can be seen that there is no ‘Rain’ or ‘Snow’ readinggs for RWIS site 582014 and site 582024. Henceforth, these two stations cannot contribute to the necessary criteria required for Ice accretion. To compliment this, two figures are drawn to show that the main contribution towards the ice accretion process is limited to few weather criterions and only few weather stations. During the entire ice event’s 68 hours, each of the precipitation type’s contribution and each weather station’s contribution was counted and shown in the Figure 57b. Figure 57a shows the three icing criterion on the x-axis and number of hours meeting icing criteria on the y-axis. It can be seen thhat freezing rain is the leading criteria out of the three. Figure 57b shows the six weather stations on the x-axis and the number of hours meeting icing criterion on the y-axis. It can be seen
115 that out of the six weather stations, the RWIS station at the VGCS Bridge and the two airports are the leading contributor.
Table 29: Station Comparison for the 2011 Winter Weather Report for VGCS Bridge from Jan 15 2011 to June 15 2011 Weather KTOL KTDZ RWIS Site RWIS Site RWIS RWIS Parameters (Toledo (Metcalf 582013 582014 Site Site Express Field (140-I-475 (141-I-75 582016 582024 Airport) Airport) @ US-23 @ I-475 (142-I- (150-I- METAR METAR Split - Split - SLM 280 @ 280 @ Lucas Co.) 4.9 Lucas) VGCS) Libbey Road) Temperature (°F) Average 39.0 39.8 29.9 40.6 35.8 39.6 Minimum -4.0 -8.0 -4.0 -3.0 -3.0 -19.1 Maximum 87.1 88.0 82.0 87.8 82.0 84.4 Precipitation Cleara 1469 1400 4041 13461 9941 6112 Rain 109 81 23 XXX 2309 XXX Snow 60 44 521 XXX 1579 XXX Others 47 1 2866 3509 67 10812
60 550
50 40 hours 330 of 40 hours 220 of 30 110
20 Number 0
Number 10
0 wet snow freezing rainfreezing fog VGCS Icing Conditions Station Weather Stations
Figure 57a: Contribution by the three criteria Figure 57b: Contribution by the six stations
Figure 57 : Contribution of the Icing Criteria and Weather Stations Thus it can be concluded that the RWIS station on the VGCS Bridge and the Freezing Rain had the most influence on the ice accumulatioon decision process.
116 Section 5.6.2.2: Comments Dashboard has done pretty well in detecting ice accumulation each time, but the analysis done on the algorithm results and University of Toledo’s response from the visual observations concludes that additional information would be needed to increase the ice fall detection speed and accuracy. This would also inccrease performance of ice accumulation detection and improve the overall system’s accuracy and reliability. For example, it was noted that after taking temperature readings on the stay and in the gap between the ice and the stay, that the atmospheric temperature readingss could be below 32F but the stay and air gap could be above 32F. This is most likely due to solar radiation, which could heat up the stay well before the air temperature increases (for example in the early morning or first sight of clear sky coverage). Following up on this assumption, data was collected from a weather station approximately 3 miles away that measures and records solar radiation, which was plotted and compared with the ice fall visual observation timeline. This plot shows that the solar radiation at this weather station started to rise close to 7:00am, which is approximately the same time the dashboard moved to R1.
Solar Radiation 2/24/2011 Ice m^2 Temp Gaugee /
Watt AM
Stay Sheath 7:04 AM 8:16 8:16 AM 9:28 AM
10:40 AM 11:52 AM Figure 58a: Solar Radiation Measurements (KOHOREGO1 Figure 58b: Measuring sheath Weather Station in Oregon, OH) temp and ice gap.
Figure 58: Solar Radiation Variation on Feb 24, 2011 Thus, it is proposed that for better ice shedding detection, a thhermocouple sensor should be installed at VGCS so that it can measure the exact temperature beneath ice. Also, installation of a sensor measuring solar radiattion could result in bettter detection of ice shedding events. In addition, since Ice shedding events are faster and crucial for the entire project, measures could be taken to decrease the time between which algorithms is run, at least while detecting ice shedding. As it was seen that when the first time sky became clear (with ice on stays), ice started melting and in less than 2 hours, most of the ice got melted.
117 Section 5.6.2.3: Study of Past Ice Events VGCS Bridge was built in the year 2007 and since then five major icing events have occurred. Research team member, Kathleen Jones has prepared a report that describes the icing events and the weather that preceded them. If the historical weather trends reported from 1955 to 1999 continued, there were likely 6-10 freezing rainstorms in the three years the bridge has been open. Freezing rain occurred for three of the first four icing events on the VGCS. “Four events in three years are likely representative of the future and years with no icing events will be unusual”, as stated by Jones. A freezing rain event requires cold air below, warm air aloft, a high to hold the cold air in place, and precipitation of liquid water. The duration of the icing event depends on how long the high pressure stays in place and the occurrence of the freezing rain (rather than ice pellets). Precipitation rates are typically low during icing events. In the January 2009 icing incident, the precipitation was just 0.06”. It is possible that some of the ice comes from super cooled drizzle. Therefore, it takes some time for the ice to accumulate. It is also possible for wet snow to accumulate on the stays when the air temperature is above freezing. This snow can turn to ice on the sheaths. Therefore, it is possible for an icing event to begin when the air temperature is above freezing. Icing on the bridge superstructure may also occur in super cooled clouds or fog. The likelihood of significant ice accumulation increases with decreasing visibility in the fog, increasing wind speeds, and the persistence of these conditions. In past events, the precipitation has been sometimes described as a wintery mix: snow, sleet, ice pellets, or rain. In the March 2008 event, the Toledo Blade said “Thunder, lightning, sleet, snow, rain, and freezing rain: There was a little of everything in the air around Toledo late Thursday and early yesterday…” (Blade March 29, 2008 cited by Jones). Sometimes, like the January 2009 icing event, there is a series of precipitation events rather than a discrete event. The basic features of the four icing events are listed in figure below.
118
Figure 59: Features of the Past Icing Events To test the fidelity of the algorithm, it is run for all the four pastt ice events. This section explains the results and analysis of the past ice events when run over the designed algorithm. (1) Ice Event Dec 12 2007 The historic data are downloaded from the sources mentioned above. The data is then run on the designed algorithm to get the result shown in the Fiigure 60 below.
119
Figure 60: Dec 12, 2007 Algorithm Performance Graph Axes of the graph: x-axis: Time in dates (least count = 1hour) left y-axis: Temperature in oF right y-axis: % of rain/snow per hour How the graph is drawn? The data from 2 Dec 2007 3:00PM – 23 Dec 2007 3:00PM is downloaded and stored in an excel file. 21 days (i.e. 504 hours) are considered for the analysis and the following graphs are obtained:
Absolute atmospheric temperature for 504 hours for all the six stations Average temperature per hour % Occurrence of rain per hour for all six stations % Occurrence of snow per hour for all six stations Exact algorithm is run over the data and the following graphs are obtained:
Number of hours when icing criteria is satisfieed Number of hours when ice-shedding criteria is satisfied
120 The above graph shows that for continuous 11 hours, icing criteria are met on Dec 9, 2007 at 22:00 and the dashboard would have moved to Y3. Again, ice-shedding criteria are met thrice on Dec 11, 2007 at 00:30am. Conclusion: Algorithm successfully caught the icing event.
(2) Ice Event Mar 28 2008 The data from 18 Mar 2008 3:00PM – 8 Apr 2008 3:00PM is downloaded and stored in an excel file. 21 days (i.e. 504 hours) are considered for the analysis. The graph below shows that for 8 out of 10 hours, icing criteria are met on Mar 28, 2008 at 02:00 and the dashboard would have moved to Y3. Again, ice-shedding criteria are met thrice on Mar 28, 2008 at 21:30. Conclusion: Algorithm successfully caught the icing event.
(3) Ice Event Dec 17 2008 The data from 7 Dec 2008 3:00PM – 28 Dec 2008 3:00PM is downloaded and stored in an excel file. 21 days (i.e. 504 hours) are considered for the analysis.
121
Figure 61: Mar 28, 2008 Algorithm Performance Graph
122
Figure 62: Dec 17, 2008 Algorithm Performance Graph The above graph shows that this icing event was not caught by the algorithm, and, therefore, some changes in the algorithm were needded. Conclusion: Algorithm did not catch the icing event..
(4) Ice Event January 3 2009 The data from 24 Dec 2008 3:00PM – 14 Jan 2009 3:00PM is downloaded and stored in an excel file. 21 days (i.e. 504 hours) are considered for the analysis.
123
Figure 63: Jan 03, 2009 Algorithm Performance Graph The above graph shows that for continuous 3 hours, icing criteria are met on Dec 31, 2008 at 01:00 and the dashboard would have moved to Y1. Again, ice-shedding criteria are met thrice on Jan 2, 2009 at 02:00am. Conclusion: Algorithm partially caught the January 3, 2009icing event. The results of all the four ice events are tabulated below. Possible solution: As it can be seen from the graph that on 31 Dec 2008, 01:00, there were 3 incidents (hours) when icing criteria was met and the dashboard moved to Y1. After then, the temperature was constantly below freezing that could have prevented the accumulated ice from melting. Thus, a possible modification in the algorithm may be introduction of a bidirectional function for level transitions (among Y1, Y2 and Y3) rather than having absolute function. The result of the above analysis is tabulated below::
Table 30: Overall Performance of Dashboard on Past Icing Events Icing event Ice accretion response Ice shedding response by Comments declared by algorithm algorithm 12 Dec 2007 Y3 on Dec 9 2007, 22:00 R3 on Dec 11 20007, 00:30 Event successfully caught 28 Mar 2008 Y3 on Mar 28 2008, 02:00 R3 on Mar 28 2008, 21:30 Event successfully caught 17 Dec 2008 None None Unsuccessful 3 Jan 2009 Y1 on Dec 31 2008, 01:00 R3 on Jan 2 2009, 02:00 Partial successful
124 Section 5.7: Conclusions The automated ice detection and monitoring dashboard for the VGCS Bridge was developed, implemented, successfully tested, and now being used by researchers and ODOT officials. This Section deals with the performance of the new system and few suggested recommendations that were learnt during the entire project. To start with here listed the primary objectives that are implemented in the dashboard:
Add weather data to existing VGCS web interface Add new stay- mounted camera views to existing VGCS weather interface Develop algorithm to monitor ice event Develop user friendly check engine lights to monitor ice on the Bridge for: o Ice onset o Ice shedding Develop reporting function for ODOT to verify the alerts and declare an event Develop export function for historical data archived Run calibration studies based on historical data Once the system is implemented, it was tested to check the fidelity of the system. Testing also helps locating certain areas where the performance could be improved. In Kathleen Jones report, the main reason behind ice accretion is the freezing rain. As it can be seen in the previous Section, freezing rain was the main reason behind 24th Feb 2011 ice event. The basis of this system is the smart mix of the automated algorithm and the visual observations, which helped aid in training the system for more optimal performance. The system uses an intelligent decision making process based upon initial criteria from past weather data analysis with parameter adjustments made after visual observations. Dashboard has done pretty well in detecting ice accumulation each time, but the analysis done on the algorithm results and University of Toledo’s response from the visual observations concludes that additional information would be needed to increase the ice fall detection speed and accuracy. This would also increase performance of ice accumulation detection and improve the overall system’s accuracy and reliability.
125 Chapter 6: New Local Weather Sensor Testing Section 6.1: Introduction The initial dashboard processed data from an existing array of sensors in the RWIS and at the local airports. The bridge has its own microclimate and this array did not capture some of the basic information about the stays or conditions on the bridge. Therefore, a local weather station with sensors tailored to monitoring icing events was added on the bridge. Stay temperature monitoring commenced in the winter of 2012-2013 and the entire suite of instrumentation on the tower was available in the winter of 2013-2014. In the winters of 2010-2011 and 2011-2012, the existing monitoring system has been able to capture icing events only to a satisfactory level of timeliness and reliability. As the Veteran’s Glass City Skyway has unique stays with stainless steel sheaths and local weather that contributes to the icing problem, the need arose for establishing a more accurate microclimate station. It is not easy to monitor the presence of ice and its thickness on the sheaths without appropriate ice sensors, thus, the most basic step towards improving the existing dashboard algorithm was the inclusion of new sensors on the bridge and their data into the algorithm. “Ice accumulation on the bridge stay sheaths is a slow process and is difficult to infer if there are not accurate sensor measurements of it available on the bridge. The ability to measure precipitation rate, the temperature on the sheath, and solar radiation on the sheath or the bridge, should improve the speed and performance in making critical decisions concerning the safety of the traveling public.” (Kumpf et al. 2011). This chapter describers the new sensors and their laboratory testing.
Section 6.1.1: Geokon Thermistors Freezing precipitation, defined by the American Meteorological Society's Glossary of Meteorology as freezing rain, freezing drizzle, and freezing fog (Glickman, 2000), can have a devastating effect on many industries, including transportation, energy, agriculture, and commerce. After examining these data, some reports of freezing precipitation and ice pellets with a surface temperature greater than 0°C were found. Surface temperatures associated with each observation of freezing precipitation were examined by plotting a cumulative frequency distribution of the temperature data (not shown). Almost none (<0.1%) of the freezing rain and freezing drizzle reports occurred at temperatures exceeding 4°C, while ~1 % of ice pellet reports did. Therefore, it was decided that only those with a temperature ≤4°C would be retained. We decided that we would also like to measure the surface temperature of the stainless steel sheath directly in order to better characterize the environment during ice accumulation. In addition, it was proposed that for better ice shedding detection, a temperature sensor can measure the exact temperature beneath ice. See Section 5.6.2. The structural monitoring system already deployed on VGCS during its construction included many temperature sensors which had a proven performance and reliability for many years. Geokon, the manufacturer of the vibrating wire sensors embedded within
126 the concrete segments of the bridge, always includes a thermistor adjacent to the sensor which is integral to the electronics of the cabling assembly. Thermistors have a known standardized, but nonlinear change in resistance with temperature and the data logger can readily measure this change by applying a known current and measuring the voltage on the cable. We decided to deploy their more precise version of this thermistor directly upon the stays themselves, so as to have a complete picture of the thermal environment of the structure. Section 6.1.2: Dielectric Wetness Sensor For subtle changes in precipitation, dielectric wetness sensors have been effective. Recently, a few tests were made on the LWS-L Leaf Wetness Sensor. These sensors are based on dielectric; as water has higher dielectric than ice, it can be used not only for wetness check, but potentially to determine type of precipitation. Research Team member, Andy Reehorst at Icing Branch, NASA Glenn Research Center ran the Decagon Leaf Wetness Sensor on his building’s roof for a month. He observed some cases where the temperature crossed the 0C line and the resulting water phase change has an impact on the leaf wetness sensor’s measurements. In the figure below, his comparison of the leaf wetness sensor output (Volts, scale on left), the air temp (National Weather Station (NWS) hourly measurements and a probe at the LWS height (about 3 feet above roof), temp scale on right side), humidity (from NWS and his roof sensor, scale on left), NWS precipitation (on right scale), and a binary (precipitation/no precipitation) Kemo rain sensor can be seen. “To me, this data points out the need for having temp measurements co-located with the leaf wetness sensor. But the LWS does seem quite sensitive to precipitation (particularly above freezing) and does a good job in showing persistent moisture.” (Andy Reehorst, 2012) Section 6.1.3: Solar Radiation or Sunshine Sensor As elaborated before, solar radiation has been a primary cause for ice shedding. Different types of solar radiation sensors have been used over the years by climatologists. One common approach has been to have two sensors, one measuring radiation from the whole sky -global irradiance, the other measuring the whole sky apart from the sun -diffuse irradiance. Another method uses an array of pyranometers, with different fixed orientations, hence different views of the sun and sky. The known position of the sun combined with the sensor orientation is used to solve for values of global and diffuse from the differing sensor outputs. Another well-established meteorological parameter is sunshine duration, measured using the Campbell-Stokes recorder. This uses a glass sphere to focus the direct solar beam onto a recording chart, causing a burn, which indicates the duration of bright sunshine. When Delta-T Devices, UK brought the Sunshine Sensor BF3 to the market, a BF3 sensor was installed on the roof of a six-story building in the Merchiston Campus of Napier University, Edinburgh from February 22–July 3, 2001 to evaluate the
127 performance of this new device. Horizontal global and diffuse irradiance data were collected from the BF3. (Wood et al. 2003) According to their observations, the BF3 provides a reliable straightforward measurement of global and diffuse irradiation, without needing polar alignment or regular adjustment. It also provides a measure of sunshine hours that is within the WMO accuracy requirements, and is significantly more accurate than the Campbell- Stokes recorder. Section 6.1.4: Rain Tipping Bucket Tipping buckets are one of the most commonly used devices for precipitation measurement. Several studies were done regarding the performance of rain tipping buckets. The most basic argument against their accuracy was they do not measure rain rate; they provide only a rough estimate of the quantity of water accumulated in one minute by counting the number of tips. A new algorithm was proposed to extract rain rate from data gathered with modern rain gauges (D'Amico et al. 2013), and test its performance against an extensive database collected over a period of eight years. This study was used as a backdrop to model our rainfall amounts to rates. The accuracy of rain bucket is worse at the high rain rates. Several studies recommended a dynamic calibration to account for the nonlinear behavior of the gauge, especially at the high rain rates. The water from the rainfall that falls into the funnel cone does not fall straight to the cone's outlet directly but water flows like a cyclone around the cone, especially at high rain rate. From the top view of funnel, when the cyclone water rotates itself, follow the circumference of a cone and flow through the outlet (Lelomphaisarl, 2012). A small obstruction was implemented. According to his research a new tipping bucket with a completely modified obstacle sheet gives better accuracy than the original obstacle sheet with only small modification.
Section 6.1.5: Goodrich Ice Detector There has been rigorous research done on the various icing sensors and their utility and accuracy for weather monitoring. An aviation routine/special weather report (METAR/ SPECI) remark was developed (Ryerson and Ramsey, 2006) that would report quantitative ice thickness at over 650 locations during ice storms using new algorithms developed for the Automated Surface Observing System (ASOS). These ASOS sites have received the Goodrich Sensor Systems (formerly Rosemount) 872C3 icing sensor, providing the system with the ability to report freezing rain, but with no capability to provide quantitative reports of ice accretion. The ASOS is currently programmed to report icing events only when they are associated with freezing rain. However, the ASOS icing sensor is also known to detect ice accretion from freezing drizzle, wind- driven mist that freezes on elevated objects, freezing fog, and hoarfrost. Here Ryerson and Ramsey studied one serious problem about freezing rain that is how to measure the amount accumulated. Glaze ice accretions vary significantly not only over short geographic distances, but also with the shape and orientation of structures on which the ice gathers, the thermal properties of those structures, and small-scale
128 local variations in wind speed and direction. Overall, it is difficult to measure ice amount, even on structures as simple as tree limbs or wires. Although Ryerson and Ramsey recognized that ice accumulation varies significantly with location because of spatial variations in meteorological and topographical conditions and the specific thermal characteristics of the accretion surface, they suggested that the ASOS ice-detection system based on Goodrich vibrating probe ice detector will now provide a consistent baseline of ice amount information. Section 6.2: Geokon Thermistor 3800-2-2 Geokon provides the model 3800 thermistors which are basically designed to measure temperatures in rock, soil and concrete dams. The sensors behave as resistors with high negative temperature coefficient of resistance. The beads are made from a mixture of metal oxide encased in epoxy or glass (Geokon Installation Manual, 2009). The thermistors were customized to be encapsulated in a very small stainless steel housing to maintain uniformity in surface characteristics in terms of deploying them on the stainless steel stay sheath. The cable is forty feet long to be able to run down the length of the tower without splicing. The model 3800-2-2 was chosen for its superior accuracy. Thermistors are semiconductors behaving as resistors with a high negative temperature co-efficient of resistance. The cable effects are not significant due to high change in resistance. They give non-linear output represented by the Steinhart-Hart log equation: T = 1 / [ A + B(ln R) + C(ln R)3 ] – 273.2
Where; T = Temperature in ⁰C, ln R = Natural logarithm of thermal resistance, coefficients A = 1.4051X10-3, B = 2.369X10-4, C = 1.019X10-7. A, B, and C are the Steinhart-Hart coefficients which vary depending on the type and model of thermistor and the temperature range of interest. Steinhart & Hart performed 100 different relationships between resistance and temperature using two to five fitted constants. A multiple regression program was run to test the relationship, and of the few reasonably good fits the above equation was consistently the best. An extensive examination of calibration functions has yielded this function suitable for calibration curves for precision thermistor temperature measurements. This equation is often used to derive precise temperature using a thermistor since it provides a closer approximation to actual temperature than simpler equations, and is useful over the entire working temperature range of the sensor. It is recommended to workers making precision measurements as its properties have been examined for a variety of data and a variety of thermistors. The coefficients used for the equation above is the same used by Geokon Inc. and Canary Systems Multilogger software for temperature measurement using the vibrating wire gages.
129 Section 6.2.1: Laboratory experiment on temperature measurement using Geokon Thermistors Objective: To measure room and freezing temperature using Geokon Thermistor Probes (Model 3800-2-2), estimate their accuracy and precision using different methods. Apparatus: Geokon Thermistors 3800-2-2 (16), Campbell Sciientific CR10X Datalogger (1), Geokon VW DSP Interface (1), Campbell Scientific Relay Multiplexer AM416 (1), Geokon Readout Box GK 404 (1), Standard Thermometer (1), Serial to USB Interface (1), Dell Latitude E6510 Laptop (1), Canary Systems Multilogger Software. An instrument commonly used to measure surface ttemperature on a long term basis is a thermistor. In our experiment we chose thermistor probes maanufactured by Geokon Inc. which are known for their small size, robustness and high degree of stability with a long lifespan. They have a wide operating range of measuring temperature from -50⁰C to 70⁰C. These thermistors are made from metal oxides encased in epoxy, suppliedd inside a stainless steel housing already potted on the end of a cable (Geokon Installation Manual, 2009).
Figure 64: Geokon 3800-2-2 Thermistor Figgure 65: Naked Thermistor Bead (photo credits, John Flynn, Geokon Inc.)
Experiment The experiment was carried out on two tiers. For all of these, the Geokon thermistors were wired to the temperature (higher) channels of a Campbell Scientific AM416 Relay Multiplexer which was then connected to a Geokon Vibrating Wire Digital Signal Processor (VW DSP) and finally recorded (and programmed) using a Campbell CR10X Datalogger. A laptop was used to send program and collect data using the Multilogger software from Canary Systems.
130
Figure 66: Canary Systems Multilogger Software The experiments can be elucidated as follows: Test for accuracy: The first setup was used to test eight thermistors all measured by Multilogger’s built-in settings for Geokon VWGs at room temperature to check for consistency and accuracy. The factory standards specifiy the thermistors to have an accuracy of +/- 0.2 ⁰C. The thermistors were connected to the first eight channels of the multiplexer and read for a few hours. It was seen that all of the thermistors maintain a similar trend and record similar temperature over a small or extended time period. The largest deviations are around 0.4 ⁰C. We could conclude that the factory specification of their accuracy is dependable.
131 Geokon Thermistors Temperature Measurement Trend 25 24.5 (C) 24 23.5 23
Temperature 22.5 22 1 3 5 7 9 11 13 15 17 19 21 23 25 27 29 31 33 35 37 39 41 43 Time (min)
TH1 TH2 TH3 TH4 TH5 TH6 TH7 TH8
Figure 67: Measurement trend of eight thermistors Test for precision at freezing temperature: Our primary objective to use the thermistors is to measure the stay-sheath temperature on bridges during freezing conditions. Thus, our second test was conducted to observe the precision of the temperature measured by the thermistor probes as recorded by a.) CR10X datalogger , b). Geokon GK404 hand-held meter directly, versus that c). measured by a standard thermometer. This test required four thermistors which were taped together and immersed in a cup of tap water and then put in the freezer and kept overnight. It was observed that the temperatures measured by all the four thermistors were in agreement and that they accurately measured temperature even at sub-zero temperature. All the thermistors were chosen individually and read by the hand-held GK 404 readout box. The temperature of the frozen water was also measured by the analog thermometer immersed in it. It was seen that the readings registered by the data logger were identical to those read by the hand-held meter and quite similar to those recorded by the standard thermometer. However, the water did not freeze at the freezing point in a few instances when inside the freezer.
132
Figure 68: Thermistors kept in freezer Figure 69: Thermistors immersed in water left to freeze
Figure 70: Readings simultaneously noted Figure 71: Stanndard thermometer by handheld GK 404 immersed in setup to record temperature Table 31: Comparison of readings taken by all 3 methods Measurement Type Time Temperature (min) (⁰C) Geokon Thermistor (read by Data logger) 0 8.5 5 7.8 10 7.2 Geokon Thermistors (read by hand-held boxx) 0 8.4 5 7.6 10 7.1 Standard Thermometere 0 8.7 (47.5 ⁰F) 5 8.1 (46.5 ⁰F) 10 7.3 (45 ⁰F)
133 Note: The readings noted by standard analog thermometer are subject to resolution issues due to error of the eye.
Thermistors put inside freezer
Freezer opened (thermometer test)
Thermmistors taken out of freezer
Figure 72: Thermistor Characteristics at Freezing Conclusion: The Geokon thermistors chosen for measuring stay-sheath surface and stay-sheath & ice interface temperature seem to be a reliable cchoice for surface temperature measurement over a broad temperature range. They follow tthe manufacturer’s specifications with respect to accuracy and seem to be a convenient choice in terms of cost, ease of mounting and robustness. They seem to be fairly precise in recording the surface temperature at freezing temperature as well.
134 Section 6.2.2: Installation of Geokon Thermistors 3800-2-2 at the VGCS on Stays 8 & 20
Figure 73: Side View of Gage Locations at VGCS
MUX locations at 2704, 2741, 2806 and 2828.
New thermistor locations on Stay 20 (Span7) and Stay 8 (Span 28).
Mounting: There are four existing multiplexer locattions (White NEMA boxes) at Span 27 Segment 04 (2704), Span 27 Segment 41 (2741), Span 28 Segment 06 (2806) and Span 28 Segment 28 (2828). On March 6th 2012, a team from University of Cincinnati Infrastructure Institute reached Toledo to mount six Geokon Thermistors on Stay 20 and six on Stay 8. Customized mounts were made by the team at University of Toledo to appropriately fit the sheath facing and outer facing thermistors in a block fabricated to hold well on the sheath surface.
135
Figure 74: Custon Thermistor Mount Fabricated for Installing on Stay Surface The thermistors in the mount assembly were set in different positions on the stays and a metal plate band clamp was used to hold the mountts in place.
Figure 75: Thermistor Placed on East Side of Figure 76: Thermistors Placed on Upper Side Stay of Stay
136
Figure 77: Far View of Thermistor IInstallation of Stay
Figure 78: Thermistor Cables Being Routed to Multiplexer Inside White Box
137 The six thermistors were mounted all around the circumference of the stay sheath in clock arms of 12 (upper), 3 (east), 6 (lower) and 9 (west). While there is one thermistor each on the west and lower sides touching the stay sheath surface, there are two thermistors placed on each of the Upper and East side one touching the stay surface and the other facing Outward. The following table describes the gage nomenclature: Table 32: New Stay Thermistors List Gage Name MUX/Channel
7X20TUO MUX5/Ch1
7X20TUS MUX5/Ch2
7X20TWS MUX5/Ch3
7X20TLS MUX5/Ch4
7X20TEO MUX5/Ch5
7X20TES MUX5/Ch6
8X08TUO MUX6/Ch1
8X08TUS MUX6/Ch2
8X08TWS MUX6/Ch3
8X08TLS MUX6/Ch4
8X08TEO MUX6/Ch5
8X08TES MUX6/Ch6
Figure 79: Stay Sheath Cross Section Showing Thermistor Positions Each of these set of six thermistors were connected to a new multiplexer that was installed inside the existing white NEMA enclosures. Thermistors on Stay 20 were wired
138 to a new 416 Relay Multiplexer which was screwed to the backplane inside the NEMA box at location 2741. Similarly thermistors on Stay 8 were wired to a new AM16/32B Multiplexer which was screwed to the backplane inside the NEMA box at location 2828. The figure gives the wiring diagram used for this upgrade. Initial Observations: Temperature data collection to MySQL database from the twelve stay thermistors at Veteran’s Glass City Skyway using Campbell Scientific’s Loggernnet software started on March 6th 2012 at 1:00 PM. This data was compared to temperature data collected from the vibrating wire gage embeddded segment thermistors, panel temperature of Campbell Scientific datalogger CR100X inside the cabinet, air temperature data from local Road Weather Informaation System (RWIS) station (142-I- 280) at VGCS and Meteorological Terminal Aviation Routine (METAR) data from two local airports: Toledo Express Airport (KTOL) and Metcalf Field Airport (KTDZ) for a period of 16 days from, March 6 to March 22, 2012.
Figure 80: Stay 20 Thermistors Temperature Trend
139
Figure 81: Stay 8 Thermistors Temperature Trend
140 Table 33: Sky Cover and Precipitation During the Period March March March March March March March March March March March March March March March March March Date 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22
Sky Cover 0.5 0.1 0.8 0.2 0 0 0.7 0 0 0.3 0.3 0.2 0.4 0.4 0 0 0
Precipitation 0 0 0.22 Trace 0 0 1.32 0.11 0 0.81 Trace 0 Trace 0 0 0 0 (inches)
141 After inspecting the temperature trend during the 16 day period it was observed that all the sources measured temperature in unison at night. Solar radiation during the daytime would cause differences in the temperature recorded depending upon the location of the thermistors. Some of the general observations gathered were:
Upper stay thermistors record much higher temperature than lower stay thermistors. RWIS & METAR data show similar temperature trends. METAR data seem to have higher maximums and lower minimums. They both lag with respect to temperature of up to 3hrs or 20 F when compared to the stay thermistors. Embedded VWG segment thermistors have significantly lower maximum and higher minimum. In order to gain deeper insight into the temperature characteristics of the individual thermistors we decided to choose few example days and have a magnified view of their trends. March 15 and March 16 seemed to appear to represent more days in the bracket and hence chosen for the purpose. The weather data for those days were obtained from the National Weather Service which included the sunrise, sunset, maximum and minimum temperatures, sky cover and precipitation in inches.
Table 34: Weather Report on March 15 Time Conversion Data Logger Clock RWIS Data Collected Local Time (EDT) (EST) (GMT) GMT + 04:00:00 GMT + 05:00:00 National Weather Service Report Local Information Comparison (March 15) Mean RWIS 2016 Sheath Parameter Time (EST) Value Time (EST) Air Thermistor Temperature Temperature
Sunrise 6:46 AM 5:43 PM 73.9 F 75 F Sunset 6:42 PM
Maximum 5: 43 PM 78 F Temperature 4:29 AM 62.6 F 62.5 F Minimum Temperature 4:29 AM 58 F
Average Sky 0.3 * Sky Cover: The total amount of clouds Cover in fraction where 1.0 is equivalent to a Precipitation 0.81 inches completely overcast sky
142
The magnified view of these example days’ 24 hour characteristics provided the hourly response of each thermistor in detail. These were the general observations:
At night, all sources seemed to converge w.r.t their temmperature recording. It could be observed that in the early morning with sunrise, the East thermistors showed the steepest and most sudden rise in temperature. They were followed by the Upper thermistors more towards the noon. During the afternoon till late evening, the West side thermistors were the warmest.
Figure 82: Characteristics for Stay 20 Thhermistors on March 15
143
Figure 83: Characteristics for Stay 20 theermistors on March 9 & 10 Since the installation of the thermistors, we did not have any icing (ice accumulation or shedding) event on the bridge in that winter. However, there was an occasion where the temperature was low enough and suitable for such an icing event had there been some kind of precipitation simultaneously. We term this event as “possible freeezing event” that occurred between March 9th evening around and March 10th morning. On March 9, 2012 the temperature measured by the stay thermistors fell under 32 F around 7:30 pm and remained under through March 10 night till the morning when the temperature rose above 32 F with sunrise at 6:55 am. As usual, the stay thermistors and the local RWIS station recorded similar temperaature all night. However, after sunrise the East stay thermistors showed significannt rise in temperature compared to other thermistors and crossed melting point (32 F) around 8:15 am. On the contrary, the local RWIS temperature read melting point three hours later at around 11:15 am. There were times where it consistently had a delta of about 20 F with the warmest thermistors. Incidentally, it was also noticed that the thermistors showed such fast and overwhelming rise in temperature (around 11 F in 30 minutes) that the current sampling rate of 30 minutes failed to track critical points. On May 17, 2013 a trip was made again to the Veteran’s Glass City Skyway bridge, where the system was upgraded from CR10X data logger to CR1000, using CRBasic program instead of Canary Multilogger for data collection, and the sampling time was lowered to 10 minutes. This trip will be discussed in details later in Chapter 4.
144 Section 6.3: LWS-L Dielectric Leaf Wetness Sensor Leaf Wetness sensors have been developed to estimate by inference the wetness of nearby leaves. The LWS-L measures the leaf surface wetness by measuring the dielectric constant of the sensor’s upper surface. The sensor is able to detect miniscule amounts of water or ice.
Figure 84: Leaf Wetness Sensor functional diagram Section 6.3.1: Laboratory experiment on measurement of output voltage using LWS-L Leaf Wetness Sensor. Aim: To measure output voltage of the leaf wetness sensor for a fixed excitation and determine appropriate threshold for wetness conditions. Apparatus: LWS-L Dielectric Wetness Sensor (1), Campbell Scientific Datalogger CR1000 (1), Serial to USB Interface (1), PC, Loggernet Software. Operating principle: The LWS-L measures the dielectric constant of a zone approximately 1cm above the upper surface of the sensor. The dielectric constant of water and ice are much higher than air, so the meassured dielectric constant is strongly dependent on the presence of moisture or frost on the sensor surface. The sensor sends a millivolt signal proportional to the dielectric measurement of the zone, which in turn is dependent on the presence of moisture or frost on its surface. Experiment: The leaf wetness sensor was connected to the CR1000 datalogger and provided switched excitation voltage of 2.5mV. The program was written in Campbell's CRBasic Editor where output voltage was set to be recorded with a scan rate of once per minute.
145 The leaf sensor was exposed to different kind of wetness conditions and the output was recorded accordingly. There were basically three experiments: 1. Amount of wetness test, 2. Impurity test and 3. Freezing conditions test.
Figure 85: Experimental setup of data logger CR1000 with LWS-L Leaf Wetness Sensor Wetness Test: Various kinds of experiments were conducted on the leaf sensor to emulate real life conditions of drizzle, light rain, heavy rainfall and freezing rain and the outputs noted accordingly.
Table 35: Wetness Test Time Experiment Output (mV)
11:30 AM Dry leaf surface 262
11:35 AM Light sprinkle 335
11:57 AM Medium sprinkle 515
12:06 PM Heavy sprinkle 597
1:16 PM Light sprinkle 380
2:50 PM Soaked napkin on surface 710
2:57 PM Partially Immersed in water 746
3:21 PM Water level increased 759 Figure 86: Droplets of Water Sprinkled on Leaf
146
Table 36: Impurity Test Time Experiment Output (mV)
11:00 AM Immersed in cup of water 788
11:05 AM Salt added to water 1030
2:05 AM Dry napkin on surface 264
Figure 87: LWS-L partially immersed in cup of water
Table 37: Impurity Test Time Experiment Output (mV)
3:45 PM Dry leaf kept in freezer 262
4:40 PM- Immersed in cup of water 298 - 300 10:30 AM (inside freezer)
Figure 88: LWS-L immersed in cup left to freeze
147
Figure 89: LWS Wetness Test
Figure 90: LWS Freezing Temperature Test
148 Observations: The different experiments yielded a handful of resullts. These can be explained as follows- Dry Leaf Output: Unlike said in manual, it has been verified through Decagon rep and the experiments that the leaf gives an output around 250-260 mV when dry. Wetness Threshold: As a few droplets of water or a lot of frostt catapults an output above 300 mV, it can be safe to consider this threshold. Scan Rate: As the leaf gives output w.r.t the quantity of wetness present on it during sampling, lower the sampling rate, the better. Scan rate of 1 min was used for the experiment. However, the sampling rate to be implemented into the algorithm was set at 10 minutes during field installation. Leaf Angle: The leaf surface is hydrophobic. It holds a bit of weetness for a few minutes while most drains off and the rest evaporates. Keeping it horizzontal helps capture precipitation better unless the scan rate is lowered to a few seconds. Dust & Impurities: With the salt test, it was observed that the presence of impurities on the leaf surface raises the output significantly. It doesn’t however affect much during dry conditions. Section 6.4: Sunshine Sensor BF5 The Sunshine Sensor BF5 is mostly used for meteorological “Global”, “Direct” and “Diffuse” solar radiation and sunshine duration measurements .The BF5 may be installed at any latitude and at any polar angle i.e. relative to North. It consists of an array of seven photodiodes encapsulated in a hemispherical dome with a shaded pattern. The following diagrams are good representatives of its constructiion:
Figure 91: Sunshine Sensor BF5 (side view) and Detailed Construction
149 Section 6.4.1: Laboratory experiment on measurement of solar radiation using Sunshine Sensor BF5. Aim: To measure solar radiation energy using sunshine sensor and determine appropriate threshold for sunshine status conditions. Apparatus: Delta-T Devices Sunshine Sensor BF55(1), Campbell Scientific Datalogger CR1000 (1), Serial to USB Interface (1), PC , Loggernet Software. Experiment: The sunshine sensor was connected to the CR10000 datalogger and provided voltage of 5V from the datalogger power supply. The program was written in Campbell's CRBasic Editor where the total radiation and diffuse content of the radiation were meant to be recorded every 15 mins and the output was calibrated in terms of Energy (Watt/m2). The photos below depict how the sensor waas exposed on an open deck (phhoto credits- Dr. Arthur Helmicki, University of Cincinnati Infrastructure Institute)
Figure 92: Sunshine Sensor BF5 Set Up on a Deck for Unobstructed Exposure to Solar Radiation Operating Principle: Uses array of 7 photodiodes with unique computer generated shading pattern to measure incident solar radiation in a way that: at least one photodiode is fully exposed to solar radiation beam at least one completely shaded both receive equal amount of diffuse sunlight from rest of sky hemisphere.
A microprocessor calculates: i. Total(Global) and ii. Diffuse components of the radiation and iii. determines Sunshine Status
150 World Meteorological Organization (WMO) specified 120W/m2 as the threshold for Sunshine Status. Feature doesn't work with CR1000 datalogger, so it was programmed manually according to manufacturer's formula: Total/Diffuse > 1.25 and Total > 25 Watt/m2. Observations: The direct and hence total radiation has a giant leap rigght after sunrise. It falls during sunset.
Sky cover (presence of passing clouds) causes drop in the solar radiation significantly.
Sunshine Status is a binary 1 when global radiation is above 120W/m2 all day from an hour after sunrise and before sunset.
During sunrise & sunset the direct beam is negligible, so most of the global radiation comprises of diffuse components. These results were compared to the daily weather report from weather.com as shown in the table below and were found to be pretty consistent with each other.
Figure 93: Solar radiation characteristics over an extended period of 16 days
151 NOTE: i. There was shadow provided by canopy to the east and westt of the deck which contributes to the low cut of the plot during morninggs and evenings. ii. Every evening the backyard was lit for dog walking, which attributes to the small peak observed during late afternoons. After the general observation, two typical example days were chosen to get a magnified view of the daily characteristic of the solar senor with respect to different times of the day. One was partly cloudy, another was mostly sunny and clear. June 23, 2012 had scattered clouds and hence was chosen as a typical partly cloudy day. As expected, it was seen the solar radiation to rise quickly soon after sunrise and fall after sunset. Clouds would cause significant drop in solar radiation during the day.
Figure 94: A typical partly cloudy day chosen to see the daily solar radiation characteristics June 26, 2012 was a very bright and sunny day according to weather staation data. So this day was chosen as an example day for solar radiation characteristics for clear conditions. As expected most of the day had very high and consistent sollar radiation, peaking around noon to close to 1000 Watt/m2.
152
Figure 95: A typical clear sunny day taken as example to see the daily solar radiation characteristics
Section 6.5: Met One Rain Tipping Bucket Introduction: The Met One Rain or Heated Snow Gage is a dual chambered tipping bucket. Rain is metered into a collection funnel on top of the bucket. There is a mercury switch inside the bucket which tips for each 0.01 inch precipitation collected. It has adjustable feet to ensure proper mounting with bracket CM240. There is a primary screen on the funnel that prevents debris from clogging. The cable length is forty feet and has Santoprene rubber cable jacket for better resistance to UV degradation and moisture. The rain collected is discharged through the rain gage base. There is a thermostat based heater inside that can melt snow into water before measurement.
153
Figure 96: Rain Tipping Bucket (from top left clockwise) distant view, top view and inside viiew Section 6.5.1: Laboratory experiment on measurement of precipitation using Tipping Bucket Aim: To measure precipitation amount using Met One Rain Tipping Bucket and determine appropriate sampling rates and threshold for conditions. Apparatus: MetOne Rain Tipping Bucket (1), 100 mL Gessler Buret (1), Campbell Scientific Datalogger CR1000 (1), Serial to USB Inteerface (1), PC, Loggernet Software. Operating Principle: Collects rainfall in the 12 inch collection funnel and meters the rain into tipping bucket. Snowfall captured in collection funnel aand melted by thermostatically controlled heating element. For every 0.01 inches of rain/snow-water eq., the tipping bucket assembly tips due to gravity and activates a mercury reed switch. A momentary contact closure for each increment of rainfall is recorded by datalogger pulse channel. Water drains out base of the gage after tipping.
The tipping quantity is determined by the equivalent volumme of liquid collected in a given surface areea. Met One Conversion Factor: 3 2 Tip to in : Catch orifice area (πd /4)) X increment in inches 3 = 113.04 X 0.01 in = 18.52 ml
154
Figure 97: Rain Bucket lab Figure 98: Gessler Buret
Experimental Setup: The rain tipping gage was connected to the CR1000 datalogger and provided excitation signal of 5V from the data logger power supply. The program was written in Campbell's CRBasic Editor where the pulse (signal return) was counted every second to record every time the bucket tips without missed detection. The data was totalized every i. 5 minutes and ii. 30 minutes to determine the amount of precipitation during the period. Table 38: Rain Bucket Lab Experiment 1 with 5 Minute Sampling Rate Record Volume of Water Poured (mL)* Precipitation Record (inches) 1 0 0 2 45 0.02 3 45 0.02 4 20 0.01 5 0 0 6 0 0 7 0 0 8 0 0 9 80 0.04 10 60 0.03 Full Data Shown in Graph.
155
Table 39: Rain Bucket Lab Experiment 2 with 30 Minute Sampling Rate Record Volume of Water Poured (mL)* Precipitation Record (inches) 1 40 0.02 2 200 0.1 3 300 0.16
It has been observed that the device follows the factory calibration standards where each tip = 18.53 ml of water. During the first few readings the quantity of liquid let in is marginally more than that tipped out, because of adhesion wheere some water sticks to the funnel. The result obtained is fairly consistent irrrespective of fast/continuous or slow/intermittent passage of water into the bucket using the buret. One tip can be considered a threshold check for precipitation in the algorithm as each tip equals to 18.53 ml of water.
Figure 99: Rain Bucket accuracy experiment (actual vs ttipping volume) Section 6.6: Goodrich Ice Detector The 0872F1 Goodrich Ice Detector detects ice accumulation on an ultrassonic axially vibrating tube and communicates the associated frequency chhanges through an RS-232
156 or digital current loop data link. The 0872F1 is mounnted on a pole and is designed to operate continuously in an outdoor environment. The 0872F1 consists of four functional assemblies: a Main circuit card assembly (CCA), an Output Interface CCA, a Filter assembly, and a Strut and Probe assembly.
Figure 100: The Goodrich Ice Detector (external and function diagrams) Section 6.6.1: Laboratory experiment on measurement of ice presence/thickness using Goodrich Ice Detector 0872F1 Aim: To measure ice thickness using Goodrich Ice Detector and determine appropriate sampling rates and de-icing threshold. Apparatus: 0872F1 Ice Detector (1), MicroCare Anti-Stat Spray (1), Sliding Calipers (1), Campbell Scientific Datalogger CR1000 (1), Serial to USB Interface (1), PC, Loggernet Software. Operation Principle: The Goodrich Ice Detector 0872F1 measures precipitation transitions between solid and liquid states. It consists of a sensing element that is exposed to the environment. Ice builds up on this element when it is exposed to icing conditions causing change in its mass. This mass change causes a shift in vibration frequency in an ultrasonic axially vibrating tube. The associated frequency changes are communicated through data link. Ice Thickness is calculated as (manufacturer-specified linear relationship): Ice Thickness = -0.00015*Frequency + 6 (in inches)
157 Experimental Setup: The ice detector was connected to the CR1000 datalogger and provided signal of 115V from the mains power supply. The program was written in Campbell's CRBasic Editor where the frequency of the probe was counted every minute. The data was recorded every 1 minute to determine the corresponding amount of ice thickness.
158
Figure 101: Ice Detector Mounted for Figure 102: Microcare Anti-Stat Freezing Spray Experiment
Figure 103: Probe before Spraying Figure 104: Probe After Spraying
Observation: The device follows the factory calibration based on the specified linear relationship between change in frequency and corresponding cchange in ice thickness. The threshold to trigger de-icing cycle was set at 0.004 inches and it proved to be able to
159 de-ice the probe completely during the heating time. The de-icing cycle lasts longer (as the probe remains warm) than the heat-time and keeps the probe frost-free for a couple minutes. The sampling rate chosen was 1 minute and it proved to be quite effective in determining presence of ice. An experiment was run using sliding caliperrs to measure ice thickness. This was done in order to manually determine the relationship between thickness of ice forming on probe and that measured by sensor, but it wasn't effective due to quick ice melting of thin layer of ice.
Figure 105: Frequency/Ice thickness characteristics of 0872F1 durinng freezing spray experiment
Table 40: Caliper Test Probe Thickness Delta for Ice Sensor Data for Caliper Test (inches) Thickness (inches) Ice (inches) Zero Reading (no ice) 0 0 = 0.2425 0.2500 0.075 0.0187 0.2550 0.125 0.0254 0.2580 0.155 0.0305 0.2650 0.225 0.0297
The thickness (diameter) of the probe was Figure 106: Thickness measured using sliding calipers with and measurement using calipers ithout the i w ce coating. The results obtained were erratic as: The ice melted too fast at room temperature and the caliperrs couldn't fit onto the ice layer without shaving it.
160 Initial results: The chief motivation behind lab experimentation of the different sensors was to find a meaningful conversion of the parameters recorded by the different sensors into comprehensive values, and also to test and calibrate them, if necessary. The following table is a summary report of the behavior of the new icing sensors.
Table 41: Icing Sensors Initial Observations Sensors Observations Stay Thermistors All stay thermistors measure temperature in unison at night. Solar radiation (daytime) causes sufficient spread in temperature recorded depending upon thermistor location. RWIS air temperature can have a delta of about 20 F during sunrise. Ice Detector The vibration frequency decreases with increased mass of ice on sensor probe. Correctly determines ice thickness as verified with calipers. Solar Sensor The direct, and hence, total radiation has a giant leap right after sunrise. It falls dramatically during sunset. Sky cover (presence of passing clouds) causes drops in solar radiation significantly. Sunshine Status is usually above 120 W/m2 (WMO threshold) all day from an hour after sunrise and before sunset. Rain Bucket The device follows the factory calibration standards where each tip = 18.53 mL of water or 0.01 inches of rainfall. Leaf Wetness The leaf gives an output around 260-270 mV when dry. It gives an output of ~ 300 mV when it is frosted and higher when there is rainfall or snow.
Section 6.7: Conclusions Laboratory experiments have documented and verified the proper operation and calibration of a new sensor suite to be deployed on the VGCS bridge. These include:
Geokon Thermistor 3800-2-2, for stay surface temperatures
LWS-L Dielectric Leaf Wetness Sensor, for detecting water/moisture with some classification as ice, light rain/snow, or heavy snow/rain.
Sunshine Sensor BF5, for detecting and quantifying solar radiation and approximate level of cloud cover
Met One Rain Tipping Bucket, for detecting and quantifying precipitation
Goodrich Ice Detector, for detecting and quantifying ice thickness
161 Chapter 7: Field Study of Temperature Effect on Stay Sheaths Section 7.1: Introduction An icing experiment station was set up at the University of Toledo’s Scott Park Campus in order to better understand the nature of icing events, conduct icing experiments regardless of the natural precipitation and minimize the risk to personnel. Three full scale sheath specimens were used for experiments in the winter of 2012-2013. The initial experiments demonstrated that icing and shedding similar to that obbserved on the VGCS could be replicated on the stay specimens. After the experiment station was operational, the VGCS experiments carried out included sensor studies, coating tests, deicing fluid tests and heating tests. When the weather was warm, experiments were carried out inn the University of Toledo icing wind tunnel. Experiments performed in the icing wind tunnel include coating and anti/deicing fluid. Section 7.2: Design of Icing Experiment Station Scott Park Setup and Background For designing an effective long term anti-icing/deicing technology for VGCS, more information needed to be gathered. The information required includes the following: data to make an accurate thermal model of cable stays, determining the thermal properties of the sheath, information about efficacy of chemicals on the stays, and information about coating and internal heating. Performing these experiments on the bridge is inefficient and dangerous. An icing experiment station was designed, built and initially operated during the winter of 2012-2013. Three full scale sheathing specimens 10ft long with the same diameter, material, and reflective brushed surface as the VGCS stays have been set up on the outdoor concrete pad at University of Toledo Scott Park campus. Figure 107 shows a screen shot of the concrete pad at Scott Park which is taken from Google Earth.
Scott Parkk’s Concrete Pad
Figure 107: Google Earth Screenshot of Scott Park
162
Figure 108: Experimental Setup On the bridge the stays have 82 to 156 prestressing epoxy coated structural strands depending on the location of stays. In the test set up, 120 un-ttensioned strands were used. As shown in Figure 79, on the bridge at mid-span, the epoxy coateed strands are at top of the cross-section. In the test configuration,, the strandds rest at the bottom of the specimen due to safety reasons. As shown in Figure 108, the experimental setup contains north facing, south facing, and horizontal specimens oriented north-south to simulate the bridge conditions as realistically as possible. Data Acquisition System and Sensor Setup The specimens were instrumented with designed thermistors to get the temperature at the sheath surface, which consist of an array of flat thermocouples installed both outside and inside the sheath in order to get surface temperature of specimen. Additionally, photo sensors to track sun location, a pprobe thermocouple for getting the inside temperature of the specimen at different locations, an array of ice detectors to determine the difference between liquid water and ice, and a local weather station to collect the ambient temperature, wind speed, wind direction, and amount of rain and snow were used throughout the outdoor experimentts. Figure 109 shows several sensors which were installed on the upper part and east side of the specimen to monitor icing behavior of specimen.
163 Leaf Wetness Sensor Goodrich Ice Detectoor
Photo Sensor
Figure 109: Sensors on South-faced Specimen For collecting data, a wireless sensor monitoring system with three nodes and one base station was used, as seen in Figure 110.
Figure 110: Data Acquisition System Section 7.3: Design of the UT Icing Tunnel and Design UT Icing Tunnel Background To test superhydrophobic coatings efficiently, an icing tunnel was designed to produce icing scenarios similar to a natural environment. The icing tunnel was designed and built by University of Toledo master student, David Whitacre. There are two main parts of the icing tunnel; which include a freezing room and a tunnel system, as seen in Figure 111 (Whitacre, 2013) and 112. The square 2.4 meter freezing room was built to support the cooling unit which was installed on the roof, also seen in Figures 111 and 112. All walls of the room were insulated to reduce temperature gains; considering that ice testing was performed in this system, gains or increases inn temperature from the outside air,
164 which is significantly warmer than the circulating air within the icing tunnel system, are not desired. The cooling unit has the capability to maintain a constant temperature as low as -20 °C. The freeze room has an outlet and inlet to allow the supercool air to travel through the tunnel system as a closed loop system. A multi-speed fan was installed inside the tunnel system for the user to be able to adjjust wind speed during experiments. Additionally, temperature was adjustable, thus, allowing various conditions suitable for icing to be achieved (i.e. high winds and very cold temperature (-20 °C), light winds and very cold temperatures (-20 °C), high winds and cold temperature (-10 °C), etc.).The tunnel system also included a test section, which is the area where an object may be placed in order for experiments to be performed.
Cooling Unit Test Section
Tunnel System
Freezing Room
Figure 111: SolidWorks Design for the UT Icing Tunnel
Cooling Unit Test Section
Tunnel System
Freezing Room
Figure 112: UT Icing Tunnel
165 Test Section This section consists of a 30 cm diameter optically clear tube as shown in Figure 113. This test section allows users to take pictures of a specimen during an experiment. It also includes a misting system, camera, and specimen mounting systems, which will be presented below.
Figure 113: Testing Section of the UT Icing Tunnel Misting System The icing tunnel can create both rime and glaze types of ice by the use of a misting system. This system uses nozzles to produce a fine mist, very small water droplets that get blown onto the testing specimen at variable speeds. In order for the mist to reach the specimen at a proper distance to allow the water droplets to freeze in a manner that resembles field conditions, the misting system was installed inn front of the testing section roughly X , as seen in figure 114. The setup consists of three misting nozzles that are hooked up to standard city water, which is adequate because it operates in low pressure. . The misting array seems to cover the tesst section diameter adequately. As previously mentioned, the icing tunnel is designed to decrease in temperature to as low as -20 °C. This allows the testing object and water droplets to be supercooled, and thus, ice accretion is allowed to occur when the droplets ccontact the surface of the testing specimen. The misting system allows the user to change the nozzle size to either a 0.012, 0.014, or 0.016 inch orifice. Having different nozzle sizes allows the user to perform experiments with different droplet sizes and flow rates. The flow rate associated to each nozzle is between 0.007 and 0.011 gallons per minute (gpm). (Whitacre 2013) The nozzles were tested for droplet size using Partiicle Image Velocimetry (PIV) method. The result shows that 0.012, 0.014, and 0.016 inch orifice nozzle distributes droplet size of approximately 50, 40, and 42 micrometers (micron), respectively.
166 Camera
Misting System
Figure 114: Misting System in the Testing Section Camera System A Panasonic HX-A100D camera was installed within the testing section, shown in Figure 114, to allow the user to clearly document each experiment. The camera lens maintains roughly the same temperature as inside the testing section, which ultimately eliminates any difficulty in taking pictures due to the possible formation of condensation on the testing section tube from the temperature difference. The camera seen in Figure 115 has a high definition 1080P video recorder and is waterproof up to 5 feet of watter. The camera does not have a video screen, however, the Panasonic Image application can be installed on a smartphone or tablet. This application allows for a real time video to be streamed to the device it is installed on via Wii-Fi. The application also allows the user to control recording, take pictures, and change the settings of the camera. (Panasonic 2013)
Figure 115: Panasonic HX_A100D Camera (Panasonic 2013) Mounting System Additionally, located in the test section is a mounting system. This system allows a testing specimen to be mounted across the testing ssection as seen in Figure 116. The mounting system also includes a load cell which is used to detect lift and drag forces as well as a dial to change the angle of the object during the experiment. However, the
167 experiments performed for this project do not concern the lift and drag forces and therefore, the data acquisition system will be neglected for the experiments.
Figure 116: Mounting System of Testing Section Section 7.4: Icing Accretion and shedding Experiments att Scott Park Formation of ice is a long process typically takes at least 8 to 10 hours. The type of accumulated ice highly depends on weather conditions. On the other hand, ice shedding would be a fast process. In the last icing event, 90 peercent of ice shed in 45 minutes (Nims, 2010). The goal of laboratory icing experiments was to simulate icing occurrences on the VGCS and to tracking the icing behavior of the sheathing. The experiments were conducted outdoors in cold weather. Icing was simulated by spraying a mist of water with a temperature of approximately 34° F at constant intervals.
Figure 117: Spraying a Mist of Water on North-faced Specimen
168 The water was applied slowly enough that the water and latent heat of transformation did not raise the specimen temperature above freezing. Figure 118 shows a ½ inch thick accumulation of ice on the north facing specimen.
Figure 118: Pattern of Ice Accumulation on Outdoor Tests Simulating shedding was trickier than that of formation. The target was to simulate ice falling in a fashion similar to shedding that occurs on the bridge. Based on the past reports of (Belknap, 2011), (Jones, 2010), and (Nims, 2010), ice shedding is more likely to happen when air temperature is above freezing and with a clear sky. A clear sky allows solar radiation to warm up the specimens. The experiments have been done to prove falling criteria and to create a baseline of speecimens’ behavior before, during, and after shedding. Solar radiation makes a dramatic difference between speecimen and ambient temperatures and causes liquid water to form under the ice on the specimen.
Presence of liquid water before shedding
Figure 119: Water beneath the Ice Layer before Shedding As the ambient temperature increases above freezing ice falls in large curved chunks due to gravitational loading. Icing experiments have shown that at least 1/4” thickness is required for shedding of an ice layer. This is consistent with an observed icing event in 2011 (Section 3.3.2). Figure 120 shows the steps of shedding that a large chunk of ice
169 on a stay during simulation experiments goes through. A video of this shedding experiment is on the VGCS website.
Figure 120: Ice Shedding Steps Figure 121 shows the temperature monitoring of soouth-faced specimen from the 15th to the 18th of February 2013. As shown in that figure, the specimen accurately simulates the icing behavior as it occurs on the VGCS.
170
Figure 121: Stay’s Behavior in Icing Test – 2/155 to 2/18
Figure 122: Stay’s Behavior in Icing Test – 2/20 to 2/22 Figure 122 monitors the second simulation test which was done during the 20th to the 21st of February 2013. Part A of this figure focuses on the ice accumulation scenario. An ice layer started to accumulate on the specimenn at 21:10. Increasing the ice thickness was difficult due to the latent heat which was caused by additional water freezing. Water spray was controlled during the accumulation process to prevent an ice
171 layer from washing away. The sudden increase and decrease of temperature in part A illustrates the change in specimen temperature when water is applied. Part B of the graph illustrates ice persistence behavior of the specimen. The outdoor weather temperature was 21° F during the night and stayed below freezing throughout the day. Part C shows the ice shedding process. The temperature of both the top and the east surface of the specimen increased dramatically at sunrise. The sun was out and solar radiation caused the presence of liquid water between ice layer and specimen. Ice shedding happened at 13:25. Part D again shows ice accumulation scenario of the specimen. There is a coincidence between part A and D. This demonstrates the consistency of the icing behavior of the specimen. Monitoring and comparing the data collected during several icing simulation tests show that it is possible to reproduce ice accretion and shedding in the same manner as it occurs on the VGCS at a high confidence level. During February and March 2013, approximately 12 icing events were simulated while no natural icing events occurred (Appendix A). Section 7.5: Thermal Experiments at Scott Park Because of the large empty area in the stays, one of the most interesting applicable technologies for the VGCS is internal heating. Designing of heating system requires an accurate thermal model of stainless steel stays. There is simply no way to establish an accurate thermal model of VGCS’s stays without experimentation. The primary uncertainty in the development of the thermal model is the convective heat flow inside a stay section. To examine the basic feasibility of heating, preliminary thermal experiments were carried out on the specimen with the same thermal mass as the VGCS stays. For the thermal experiments, the pipe was instrumented with flat probe thermocouples to collect temperature in points of interest. Flat thermocouples were installed on beginning, end, and midsection of the pipe’s surface. Three tiny holes were made in the specimen for collecting the inside temperature of the pipe at the specified locations. An anemometer was used for collecting the temperature and air speed of the pipe’s inlet and outlet. Ice detectors were used to distinguish liquid water from ice. A 70,000 BTU forced air space heater was used as a heat source.
172 Sensor Locations Heater
Figure 123: Thermal Experiment Setup On the bridge at mid-span, the epoxy coated strands are at the top of the cross- section. In the test configuration, the strands are at the bottom of the specimen. This is thermally acceptable to have the strands at the bottom of the specimen considering conduction through the sheath is more significant than convection inside the sheath. The goal was to understand pipe’s response in anti-icing and the ice accumulation pattern when heat flows inside the pipe. Thermal tests were conducted to understand the behavior of stainless steel sheath in both deicing and anti-icing. The speciimen showed a promising performance in the deicing strategy rather than anti-icing strategy.
Figure 124: Strands Configuration in Thermal Tests The goal of the first thermal experiments was to gain a better understanding of the general thermal response for the sheath during deicing, how heat distributes through the pipe, and the ice melting pattern. The initial test included accumulating ½ inch thick layer of ice on the specimen and then blow hot air innside the pipe until all the ice melted.
173 Figure 125 shows the melting pattern in thermal deicing test. The heating system successfully melted the accumulated ice without shedding.
Figure 125: Deicing Pattern in Thermal Test The second test was an anti-icing thermal experiment. In this particular test, the clear specimen with no ice was heated just above freezing and then a mist of water was sprayed onto the specimen until the formation of ice occurred. Detailed interpretation of the data is presented by Knot (Likitkumchorn, 2014).
Figure 126: Accumulated Ice in Anti-icing Thermal Test These experiments established the basic feasibility of using heat as an active anti- icing/deicing strategy. Future experiments should haave better dducting to control inlet and outlet air flow. It is also suggested that the epoxy coated strands should be installed with the same configuration as VGCS to have the similar thermal effects.
174 Section 7.6: Anti/de-icing Fluid Experiments at Scott Park Chemicals were considered as a second anti-icing/deicing strategy for the VGCS. There are concerns about the efficacy and the effect on the stay appearance from the chemicals. Anti-icing/deicing experiments to determine the effiicacy and behavior of chemicals were conducted. Beet Heat is an organic based material, which is made of refined molasses carbohydrate, NaCl, CaCl2, KCl, and MgCl2. The efficacy of the chemical for deicing of pavements was proved by ODOT (Trademarkia, 2011). In the first test, the Beet Heat concentrate was appllied with a manual sprayer on half of the specimen and a mist of water was sprayed onto the specimento see the efficacy of that anti-icing strategy. Figure 127 shows how ice accumulated on the specimen in the presence of the chemical.
Figure 127: Formation of Ice in Chemical Anti-iccing Test As shown in Figure 128, the chemical first melts lower layers of ice to water, but since the water does not roll and/or blow off the sheath, accumulated water suddenly turns to ice again. In the deicing chemical test, a 1/8 inch thick layer of ice on the specimen was accumulated, then a drip tube system was used to flow the Beet Heat on the ice layyer. As shown in Figure 128, the chemical just melts a narrow rivullet through the ice due to low viscosity.
175
Figure 128: Drip Tube System used in Chemical Deicing Test Beet Heat was used as a chemical in preliminary anti-icing/deicing tests. Beet Heat showed unpromising performance during anti-icing/deicing experiments. It is suggested to use a chemical with design characteristics for having a better performance in deicing tests or add detergent to have more viscosity. Drip ttube system is supposed to mount on the VGCS stays for distributing the chemicals in anti-icing strategy. This system can affect the aerodynamics of the stays, which should also be considered before installation. Section 7.7: Coating Experiments at Scott Park Coatings are supposed to reduce the adhesion strength of ice to the stay surface and are often considered as an anti-icing or passive technology. To investigate the efficacy of a coating for the icing problem of VGCS, a revolutionary type of coating was tested. The coating is considered as a surface treatment that can make surfaces extremely water repellant and ice phobic. The specimen was instrumented by thermocouples in order to the see the thermal behavior of stainless steel sheath with the presence of a coating. Hydrobead was sprayed to one side of the specimen, and then a mist of water was sprayed onto the specimen to see the efficacy of the super hydrophoobic coatingg (Hydrobead, 2013).
176
Figure 129: Hydrobead Sprayed on Half of the Specimen Hydrobead caused water to bead into small dropletss. Due to brushed surface of sheath, small water droplets did not roll and/or blow off the ccoated surface, but rather suddenly turned to ice.
Figure 130: Water Droplets due to Hydrobead Figure 131 shows the behavior of sheath that is covered half by hydrobead and half without hydrobead in coating test. Ice started to accumulate on the specimen by spraying a mist of water with the temperature approximately 34° F. Ambient temperature was 23.4° F. The graph starts with a sudden raise in the specimen’s temperature due to latent heat which caused by freezing water and drops smoothly. On the right side of the figure, the gap shows that hydrobead moves the freeezing water away from the sheath.
177
Figure 131: Specimen’s Behavior in Coating Test The efficiency of the super ice phobic coating was evaluated during outdoor tests on the stainless steel and HDPE specimens. Hydrobead caaused the water to bead into small droplets and ice built up on the specimen. Hydrobead also changed the ice structure and build up rate on the stainless steel specimen. Overall, hydrobead did not significantly impede the build-up of ice and it raised durability concerns in long term. Other concerns with this water repellant and ice phobic coating are: discoloration of the shiny surface of the VGCS stays, attraction of dirt, maintenance cost, and renovation after ice events. Aftf er approximately one month, the hydrobead had a gummy appearance on the stays. Section 7.8: Coating Experiments using Icing UT Tunnel Section 7.8.1: Testing Procedure In order to do the experiments in the icing tunnel, the cooling unit was turned on and provided adequate time to reach a temperature of -5.5 °C, whiich was found to be the typical temperature for a freezing rain storm. The test was started by mounting a coated specimen into the test section and then turning on the fan speeed to be approximately 8.8 m/s, which was found to be the typical wind speed for a freezing rain storm. After allowing the system to stabilize, the camera was turned on and started recording while the misting system was installed into the test sectioon. Once everything was installed, the misting system was turned on and allowed to run throughout the test. The camera recorded and saved the footage of the entire experiiment on a memory card, thus, allowing the user to remove the memory card in order to replay the videoo This permitted both the user and the rest of the research team to have a better understanding of the effects the coating had on the relationship between ice formation and the test specimen.
178 The length of each test was 10 minutes long. After the test was done, the misting system and camera were turned off, the coated specimen was removed from the test section, and the ice accumulation was measured. The next coated specimen was then placed into the test section and the same procedure was repeaated. This test procedure was completed for an uncoated specimen as well as Hydrobead, PhaseBreak TP, and Boyd WeatherTITE coatings. Section 7.8.2: Experiments – Icing Progression Uncoated Specimen The first test was done by misting 40 micron supercool droplets on an uncoated specimen. The purpose of this test was to comparee the uncoated specimen to the hydrophobic coated specimens in order to determine whether the coatings would help reducing the rate of ice accumulation. As mention in the procedure section, the specimen experienced a temperature and wind speed of -5.5 °C and 8.8 m/s, respectively, which simulated a freezing rain storm. The figures below show the progression of the ice accumulation.
Figure 132: Uncoated - 40 Micron - 0:00 min As seen in figure 132, the uncoated specimen was installed into the test section. The dark line drawn across the specimen is representative of stagnation points along the specimen.
Figure 133: Uncoated - 40 Micron - 0:15 min After the mist system was turned on, droplets began to form on the surface of specimen as seen in figure 133.
179 Water Droplets
Water Puddle
Figure 134: Uncoated - 40 Micron - 0:30 min As seen in figure 134, droplets were pushed to the ttop and bottom of the specimen while water puddled along the stagnation line.
Water Droplets
Frozen Puddle
Figure 135: Uncoated - 40 Micron - 0:45 min As time passed, the puddle along the stagnation line begin to freeze while the droplets on the top and bottom of the specimen were still in liquid form.
Water Puddle
Frozen Puddle
Figure 136: Uncoated - 40 Micrron – 1:00 min
180 Water Puddle
Frozen Puddle
Figure 137: Uncoated - 40 Micrron – 1:30 min The frozen puddle continued to accumulate uniformly along the stagnation line while the water droplets on the top and bottom of the specimen began to puddle as shown in Figures 136 and 137.
Frozen Puddle
Water Droplets
Figure 138: Uncoated - 40 Micrron – 2:00 min The ice continued to accumulate uniformly on the first layer of ice at the stagnation line. The puddles formed on the top and bottom of the specimen began to freeze similarly to what happened at the stagnation line. However, there was still water droplets forming on the bottom of the specimen.
Icicles Figure 139: Uncoated - 40 Micrron – 4:00 min
181 Figure 139 shows that the whole specimen was covered with ice and the accumulation continued increasing as time passed. The ice at bottom of the specimen starts forming icicles due to the gravity.
Uneven Surface
Icicles Figure 140: Uncoated - 40 Micrron – 6:00 min
Uneven Surfacce
Icicles Figure 141: Uncoated - 40 Micrron – 8:00 min Droplets of water had frozen on the earlier layer of uuniform ice, thus, causing uneven ice accumulation. This pattern started from the right side of the specimen, as see in Figures 140 and 141. The thickness of ice accumulaation continued to incrrease and the length of icicles increased as well.
Figure 142: Uncoated - 40 Micron – 10:00 min
182 Almost 50% of the surface of the specimen is covered with an uneven surface of ice, particularly the right side as seen in Figure 142. Also, several icicles were present and became very long by the end of the test.
Figure 143: Uncoated - 40 Micron – After Test At the completion of the test, the specimen was removed fromm the test section. The thickest part of the ice was measured to be approximately 6.5 mm. The specimen then was placed in room temperature (approximately 21 °C) and left to shed. After some time, the ice shed as one big sheet as seen in Figure 144.
Figure 144: None Coating - 40 Micron – Shed Ice Sheet The uncoated specimen then was tested with two other nozzle sizes using the same procedure as above. The droplet sizes of the nozzles were 42 and 50 micron. Both tests showed ice accretion the surface of the specimen. The thickest part of the ice was measured to be approximately 5.5 and 5 mm, respectively, for 42 and 50 micron droplet size nozzles.
183 Hydrobead The first coating tested was Hydrobead, which came in a spray can. Thiss is a hydrophobic coating that is clear in color as seen in Figure 145. The coating was sprayed on the entire surface of the specimen. The specimen was allowed to dry for 20 minutes as instructed in the manual prior to installation in the test section.
Figure 145: Hydrobead-Coated Specimen The figures below illustrate the progression of the ice accumullation for Hydrobead coating test experiencing 40 micron supercool water droplet siize.
Figure 146: Hydrobead – 40 Micron – 0:00 min As seen in Figure 146, the Hydrobead-coated specimen was installed into the test section. Similar to the previous test, the dark line drawn across the specimen is representative of stagnation points along the specimen.
184
Figure 147: Hydrobead – 40 Micron – 0:15 min The water droplets began to appear on the surface of specimen once the misting system was turned on as shown in Figure 147.
Figure 148: Hydrobead – 40 Micron – 0:30 min At 30 seconds, the small water droplets began to form into biggger droplets or puddles. The bigger droplets move to the top and bottom of the specimeen while the smaller droplets remain near the stagnation line.
Water Droplet
Frozen Drooplet
Figure 149: Hydrobead – 40 Micron – 0:45 min
185 Water Droplet
Frozen Droplet
Figure 150: Hydrobead – 40 Micron – 1:00 min
Water Droplet
Frozen Droplet Figure 151: Hydrobead – 40 Micron – 1:30 min As time passed, the smaller droplets along the staggnation line began to freeze while the bigger droplets on top and bottom of the specimen were still in a liquid state, which can be seen in Figures 149 to 151.
Frozen Droplet
Frozen Droplet
Figure 152: Hydrobead – 40 Micron – 2:00 min The frozen droplets continued to accumulate along the stagnation line while the water droplets on top and bottom of the specimen began tto freeze.
186 Icicles
Figure 153: Hydrobead – 40 Micron – 4:00 min Figure 153 shows that most of the specimen’s surface is covered with frozen droplets causing an uneven ice accumulation. The ice at bottom started forming icicles due to the gravity.
Figure 154: Hydrobead – 40 Micron – 6:00 min
Figure 155: Hydrobead – 40 Micron – 8:00 min More water droplets froze on the previous layer of ice, thus, thhe specimen was experiencing ice accretion as seen in Figure 154 and 155.
187
Figure 156: Hydrobead – 40 Micron – 10:00 min At the end of the experiment, the surface of the specimen was completely covered by an uneven ice layer and the icicles located at the bottom have become long and orienteddownstream.
Figure 157: Hydrobead – 40 Micron – After Test Similar to the previous test, the specimen was removed from the test section once the test was completed. The thickest part of the ice was measured to be approximately 10 mm, respectively. The ice sheet, shown in Figure 158, shed from the specimen when placing it in room temperature for a period of time.
188
Figure 158: Hydrobead – 40 Micron – Shed Ice Sheet The Hydrobead-coated specimen was then tested with two other nozzle sizes using the same procedure. The droplet sizes of the nozzles were 42 and 50 micron, respectively. Both tests showed ice accretion on the surface of the specimen. The thickest part of the ice was measured to be approximately 6.5 mm, resspectively, for both 42 and 50 micron droplet size nozzles. PhaseBreak TP The next coating tested was PhaseBreak TP, a superhydrophobic coating with greyy color as seen in Figure 159. This coating consisted of 2 parts; a primer and TP. Both parts were applied on a specimen using a foam brush and were given time to cure between coats as instructed in the manual. After the coating was fully cured, the specimen was installed in the test section and the procedure described in section 6.3 was followed. The figures below portray the progression of the ice accumulation for PhaseBreak TP coating experiencing 40 micron supercool water droplets.
Figure 159: PhaseBreak TP – 40 Micron – 0:00 min The PhaseBreak TP-coated specimen was installed into the test section as seen in Figure 159. Once again, the dark line drawn across the specimen is representative of stagnation points along the specimen.
189
Figure 160: PhaseBreak TP – 40 Micron – 0:15 min
Figure 161: PhaseBreak TP – 40 Micron – 0:30 min After 30 seconds of misting, droplets of water appeaared on the surface of the specimen. The droplets kept moving on the surface of the specimen due to wind passing through the test section.
Water Droplet
Frozen Droplet
Figure 162: PhaseBreak TP – 40 Micron – 0:45 min
190 Water Droplet
Frozen Droplet
Figure 163: PhaseBreak TP – 40 Micron – 1:00 min As time passed, the water droplets along the stagnation line beegan to freeze while the droplets on the top and bottom of the specimen were still in a liquid state, which can be seen in Figures 162 and 163.
Water Droplet
Frozen Droplet Icicles Figure 164: PhaseBreak TP – 40 Micron – 1:30 min
Water Droplet Frozen Droplet
Icicles Figure 165: PhaseBreak TP – 40 Micron – 2:00 The frozen droplets continued to accumulate along the stagnation line while the water droplets on the bottom of the specimen began to freeze, creating icicles. However, droplets of water still appear on the top of the specimen.
191 Frozen Droplets
Figure 166: PhaseBreak TP – 40 Micron – 4:00
Frozen Droplets
Figure 167: PhaseBreak TP – 40 Micron – 6:00 While more ice accumulated along the stagnation line and the bottom of the specimen, the water droplets on top began to freeze; this can be seen in Figures 166 and 167.
Figure 168: PhaseBreak TP – 40 Micron – 8:00
192
Figure 169: PhaseBreak TP – 40 Micron – 10:00 More droplets of water froze on the initial layer of ice as time elapsed. Both the thickness of ice accumulation and the length of iciclles located along the bottom of the specimen continued to increase as time passed. By the end of the test, the surface of the specimen is completely covered by an uneven ice layer and the icicles became really long and oriented downstream.
Figure 170: PhaseBreak TP – 40 Micron – After Test The specimen was removed from the test section after the test was completed. The thickest of the ice was measured to be approximately 10 mm, respectively. The ice melted to a thin and small sheet, shown in Figure 171, and shed from the specimen at room temperature.
193
Figure 171: PhaseBreak TP – 40 Micron – Shed Ice Sheet The PhaseBreak TP-coated specimen was then tested with two other nozzle sizes using the same procedure. The droplet sizes of the nozzles are 42 and 50 micron, respectively. Both tests showed ice accretion on the surface of the specimen. The thickest part of the ice was measured to be approximately 6.5 and 5 mm, respectively, for 42 and 50 micron droplet size nozzels. Boyd WeatherTITE The last coating tested was provided by Boyd Coating Research Company, Incorporated and is their WeatherTITE coating. It iss a white suuperhydrophobic coating. There were three parts to this coating, each part caame in a separated container labeled part A, B and C. The three parts were mixed and applied on a specimen using a foam brush. It was then given time to cure as instructed in the manual. After the coating was fully cured, the specimen was installed in the test section. The figures below show the progression of the ice accumulation for WeatherTITE experiencing 40 micron supercool water droplets.
Figure 172: WeatherTITE – 40 Micron – 0:00 min The WeatherTITE-coated specimen was installed into the test section as shown in Figure 172. The dark line drawn across the specimen is representative of stagnation points along the specimen.
194 Water Droplet
Water Puddle
Figure 173: WeatherTITE – 40 Micron – 0:15 min After 15 seconds of misting, droplets of water appeaared on the surface of the specimen as seen in Figure 173. The droplets were pushed to the top and bottom of the specimen while water puddled along the stagnation line.
Water Droplets
Frozen Puddle
Figure 174: WeatherTITE – 40 Micron – 0:30 min
Water Droplets
Frozen Puddle
Figure 175: WeatherTITE – 40 Micron – 0:45 min As time passed, the puddle along the stagnation line began to freeze while the droplets on top and bottom of the specimen were still in a liquid form as shown in Figures 174 and 175.
195 Water Drooplets
Ice
Figure 176: WeatherTITE – 40 Micron – 1:00 min
Water Droplets
Ice
Figure 177: WeatherTITE – 40 Micron – 1:30 min After 1:30 minutes, the ice continued to accumulate uniformly along the stagnation line while the water droplets on top and bottom of the specimen began forming into bigger droplets or puddles as shown in Figures 176 and 177.
Water Droplet Ice Dropllet
Icicle
Figure 178: WeatherTITE – 40 Micron – 2:00 min
196 Water Droplet Ice Droplet
Icicle
Figure 179: WeatherTITE – 40 Micron – 3:00 min The ice continue to accumulate uniformly on top of the first layer of ice at the stagnation line. Some of the big water droplets on top of the specimen began to freeze and the icicles started to build up on the bottom of the specimen, which can be seen in Figure 178 and 179.
Ice Droplet
Icicle
Figure 180: WeatherTITE – 40 Micron – 4:00 min
Ice Droplet
Icicle
Figure 181: WeatherTITE – 40 Micron – 6:00 min As time passed, more layers of ice accreted at staggnation line. All big droplets of water on the top of the specimen froze as shown in Figures 180 and 181. The icicles also become longer.
197 Uneven Ice Surface
Figure 182: WeatherTITE – 40 Micron – 8:00 min
Uneven Ice Surface
Figure 183: WeatherTITE – 40 Micron – 10:00 min The thickness of ice accumulation continued increaasing and so did the length of icicles. By the end of the test, the surfr ace of the specimen is covered by an uneven ice layeer due to water droplets freezing in a previous layer. Otherwise, the water droplets that freeze quickly rather than spreading out create an uneven surface for the ice accretion. The icicles also become long and oriented downstream.
Figure 184: WeatherTITE – 40 Miccron – After Test
198 After the test, the specimen was removed from the test section. Figure 184 shows the ice accretion on the specimen after the test. The thiickest part oof the ice was measured to be approximately 8 mm, respectively. The ice sheet, shown in Figure 185, shed from the specimen while placing it at room temperature.
Figure 185: WeatherTITE – 40 Micron – Shed Ice Sheet Section 7.8.3: Result Summery of Icing Tunnel Coating Tests After performing superhydrophobic coating tests, the table below was gennerated to compare the result of each coating and droplet size.
Table 42: Approximated ice thickness comparison of coatings and droplet sizes
Coating None Hydrobead PhaseBreak TP WeatherTITE Droplet Size 40 micron 6.5 mm 10.0 mm 10.0 mm 8.0 mm 42 micron 5.5 mm 6.5 mm 6.5 mm 9.5 mm 50 micron 5.0 mm 6.5 mm 5.5 mm 9.5 mm
The result show that none of the coatings can prevent the ice from accumulating on the specimen. It is noticeable that the coated specimens tend to have higher ice accumulation. This is due to ice droplets freezing quickly rather than spreading out during the tests causing an uneven surface which can trap more water. Once the water freezes, it becomes a thicker layer of ice on the surrfface. On the other hand, the ice formed evenly on the surface of the uncoated specimen. Due to the uniform surface, the water tends to slide off the surface, instead of getting trapped. This causes lower ice accumulation on the specimen. In addition, the results of the experiments also show that smaller droplet sizes causes higher ice accumulation. This is true for most cases during the coating experiments. The only coating that does not have this trend is the WeatherTITE coating. This could be due to a different chemical makeup of the coating.
199 Section 7.9: Field Experiment Trips On February 16 and February 20, it was expected of temperatures to go down to the freezing point at Toledo and hence a trip was set up to expose the sensors to a possible icing environment for a more practical test before actual installation. The leaf wetness sensor, ice detector and stay thermistors were used in this field test. The team from University of Toledo set up stay sheath specimens and exposed them to freezing temperatures. A garden hose sprinkler and Windex bottles were used to simulate rain.
Table 43: Event History (February 16, 2013) Time Event February 15, 5 – 6 pm Set up ice detector, leaf wetness sensor, sunshine sensor, and thermistors. Set up CR1000 white boxes. February 16, 4 am Programmed CR1000s. UT team set up their spraying equipment. February 16, 4:58 am Started collecting data. February 16, 5 – 6:45 am Water sprayed at random intervals. Photos captured and results noted. February 16, 6:45 – 8:30 Sensors observed, photos taken. am Left site at 8:30 am. February 16, 8:30 – 4:10 Data collected without physical observation. Started packing at 4 pm. pm February 16, 4:40 pm Left site for UC.
The UCII team reached Toledo on February 15th evening and set up the sensors and the data acquisition system for the next morning. The team from University of Toledo had already set up a specimen stay sheath identical to those on the bridge on a concrete pad. Geokon thermistors, same model as those already installed on the VGCS stays were clamped on the stay sheath. Similarly the leaf wetness sensor was also zip-tied on top of the specimen. The ice detector was set up right beside the specimen stay to be exposed to similar simulated weather conditions. The photos below show the initial set up of the sensors on the experiment pad at Toledo, just a few miles away from the Veteran’s Glass City Skyway bridge (photo credits – Jason Kumpf, University of Cincinnati Infrastructure Institute).
200
Figure 186: Stay Specimens at Figure 187: Data-logging System FFigure 188: Sunshine Sensor Setup Different Angles and Orientations Setup
Figure 189: Ice Detector Placed Figure 190: Stay Thermistors Zip- Figure 191: Leaf Wetness Sensor Right Beside Stay tied on Sheath Taped on top of Specimen
201 The different sensors came up with interesting resuults. The ice detector measured ice accurately on its probe. The following photos explaiin its condition at different times on February 16.
Figure 192: Ice Detector at Various Times Throughout the February 16 Experiment
202
Figure 193: Leaf Wetness Sensor at Various Times Throughout the February 16 Experiment It was observed that with each spraying event, the the temperature of the sheath rose up. This was because the stored water temperature was above the air/stay temperature which were below freezing. Shortly after each spraying, the water froze on the stay and the 0872f1 probe. Initially the deicing threshold was kept low at 0.02 inches of ice, so heating syscle was triggered early. Later it was reprogrammed to 0.15 inches. As the stored water was warmer, most of it drained/trickled dowwn the ice detector probe before instantaneously freezing. The sun came out at 7:28 am, the temperature recorded by the thermistor rose, and the accumulatted ice started melting.
203
Figure 194: Ice Detector Characteristics (Toleddo experiments on February 16) Similarly, there were three thermistors placed strategically touching the sheath and facing outwards and their behavior were noted. All the thermistors recorded higher temperature with each event of spraying. This was because the stored water was warmer than the stay temperature. As the water raised the stay temperature, the sheath facing theermistors recorded elevated temperatures.
204
Figure 195: Characteristics of stay thermistors (Toledo, February 16)
Figure 196: Leaf Wetness Sensor ice melting characteristics
205 During the spraying experiments, water froze on the leaf wetness sensor as well. As a result there was a constant dielectric of about 310 mV when a layer of ice was on its surface. Later, with sunrise, the characteristics of the leaf wetness sensor were compared to that of the radiation results obtained from sunshine sensor. It could be seen that with each peak (stronger sunshine), the dielectric also showed higher peaks due to melting of ice. A second trip was made to Toledo on February 20th for another set of tests. The day was carefully chosen as the temperature was supposed to go under freezing point and was considered optimum for icing experiments. On February 20 afternoon, a team from UC went to Toldeo and set up the sensors and data acquisition system. The data loggers were programmed, and the spraying equipments were set up. Around 9 pm, the spraying experiments were started.
Table 44: Event History (February 20-21, 2013) Time Event February 20, 7:30 pm Set up ice detector, leaf wetness sensor, sunshine sensor and thermistors. Setup CR1000 & white boxes. February 20, 8:30 – Programmed CR1000s. 9:00 pm UT team set up their spraying equipment. February 20, 9:10 pm Started collecting data. Spraying started at 9:13 pm. Water direction changed at 9:19 pm. February 20, 10:20 pm Left site. UT team stopped hose around 11:00 pm. February 21, 4:00 am Back to site, more photos taken. Ice detector de-iced. February 21, 5:37 – Ice detector programmed and deployed. 5:50 am Leaf sensor experiment using bug spray. February 21, 6:03 am Spraying began again. Close observation done to check icing/de-icing characteristics of ice detector. Spraying stopped at 6:34 am. February 21, 7:12 – Ice detector disconnected. Other sensors disconnected by 8:40 am as 8:40 am shedding chances were low due to sky cover.
206 These photos below are a good exhibit of the icing eexperiments setup at Toledo on February 20 – 21(photo credits – Jason Kumpf, Uniiversity of Cincinnati Infrastructure Institute) These photos demonstrate the spraying experimentts:
Figure 197: LWS-LS with Different Slants Figure 198: Top & Side Thermistors Setup
Figure 199: Ice Detector Setup Figure 200: First Spray Shower
207 Figure 201: Garden Hose mount on ladder (left) & hand held (right) for experiment on ice detector & leaf sensors
Figure 202: Ice Detector at Various Times during Experiment (Left and Middle during ice accretion; right during deicing) Quite similar to February 16th experiments there were three thermistors placed strategically touching the sheath and facing outwards and their behavior were noted. All the thermistors recorded higher temperature with each event of spraying. This was because the stored water was warmer than the stay temperature. As the water raised the stay temperature, the sheath facing theermistors recorded elevated temperatures.
208
Figure 203: Stay thermistor characteristics (Toledo experiments February 20 – 21) There were two dielectric leaf wetness sensors that were placed on the stay at different slants. Due to the warm water being sprayed on them, they captured high dielectric in the beginning. As the water froze on the leaves, they reported dielectric around 310 mV. They were left overnight covered with thick layer of ice. Next morning, both sensors were manually deiced and exposed to water sprinkler again. They showed similar characteristics, except the tilted leaf wetness sensor drained off its water faster and hence showed lower dielectric.
209
Figure 204: Leaf Wetness Sensor Characteristics (Toledo, February 20 – 21) Initially smaller bug spray bottles were used for testing ice accumulation on the ice detector probe. It was observed that with each little spraying event, the ice detector recorded proportional increase in thickness of ice on its probe. Shortly after spraying was started using garden hose, the water froze on the stay and on the 0872F1 probe. Initially the deicing threshold was kept low at 0.06 inches of ice, so heating cycle was triggered early. Later it was reprogrammed to 0.15 inches. During ice accumulation on the probe, a set of calipers were used to manually determine the ice thickness and get an estimate of the ice detector characterisctics. It was noted that the human errors aside, the calipers gave accurate readings.
210
Figure 205: Ice Detector characteristics (Toledo, February 20 – 21) Section 7.10: Conclusions The experiments on the sheathing specimens at the field station and in the laboratory coupled with the literature review lead to the concluusion that the proposed sensors functioned as desired and they were recommend for installation. The Installation and initial behavior in service of the sensors is discussed in chapter 8. The experiments at the field station and in the laboratory indicated that none of the active control measures considered at this time - coatings, fluids and heating - are not viable at this time. Outdoors experiments have documented and verified the proper operation and calibration of a new sensor suite to be deployed on the VGCS. These include:
Geokon Thermistor 3800-2-2, for stay surface temperatures
LWS-L Dielectric Leaf Wetness Sensor, for detecting water/moisture with some classification as ice, light rain/snow, or heavy snow/rain.
Sunshine Sensor BF5, for detecting and quantifying solar radiation and approximate level of cloud cover
Goodrich Ice Detector, for detecting and quuantifying ice thickness The tests on the coating, fluids and heating in combination with the earlier background work reveled there is no active technology that is currently economically feasible for the VGCS.
211 The coating tests at the field station and in the laboratory revealed that the coatings studied did not perform the basic ice prevention function desired. Additionally, they did not age gracefully. If they are durable enough to last a winter, they may discolor or become gummy. Therefore because of poor functionality, low durability and likely installation expense, the existing coatings were deemed infeasible.
212 Chapter 8: Deployment of New Sensors and Upgrade of the Dashboard Section 8.1: Introduction The existing sensor array on the bridge was not adequate to accurately characterize the local conditions on the bridge or the stays. The gaps in information from those sensors are presented in Table 2. Where suitable commercial sensors were available, they were procured and deployed on the bridge. The performance of the commercial sensors was verified in the laboratory and the field as described in chapters 6 and 7. The thermistors were mounted directly on the stays and their position and installation is described in chapter 6. The other sensors as well as a pan-tilt-zoom video camera were installed along with the weather tower on the bridge. The weather tower was installed in the spring and summer of 2013 so it could be tested and made available for the 2013-2014 icing season. This chapter describes the location and design of the tower, the deployment of the instruments and camera, as well as the upgrade of the dashboard to incorporate the new sensors. Also, an assessment of the upgraded dashboard’s performance during the following winter (2013-2014) and recommendations to minimize false alarms are presented. Section 8.2: Self Supporting Instrumentation Tower Design Section 8.2.1: Tower Design The weather tower to support the sensors and camera was located on the south eastern end of the bridge. It was on placed on the eastern side so that it would not be in the shadow of the stays for the prevalent easterly ice storms. It was located near the RWIS on the south end to simplify connections and was located so that the sensors were above the stays so no ice would fall on them. The arrangement of the sensors on the tower was chosen such that as much as practical they would not fall in each other’s wind, rain or solar shadow assuming the storm came from the east. A ROHN self-supporting tower was chosen to support the instruments on the VGCS. The design of the tower was in accordance with national standard ANSI/EIA-222-F and ANSI/TIA-222G “Structural Standards for Steel Antenna Towers and Supporting Structures”. . In the Rohn design report, tower elements were analyzed as three dimensional beam models. A beam element is considered as a two node element with three degrees of freedom at each nodes (two translation and one rotation). The performed calculations for the anchorage system are based on the following maximum factored reactions: maximum download reaction of 21.3 kips, maximum uplift reaction of 20.4 kips, total shear reaction of 1.73 kips, and an over turning moment of 25.21 kip-ft. The Rohn tower is divided into three sections of roughly equal height, which are the 45GSR for the upper and middle sections and the 45GSRH for the lower section. The reason for this is due to the lower section requires stiffer diagonal and horizontal sections. The Rohn design assumed the base of the tower was 120 feet above the ground. The following design criteria was used: 360 degree wind orientation per 30 degree increment, basic wind speed (no ice) = 90 mph, basic wind speed (with ice) = 40
213 mph, design ice thickness = 1.0", exposure category = C, structure classification = II, topography category = 1. The maximum compressive and tensile forces in legs are 20.4 kips and 19.6 kips, respectively. When considering a governing compressive capacity of 44.4 kips and a governing tensile capacity of 51.7 kips, the demand to capacity ratio was determined to be 0.46. This shows the tower has capacity well beyond the design loads to accommodate future instrumentation. The demand to capacity ratios, 0.29 for the 45GSRH diagonal and 0.24 for horizontal sections, show this standard tower design is not being pushed to its limit. The Rohn tower has been installed on the east side of the VGCCS. As shown on Figures 18, 19 and 20, typically most of the ice accumulates on the east side of the stays. On the west side, stays are usually bare except for occasional frozen rivulets. This reflects the conditions during the most likely icing event. Section 8.2.2: Anchorage System Design The AASHTO Standard Sign Support Specification was used for the anchorage system design. This specification is the most appropriate because it adddresses the mounting and general tower concerns. However, this is an Allowable Stress Design (ASD) specification whereas the Rohn loads are maximum load envelopes in a Load and Resistance Factor Design (LRFD) format, therefore, it was necessary to formulate the loads into an ASD format. When possible, nominal sstrengths are computed in an LRFD format and compared to the required strengths. When AASHTO Standard Sign Support Specification did not provide guidance in designing of the tower, AISC 14th, and ACI- 318 were used. A typical double nut connection was chosen as the anchorage system. Design loads were based on the ROHN report, which considered strength and service limit states. The following failure modes were considered for design criteria: bolt failure, tensile strength of concrete, base plate failure, fatigue failure of base plate and anchor bolt, failure of plate washer, and bearing strength of concrete. The anchorage design calculations are presented in the (Ali, 2013). The toower was installed in the spring of 2013 and the instruments were installed shortly thereafter.
Figure 206: Tower Anchorage System
214 Section 8.3: VGCS Ice Sensors Bridge Installation trip (May 16-17, 2013) On May 15, 2013 a team from University of Cincinnnati Infrastructure Institute made a trip to Toledo in order to install all the ice sensors on the tower. Contractors from US Utilities had finished building a standard 45 GSR meeteorological tower near stay 19.
Figure 207: Rohn’s Weather Tower Drawing A traffic controller cabinet supplied by ODOT was erected at the tower base to house the data loggers and power supplies for the camera and ice sensors.
Figure 208: Tower mounted near stay 19
215
Figure 209: Initial Plan by UT Research Team for Tower Mounting On May 16, 2013, the URT gathered at Veteran’s Glass City Skyway to mount the sensors on the tower. Horizontal cross-arms were fixed on to the tower at different heights to support the sensors. The UCII crew was divided into three groups to: a) set up the data loggers inside the tower cabinet, b) ride a bucket truck and mount the sensors at different elevations, and c) Upgrade the data acquisition system with neew hardware and new sampling rate.
216 Table 45: Summary of VGCS Sensor Installation Trip
Crew Position Duties 1. Tower Cabinet Setup white backplanes on shelves. Receive cables from conduit. Connect cables from each sensor to CR1000 and push new program. Reprogram with appropriate sampling rate. Setup IP address for each data logger. 2. On the Tower (Each Sensor Location) Connect vertical cross‐arm to tower. Install the gage on the arm. Let the cable down through the conduit. Repeat for all sennsors. 3. Old/Main Cabinet Location Upgrade system to CR1000. Reprogram with new scan rate. Setup new IP conffiguration.
MOUNTING TECHNIQUES Different techniques were used to mount the sensors on the tower at different elevations. The ice detector was at the topmost level, followed by solar radiation sensor, rain tipping bucket and leaf wetness sensor. Leaf Wetness Sensor: The leaf wetness sensor waas placed flat with the dielectric surface facing upward on the cross arm and firmly attached to it using zip-ties.
Figure 210: Leaf Wetness Sensor Zip-tied to Cross-arm
217 Rain Tipping Bucket: The rain bucket was screwed on to a leveling bracket which sat on a vertical pipe. This pipe was mounted on the crrooss arm using cross-over T joints with OD 1.9” for 1.5” IPS cross arms.
Figure 211: Rain Bucket mounted on cross--arm using leeveling bracket Sunshine Sensor: The sunshine sensor was screwed to a horizontal plate with holes drilled at its corner to insert U-bolt and and firmly attached to the cross-arm.
Figure 212: Sunshine Sensor attached to cross-arm with steel U-bolts
218 Ice Detector: A plate was attached to the body of the ice detector, which was tied to stainless steel band clamps circling a 1.5 IPS pipe. This pipe was mounted on the cross arm using cross-over T joints with OD 1.9” for 1.5 IPS cross arms.
Figure 213: Ice Detector Mounted using Steal Worm Band Clamps Figure 214: Ice Detector Mounted Close Up
Figuure 216: CR1000 Datalogger Setup Figure 215: Sensor Cable Conduit Insider Tower Cabinet
219
Figuure 218: Completed New Weather Figure 217: Close up of Weather Tower Station Near Stay 19
After the sensors were physically installed, and all firmware configurations settled, UCII started collecting local weather data at the bridge starting on May 17, 2013.
IP Configuration All the new data loggers had to be mapped to the IP addresses provided by ODOT IT personnel.
Existing DL (Strain Gages, Stay Thermistors) – 10.102.90.236: port 6783 DL1 (Ice Detector) – 10.102.90.237: port 6783 DL2 (Leaf Wetness Sensor) – 10.102.90.238: port 6783 DL3 (Sunshine Sensor) – 10.102.90.239: port 6783 DL4 (Rain Bucket) – 10.102.90.240: port 6783
All pointed to the same external IP 156.63.133.93 Note: DL = Data Logger
Section 8.4: Changes to the Ice Accumulation Algorithm In chapter 5, the authors discussed the parameters and weather stations considered for developing the accretion algorithm. The following were added to the previous list of voters in the algorithm. Local Stations: There are two local stations having five new sensors that contribute to the evaluation of the accumulation conditions and hence the algorithm:
220 Accumulation Determining Stations 1. Accu_Local1: - Stay thermistors and dielectric wetness sensor (leaf wetness sensor) 2. Accu_Local2:- Stay thermistor and rain tipping bucket 3. Accu_Local3 :- Ice detection sensor The ice detector comprises of a single station alone for simplicity of the algorithm because of its superior reliability and criticality. Data Update Time: An important factor to consider in the ice determination algorithm is that each of the weather stations have a distinct data update time. In an hour, the RWIS stations update data about 4-6 times an hour, while the METAR stations do it once or twice. The new local sensors have a sampling rate of 10 minutes so as to read them at a pace that matches those of the RWIS. Station Individual Weights: According to their importance each existing station had been assigned a weight for its contribution towards triggering an alarm. The closest weather station (RWIS 142-I-280) had a weight of 0.3. Both airports report additional data with high reliability and, hence, also have a weight of 0.3 each. The other three RWIS stations each have a weight of 0.1 because they are farther from the bridge. The new Accu_Local 1, 2, and 3 stations each have a weight of 0.3. Threshold Weights: Various simulations had been previously done to obtain an appropriate threshold for triggering ice accumulation. The threshold was set at 0.3 so that either of the two METAR airports or the local RWIS station could trigger ice accumulation alert alone. The other three distant RWIS stations can only do the same, when in unison. Each of the new UCII stations included in the proposed algorithm can also trigger this threshold by itself. Ice Accumulation Algorithm: Sensors in any environment can occasionally misread the actual measurement. The algorithm evaluates all eight weather stations’ (including four RWIS, two METAR, and new UCII stations Accu_Local1, Accu_Local2, Accu_Local3) records for the last hour. Only if a certain percentage of the total records from the last hour meets any of the ice accumulation criteria, then the station has satisfied the icing and is given a Boolean value 1. However, if this condition is not satisfied by a weather station, the respective station is provided a Boolean value 0. Initially, the percentage of satisfying records is 50% for the new sensors, and 80% for the RWIS / METAR stations. This is then used to find the conditions favorable to ice accumulation by multiplying the Boolean value of each weather station (0 for not met, 1 for met) with the station weight and summing each result.
221 If the total weight calculated, as above, is greater than a set threshold (0.3), we consider that conditions favorable to icing have been met for the last hour. In the existing algorithm, there have been separate functions used for each station. Hence there is a function for the four RWIS stations, one for the two METAR stations. Similarly, three new functions have been introduced for the new sensors. This kind of modularity helps upgrade the existing algorithm without affecting it; that is, the existing decision algorithm for the initial UCII dashboard (seen in Chapter 6) remains the same and will work exactly the same as before, but now it is complemented by a similar algorithm employed with the new sensors which can also vote and cause the threshold to be met. The table in the next page describes the modularity of the modified ice accumulation algorithm.
Figure 219: Flowchart of existing Ice Accumulation Algorithm (Agrawal, 2011)
222 Table 46: Ice Accumulation Station Functions
Ice Accumulation Stations
Function Name Type Measurement Parameters
Precipitation: Rain, Snow 1. RWIS Web (4 Locations) Temperature: Air Temp. Precipitation Rain, Snow (various types)
Existing 2. METAR Web (2 Locations) Temperature: Air Temp. Precipitation: Dielectric Constant (mV) to Stay Thermistors, Leaf 3. LocalStation1 wetness check conversion Wetness Sensor Temperature: Stay Temp. Stay Thermistors, Rain Precipitation: Water collected (inches) 4. LocalStation2 Bucket Temperature: Stay Temp. Proposed 5. LocalStation3 Ice Detection Sensor Ice Presence check directly.
Modular: 3 new independent station functions for ice accumulation criteria
The flowchart below is a good representation of the proposed ice accumulation algorithm. So far we have been using test thresholds for each of the new sensors. The proposed threshold of ice thickness by ice detector is 0.15 after which melting cycle starts. The threshold for leaf wetness sensor is 300 mV and for rain bucket is a rate of 0.1 inches/hr. These thresholds are subject to training in the real world and with more icing events and data analysis, we can settle with a more and more appropriate threshold for each of the new sensors’ icing criterion.
223
Figure 220: Flowchart for revised ice accumulation algorithm State Transitions: The determination of possible ice conditions (“yes”=1 or “no” = 0) happens each data collection cycle as stated. However, historically it has been noted that persistent ice accumulation over hours led to ice shedding. To determine if ice event conditions persist and to what extent required the development of a set of states and corresponding conditions for transition between states. Different states used in the Ice accumulation process are: C: ‘Clear’–No Ice present Y1: ‘Yellow 1’–Icing possible Y2: ‘Yellow 2’–Icing likely Y3: ‘Yellow 3’–Icing highly likely A: ‘Alert’–Ice confirmed
Once at Alert, the ice shedding algorithm gets initiated, to see if there is possibility of ice falling from the stays. If not, it remains at this state, unless there is a report after visual verification.
224 Section 8.5: Changes to the Ice Shedding Algorithm The shedding algorithm was revised to include some of the new sensors in addition to the 6 off-bridge stations used before. The new sensors used are reffered to as “Local Stations”. Local Stations: There are two local stations having two sensor types that contribute to the shedding conditions and hence the algorithm: 1. Fall_Local1 :-Stay Thermistor 2. Fall_Local2 :- Sunshine Sensor Data Update Time: An important factor to consider in the ice determination algorithm is that each of the weather stations has a distinct data update time. The RWIS stations update data about 4-6 times an hour, while the METAR stations do it once or twice. The local sensors have a sampling rate of 10 minutes. Station Individual Weights: According to the importance of each station, they had been assigned weights for their contribution towards triggering an alarm. The closest weather station (RWIS 142-I-280) has, thus a weight of 0.3. Similarly, both the airports report additional data and hence also have a weight of 0.3 each. The other three RWIS stations each have a weight of 0.1. The new Fall_Local 1 and 2 stations based on local icing sensors on the bridge are designed to have a weight of 0.3 in the algorithm. Threshold Weights: As before, the threshold was set at 0.3 so that either of the two METAR airports or the local RWIS station could trigger ice accumulation alert alone. The other three distant RWIS stations can only do the same, when in unison. Each of the local stations included in the revised algorithm can also trigger this threshold by itself. Ice Shedding Algorithm: As previously mentioned, sensors in any environment can occasionally misread the actual measurement, thus, each of the eight weather stations’ (including four RWIS, two METAR, and new local stations Fall_Local1, Fall_Local2) records are evaluated for the last hour. Only if certain percentage of the total records from the last hour meets any of the ice shedding criteria, then the station has satisfied the icing criteria as for the last hour and is given a Boolean value of 1. However, if this condition is not satisfied by a weather station, the respective station is provided a Boolean value 0. The percentage is 50% for the local sensors, and 80% for the RWIS / METAR stations. This is then used to find the conditions favorable to ice shedding by multiplying the Boolean value of each weather station (0 for not met, 1 for met) with the station weight and summing each result. If the total weight calculated, as above, is greater than a set threshold (0.3), we consider that shedding conditions have been met for the last hour.
225 In the existing algorithm, there have been separate functions used for each of the siix stations, hence, there is a separate function for the four RWIS stations and the two METAR stations. Similarly, two new functions have been introduced for the new local sensors. This kind of modularity helps upgrade the existing algorithm without affecting it. The table in the next page describes the modularity of the modified ice shedding algorithm.
Figure 221: Flowchart of existing Ice Shedding Algorithm (Agrawal, 2011)
Table 47: Ice Fall Station Functions in alggorithm
Ice Shedding Check Stations
Function Name Type Measurement Parameters
1. RWIS Web (4 Locations) Temperature: Air Temp.
Sky cover: Clouds/visibility.
Existing 2. METAR Web (2 Locations) Temperature: Air Temp.
3. LocalStation1 Stay Thermistors Temperature: Stay Temp.
4. LocalStation2 Sunshine Sensor Sky cover: Solar Radiation (Watts/m2) Proposed
Modular: 2 new independent station functions for ice accumulation criteria
226 The flowchart below is a good representation of the revised ice shedding algorithm. So far test thresholds have been used for each of the new sensors. The solar radiation threshold is set at 125 Watt/m2. With more icing events and data analysis, a more and more appropriate threshold for each of the new sennsors’ icing criterion can be set.
Figure 222: Flowchart for revised ice shedding algorithm Section 8.6: Changes to the Dashboard In previous sections of this chapter, the design and functionality of the algorithm was discussed. This chapter explains how the algorithm is implemented for front end user. The main website had several tabs: 1. Dashboard 2. Map (Weather Data by location) 3. History 4. Documentation 5. Plotting
227 Section 8.6.1: Dashboard Main Panel The inputs from the sensors go into a Python algorithm. The implementation of the Python algorithm was exhibited through the dashboard was divided into few parts in order to integrate data from and manage connectiviity between separate systems. The function of the first part was to automate the collection, archiving and data mining of weather measurements of the RWIS stations, airpoort METAR and weather tower sensor data. These processes needed to be abiding and sturdy so that we have a reliable set of weather measurements from all the appropriate sources. The software was finally developed to carry out the automated evaluation prroocess and put on schedule to run once per hour. Upgrade of the dashboard: A new module has been added to the dashboard algorithm that determines the functional status of each remote weather station. It reports when any of the six stations has not been reporting enough data in the past one hour. At the bottom of the main panel is the table that shows the funnctional status of each of the four RWIS and two METAR stations. The greenn symbol turns red for each station not reporting data. The main panel also includes the links to all other pages of the dashboard. Please find below a screen shot of the web site showing the mmain dashboard panel:
Tabs of ice monitoring website
Speedometer dial
Ticker for last 48 hours record
Monitor of Monittor: Weather station status
Figure 223: Dashboard Main Panel
228 Section 8.6.2: Map (Weather Data by location) As before, the icing monitor website includes an interactive map of the weather stations. It contains pop-up balloons for each weather station where current sensor readings are shown and historical readings can be plotted on a timeline. The map also provides view of the cameras installed on the bridge. Google Maps API has been used for this application. Google Maps is a free web mapping service provided by Google, which offers street level maps for pedestrians, cars, and public transportation. The Google map on the dashboard is used as a graphical interface providing the particulars about the various weather sites being monitored for determining the icing conditions at VGCS. It also contains the location of various sites, their past/present weather conditions along with their source links. Initially, this weather map comprised of the few RWIS, METAR and local stations surrounding the bridge. After successful installation of the various new ice sensors on the bridge, new station have been added to the weather map comprising information about these UCII sensors and newly installed camera.
229
Figure 224: Example Snapshot of Weather Map, with Pop-up for Ice Detector When the map tab is clicked, the Google map openns with several markers corresponding to the different stations. There are two green balloons written “A” on them. These are the two Airports namely KTOL (Toledo Express Airport) and KTDZ (Metcalf Field Airport. The four red balloons written “R” on them are the four RWIS stations. There are two yellow balloons written “L” on it. These are the local weather stations near to the VGCS Bridge. Another green marker with “U” written on it represents the UCII sensors station. In the UCII marker, on clicking any of the ice sensors provide detailed information through a 48 hour plot. Here are examples:
230
Figure 225: Last 48 hour report of Solar Sensor (Global Radiation)
Figure 226: Last 48 hour report of Leeaf Wetness Sensor
Similarly, the last 48 hour plots for the ice sensor and the rain sensor can be obtained by clicking the appropriate links in the marker. Section 8.6.3: New Sensors Plotting The plotting tab has been upgraded to include the local sensors. This plotting tab has different sections as illustrated in the Figures 227 - 232 below. Each plotting section can be populated with the different stations/sensors it consists of. The user can select the range of the start and end date for this plotting function. Accordingly, the data for the different sensors can be obtained for the desired time range. At present the different sections in the ice moonitor plotting tab are:
231
Figure 227: Stay 20 Thermistors plot (January 1 – July 1)
Figure 228: Stay 8 Thermistors plot (January 1 – July 1)
232
Figure 229: Ice Detector plot (June 1 – July 1)
Figure 230: Leaf Wetness Sensor plot (June 1 – July 1)
233
Figure 231: Rain Tipping Bucket plot (June 1 – July 1)
Figure 232: Sunshine Sensor plot (June 1 – July 1)
234 Section 8.7: Insights Gained from the Operation of the Upgraded Dashboard Since the installation of the new ice sensors by University of Cincinnati Infrastructure Institute in May 2013, we have been able to obtain data for one season of winter, December 2013 – April 2014. To test the performance of the new sensors before being installed, rigorous testing methods had been implemented in the lab and in the field. Those experiments helped us understand the nature and anticipate the usefulness of the sensors when deployed in the real world. However, to establish ground truth of the actual performance of the sensors, the training data obtained this winter is very important. This winter had been quite eventful as far as wintry conditions are concerned. We have had multiple occasions where our upgraded dashboard algorithm triggered ice accumulation warnings, mostly at moderate levels of Y1 (level 1 ice accumulation), few Y2 (level 2 ice accumulation) and Y3 (level 3 ice accumulation). There has been no incidence where the dashboard warned ice shedding. The remote weather sensors from RWIS and METAR stations had helped our ice event determination algorithm the past few winters. However, that data wasn’t short of various issues including time delay, inaccuracy and lack of local supervision. The new sensors – ice detector, leaf wetness sensor, solar radiation sensor, rain tipping bucket and stay thermistor have paved way to a new level of microclimate information at the Veteran’s Glass City Skyway bridge. Each of them has been obtaining different and crucial data for ice accumulation and/or ice shedding event determination. The stay thermistors had been installed on the bridge in April 2013. However, the other sensors were installed more recently and were exposed to icing conditions on the bridge for the first time this past winter. The collection of ground-truth data will enable calibration of remote-sensing data for the years to come, and aid in the interpretation and analysis of the weather parameters. The events triggered by the sensors this winter are enumerated in the following table:
235 Table 48: Chronology of winter 2013/2014 icing event triggers Date Highest Stations/Sensors Triggered Dashboard Status 12/09/2013 Alert KTOL, ODOT Report 12/11/2013 Y1 Stay Thermistor + Dielectric Sensor 12/12/2013 Y1 Stay Thermistor + Dielectric Sensor 12/13/2013 Y2 RWIS2016, Stay Thermistor + Dielectric Sensor 12/14/2013 Y3 Stay Thermistor + Dielectric Sensor 12/23/2013 Y1 RWIS2016 12/29/2013 Y1 KTDZ, RWIS2016 01/01/2014 Y1 Stay Thermistor + Dielectric Sensor, RWIS2016 01/04/2014 Y1 RWIS2016 01/05/2014 Y2 RWIS2016, Stay Thermistor + Dielectric Sensor, KTDZ 01/09/2014 Y1 KTDZ 01/10/2014 Y2 Stay Thermistor + Dielectric Sensor, RWS2016 01/17/2014 Y1 RWIS2016 01/20/2014 Y1 Stay Thermistor + Dielectric Sensor, RWIS2016 02/20/2014 Y2 Stay Thermistor + Dielectric Sensor, RWIS2016 03/08/2014 Y1 Stay Thermistor + Dielectric Sensor,RWIS2016,RWIS2013, KTOL, KTDZ 03/12/2014 Y2 KTOL, KTDZ 03/29/2014 Y1 KTDZ 04/03/2014 Y2 Stay Thermistor + Dielectric Sensor, Stay Thermistor + Rain Bucket, KTDZ, Ice Detector 04/15/2014 Y1 KTOL, Stay Thermistor + Rain Bucket *Note: The weight of the station for leaf wetness sensor + stay thermistors was reduced to 0.1 in early January, thus reducing false alarms. Section 8.7.1: Ice Events (Winter 2013/2014) Thanks to the polar vortex/jet stream, this winter provided decent opportunities for the icing sensors to be tested out in the real world where there is dust, debris and reduced access to maintenance. There were several days where the temperature was way below freezing temperature, and there was ample precipitation. We had multiple ice accumulation incidents; however, there were no prominent instances of ice shedding. Some of the significant days are enumerated below:
236 December 9, 2013 The first time this winter layer of ice of significance was found on the stays was on December 9. Matt Harvey, ODOT, checked the stays at 2:10 am and reported thin layer of ice on all cables. Temperature of cables was 24-227°F. The air temperature was 31°F. There was light freezing rain and light snow. Again at 10 am, he reported, “As of 9:32 a thin layer of ice in on the top 2/3 of the cables. No ice on pylon.” David Canavel, ODOT, cleared the bridge of any ice by visual inspection in the afternoon. During December 9, 2013 there was ample precipitaation at freeezing temperatures as recorded by the leaf sensor. Although mostly quiet all winter, the ice detector did record up to 0.12 inches of ice on the day of the icing event. It did not trigger the dashboard, although the threshold was set at a much lower value of 0.05 inches. There were some questions that arose about the dissappearancce of the layer of ice off the sheath, whether it could be attributed as shedding, melting, evaporation or sublimation. No chunks or fragments of ice were observed underneath the stays. However, both the leaf wetness sensor and the ice detector weere impressive in the sense it seemed that plots of both the graphs (as shown below) were showing both increasing trend until about 3 am, which coincides with "snow" data both at the airports and observations by ODOT. Similarly, both sensors showed deecreasing values thereafter, which coincides again with the airports not reporting "snow" and observations by ODOT on the ground.
Figure 233: Ice Detector & LWS Characteristics during Ice Event, December 9, 2013
237 December 13 & December 14, 2013 The rest of the month of December saw few minor accumulation events. A lot of ice accumulation alerts were triggered between December 13 and December 14, 2013. Initially it was just the local RWIS2016 sensor that reported freezing rain. This was followed by consistent simultaneous reporting of high dielectric and sub-zero temperature by the leaf wetness sensor and the stay thermistors respectively. The dashboard reported level Y3 at 4 pm on December 14. This was a significant cause for concern and called for some visual report, this being among the first events for the new ice sensors. However, Matt Harvey, ODOT reported, “Found no ice, very little snow on cables. Cable temperature at 4:47 am was 19.5-23°F. Pylon temperature was 21°F”. January 5, 2014 There were just a few sporadic triggers on the 1st of January, but January 5th marked the first true ice event of the year 2014, as there was a lot of ice accumulation warnings from the dashboard on this day. The local RWIS2016 reported freezing rain in the previous evening, which was followed by snow as reported by the Toledo Metcalf Airport. The leaf wetness sensor also triggered higher dielectric. The dashboard stayed at Y2 for three hours. January 10, 2014 On this day the icing alarm was triggered by the local RWIS2016 station, and the Leaf Wetness Sensor with stay thermistors. Initially freezing rain was reported by RWIS2016 which was followed by snow. The dielectric LWS reported higher dielectric. It was, in fact one of the days when it recorded about 800-900 mV dielectric which usually signifies rainfall. The rain bucket tipped twice in long intervals, and recorded lower precipitation than its threshold (set at 0.05 inches) thus failing to set its flag to true for the ice accumulation trigger. Trace of ice was measured by the ice detector as well, although that didn’t trigger the alarm. According to National Weather Service, there were 0.46 inches of precipitation and 0.6 inches of snow. February 20, 2014 The dashboard started reporting icing conditions since 1 o’clock in the morning. This was triggered by the local RWIS sensor 582016 which reported freezing rain. At 6 am the status moved to Y2 and continued staying at Y2 for 3 more hours. The RWIS 2016 stopped reporting freezing rain activity and the status came down to Y1 at 9 am. This time both the airports signaled freezing rain conditions. The wetness sensor also reported higher dielectric.
238
Figure 234: VGCS Icing camera view before noon On this day the ice detector measured 0.06 inches oof ice, but interestingly, this did not trigger the dashboard although the threshold was set at 0.05 inches. However, this was not a case of missed detection as there was just this one spike above the threshold, not satisfying the condition of having enough records in one hour to set the icing flag. Instead, the ice thickness quickly fell to 0.01 inches. Since the heat time recorded by the sensor was zero (its de-icing threshold being set at 0.15 inches), shedding happened right on the device probe naturally. Wind or bridge vibrations may be a possible cause. The outer stay thermistors dwelled under 32°F longer than other thermistors possibly
239
Figure 235: Ice Detector & Leaf Wetness Sensor characteristics on February 20 March 12, 2014 The March 12 ice accumulation event was mostly driven by fog in the morning. According to predictive analysis by Dr. Nims, fleckss of ice and then a slight build up on the top was expected. Until 2 pm, the snow was dry enough that it was not sticking to the sides of the VGCS or PMB specimens even though the wind was steady. With progression of time, and the temperature dropping and the snow slowing down, both the sticking efficiency and the volume of snow in the air was getting lesser. The airports continued to report cold and fog which is what had been triggering the dashboard alerts. The ice detector didn’t show icing. The cameras appeared to be snow covered and hence were inconclusive as well. April 3, 2014 The dashboard recorded its first Y1 at 11 am and Y2 at 5 pm. The ice dettector BF5 continued to register no ice. The RWIS & METAR weather stations recorded air temperature staying higher than the stay temperatuure. Temperature of the stays dipped and precipitation continued. Precipitation was confirmed in uniison by both the leaf wetness sensor and the rain tipping bucket. The airport at Toledo Metcalf also reported some freezing rain. As a result, the dashboard went up to Y2. If this continued, there was a strong possibility of reaching Y3 status. Personal communication was made between UCII and UT researchers. According to Dr. Douglas K Nims, University of Toledo, "I spoke with Dave Kanavel. He had checked the stays about 3 pm and 3:45 pm. He observed a thin layer of ice that did not wrap around the stays (~1/32') at 3 pm and it was gone when he crossed back over at 3:45 pm. The
240 ice extended about 30 feet up the stays. The ODOT weather service forecast is for the temperature to rise to 40°F by midnight." Some shedding was observed as well later this day. Shedding was observed. According to Clint Mirto, student, University of Toledo, icing was occurring on VGCS. As large as a five feet piece of ice was reported to fall off around 4 pm. This was another event of significance as the ice detector triggered the alarm the only time this winter. It recorded 0.16 inches of ice at 11:50 am. The deicing threshold was set at 0.15 inches and the heater melted the ice, but it recorded 0.13 inches of ice again at 1 pm. Interestingly, the ice shed off the probe, but not because the heater went into de-icing cycle this time. The ice detector recorded zero seconds of heating time which means the heater did not get turned on. The shedding of the ice on the probe can possibly be attributed to natural causes like wind or vibrations. The shedding/melting of ice on the stays around 3 pm – 4 pm can be attributed to the rising temperature of the stays above 32°F and little peaks of solar radiation (~120 Watt/m2). This was a good day for observation of the heated rain/snow gage performance. The rain bucket too tipped constantly several times during the heavy showers. According to National Weather Service, there was 1.15 inches of precipitation on April 3rd, 2014. April 15, 2014 April 15th has been marked as the last day of icing alert recorded this winter season. There were two alerts of Y1. This was also the second time the rain tipping bucket was the cause of an alarm. The rain bucket tipped few times between midnight and 3:20 am. The local airport at Toledo (KTOL) also set the alarm due to snow. The ice detector measured very mild accumulation after midnight. According to National Weather Service, there was 0.13 inches of precipitation on this day.
241
Figure 236: Ice detector & Leaf wetness Sensor characteristics on April 3
Figure 237: Rain Tipping Bucket & Leaf Wetness Sensor characteristics on April 3
242
Figure 238: Solar Radiation & Stay Thermistor 8X08TWS characteristics on April 3 Table 49: Web Report Tool: Sample Icing Events and Comments, Deecember 2013
243 Section 8.7.2: Sensor Performance All the sensors (remote RWIS/METAR stations as well as the new local sensors) gave us useful information this past winter. There were several instances of ice accumulation alarms and we can say, it has been but a busy winter at the Veteran’s Glass City Skyway bridge. However, in spite of the added sensors, there were cases where we had false alarms, possible missed detections of shedding, as well as loss of data during a large power outage in January 2014. It was observed that the majority of the ice accumulation events were triggered by the leaf wetness sensor and the local RWIS station at VGCS (RWIS2016). A good number of these were false alarms; the RWIS2016 sensor broke down a few times. The RWIS 582016 has had a history with its rain sensor being stuck during freezing rain. This year RWIS 2013 and KTDZ also were down a few times, not providing data seamlessly. Similarly, the local sensors albeit more dependable, were not perfect either. Dielectric wetness sensor The dielectric leaf wetness sensor triggered the dashboard many times, some of which were false triggers. This has to do with the threshold being used in the algorithm mostly. As the sensor is sensitive to very low quantity of moisture, raising its threshold in the algorithm or forming a band of freezing conditions(290-310 mV) after high dielectric activity( >400 mV) can help reduce these false alarms. However, it gave us vital triggers on events such as December 14 and February 20th when the ice detector did not measure any ice. Due to its high frequency of false alarms, the weight of the sensor station was lowered from 0.3 to 0.1 in the ice accumulation algorithm in early January following which this problem was resolved. Nevertheless, the current threshold used for the LWS leaf wetness sensor needs to be adjusted apropos to visual inspection and more training data in the future. Stay Thermistors The stay thermistors are beyond doubt some of the most crucial sensors on the bridge. This is because the icing algorithm requires temperature data in addition to the precipitation data provided by sensors like leaf wetness sensor and rain tipping bucket. The stay thermistors provided valuable data for the stay temperature throughout the winter. This data was necessary as knowledge of stay temperature is more instrumental for quicker determination of icing events which the remote RWIS/METAR data do not provide. The thermistors placed at different positions and directions gave us appropriate data of temperature of the stays. During melting conditions, it was observed that sheath thermistors got warmer sooner that the outer thermistors due to latent heat required to melt the ice. Here the sensor chosen 7X20TUS is the thermistor directly lying on the sheath while the sensor 7X20TUO is facing more outwards.
244
Figure 239: Leaf Wetness Sensor characteristics winter 2013/14
Figure 240: Stay Thermistor characteristics winter 2013/14
245
Figure 241: Sheath thermistors warming faster than outer (March 4, 2014) Rain Tipping Bucket The heated rain/snow tipping bucket mostly gave inaccurate results all winter as it was devoid of regular maintenance and was possibly down. The primary screen of the device needs to be removed during snowfall and reeinstalled after snowfall to protect it from dust and debris. Removal of the screen enables the snow to reach the heated filament and melt, so that the equivalent precipitation can be measured by the tipping gage. The tipping bucket inside consists of a hinge that also neeeds to be calibrated from time to time so that it tips accurately without getting stuck. The rain bucket always gave accurate data in the lab experiments, and has performed well by itself recently on icing events of April 3rd and April 15th, 2014.
246
Figure 242: Rain Tipping Bucket characteristics winter 2013/14 Ice Detector 0872F1 Although the most popular candidate for ice thickness determination, the ice detector hardly recorded much accumulation earlier this winter, as it is basically apt for fast ice accumulation through freezing rain than through slow snow accumulation as the snow glides down before freezing on its almost vertical probe. It has automated heating feature, which is crucial to its functions. We have had events where the ice on the sensor probe shed off naturally without the heater being turned on at the said threshold of 0.15 inches (February 20 & March 4 events) which may not have occurred due to automated heating. Deeper analysis needs to be done for events like that on December 9, 2013 when the algorithm had a case of missed detection although tthe ice detector measured 0.12 inches of ice. Since automated heating adversely affects the reading of ice thickness, more research needs to be done to use a suitable de-icing threshold and adjustment of the algorithm appropriately (for instance use cumulative ice thickness). In a few cases, the ice on the sensor probe got removed because of shedding probably due to natural causes like wind or vibration. This cases need to be investigatted. Also, since the ice may disappear due to automated heating or just plain shedding, ice thickness measurement should be done by measuring cumulative ice accretion on the probe. Besides measurement of ice thickness, the sensor ccould also prove valuable for shedding events.
247 New ideas were brought to the table about utilizing the sensor also for ice shedding vis- à-vis the reduction of ice on the probe. However, this brings to question the elimination of its automated heating which may affect its accuracy in measuring ice thickness and its lifespan and hence is subject to further discussion. It is important to note that the ice thickness indicated on the ice probe may not exactly equal that found on the stay sheath as they are two very different structures with different materials, size, shape and orientation; hence, some calibration of this must be achieved using field observations during the past and future events. More icing events and a larger training set will help determine efficacy of the sensor.
4
Figure 243: Ice Detector characteristics winter 2013/14 Sunshine Sensor BF5 Because of how solar radiation can affect ice shedding even at lower temperatures, this sensor is a great potential predictor of ice shedding. The solar radiation sensor gave us excellent data of year round sunshine, and we have set its sunshine threshold for icce shedding at 120 Watt/m2. However, the sensor’s effectiveness for ice shedding, its threshold and adjustment to the ice shedding algorithm can only be achieved once we have enough training data from shedding events. We are yet to receive a shedding event and decipher the threshold number of our sunshine sensor that could accurately trigger alarm for ice shedding event on stays.
248
Figure 244: Solar radiation Sensor characteristics winter 2013/14 Section 8.7.3: Issues and Observations from Winter Performance The effect of the new sensors can be further internalized by observing their frequency of occurrence in the icing alarms and is explained in the graph below:
Figure 245: Relative distribution of alarmss triggered by new sensors As we discussed before, the weight of the leaf wetness sensor had to be lowered due to its high frequency triggers. This problem will be ameliorated once we can settle on the
249 most appropriate threshold value for the sensor. The rain bucket being down most of the season also does not have a large share of the dashboard triggers. The ice detector set the alarm off only twice as it has a high threshold of 0.05 inches. Since this device is primarily being used for ice determination and not thickness in the algorithm, we could either lower its threshold ( although about 0.2 inches ice or more is considered hazardous) or quickly send our warning level to Y2 (secondary alarm for accumulation). For example, the icing event of April 3rd occurred after only a few hours, which in comparison to previous events was relatively quickly. We did not reach Y3, which requires persistence of icing conditions for several more hours. One possible solution to this faster icing scenario is to define another threshold for the icing sensor that would push us to Y3 immediately regardless of the normal alarm algorithm. This would override the normal process in cases of fast accumulation of ice as on April 3rd. Similarly, for ice shedding sensors – stay thermistors and solar radiation sensor, the individual sensor threshold flag was set numerous times, but we do not have any useful data for deliberation as the overall shedding conditions were not met and hence the shedding alarms were not triggered. If we look back at table 17, there have been many events were the remote weather stations missed an event but the new icing sensors caught them and vice versa. There is no doubt that introduction of the new sensors have strengthened our dashboard algorithm by complementing & supplementing the existing remote weather stations with redundancy. The ultimate goal is to have as many sensors reporting similar icing conditions accurately and eliminate any requirement of human intervention. Despite their minor shortcomings which can only be improved by our learning of their long term characteristics, the new sensors have provided us much more redundancy and control over our dashboard algorithm. Tuning the dashboard algorithm most accurately would depend upon these sensors’ performance and their longevity, and these can be better validated with time and a larger training set. Section 8.8: Conclusions A new sensor suite has been deployed on the VGCS bridge. These include:
Geokon Thermistor 3800-2-2, for stay surface temperatures
Decagon LWS-L Dielectric Leaf Wetness Sensor, for detecting water/moisture with some classification as ice, light rain/snow, or heavy snow/rain.
Delta-T Devices Sunshine Sensor BF5, for detecting and quantifying solar radiation and approximate level of cloud cover
Met One Rain Tipping Bucket, for detecting and quantifying precipitation
Goodrich Ice Detector, for detecting and quantifying ice thickness
250 In addition, the dashboard has been modified to account for and utilize these new sensors for predication of ice accumulation and shedding. The following objectives for phase II have been successfully implemented.
New module has been added in the algorithm to test the functioning of all the remote weather stations. This monitor of the monitor concept can be viewed through a new table in dashboard main panel which reports station status as green (working) or red (not working). To increase the speed, redundancy and accuracy of the algorithm we installed different icing sensors such as stay thermistors, ice detector, dielectric leaf wetness sensor, rain tipping gage, solar sensor etc. on the bridge location itself In order to decide ice events more accurately, there are some parameters that are critical, such as solar radiation, type of wetness, thickness of ice etc. New sensors now provide us this data. As we have active control over our data logger programs scheduling data collection, we can now change different thresholds, sampling rates etc. as per our requirements. As we required a much faster system to catch quick ice shedding incidents, our local sensor system can now collect data in seconds, if required.
251 Chapter 9: Ice Presence and State Sensor Development Section 9.1: Introduction There is no commercial available sensor for the ice presence and state. There are two primary motivations for developing this sensor. 1) The Goodrich ice detector cannot sense ice persistence. The ice detector can detect the accretion of ice and by using an integration strategy to adjust for the ice detector heating cycles can estimate the initial thickness of the ice (Ryerson has a paper on this. It could be cited here.). However, the ice may sublimate or fall off the sensor tip before it releases from the stay. Thus, the ice detector has limited utility in revealing ice persistence. 2) Liquid water beneath the ice is a precursor to ice shedding. Coupled with stay surface temperature from the stay thermistors, recognizing the presence of water beneath the stays is expected to be a powerful predictor of imminent shedding. Such knowledge would give ODOT time before a shedding event to change the traffic patterns. Therefore, an effort to develop an ice presence and state sensor was undertaken. A simple resistance based sensor was developed. Laboratory and outdoor tests were successful in differentiate the state of water on the sensor as it changed leading up to an experimental ice shedding event. The sensor cannot reliably provide an indication of the thickness of the ice. The sensor is ready for deployment. Section 9.2: Ice Presence and State Sensor Laboratory Testing The Ice Presence and State Sensor (hereafter referred to as the “UT icing sensor”) was successfully developed using the differences in the electrical resistance properties of water, ice, and slush. Pure water will not conduct electricity to any measurable degree. However in naturally occurring water, the impurities in the water permit some conduction. Ice can also conduct electric current like water. But the conductivity is much reduced because the ion motion through the solid is thousands of times smaller than the motion of the same ions in the liquid water. Therefore, water is a much better conductor than ice. Finally, slush, which is the mixture of water and ice, is also conductive. It does not allow conduction as well as water, but permits much more conduction than ice. By using this basic property (conductivity), this sensor can detect the state of a medium whether it is water, ice, or slush. Section 9.2.1: Sensors and Data Acquisition System Conductivity is the reciprocal of resistivity: conductivity = 1/resistivity. Therefore, it is inversely proportional to resistance. UT icing sensor is developed to detect the resistance of a medium present on the surface of the sensor. Figure 246 shows the schematic circuit of the UT icing sensor. R1, R2, Eo, and Ei represent the fixed resistor, variable resistor, output data, and input voltage, respectively. R2, also called electrode spacing, is the area where sensor detects the state of a medium as seen in Figure 247.
252 Eo is connected to data acquisition system which transfers data to a computer. Ei is connected to a power supply which provides a direct current of 5 Volts.
Electro Spacing
Figure 246: UT Icing Sensor Circuit
Thermocouple
Variable Resistor (R2)
Figure 247: Electro Spacing Area of the UT Icing Sensor As seen in figure 247, a K-type thermocouple is attached to the UT icing sensor. This thermocouple plays a very important role in determinning the state of a medium. The criteria has been created as if the temperature of the medium is below -2 °C, it is considered to be ice. If the temperature of the medium is above 2 °C, it is considered to be water. And finally, if the temperature of the medium is between -2 and 2 °C, it is considered to be slush. National Instrument devices were chosen for the datta acquisition system for the UT icing sensor as shown in Figure 248. The temperature data and resistance data are transferred through NI USB-TC01 Thermocouple Measurement and NI USB-6210 devices respectively. Both devices are connected to a LabView program which interprets and shows the data on its dashboard as shown in Figure 249. In the dashboard, the high temperature limit for ice and low temperature limit for water can be set according to the rule. It also shows the possible state of the medium whether it is water, slush, or ice. The rules are mentioned before, if the temperature of the medium is below -2 °C, above 2 °C, and between 2 and -2°C, it is considered to be ice, water and slush respectively. In
253 addition, the dashboard presents the real time plots of resistance (kOhm) and temperature (°C) according to time (s). The dashboard is also automatically saved the data into an excel files once the user clicks on the stop button.
Direct Current Power NI USB-6210 Supply UT Icing Sensor
NI USB-TC01
Figure 248: UT Icing Sensor Connected to Data Acquisition System
Figure 249: Dashboard of UT Icing Sensor Section 9.2.2: Design of Experiments The purpose of the initial experiment was to measure resistances for three mediums; water, slush and ice. All measurements are done in a controlled room temperature at 22 °C. The resistance of ice is known to be the highest of all, on the other hand, water is known to have lowest resistance. The experiment is done using two UT icing sensors. Figure 250 and 251 shows UT icing sensor with 1-mm and 7-mm electro spacing respectively.
254 7 mm 1 mm
Figure 250: 1-mm Electro Spacing UT Icing Figure 251: 7-mm Electro Spacing UT Icing Sensor Sensor The UT icing sensor and thermocouple were glued onto the bottom of a plastic container. Then, ice, slush, and water are filled in the container to measure the resistance of each medium. Figure 252 and 253 shows water and ice measurement respectively. Other than measuring the resistance of water and ice, slush’s resistance is also to be measured. Water and ice are blended in the curtain ratio to make slush. This ratio is corresponding to weights of ice and water. In this experiment three types of slush are tested; 75% slush, 50% slush, and 25% slush. Figure 254 shows 75% slush which is the mixture of 75% ice and 25% water. This type of slush has the leeast liquid water content (LWC) of all the slushes. Figure 255 shows 50% slush which is the mixture of 50% ice and 50% water. This type of slush has medium LWC. And finally, Figure 256 shows 25% slush which is the mixture of 25% ice and 75% water. This type of slush has the most LWC of all the slushes.
255
Figure 252: Water Measurement Figure 253: Ice Measurement
Figure 254: 75% Slush Measurement Figure 255: 50% Slush Measurement
Figure 256: 25% Slush Measurement Furthermore, the measurements were made for three different thicknesses in each medium: 6 mm (0.25 in), 13 mm (0.50 in), and 19 mm (0.75 in)). Figure 257 to 259 shows examples of measuring resistance of ice for three thicknesses. The blue lines repressent the thicknesses and the medium needs to be filled up to the line to get an accurate measurement. The similar fashion measurements were also done for water and slush.
256
Figure 257: Ice Measurement at 6 mm Figure 258: Ice Measurement at 13 mm thickness thickness
Figure 259: Ice Measurement at 19 mm thicknness
Section 9.2.3: Laboratory Test Results UT Icing Sensor with 1-mm Electrode Spacing. The first set of resistance measurements was done using UT icing sensor with 1-mm electrode spacing, the sensor can be seen in Figure 250. The results are shown below.
257 4000000
3500000
3000000 ) 2500000 Ohm () (
2000000 6 mm Thickness 13 mm Thickness 1500000
Resistance 19 mm Thickness 1000000
500000 ~200000 0 ‐10 ‐8 ‐6 ‐4 ‐2 Temperature ( C)
Figure 260: Resistance of Ice for 1-mm Electro Spacing Sensor The figure above shows comparison of resistances of ice in three different thicknesses corresponding to temperatures. Three measurements were donne while letting the ice melt at room temperature. The ice which has a temperature of -2 °C has resistance as high as approximately 3,500 kOhm and as low as approximateely 200 kOhm. Figure 261 shows the screenshot of the monitor while measuring the resisstance of ice. The “possible state” section shows that ice is detected (as indicated by the red arrow).
Figure 261: Dashboard Screenshot of Ice Measurement
258 80000
70000
60000
50000 (Ohm)
40000 75% Slush 6 mm Thickness Resistance Drop 75% Slush 13 mm Thickness 30000
Resistance 75% Slush 19 mm Thickness 20000
10000
0 0 20 40 60 80 100 120 Time (s)
Figure 262: Resistance of 75% Slush for 1-mm Electro Spacing Sensor Figure 262 shows comparison of resistances of 75% slush in three different thicknesses corresponding to time. Three measurements were done while letting slush melt at room temperature for 120 seconds. The slush which was at a temperature range of -2 and 2 °C has resistance as high as approximately 70 kOhm and as low as approximately 10 kOhm. The resistance drop seen in the graph shows that the sensor detects part of slush that has a higher liquid water content (LWC). Figure 263 shows the screenshot of the dashboard while measuring the resistance of 75% slush. The possible state section shows that slush is detected.
Figure 263: Dashboard Screenshot of 75% Slush Measurement
259 1400
1200
1000 (Ohm) 800 Resistance 50% Slush 6 mm Thickness 600 Drop 50% Slush 13 mm Thickness 400 Resistance 50% Slush 19 mm Thickness 200
0 0 20 40 60 80 100 120 Time (s)
Figure 264: Resistance of 50% Slush for 1-mm Electro Spacing Sensor Figure 264 above shows comparison of resistances of 50% slush in three different thicknesses corresponding to time. Similar to the previous experiment, three measurements were done while letting slush melt at room temperature for 120 seconds. The slush which was in a temperature range of -2 annd 2 °C has resistance as high as approximately 1,300 Ohm and as low as approximattely 250 Ohm. It also shows resistance drop which means the sensor detects the part of slush that had higher LWC at that time. The 50% slush of 13 mm thickness shows low resistance because the sensor exposes to a very high LWC. Figure 265 shows the screenshot of the dashboard while measuring the resistance of 50% slush. The possible state section shows that slush is detected.
Figure 265: Dashboard Screenshot of 50% Slush Measurement
260 300
250
200 (Ohm)
150 Resistance Drop 25% Slush 6 mm Thickness 25% Slush 13 mm Thickness 100
Resistance 25% Slush 19 mm Thickness
50
0 0 20 40 60 80 100 120 Time (s)
Figure 266: Resistance of 25% Slush for 1-mm Electrode Spacing Sensor Figure 266 shows comparison of resistances of 25% slush in three different thicknesses corresponding to time. Three measurements were done while letting slush melted in a room temperature for 120 seconds. The slush which is in a temperature range of -2 and 2 °C has resistance as high as approximately 270 Ohm and as low as approximately 250 Ohm. The resistance drop seen in the graph shows that the seensor detects part of slush that has higher LWC. The 25% slush at all thicknesses shows low resistance because this type of slush has very high LWC. Figure 267 shows the screenshot of the dashboard while measuring the resistance of 25% slush. The possible statte section shows that slush is detected.
Figure 267: Dashboard Screenshot of 25% Slush Measurement
261 250
200 (Ohm) (Ohm) 150
Water 6 mm Thickness 100 Water 13 mm Thickness
Resistance Resistance Water 19 mm Thickness 50
0 0 20 40 60 80 100 120 Time (s)
Figure 268: Resistance of Water for 1-mm Electro Spacing Sensor The graph above shows comparison of resistances of water in three different thicknesses corresponding to time. Three measurements weree done in a room temperature for 120 seconds. The water which is in a temperature range above 2 °C has resistance as high as approximately 207 Ohm and as low as appproximately 188 Ohm. In this case, there is no presence of resistance drop because water has uniform property throughout its body. Figure 269 shows the screenshot of the dashboard while measuring the resistance of water. The possible state section shows that water is detected.
Figure 269: Dashboard Screenshot of Water Measurement
262 UT Icing Sensor with 7-mm Electrode Spacing. The second set of resistance measurements was done using UT icing sensor with 7-mm electrode spacing, the sensor can be seen in Figure 251. The same procedure was performed as 1-mm electrode spacing sensor. The results are shown in the following figures.
6000000
5000000
4000000 (Ohm)
3000000 6 mm Thickness
2000000 13 mm Thickness
Resistance 19 mm Thickness 1000000
0 ‐8 ‐7 ‐6 ‐5 ‐4 ‐3 ‐2 Temperature ( C)
Figure 270: Resistance of Ice for 7-mm Electro Spacing Sensor
200000 180000 160000 140000 (Ohm)
120000 100000 75% Slush 6 mm Thickness 80000 75% Slush 13 mm Thickness 60000
Resistance 40000 75% Slush 19 mm Thickness 20000 0 0 20 40 60 80 100 120 Resistance Drop Time (s)
Figure 271: Resistance of 75% Slush for 7-mm Electro Spacing Sensor
263 700 600 500 (Ohm) 400 50% Slush 6 mm Thickness 300 200 50% Slush 13 mm Thickness
Resistance 100 50% Slush 19 mm Thickness 0 0 20 40 60 80 100 120 Time (s)
Figure 272: Resistance of 50% Slush for 7-mm Electro Spacing Sensor
350
300
250 (Ohm) 200 25% Slush 6 mm Thickness 150 25% Slush 13 mm Thickness 100
Resistance 25% Slush 19 mm Thickness 50
0 0 20 40 60 80 100 120 Time (s)
Figure 273: Resistance of 25% Slush for 7-mm Electro Spacing Sensor
250 245 240 235
(Ohm) 230
225 Water 6 mm Thickness 220 Water 13 mm Thickness 215
Resistance Water 19 mm Thickness 210 205 200 0 50 100 Time (s)
Figure 274: Resistance of Water for 7-mm Electro Spacing Sensor
264 The result of 7-mm electro spacing sensor shows very similar trend compared to the previous experiment. The resistance of ice is the highest of all, above 400000 Ohm. On the other hand, the resistance of water is the lowest and that of slush is in between. The slush’s resistance depends on its LWC. The result indicates that the higher LWC, the lower the resistance as seen in the result of 25% slush having lower resistance than 50% and 75% slush respectively. Figure 275 is provided to present resistances of all mediums at the 6 mm thickness and 7 mm electro spacing. It clearly shows that each medium has its own resistance range and this trend is similar for other cases.
10000000
1000000
100000 Ice (Ohm)
75% Slush 10000 50% Slush
Resistance 25% Slush 1000 Water
100 0 20 40 60 80 100 120 Time (s)
Figure 275: Resistances for 6-mm Thickness and 7-mm Electro Spacing Sensor However, the results show that the resistance is not dependent on the thickness of a medium. By analyzing Figures 270 to 274, it appears that the thickness does not play a significant role in the medium’s resistance. Overall, the result from this experiment is very valuable and was be applied to the full scale experiment which will be discussed in the next section. Section 9.3: UT Icing Sensor in Full Scale Experiments Since the performance of the UT icing sensor in the laboratory was promising, the sensor was installed in a full scale experiment. The goal of experiments was to simulate the ice and wet snow formulation on the specimens and track the behavior of those mediums in the more realistic environment. The full scale experiment was initially designed in winter 2012-2013. For the efficiency and safety issues, the experiment station was located at the Scott Park campus of the University of Toledo instead of at the VGCS. Three 10 ft-long cable stay sheath with the same diameter and material as the VGCS stays have been set up as shown in Figure 276. In winter 2013-2014, a high density polyethylene (HDPE) specimen from another bridge was added to the experiment station as seen in Figure 277. Ice and wet snow experiments were performed at this experiment station.
265
Figure 276: VGCS Stainless Stteel Specimens
HDPE Specimen Frame Structure
Figure 277: HDPE Specimen and Frame Structure Section 9.3.1: Specimens and Data Accquisition System Setup At the experiment station, the three VGCS specimens have been set up on a concrete pad. All three specimens are align from North to South and two of them are supported by concrete blocks to have approximately 30 degree angle to the ground as seen in Figure 276 representing semi-steep stay-cables at VGCS bridge. These two supported specimens are facing in different direction, one facing North and the other facing South, as same as the orientation of the VGCS bridge. The sun radiation has an effect on this orientation which will be discuss in the result of this chapter. Figure 278 shows 120 un- tensioned strands placed inside the North facing specimen. This experimental setup is to simulate the bridge conditions as realistically as possible.
266 Strands
Figure 278: North Facing Specimen with 120 Stands Inside The HDPE specimen had a diameter of 20 centimeters. The HDPE specimen was set up in a similar orientation as the South facing VGCS specimen. Figure 277 shows the HDPE specimen supported by concrete blocks so it makes approximately 30 degree angle to the ground. The frame structure was built to be able to install sensors, such as ultrasonic and thickness sensors, for more data collection during the experiments. The UT icing sensors developed in a laboratory were installed on both VGCS and HDPE specimens. The sensors were installed approximately at the mid-span of both specimens as shown in Figures 279 and 280. Figure 280 also shows that the dielectric leaf wetness sensor is installed on the top part of the HDPE specimen. Four sets of the UT icing sensors are placed on East, Top, West, and Bottom sides of each specimen, Figure 281 represents the cross section of a specimen and its sensors setup orientation. The reason the sensors were set up as shown is because the thermal properties are need to be discovered along the circumference of the specimens while ice or wet snow occurs. This information will be very helpful to determine ice or snow accretion and shedding. Figure 282 shows the UT icing sensor attached to the surface of HDPPE specimen.
267 UT Icing Sensors LWS
Figure 279: Sensors Setup on VGCS Figure 280: Sensoors Setup on HDPE Specimen Speccimen
Top
West East
Bottom Figure 281: Cross Section and Sensor Setup Orientation of both Specimens
UT Icing Sensors
Thermocouple
Figure 282: UT Icing Sensor on HDPE Specimmen
Specimen
The UT icing sensors tested in the full scale experiments were the same as the ones used in the laboratory. However, a different the data acquisition system was used. The
268 Scott Park data acquisition system recorded voltage rather than resistancce. The new MicroStrain data acquisition system were selected because it is wireless. The wireless feature is very convenient for this type of experimennt. Figures 283 and 284 shows MicroStrain V-Link and TC-Link respectively. The same UT icing sensor circuit built for the laboratory test is connected to MicroStrain V-Link directly. This data acquisition system provides 7 mVDC and was set to read the voltage output from the UT icing sensor circuit. The commercial LWS which has similar electrical mechanism as the UT icing sensor is also connected to the MicroStrain V-LLink to be able to compare the result data of both sensors. The MicroStrain TC-Link is a wireless thermocouple data acquisition system. It works with any type of thermoccouple in which K-type thermocouples are chosen for this experiment.
Figure 283: MicroStrain V-Link Figure 284: MicroStrain TC-Link According to the schematic circuit of the UT icing sensor shown in Figure 246, R1, R2, Eo, and Ei represents fixed resistor, variable resistor, output data, and input voltage respectively. The equation of the sensor circuit is shown below.
R1 and Ei are designed to be fixed which in this case, there are 1 kOhm and 7mV respectively. The output voltage (Eo) is depended on what detected on the sensor (R2). The equation before can be further simplified as:
1
269 The final equation above shows that when Ei and R1 are fixed, Eo and R2 have an inverse relationship. According to the circuit, there are four cases can occur: 1) If the sensor detects nothing (air) which has resistance close to infinity, the output voltage will be the lowest. 2.) If the sensor detects water which has resistance as low as 200 Ohm, the output voltage will be the highest. 3.) If the sensor detects ice which has resistance is as high as 2 mega Ohm, the output voltage will be low but not as low as that of air. 4.) If the sensor detects slush (wet snow) which has been proven in the previous section that its resistance range is between that of ice and water, the output voltage will be in a range between ice and water as well. The output voltage indicated what was present on the surface of the UT icing sensor. That is the output voltage reflected the state of the water in the space between the electrodes. Because of this significant difference in voltage rannges, the state of the medium can be easily determined whether it is ice, water, slush (wet snow) or air.
Figure 285: MicroStrain WSDA-Base (Signal Receiver) Both V-Link and TC-Link send radio signals to a base station called WSDA-Base as seen in Figure 285. This base is connected to a computer via USB port and transfers the data collected to Node Commander Software. This software is consisted by a dashboard that presents real time data and configures the dataa acquisition system when needed. The collected data can be exported as an Excel file for further analysis. Section 9.3.2: Full Scale Outdoor Tests The UT icing sensor connected to MicroStrain data acquisition system was testing in a laboratory once more before deploying to a full scale experiment. The sensor was attached to a small container, shown in Figure 286, which later was filled with three
270 different states of medium (ice, slush, and water) to determine an output voltage of each state.
Figure 286: V-Link and UT Icing Sensor Figure 287 to 289 shows ice, slush, and water on the surface of the UT icing sensor while collecting sensor output data.
Figure 287: Ice Testing Figure 288: Slush Testing
Figure 289: Water Testing
271 The resultant data from this testing is presented in the Figure 290. Each medium shows a characteristic range of output. The range of water is very high, on the other hand, ice shows a very low range, but not as low as a dry surface (air between the electrodes). The output range of slush was between water and ice. Since it is a mixture of ice and water, it has a non-uniform property which results inn inconstant output as shown in the figure. As resulted, all four states clearly have different ranges of data output. Therefore, the presence of medium on the surface of the sensor can be easily determined by the ranges.
Figure 290: UT Icing Sensor Initial Test Section 9.3.3 Full Scale Experiments Result During winter of 2012 to 2014, multiple full scale experiments were done, including icing and wet snowing methods. Icing experiment on VGCS specimen which was done during the winter of 2012-2013 have been presented and discussed by Ali (Arbabzadegan 2013). This thesis focuses on the data collected via UT icing sensor from the icing experiment. In addition, wet snow experiments on HDPE and result which were done in the winter of 2013-2014 is also discussed in this thesis. Icing Experiments: The experiments were performed at the Scott Park site on cold nights, the ambient temperature was -2 to -15 °C. Water was misted slowly on the VGCS specimen as seen in Figure 291. This caused the laatent heat of transformation not to raise the specimen temperature above 0°C (Arbabzadegan 2013). Figure 292 shows the ice pattern accumulated on the specimen after misting water for total time of 8 to 10 hours. The ice thickness was measured to be approximately ½ inch.
272
Figure 291: Misting Water on VGCS Specimen
Figure 292: Ice Accumulation on VGCS Speccimen During this experiment, the data acquisition system monitored the behavior of the ice. Since the slush state was not taken in account, a higher fixed resistor (R1) was placed into the sensor circuit to expand the voltage range that could be sensed. Increasing the range permitted a more sensitive indication of the ice state. The range of ice state is wider than that found in the initial laboratory test and Figure 290.
273 40 20
35 15
30 10 UT Icing 25 5 Sensor) Thermocouple 20 0 (Icing
15 ‐5 Data
Raw 10 ‐10
5 ‐15 AB C D 0 ‐20 00:00 21:00 18:00 15:00 12:00 09:00 06:00 03:00 00:00
Time 02/17/2013 02/16/2013 02/16/2013 02/16/2013 02/16/2013 02/16/2013 02/16/2013 02/16/2013 02/16/2013
Figure 293: Stay Behavior in Icing Experiment The UT icing sensor and thermocouple data for icing experiment on February 16th, 2013 are presented in Figure 293. Part A in the figure shows the sensor output before starting the experiment. The icing sensor output was very low because it did not detect anything but air. The thermocouple showed the temperature of the stay which is as same as the ambient temperature. Part B focuses ice accumulation scenario. The cool water was misted starting around 5:00 AM, the UT icing sensor output shows suddenly increase due to the presence of water. Similar to the thermocouple, the temperature rises up to zero degree because of latent heat during the ice transformation. The UT icing sensor output and temperature then gradually dropped because the surface of the specimen was covered with ice. Even though the water was kept misting on the specimen and more ice built up, the sensors were not able to sense that because it can only sense what happen at the surface of the specimen. The misting was stopped before sunrise which was around 7:30 AM. Part C illustrates ice melting due to a sun radiation scenario. The sun rose around 7:45 AM. The sun radiation had significant impact on the specimen temperature. As sun rose, the specimen and increased as well. However, the ambient temperature was constant throughout that day at -2 °C. Around 10:00AM, ice sheet on the specimen started to melt due to the rise of specimen temperature. This created small water layer between ice sheet and the surface of the specimen. Again the thermocouple shows the latent heat during ice transformation. But this time, the ice turned to water. The UT icing sensor also clearly showed that it detected the presence of water. The change in specimen temperature caused the presence of ice and water throughout the day. However, the
274 specimen temperature was not high enough for shedding event. The ice sheet was still on the surface of the specimen at the end of the day. After sunset, the ambient temperature sudden decreased as shown in part D. This caused the specimen temperature to drop as well as. The output of the UT icing sensor indicated the presence of ice without water at the surface. Comparison of the ice accumulation monitoring and UT icing sensor data shows that the sensor reliably reports the presence of ice on the specimen surface. when no natural icing events occurred and the results appeared to be very similar. Section 9.4: Conclusion and Next Steps A sensor that can differentiate between ice and water on the VGCS stays has been developed. There is no similar commercial senor. The sensor is rugged and compact so it can be mounted directly on the stay. In an icing event, it will be covered by the ice and detect the presence of liquid water in the interstice between the ice layer and the stay sheath. When there is water in this space, ice fall is imminent. The sensor has been tested in the laboratory and on full scale mock-ups of the VGCS sheath outdoors. The next step is to deploy the sensor on the bridge and integrate it into the dashboard.
275 Chapter 10: Transition and Maintenance Section 10.1: Introduction In spring 2014, the dashboard had reached a stage of development where it was appropriate to transfer it to District 2. This transition was organized in a way that the District will able to operate the dashboard through the winter of 2014-2015 and beyond. The form of transition that was agreed to was a standalone computer at ODOT’s Northwood Output. The Northwood Outpost is the operations center for the VGCS. Shift supervisors. The shift supervisors work with the D02 Highway Management Administrator to make the decisions concerning the response to actual and potential icing events. Therefore, this arrangement puts the Dashboard information at the fingertips of the decision makers. The standalone system is a static system. The configuration was locked and operational at the time of transfer. Copies of all documents and the user manuals for the sensors have been given to ODOT. This user manual will be permanently archived with this report on the ODOT Office of Research Website the current link is http://www.dot.state.oh.us/Divisions/Planning/SPR/Research/reportsandplans/Pages/default.aspx. Loss of power or modifications to the system may result in loss of functionality. Section 10.2: Standalone Computer System The dashboard was frozen in the spring of 2014 and the version of the dashboard that was current at that time was transferred to a standalone system. The standalone system captures the basic functionality of the Dashboard in a desktop system. The local weather conditions and output from the sensors is going to the dashboard. Output from the camera is going directly to ODOT District 2. Figure 10-1 presents architecture of the standalone system. The “Back End” collects data from the local sensors on the bridge and the weather stations throughout the region. The bridge sensor data comes in via ODOT’s intranet. The regional weather data comes from Weather Underground. The data is operated on by the version of the algorithm that was current in spring 2014. The algorithm is described in Chapter 8.3. The “Front End” takes the output from the algorithm as well as the data stream puts them on the Dashboard. The algorithm output goes to the speedometer user interface and access to view the data directly is through the tabs visible on the Dashboard “Home Page”. The transition of the standalone system to D02 took place on June 5, 2014. The research team delivered the system to the Northwood Outpost. The research team and ODOT D02 IT configured the system and verified its operation Section 10.3: Maintenance ODOT is being trained in the operation of the dashboard and a dashboard user manual to support the transition was developed. The user manual has section on maintenance. The user manual is archived on the ODOT of Office of Research website page for final reports the “subject area” is “structures” the current link for the website is http://www.dot.state.oh.us/Divisions/Planning/SPR/Research/reportsandplans/Pages/default.aspx
276 . The maintenance section describes the activities necessary to keep the standalone system operating. The user manuals for the instruments mounted on the VGCS are stored on the standalone system and may be accessed through the Dashboard.
277
Figure 294: Flowchart for Stand Alone System Chapter 11: Conclusion, Benefits, Implementation and Future Work Under some winter conditions, ice forms on the cables stays of the VGCS. Ice accumulations have been observed at a thickness of 3/4”. The ice accumulation depends on the temperature, precipitation and duration of the storm. The accreted ice conforms to the cylindrical shape of the stay sheath. Thus, as the stays warm, the ice sheds in curved sheets. These curved sheets of ice then fall up to two hundred and fifty feet to the roadway below and may be blown or glide across several lanes of traffic on the bridge deck. The falling ice sheets require lane closures and could present a potential hazard to the traveling public. Section 11.1: Summary of Goals and Objectives The overall goal of this research was to assist ODOT in implementing an icing management procedure for the VGCS. This procedure may be active, passive or administrative. Initially, it was the team’s hope that an existing anti/deicing technology would be identified and implemented. With that in mind, phase I of the project began. After an initial review of anti/de-icing technology guided by the experts at the Army Cold Regions Research and Engineering Laboratory, it became apparent that no off-the-shelf or economically viable solution existed. Thus, the following objectives for the overall project were agreed on: 1) Identify available technologies and procedures that could be used to solve the VGCS icing problem. This requires an assessment the state of the art through a literature review and consultation with the icing experts and examining the advantages, disadvantages, and potential applicability of each identified technology on the VGCS. 2) Identify the most viable solutions. It was expected that the most practical solutions will be novel adaptions or combinations of existing solutions. For each viable solution, it was desired to develop a detailed description of the implementation, to define of required validation testing, (either in situ or offsite), to perform a benefit/cost analysis, to develop a budget for implementation and to define a time frame for implementation. 3) Develop a real-time icing condition monitor. This monitor should present data that tracks icing events in a format that is easy to understand and is useful for managing icing incidents. The monitor should also archive the data. Local condition data that is collected from the bridge should be used to increase the algorithm intelligence and error handling of the monitoring system. 4) Collect data to understand the microclimate and icing behavior of the bridge. The collected information should be sufficient to allow accurate costing of an anti/de- icing technology, resolve uncertainties to reduce the risk of deploying an icing strategy that does not work, and be useful for improving and updating the icing monitor.
279 5) Develop a sensor for determining ice presence and state. No suitable sensor exists. Therefore, development and field testing were undertaken 6) Transition the monitoring and local weather station to ODOT District 2 so that the functionality of the dashboard and the information from the icing sensors is available to the operators of the VGCS. Section 11.2: Results 1) Past icing events were reviewed, the mechanisms for icing where explored, and the basic conditions that are favorable to icing accretion and shedding were ascertained. The general weather system most often associated with major icing is warm air from the Gulf of Mexico overriding cold air from Canada. This leads to liquid water falling on a cold surface. Historically, roughly two icing events occur each year. Icing on the VGCS occurs when there is general icing in the area. There have been five major icing events on the VGCS. The last of which was in February 2011. Conditions are favorable for ice accretion when one of the following conditions occurs: iv. Precipitation with air temperature at the bridge below 32o F, or v. Fog with air temperature at the bridge below 32o F, or vi. Snow with air temperature at the bridge above 32o F. The ice accretion rate is generally slow because during an ice storm precipitation rates are low and much of the water runs off the stays. Therefore, it is expected that it will take over six hours to form an ice layer above the critical thickness of 0.25 in. However, in one instance in the spring of 2014 ice exceeded 0.25 in. in a much shorter time. Once the ice accretes on the stays and pylon, it may persist until shedding conditions occur. Temperatures above 32o F and/or solar radiation cause ice fall. The ice fall in four of the five previous events was accompanied by temperatures rising above 32o F. If the solar radiation level is high enough, ice can shed at temperatures below freezing (32 o F). If there is ice on the stay, weather conditions that cause ice fall are: iii. Air temperature above 32o F (warm air), or iv. Clear sky during daylight (solar radiation). In February of 2011, copious amounts of water flowing beneath the ice were observed when the outside temperature was several degrees below freezing. This layer of water below the ice is a precursor to ice shedding. There is a greenhouse effect that occurs in the interstice between the stay sheath and accumulated ice when the solar radiation passes through the ice and heat becomes trapped. Ice accretion and shedding do not occur simultaneously.
280 Given the unique features of the VGCS, the paucity of literature directly on point, and the urgency of addressing the problem, an expert team was selected. The expert team was the best way to quickly gain familiarity with the state of the art as well as define testing procedures and identify available test facilities. A research team with expertise in icing, icing instrumentation, icing test facilities, the bridge construction and bridge instrumentation was formed to address ice prevention and mitigation on the VGCS. The initial technology review was extended to a comprehensive review all technologies that could be identified regardless of their technology readiness level. A matrix of over 70 potential technologies was developed. The matrix describes the advantages and disadvantages of each technology. The broad categories of technologies are chemicals and chemical distribution systems; coatings; pneumatic or electrical expulsive deicing systems; modifying the design of the cables (such as by adding straking); radiant heating; interface heating; high-velocity air, water, or steam; mechanical or manual deicing methods; piezoelectric pneumatic systems; vibration, covers and robotic climbers. Ice detection systems were also reviewed. 2) To simulate icing events and use a test bed for experiments an icing field station was designed and built. It had three full scale sheath specimens ten feet long. One of these specimens included strand. The station had a local weather station and a wireless data acquisition. The initial set of experiments verified that ice accretion and shedding similar to that which occurs on the bridge could be replicated. The icing station was then used for experiments on anti/decing chemicals, anti-icing coating, heat for anti-icing and deicing, and tests of instruments. 3) The technologies that were the most viable were identified. They were: a. Deicing/anti-icing chemicals which would not present a biohazard when leached into the river such a sodium chloride; agricultural products, such as beet based deicers, and calcium chloride b. Anti-icing coatings c. Heat. The VGCS stays are mostly hollow so there is a potential to internally heat the stays. Experiments to evaluate the efficacy of each viable technology were carried out. The experiments were carried out in an icing wind tunnel and at an icing experiment station which was constructed for this project. The icing experiment station had full-scale stay sheath specimens that were 10 feet long. The anti-icing chemical experiments showed that on the stainless steel surface of the sheath the chemicals tested did not persist. Therefore, the onset of icing was only slightly delayed. The deicing experiments showed that the chemical tested was not viscous enough to sheet across the sheath surface. Rather it cut groves through the ice and was not effective in removing the ice. These results are consistent with the results in the literature. In addition, to not performing the
281 desired anti/deicing functions, chemicals would require a distribution system so they were deemed impractical. Several anti-icing coating were tested in the icing wind tunnel and at the icing experiment station. The coatings did not significantly delay the onset of ice, which stuck to the stay specimens and most did not change the shape of ice that shed. The coating that was outdoors for an extended duration of time became opaque and gummy, therefore, it would alter the appearance of the stays. These results are consistent with the results in the literature. Additionally, coating would be difficult to apply so they were deemed impractical. Introductory heating experiments were carried out at the icing experiment station. The heating was effective and deicing and partially effective at deicing. The temperature of the epoxy coated strand inside the sheath cannot be raised above 150 degrees F, so the initial thermal analysis showed the heating was a slow process. The requirement to heat each stay would require an expensive heating system. At that point, heating was deemed impractical so no advanced experiments or thermal analyses were conducted. No active or passive system was identified which had sufficient level of promise to justify detailed estimates of installation, operation or maintenance costs. 4) When it was judged that the regional weather information and the RWIS did not provide enough information to assess the microclimate and icing behavior, a local weather station was installed on the bridge. The weather station has five sensors and a weather proof camera. Thermistors are mounted on a stay in the back span and on a stay in the cantilever. A Goodrich ice detector, a solar radiation monitor, a dielectric leaf wetness moisture sensor, heated tipping rain gage and the weather proof, pan, tilt zoom camera are mounted on a weather tower of the south end of the back span. The combination of the existing sensors and the newly added sensors gives a good picture of the conditions on the bridge. Prior to deployment in the field, experiments on the sheathing specimens at the field station and in the laboratory coupled with the literature review lead to the conclusion that the proposed sensors functioned as desired and they were recommended for installation. It is especially valuable that the Goodrich ice detector can be used to provide an estimate of the thickness of the icing accumulation. The actual ice accumulation can then be compared to the critical thickness of one-quarter of an inch. The stay thermistors revealed the temperatures on the stays was significantly different than the RWIS or Metar data. They data both lag the stay temperatures by up to 3 hours or 20 F. This means air temperature is of limited use for assessing shedding. 5) To make the research immediately actionable by ODOT operations, a real-time icing condition monitor was developed. The research team designed a real-time
282 monitoring system to track icing conditions on the bridge with a straightforward interface so information on the icing of the bridge was readily available to the bridge operators. This monitoring system is referred to as the “icing dashboard” or simply “the dashboard” because the information necessary to support ODOT operations is presented on one simple visual display. The dashboard tracks the icing events in a format that is easy to understand, is useful for managing icing incidents and archives data. When conditions favorable to icing occur the dashboard alerted the research team. If the conditions favorable to icing persisted, ODOT was notified and, as required, requests for verification of ice accretion were made. The primary features that are implemented in the dashboard include:
User friendly check engine lights to monitor ice on the Bridge for: a. Conditions favorable to ice accretion b. Conditions favorable to ice shedding
Add weather data to existing VGCS web interface
Add new stay - mounted camera views to existing VGCS weather interface
Develop algorithm to monitor ice events
Develop reporting function for ODOT to verify the alerts and declare an event
Develop export function for historical data archive
Run calibration studies based on historical data The basis of this system is the smart mix of the automated algorithm and the visual observations, which helped aid in training the system for more optimal performance. The system uses an intelligent decision making process based upon initial criteria from past weather data analysis with parameter adjustments made after visual observations. Dashboard has done well in detecting ice accumulation each time, but the analysis done on the algorithm results and onsite observations from research team members and ODOT have been used to refine the algorithm as well as the interface. Additionally, local condition data and the review of recorded events that were collected from the bridge were used to increase algorithm intelligence and error handling. The improvements focused on the enhancement of the visual display, refinement of the accretion and shedding algorithms and incorporation of data for a local weather station on the bridge and reducing false alarms. The dashboard has proven to be a valuable resource for the bridge operators as well as a valuable tool for reviewing weather events. The automated ice
283 detection and monitoring dashboard for the VGCS was developed, implemented, successfully tested, and has been transferred to ODOT. 6) An ice presence and state sensor has been developed. No suitable sensor to detect the continued presence of ice or the transition from ice to water exists. Therefore, development and field testing of a suitable sensor were undertaken. The resistance based sensor detects the presence of ice and can differentiate between ice and liquid water. The sensor is designed to be mounted on the sheath and can detect the layer of water which forms beneath the ice just prior to shedding. The sensor has been tested in the laboratory and at the icing experiment station. 7) The transition of the dashboard to District Two has concluded. A local standalone computer with the dashboard on it has been provided to the District. The standalone version maintains the basic functionality of the dashboard algorithms and alert system and provides links to the icing weather instrumentation on the bridge. A person at the computer can monitor the conditions on the bridge and determine the causes of alerts. The dashboard is a complex system and to maintain the functionality of the dashboard the District should not make any changes to program. Section 11.3 Benefits The overall benefit is increased safety for the traveling public. The specific benefits of completing this project were:
The icing events in northwest Ohio for the past twenty years including the first four icing events on the bridge were reviewed. Vehicles were damaged in at least two of the first four icing events. All of the known anti/de-icing technologies were investigated. The included over 70 technologies. These technologies are described in the technology matrix. Observations and a detailed study of the February 2011 major icing event were completed. This was the fifth major icing event on the bridge since its opening. The bridge was closed for several hours. The team used the dashboard to capture the weather. Team members on the bridge were able to obtain video and images of the ice shedding. The data and images lead to increased understanding of the ice fall behavior. The explosive ice shedding was observed: all the ice fell from a stay in a minute or so. This produces a shower of large sheets of ice that can go completely across the road and into the Maumee River. The study of the past weather and icing events lead to quantitative guidelines about the weather conditions that made icing accretion and shedding likely. These guidelines form the core of the algorithms in the ice monitoring system implemented on the bridge. In response to ODOT’s request for a way to make the results of the research easily actionable by the operators of the bridge, a real-time monitoring system was implemented. This interface was design so it displayed information about the icing status of the bridge in a simple on screen format,
284 When it was identified that the existing sensor system on the bridge and in the surrounding region was not adequate to monitor the microclimate on the VGCS and icing conditions of the stays, it was decided that a suite of local sensors were required. A weather tower with local sensors and a camera as well as stay mounting brackets to attach thermistors directly to the sheath were designed and installed. The sensors were made operational and their data was incorporated into the dashboard. Experiments on anti/deicing chemicals, anti-icing coating and anti/deicing application of a heating system were carried out on full scale sheath specimens at the icing experiment station and in the icing wind tunnel at the University of Toledo. These studies coupled with the literature review demonstrated that no existing technology was appropriate for anti/decing on the VGCS. No commercial sensor for directly measuring the presence or state of ice on the sheath exists. An electrical resistance based sensor has been developed. The sensor detects the presence of ice and can detect the layer of water in the interstice between the stay sheath and ice, which is a precursor to ice shedding. This sensor has been tested in the icing wind tunnel and at the icing experiment station. The dashboard collects a comprehensive set of data from the regional and local sensors on the bridge. It records all the icing and shedding alerts, serves as a log for all the observations and has the capability of exporting and plotting the data.
In addition to the weather data, the dashboard serves as a repository of all references, reports, presentations and other documentation of this project. This allows convenient access to the information for ODOT and researchers. Section 11.4: Implementation The implementation has two primary parts: the local weather station and the dashboard. Uncertainties in the VGCS icing microclimate and the bridge icing behavior lead to the installation of a local icing weather and stay temperature sensing station. The existing RWIS sensors on the bridge were not targeted to icing behavior. They are as their name suggests focused on roadway conditions. The regional sensors remote from the bridge, such as the airport weather stations and RWIS not on the bridge, give valuable overall information, but shedding is very local to the bridge and the local conditions are critical. Therefore a local weather station was installed on the bridge. The local weather station has 5 sensor types and a camera: a solar radiation sensor, tipping bucket gage, leaf wetness sensor, an ice detector, thermistors and a weather proof pan-tilt-zoom camera. The regional sensors and the local weather station together give a good picture of the microclimate and general icing conditions on the bridge. The need to have easy to interpret actionable information about the icing environment on the bridge gave rise to the intelligent monitoring system or “dashboard”. The dashboard was placed into service in the winter of 2010-2011 and has been in service and regularly upgraded since then. In the winter of 2012-13, the stay thermistors were
285 added to the bridge and factored into the algorithms. In the winter of 2013-14, the instruments on the local weather station were added to the dashboard and the algorithms correspondingly revised. Once the dashboard was implemented, it was tested to check the fidelity of the system. Section 11.5: Transition and Long Term Maintenance The dashboard was transferred to District 2 on June 5, 2014. The transition was organized in a way that the District will be able to operate the dashboard through the winter of 2014-2015 and beyond. The system was transferred as a standalone system mounted on a single computer at ODOT’s Northwood Output. This arrangement puts the Dashboard information at the fingertips of the decision makers. The research team and ODOT D02 IT configured the system and verified its operation The standalone system captures the basic functionality of the Dashboard in a desktop system. The local weather conditions and output from the sensors on the bridge are going to the dashboard. The standalone system is a static system. The version of the dashboard that was current at that time was transferred. The configuration was locked and operational at the time of transfer. Loss of power or modifications to the system may result in loss of functionality. Copies of all documents and the user manuals for the sensors have been given to ODOT. This user manual will be permanently archived with this report on the ODOT Office of Research Website in the “structures” subject area. The current link is http://www.dot.state.oh.us/Divisions/Planning/SPR/Research/reportsandplans/Pages/de fault.aspx. ODOT was trained in the operation of the dashboard and a dashboard user manual to support the transition was developed. The maintenance section user manual describes the activities necessary to keep the standalone system operating. The individual user manuals for the instruments and data acquisition system mounted on the VGCS are stored on the standalone system and may be accessed through the Dashboard. Section 11.6: Archiving of Supporting Documents Copies of all documents and the user manuals for the project have been given to ODOT. The user manual and the complete collection of the photos and videos from the observations of the February 2011 icing event will be permanently hosted by ODOT on the Office of Research Website in the “structures” subject area. The current link is http://www.dot.state.oh.us/Divisions/Planning/SPR/Research/reportsandplans/Pages/de fault.aspx. Section 11.7: Recommendations for Future Work The research team desires to provide support to ODOT in order to minimize any possible hazard to the traveling public due to falling ice on the VGCS. Recommendations for future work fall into two categories: activities necessary to maintain the functionality of the dashboard and local weather station and activities that will improve the response to icing incidents.
286 Successfully, maintaining and operating the dashboard and local weather station so that it is useful to ODOT in the near term and long term require the following:
To ensure continued operation of the dashboard the following must occur: o Continuous power to the standalone system. o Continuous power to the sensors and loggers on the bridge. o Continuous internet connectivity to the standalone system. o Continuous internet connectivity to the sensors and loggers on the bridge. o All 3 servers Apache, MySQL and LoggerNet must be running at all times. o System clock should reflect the correct time (change between EST/EDT). o No updates on Software, Windows, no automatic updates. To ensure continued operation of the local weather station the following must occur: o The tower must be regularly inspected for structural integrity. o Instruments should be checked periodically and routine maintenance performed before each winter o The camera should be checked periodically and regularly maintained. o The local weather station connects to the ODOT intranet via the existing instrumentation backbone. The backbone and all associated loggers, multiplexers and power supplies need to be maintained.
To be useful in the long term, flexibility, portability and adaptive intelligence must be applied to the existing dashboard and local weather. More information about the condition of ice on the sheaths is recommended. Overall the recommendations can be broken into three groups: actions to enhance understanding of the ice conditions on the stays, actions to enhance the dashboard, and actions to maintain and upgrade the instrumentation backbone.
Actions to enhance understanding of the ice conditions on the stays: o The presence and state of ice on the stay are critical variable which are not sensed. The sensor the research team has developed for this purpose or an alternate commercial sensor, which may be developed in the future, should be considered for deployment. As no off-the-shelf senor exists, this will likely require some testing before deployment. Actions to enhance the dashboard. o No major icing event has occurred since the local weather station was installed. When a major event occurs, the data and observations should be reviewed and the accretion and shedding algorithms updated accordingly. o Because the transferred standalone system is frozen and complex, it will become obsolete. Arrangements to maintain the dashboard and access to the local weather station data in the long term are recommended. o The dashboard is large and complex. The operating system has over 1,000 files, 100 directories and is over 2 GB in size. As a result of this complexity, it is difficult to transition the dashboard with full functionality.
287 Restoring the full functionality of the dashboard, particularly making it accessible over the internet, will increase its utility to future researchers and operators. Consideration should be given to make the changes necessary to ensure the dashboard is more owner friendly. This may involve simplifying and streamlining the dashboard by identifying critical tasks and modifying the connectivity of the backbone. o Shedding occurs quickly. It may make sense to revise the refresh rate more often than hourly. o Allow a quick jump to an alert start if the ice detector indicates an ice accumulation of 0.25. o During an icing event the ice detector’s heater cycles. However, it is possible that by tracking the heater cycling and the accumulation between heating cycles to estimate the cumulative accreted thickness. o Incorporating some forecasting data is useful for reducing false alarms and estimating the significance of an event during accretion and predicting shedding. Adding some forecasting capability would increase the usefulness to the operators. o The dielectric leaf wetness sensor is very sensitive to moisture, alarming for all kinds of precipitation. Setting an upper bound for leaf wetness sensor could reduce false alarms. Actions to maintain and upgrade the instrumentation and data acquisition backbone. o The instrumentation and the data acquisition backbone will age and must be upgraded periodically. o Prior to each icing season the performance of the instrumentation should be reviewed and necessary repairs made.
288 Bibliography
“0872F1 Ice Detector”, February 2011, [Online], < http://s.campbellsci.com/documents/ca/manuals/0872f1_man.pdf>
AASHTO LRFD, 2010, AASHTO LRFD Buiding Design Specifications, 5th Edition with 2010 Interims. AASHTO Signs, 2011, AASHTO Standard Specifications for Structural Supports for Highway Signs, Luminaries, and Traffic Signals, Fifth Edition 2009, 2011 Revisions. ABCNews4 WCIV-TV. (2014, January 31). At long last: Ravenel Bridge completely reopens to traffic. - WCIV-TV. Retrieved May 19, 2014, from http://www.abcnews4.com/story/24599706/ravenel-bridge-closed
ACI 318-11, 2011, Building Code Requirements for Structural Concrete.
Adams, L., Weisend Jr, N., Wohlwender, T., Jul. 14 1992. Attachable electro-impulse de-icer. US Patent 5,129,598.
Admirat, P., 2008: Wet snow accretion on overhead lines. Atmospheric Icing of Power Networks, M. Farzaneh, Ed., Springer, 119-169.
Agrawal, S., 2011, Automated Ice Monitoring System for the Veteran’s Glass City Skyway. School of Electronics and Computing Systems of the University of Cincinnati.
Ackley, S. F., and K. Itagaki, 1974: “Crystal structure of a natural freezing rain accretion”, Weather, 29, 189–192.
Al-Khalil, K., 2007. Thermo-Mechanical Expulsion Deicing System - TMEDS. AIAA- 0692.
Antonio L., and Stefano M.; “Forecast: A Neural System for Diagnosis and Control of Highway Surfaces”, IEEE, Intelligent Systems and their Applications, May/June 1998, Vol.13, pp. 20-26
Baker, J. 2007, Photo credit - photos were taken at approximately 11 am on December, 11, 2007.
Battelle, 2010, Battelle Icephobic Coating PowerPoint Presentation, May 26, 2010.
Bennett, I., 1959: “Glaze: Its meteorology and climatology, geographical distribution, and economic effects”, Environmental Protection Research Division Tech. Rep. EP- 105, Headquarters, U.S. Army Quartermaster, Research and Engineering
289 Command, Natick, MA, 217
“BF5 Manual”, [Online], < ftp://ftp.dynamax.com/manuals/BF5_Manual.pdf>
Boselly III, S. E., Doore, G. S., Thornes, J. E., Ulberg, C. (1993), “Road Weather Information Systems: Research Report,” Strategic Highway Research Program, vol.1, National Research Council, Washington, DC, 1993. Bowers, P. (2014, February 4). Could the Ravenel Bridge have been winterized? Charleston City Paper [Charleston]. Retrieved from http://www.charlestoncitypaper.com/TheBattery/archives/2014/02/04/could-the- ravenel-bridge-have-been-winterized
British Broadcasting Corporation “Compensation claims after Severn bridge ice closures”, http://news.bbc.co.uk/2/hi/uk_news/wales/8537075.stm, accessed November 13, 2010.
CAV Aerospace, 2008, http://www.weepingwings.com/. Accessed November 15, 2010.
Chad Cheng S., Heather A., Guilong L., Joan K., Bryan t., and Qian L.; “An Automated Synoptic Typing Procedure to Predict Freezing Rain: An Application to Ottawa, Ontario”, American Meteorological Society, Vol. 49 Issue 5, May 2010
Claffey, K., K. Jones, and C. Ryerson, 1995, Use and Calibration of Rosemount Ice Detectors for Meteorological Research. Atmospheric Research, Vol. 36, pp. 277– 286.
Cortinas, J. V., Jr., 2000: “A climatology of freezing rain over the Great Lakes region of North America”, Man. Wea. Rev., 128, 3574-3588
Countryman Electric, "1998 Ice Strom." 2008. http://www.countrymanelectric.com/Generators/ElectricalOutages/98IceStorm/tabid/ 126/Default.aspx. Accessed February 10, 2011.
Couture, P., 2011, Smart Power Line and Phobic De-icer Concepts for Transmission- line Capacity and Reliability Improvement. Cold Region Science and Technology, Vol. 65, pp, 13-22.
Cox and Company, 2008, Electro-Mechanical Expulsion Deicing System (EMEDS). http://www.coxandco.com/aerospace/lowpower_ice_protection.html.
D'Amico, M., Jong, S.L., Riva, C., 2013, “Tipping bucket data processing for propagation application”, 7th European Conference on Antennas & Propagation, 256 – 260
Dalecky, A., K. Temple, and B. Tuccillo, 1996, Deicing Salt. Groundwater Pollution Primer, Civil Engineering Department, Virginia Tech.
290 Dalili, N., Edrisy, A., Carriveau, R., 2009. A review of surface engineering issues critical to wind turbine performance. Renewable and Sustainable Energy Reviews 13 (2), 428 – 438.
Dalle, B., and Admirat, P., 2011, Wet Snow Accretion on Overhead Lines with French Report of Experience. Cold Region Science and Technology, Vol. 65, pp, 43-51. Daniel, M. (2005, March 15). Bridge's falling ice called fluke of nature. Boston Globe [Boston].
Deb, B., 2013, Continued Weather Monitoring System for the Veteran’s Glass City Skyway. School of Electronics and Computing Systems of the University of Cincinnati.
Decagon LWS, 2010, Leaf Wetness Sensor. Brochure, www.campbellsci.com.
Delta-T Devices, 2002, Sunshine Sensor. Brochure, www.delta-t.co.uk.
Discovery Channel, 2006, The Deadliest Catch, Man vs Ice. Episode #19, Season #2, first aired 16 May 2006.
DNV, 2005, Managing cold climate risks. Classification News No. 3 2005, 09 June. Eikill, G.O., and B.C. Oftedal, 2007, Statoil ASARig design philosophy key for harsh environment drilling in Barents Sea, Haltenbanken area. Drilling Contractor, March/April 2007, pp. 83–87.
Dobesch, H, Nikolov, D and Makkonen, L 2005. 'PHYSICAL PROCESSES, MODELLING ANDMEASURING OF ICING EFFECTS IN EUROPE',
Druez, J., Louchez, S., McComber, P., 1995. Ice shedding from cables. Cold Regions Science and Technology 23 (4), 377 – 388.
Embry, G.D., R.W. Eskine, L.A. Haslim, R.T. Lockyer, and P.T. McDonough, 1990, Electro-Expulsive Separation System Shipboard Applications. Naval Engineers Journal, September, 1990, pp. 55–66.
EPA, 2000, Aircraft Deicing Operations. United States Office of Water (4303) EPA-821- R-00-016 Environmental Protection Washington, DC 20460, revised, 447 p. FAA, 2000, Hot Water Deicing of Aircraft. DOT/FAA/AR-00/40, Office of Aviation Research, 79 p.
Fagan, C., 2004, Safety of Well Testing. MMS JIP Final Project Report, Report No. 4514774/DNV, 15 Nov 2004.
Farzaneh M., and Savadjiev K.; “Statistical Analysis of Field Data for Precipitation Icing Accretion on Overhead Power Lines", IEEE Transaction on Power Delivery, Vol. 20, No. 2, April 2005
291 Farzaneh, M., Volat, C., Leblond, A., 2008. Atmospheric Icing of Power Networks. Springer Science+Business Media B.V. 2008, Ch. Anti-icing and De-icing Techniques for Overhead Lines, pp.229–268.
Feher, L., 2003, Compact Millimeterwave System for De-icing and Preventing the Formation of Ice on the Outer Surfaces of Shell Structures Exposed toMeteorological Influences. US Patent 6,642,490, 4 November 2003.
Feher, L., and M. Thumm, 2006, Design of Avionic Microwave De-/Anti-Icing Systems.in Willert-Porada, M., Microwave devices. Springer 792 p., ISBN: 3540432523, pp. 695–702.
Ferrick, M.G., N.D. Mulherin, B.A. Coutermarsh, G.D. Durell, L.A. Curtis, T.L. St. Clair, E.S. Weiser, R.J. Cano, T.M. Smith, and E.C. Martinez, 2006b, Double Lap Shear Testing of Coating Modified Ice Adhesion to Space Shuttle Component Surfaces, ERDC/CRREL TR-06-21, 62 pp.
Ferrick, M.G., N.D. Mulherin, R.B. Haehnel, B.A. Coutermarsh, G.D. Durell, T.J. Tantillo, L.A. Curtis, T.L. St. Clair, E.S. Weiser, R.J. Cano, T.M. Smith, E.C. Martinez, 2008, Evaluation of ice release coatings at cryogenic temperature for the space shuttle. Cold Regions Science and Technology, vol. 52, 224–243.
Ferrick, M.G., N.D. Mulherin, R.B. Haehnel, B.A. Coutermarsh, G.D. Durell, T.J. Tantillo, T.L. St. Clair, E.S. Weiser, R.J. Cano, T.M. Smith, and E.C. Martinez, 2006a, Double Lap Shear Testing of Coating Modified Ice Adhesion to Liquid Oxygen Feed Line Bracket, Space Shuttle External Tank, ERDC/CRREL TR-06-11, 76 pp.
Fett, R.W., R.E. Englebretson, and D.C. Perryman, 1993, Forecasters Handbook for the Bering Sea, Aleutian Islands and Gulf of Alaska, NRL/PU/7541-93-0006, Naval Research Laboratory, Monterey, California, pp. 302.
Figg, Eugene C. 2005, “Stay-Cable Cradle System”, US Patent 6,880,193 B2, April.
Fikke, S., Ronsten, G., Heimo, A., Kunz, S., Ostrozlik, M., Persson, P., Sabata, J., Wareing, B., Wichura, B., Chum, J., 2007. COST 727: atmospheric icing on structures: measurements and data collection on icing: state of the art. Meteo Schweiz.Foster-Miller, 2004, Electro-Expulsive Deicing of Shipboard Composites. NAVY SBIR Proposal.
Findlay, P. (2014, February 8). Ravenel Bridge, other spans face "ice bomb" problem. The Post and Courier [Charleston]. Retrieved from http://www.postandcourier.com/article/20140208/PC16/140209376 Finstad, K., 1998: Accreted snow loads. Repoert submitted to U.S. Army Cold Regions Research and Engineering Laboratory
292 Frank, S., 2004, A Pet-Friendly De-icer? Green Guide, 11 Feb 2004, http://www.thegreenguide.com/doc/ask/icehttp://www.thegreenguide.com/doc/ask/ic e. Accessed November 15, 2010.
Frankenstein, S., and A.M. Tuthill, 2002, Ice Adhesion to Locks and Dams: Past Work; Future Directions? Journal of Cold Regions Engineering, Vol. 16, No. 2, pp. 83–96.
Franklin, C.H., and C.O. Rogne, 1991, Ice Accretion and Persistence Measuring Instrument. Franklin Engineering Co., 46 p.
Geng X., Hu X.; "On-line Monitoring Method and Observation System of Transmission Line Icing", IEEE Conference on Industrial Electronics and Applications, ICIEA 2009
Geokon Installation Manual, “Model 3800/3810 Thermistors & Thermistor Strings”, 2009.
Glickman, T. S., Ed., 2000: “Glossary of Meteorology”, 2d éd. Amer. Meteor. Soc., 855
Gluckman, N. (2014, January 26). No easy fix to ice buildup issue on Penobscot Narrows Bridge, say experts. The Bangor Daily News [Bangor]. Retrieved from http://bangordailynews.com/2014/01/26/news/hancock/no-easy-fix-to-ice-buildup- issue-on-penobscot-narrows-bridge-say-experts/
Goodrich, 2008, Next Generation IceHawk Wide Area Ice Detection System. Brochure, www.aerospace.goodrich.com.
Goodrich, 2009, Ice Detector Model 0872F1. Brochure, www.campbellsci.com.
Govoni, J.W., and C.H. Franklin, 1992, Evaluation of a pneumatic guy-line deicing boot. US Army Cold Regions Research and Engineering Laboratory. Special Report 92- 04, 7 p.
Gregoris, D., S. Yu, and F. Teti, 2004, Multispectral Imaging of Ice. Proceedings of the 2004 IEEE Canadian Conference on Electrical and Computer Engineering, Niagara Falls, 3–5 May, pp. 2051–2056.
Guest, P., 2005, Vessel Icing. Mariners Weather Log, Vol. 49, No. 3, December 2005, 9 p.
Gulley, L., 1998, Testing of Runway Ice Control Products, Materials Compatibility. AFRL-ML-WP-TR-1999-4040. Wright-Patterson AFB, 187 p.
Guo, T.C., 2005, Microwave Deicing of Highways, Commercial Parking, and Pedestrian Crosswalks. Potomac Research proposal to TEDCO. 13 p.
Guo-Ming MA., Cheng-rong L., Jian J., Jiang-tao Q., and Yang-Chun C.; "Ice Monitoring
293 System of Transmission Lines based on Fiber Bragg Grating Sensor", IEEE, International Conference on High Voltage Engineering and Applications (ICHVE), 11-14 Oct., 2010
Habib, E., Krajewski, W.F., Krugeraccuracy, A., March-April, 2001 ,“Sampling Errors of Tipping Buckets Rain Gauge Measurements”, Journal of Hydrologic Engineering, 159-166
Haehnel R.B., 2001, Advances in icing control at corps hydraulic structures, Ice Engineering, Number 31, Novemebr 2001, http://www.crrel.usace.army.mil/ierd/tectran/IERD31.pdf, accessed November 19, 2010.
Haehnel, R.B., 2002, Evaluation of coatings for icing control at hydraulic structures. US Army Cold Regions Research and Engineering Laboratory, Hanover, New
Haehnel, R.B., and N.D. Mulherin, 1998, The bond strength of an ice–solid interface loaded in shear. Ice in Surface Waters, Proceedings of the 14th International Symposium on Ice, 27–31 July 1998, Potsdam, New York. Rotterdam: Balkema Press, p. 597–604.
Haslim, L.A., and R.D. Lee, 1987, Electro-Expulsive Separation System. US Patent 4,690,353, 1 September 2007.
Hill, E., M. Rios, G. Botura, J.T. Riley, C.J. Dumont, D. Pullin, S. Uppulari, A. Broeren, M. Selig, and D. Anderson, 2006, Investigations of Performance of PneumaticDeicing Boots, Surface Ice Detectors, and Scaling of Intercycle Ice. Report DOT/FAA/AR-06/48, 158 p.
Homola, M.C., P.J. Nicklasson, and P.A. Sundsbø, 2006, Ice Sensors for Wind Turbines. Cold Regions Science and Technology, Vol. 46, pp. 125–131.
“How to measure temperature”, [Online],
Hydrobead, 2013, www.Hydrobead.com.
IADC, 2007, Rig ASP Program, 2007: A Summary of Occupational Incidents (Industry Totals). Issued 20 June 2007, 27 p.
IDI, 2007, Deicing Systems. http://www.idiny.com/deicing.html. Accessed November 16, 2010.
IMS, 2007, Electro-Expulsive De-Icing System (EEDS) Technical Description. Ice Management Systems, Inc., 42136 Sarah Way, Temecula, California, 11 p.
294 Ingram, R., Codner, G., Gerardi, J., July 1998. Electro-magnetic expulsion de-icing system. US Patent 5,782,435.
Inkpen, S., J. Hall, C. Marshall, C. Brobeck, and C. Nolan, 1995, Device and Method for Identifying and Quantifying Layered Substances. US Patent 5,394,340, 28 February 1995.
International Organization for Standardization, 2001: Atmospheric icing of structures, ISO 12494, Geneva, Switzerland. Jarvis, A.W. (2011); “METAR/TAF Information,” US National Weather Service,
John S. Steinhart, Stanley R. Hart, “Calibration curves for thermistors”, Deep Sea Research and Oceanographic Abstracts, Volume 15, Issue 4, August 1968, Pages 497-503, ISSN 0011-7471, DOI: 10.1016/0011-7471(68)90057-0.
Jones K.F.; “A Simple Model for Freezing Rain Ice Loads”, Atmospheric Research, Vol. 46, 1998, pp. 87-97
Jones, Kathleen F, 1996, “Ice Accretion in Freezing Rain. Cold Regions Research and Engineering Laboratory (U.S.)”, United States. Army. Office of the Chief of Engineers, CRREL report, Volume 96-2
Jones K.F.; “Toledo Weather Conditions Associated with Ice Accumulation on the Skyway Stays (Draft)”, Cold Regions Research and Engineering Laboratory, April 12, 2010
Jones, K.F. and A.B. Peabody, 2006. The application of a uniform radial ice thickness to structural sections. Cold Regions Science and Technology, 44, pp 145-148.
Jones, K.F., 1998. A simple model for freezing rain ice loads, Atmospheric Research, 46, 87-97.
Jones, K.F., 2010a, Toledo weather conditions associated with ice accumulation on the Skyway stays, this can be downloaded from the project website.
Jones, K.F., 2010b, U.S. Army Corp of Engineers Ice Storm Database, http://cmep.crrel.usace.army.mil/ice/. In the GIS you can zoom in to a region and get information about individual damaging storms.
Joseph, J.C., 2002, MAIDS Microwave Aircraft Icing Detection System. Dedicated Electronics, Inc. Briefing to SAE AC-9C Inflight Icing Technology Subcommittee.
Kelly Aerospace, 2008, Thermawing. http://www.kellyaerospace.com/thermawing.html. Accessed November 16, 2010.
295 Ketcham, S.A., L.D. Minsk, R.R. Blackburn, and E.J. Fleege, 1996, Manual of practice for an effective anti-icing program: A guide for highway winter maintenance personnel. US Department of Transportation. Federal Highway Administration. Office of Engineering Research and Development. Report No.95-202, 63 p. CRREL MP 3855.
Kilinich, S. A., and Farzaneh, M., 2011, On Ice-releasing Properties of Rough Hydrophobic Coatings. Cold Region Science and Technology, Vol. 65, pp, 60- 64.
Kleissl, K., and C.T. Georgakis, 2010, Bridge ice accretion and de- and anti-icing systems: A review, presented at The 7th International Cable Supported Bridge Operators' Conference, Jiang Su, China, May 2010, can be downloaded from the project website.
Koenig, G. G., and Ryerson, C. C., 2011, An Investigation of Infrared Deicing through Experimentation. Cold Region Science and Technology, Vol. 65, pp, 79-87.
Koontz, H.S., and J.P. Forr, 1993, Electrically Heated Window. US Patent 5,182,431, 26 January 1993.
Kozlesky, J., Preliminary thermal ice prevention analysis of the Veterans Glass City Skyway Bridge stay cables, Honors Thesis, University of Toledo, December 2010.
Kuhn, E., May 2006. Maintenance with industrial rope access: Uddevalla bridge - a case study. In: IABSE Conference ’Operation, Maintenance and Rehabilitation of Large Infrastructure Projects, Bridges and Tunnels’, Copenhagen, 2006.
Kumpf, J., Helmicki, A., Nims, D.K., Hunt, V., Agrawal, S., 2012, “Automated Ice Inference and Monitoring on the Veteran’s Glass City Skyway Bridge”, Journal of Bridge Engineering, ASCE, 17:975-978
Laforte, C., J. Laforte, and J.C. Carrière, 2002, How a Solid Coating Can Reduce the Adhesion of Ice on a Structure. In Proceedings of the International Workshop on Atmospheric Icing of Structures, Brno, Czech Republic, June 2002, 6 p.
Laforte, J.L., M.A. Allaire, and J. Laflamme, 1998, State-of-the-art on power line de- icing. Atmospheric Research, Vol. 46, pp. 143–158.
Laursen, E., Zwieg, T., 2007. Kan anvendelse af nanoteknologi forhindre overisning p°a storebæltsbroen. In: Dansk Vejtidsskrift, feb 2007.
Leblond, A., Montambault, S., St-Louis, M., Beauchemin, R., Laforte, J., Allaire, M., 2002. Development of new ground wire de-icing methods. In: Proceedings of the 10th International Workshop on Atmospheric Icing of Structures, Brno, Czech Republic. Vol. 17. p. 20.
296 Lewlomphaisarl, U. ; Saengsatcha, P., 2012, “High Accuracy Tipping Bucket Rain Gauge”, Proceedings of SICE Annual Conference (SICE), 372 – 375
Long J., Lin J., and Lin K.; “Precipitation Prediction Modeling Using Neural Network and Empirical Orthogonal Function Base on Numerical Weather Forecast Production”, IEEE, Sixth World Congress on Intelligent Control and Automation, 2006. WCICA 2006
Løset, S., 1985, Investigation and research on anti-icing and de-icing devices for marine application. In Proceedings, International Workshop on Offshore Winds and Icing, Halifax, Nova Scotia, 7–11 October 1985. Edited by T.A. Agnew and V.R. Swail, pp. 95–101.
Lozowski, E., K. Szilder, and L. Makkonen, 2000, Computer simulation of marine ice accretion. Philosophical Trans of the Royal Society: Mathematical, Physical and Engineering Sciences, Vol. 358, 2811–2845.
Luming L., Xin S., Xiaobo M., Zhiguo Z., Xue C., and Min Z.; "Real-time Ice Monitoring on Overhead Power Transmission Lines with FBG Sensor", 2010 Second IITA International Conference on Geoscience and Remote Sensing, 28-31 Aug., 2010
Lyle, D., 2001, Prepare for the Worst Offshore. Hart’s E&P, August 2001, pp. 69–72.
Makkonen L., and Kari A.; “Climatic Mapping of Ice Loads on Airport Weather Observations”, Atmospheric Research, Vol. 36, Issues 3-4, May 1995, pp. 185-193
Makkonen, L., and Wichura, B., 2010: Simulating wet snow loads on power line cables by a simple model. Cold Regions Science and Technology, 61, 2-3, 73-81.
Makkonen L.; “Modeling Power Line Icing in Freezing Precipitation”, Proc. 7th International Workshop Atmospheric Icing Structures, Chicoutimi, QC, Canada, 1996, pp. 195–200
Makkonen, L., 2000. Models for the growth of rime, glaze, icicles and wet snow on structures. Philosophical Transactions: Mathematical, Physical and Engineering Sciences, 2913–2939.
“Manual – Leaf Wetness Sensor”, [Online]
“Manual on the use of thermocouples in temperature measurement”, ASTM Committee E-20 on Temperature Measurement, Subcommittee E20.04 on Thermocouples, ASTM International, 1981.
Martin, A.N., 1991, Microwave Deicing for Aircraft Engine Propulsor Blades. US Patent 5,061,836, 29 October 1991.
297 McCurdy, J.D., C.R. Sullivan, and V.F. Petrenko, 2003, Resonant 33 kV, 100 kHz Power Source for Deicing Power Transmission Lines Using Optimized Resonant Inductors. Final Report to the Army Research Office, Report No. ARO 39152 10-EV, 6 p.
Meiszner, P. (2013, December 20). Port Mann Bridge “ice bomb prevention system” debuts during today’s snowy weather. Global News. Retrieved May 19, 2014, from http://globalnews.ca/news/1044434/port-mann-bridge-ice-bomb-prevention- system-debuts-during-todays-snowy-weather/ Menini, R., Ghalmi, Z., Farzaneh, M., 2011, Highly Resistant Icephobic Coating on Aluminum Alloy. Cold Region Science and Technology, Vol. 65, pp, 65-69.
Meteorological Assimilation Data Ingest System, NOAA Research, Predicting Climate Variability and Extreme Events, (http://www.research.noaa.gov/climate/t_prediction.html)
“MetOne Rain Gage Models 380 and 385”, [Online], < http://s.campbellsci.com/documents/us/manuals/met1.pdf>.
Minsk, L.D. Heated Bridge Technology: Report on ISTEA Sec. 6005 Program. Washington, .C.: Federal Highway Administration, 1999.
Minsk, L.D., 1985, Measurement of icing on offshore structures. In Proceedings, International Workshop on Offshore Winds and Icing, Halifax, Nova Scotia, Oct. 7- 11, 1985. Edited by T.A. Agnew and V.R. Swail, p. 287–292. Publisher: Downsview, Ontario, Atmospheric Environment Service. Canada. CRREL MP 2010.
Mishra, S.K., 2001, A Mixture of Snow and Ice. Roads & Bridges, Vol. 39, Number 12, 4 p.
Mulherin, N.D., and R.B. Haehnel, 2003, Progress in Evaluating Surface Coatings for Icing Control at Corps Hydraulic Structures. US Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire, Ice EngineeringInformation Exchange Bulletin, TN-03-4, 7 p.
Multilogger Software User’s Guide, Canary Systems, Version 5.0.
Nadine, R., Volkmar, S., 2009. Anti-ice technologies - coating concepts and evaluation. In: IWAIS.
NASA, 2002a, Aircraft Anti-Icing Heaters Made from Expanded Graphite. NASA Tech Briefs, LEW-16895, 14 p..
NASA, 2002b, Deicing and Anti-Icing Unite, Spinoff 2002, pp. 74–75.
NASA, 2006, Preventing Ice Before It Forms, Spinoff 2006, pp. 42–43.
298 NASA, 2010a, http://www.nasa.gov/topics/aeronautics/features/trl_demystified.html, accessed November 17, 2010.
NASA, 2010b, http://esto.nasa.gov/files/TRL_definitions.pdf, , accessed November 17, 2010.
National Research Council. Transportation Research Board. Special report, No. 185, p. 303–307.
Natural Gas Technologies Center, 2007, Radiant Aviation Services—Infrared Heater Testing. Report Version 2, Project No. 222706, Presented to Radiant Aviation Services Inc. (RAS), 26 p.
NCEP, 2008, http://polar.ncep.noaa.gov/marine.meteorology/vessel.icing/. NOAA, 2007, Dixon Entrance, AK to Cape Spencer, AK. Coast Pilot #8, pp 119–135. Accessed November 16, 2010.
Net Resources International. "Svinesund Road Bridge, Sweden / Norway." http://www.roadtraffic-technology.com/projects/svinesund/. Accessed February 10, 2011.
NOAA’s National Weather Service, Aviation Digital Data Service (ADDS), METAR tutorial (http://aviationweather.gov/adds/metars/)
NOAA’s National Weather Service, Aviation Digital Data Service (ADDS), PIREP tutorial (http://aviationweather.gov/adds/pireps/)
NOAA’s National Weather Service, Aviation Digital Data Service (ADDS), Supplementary Icing Information (CIP/FIP) (http://aviationweather.gov/adds/icing/)
Numata, H., Kitada, K., 2008. Measures against snow accretion on suspension bridges in snowy regions: Hakucho bridge. In: The 6th International Cable Supported Bridge Operators’ Conference (ICSBOC). pp. 121–125.
NWS, 2010, National Weather Service, http://www.nco.ncep.noaa.gov/pmb/nwprod/analysis/
Nygaard, B.E.K., Ágústsson, H., and Somfalvi-Tóth, K. "Modeling Wet Snow Accretion on Power Lines: Improvements to Previous Methods. Using 50 Years of Observations," J. Appl. Meteor. Climatol., in press, 2013.
Ocean Rig, n.d., Eric Raude. brochure, 7 p. OCEAN RIG ASA P O BOX 409, FORUS NO-4067 STAVANGER, NORWAY TEL +47 51 96 90 00 FAX +47 51 96 90 99 [email protected], www.ocean-rig.com. Accessed November 16, 2010. Oilrigdisasters, 2008, http://www.oilrigdisasters.co.uk/. Accessed November 16, 2010.
299 Ohio Department of Transportation, “Road Weather Information System (RWIS),” Office of Maintenance Administration,
Ohio Department of Transportation, Road Weather Information System (RWIS), Office of Maintenance Administration, May 5th, 2011
Oregon Ohio Weather (2011), “Weather Station Web Site,”
OSHA, 2004, Target Industry Profile: Oil and Gas Field Services, Industry Group 138, 15 p.
Pakarinen, R., 2006, A Big Drill for the Big Chill, Inside Marine 3.06, pp. 62–63.
Parent, O., Ilinca, A., 2011, Anti-icing and De-icing Techniques for Wind Turbines: Critical Review. Cold Region Science and Technology, Vol. 65, pp, 88-95.
Pavek, J., 2001, Magnesium Chloride as a De-Icing Agent. http://www.usda.gov/rus/electric/engineering/2001/magnesium_chloride.htm. Accessed November 16, 2010.
Penobscot Narrows Bridge and Observatory. (2014, December 5). Wikipedia. Retrieved May 19, 2014, from http://en.wikipedia.org/wiki/Penobscot_Narrows_Bridge_ and_Observatory
Petrenko, V.F., 2005, Systems and Methods for Modifying an Ice-to-Object Interface. US Patent 6,870,139 B2, 22 March 2005.
Petrenko, V.F., 2010, personal communications, September 2009 – June 2010.
Petrenko, V.F., 3 videos: de-snowing a bridge cable, de-icing Uddevalla Bridge , and de-icing Fallande Is Bridge, provided by e-mail April 2010, posted on project website.
Petrenko, V.F., and C. Sullivan, 2003, Methods and Systems for Removing Ice from Surfaces. US Patent 6,653,598 B2, 25 November 2003.
Petrenko, V.F., and C. Sullivan, 2006, Low-Frequency De-Icing of Cableways. US Patent 7,038,125 B2, 2 May 2006.
Petrenko, V.F., and C.R. Sullivan, 2007, High-Frequency De-Icing of Cableways. US Patent 7,164,100 B2, 16 January 2007.
300 Petrenko, V. F., Sullivan, C. R., Kozlyuk, V., Petrenko, F. V., Veerasamy, V., 2011, Pulse Electro-thermal De-icer (PETD). Cold Region Science and Technology, Vol. 65, pp, 70-78Petrenko, V.F., and S.G. Qi, 1999, Reduction of Ice Adhesion to Stainless Steel by Ice Electrolysis. Journal of Applied Physics, 86(10), p. 5450– 5454.
Petrenko, V.F., and Z. Courville, 2000, Active De-icing Coating for Aerofoils. Proceedings of the American Institute of Aeronautics and Astronautics (AIAA) 38th Aerospace Sciences Meeting and Exhibit, Reno, NV, 10–13 January 2000. Originator No: AIAA-2000-063, 25 p.
Petrenko, V.F., M. Higa, M. Starostin, and L. Deresh, 2003, Pulse Electrothermal De- Icing, Proceedings of the Thirteenth (2003) International Offshore and Polar Engineering Conference, Honolulu, Hawaii, May 25–30, 2003, pp. 435–438.
Pohlman J.C., and Landers P.; "Present State-of-the Art of Transmission Line Icing", IEEE Transactions on Power Apparatus and Systems, Vol. PAS-101, No. 8 August 1982
Ramsay, A.C., 1997: Freezing rain detection and reporting by the Automated Surface Observing System (ASOS). Preprints, First Symposium on Integrated Observing Systems, Long Beach. RAS, 2008, http://www.radiantenergycorp.com/. Accessed November 16, 2010.
Reehorst, Andrew, February 2012, Personal Communication
Rindels, S., 1996, Deicing Materials for Slick Sidewalks and Roads. Horticulture and Home Pest, 13 December 1996, pp. 177–178.
Robbins, C., and J. V. Cortinas Jr., 1996: “A climatology of freezing rain in the contiguous United States: Preliminary results”, Preprints, 15th Conf. on Weather Analysis and Forecasting, Norfolk, VA, Amer. Meteor. Soc., 124-126
Ruggi, E., Pole, C., 1998. Deicing with a mobile infrared system. Tech. rep., Montreal, Quebec: APS Aviation Inc.
Ryerson, C. C., 2013, Icing Management for Coast Guard Assets. Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire 03755.
Ryerson, C. C., 2011, Ice Protection of Offshore Platforms. . Cold Region Science and Technology, Vol. 65, pp, 97-110.
Ryerson, C.C., 1995, Superstructure Spray and Ice Accretion on a Large US Coast Guard Cutter. Atmospheric Research, Vol. 36, pp. 321–337.
301 Ryerson, C.C., 2009, Assessment of Superstructure Ice Protection as Applied to Offshore Oil Operations Safety, Project number 597, U.S. Department of Interior, Mineral Management Service Program, March 2009..
Ryerson, C.C., and A. Elíasson, 1993, Ice Detector Prediction of Transmission Line Ice Loads in Northwestern Iceland. In Proceedings of the 6th International Workshop on Atmospheric Icing of Structures, Budapest, Hungary, pp. 189–194.
Ryerson, C.C., and A. Gow, 2000, Crystalline Structure of Ship Superstructure Spray Ice. Philosophical Trans of the Royal Society: Mathematical, Physical and Engineering Sciences, Vol. 358, 2847–2871.
Ryerson, C.C., and A. Ramsay, 2007, Quantitative Ice Accretion Information from the Automated Surface Observing System (ASOS), Journal of Applied Meteorology and Climatology.
Ryerson, C.C., and P. Longo, 1992, Ship Superstructure Icing: Data Collection and Instrument Performance on the USCGC Midgett Research Cruise. US Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire, CRREL Report 92-23, 142 pp.
Ryerson, C.C., G. Koenig, F. Scott, and J. Tantillo, 2004, Infrared Aircraft Deicing Performance Assessment. In Gulley, L., and J. Davilla, n.d., Survey and Evaluation for Alternative Deicing Energy Sources. Air Force Institute for Operational Health (AFIOH), Contract No. F41624-01-D-9007, Task Order 0007, 74 p.
Ryerson, C.C., K. Claffey, and G. Lemieux, 1994, Surface Hoarfrost Measurement and Climatology. In Proceedings of the 51st Eastern Snow Conference, pp. 121–130.
Ryerson, C.C., M. Wyderski, D. Tarazano, and J. Davila, 2003, Objective Evaluation of Fluid, Blower, and Infrared Ground Deicing Technologies. AIAA-2003-0567, American Institute of Aeronautics and Astronautics 41st Aerospace Sciences Meeting and Exhibit, 6–9 January, Reno, 10 p.
Ryerson, C.C., T. Gilligan, and G. Koenig, 1999, Evaluation of Three Helicopter Preflight Deicing Techniques, 37th AIAA Aerospace Sciences Meeting and Exhibit, AIAA-99-0499, 11–14 January, Reno, 9 p.
Ryerson, C.C., T. W. Gilligan, G.G. Koenig, evaluation of three Helicopter Preflight eicing techniques, U.S. Army Cold Regions Research and Engineering Laboratory, Hanover, NH, http://www.crrel.usace.army.mil/library/conferencepapers/MP- 5296.pdf accessed November 19, 2010.
SAE, 1997, Deicing System, Pneumatic Boot, Aircraft, General Specification for. SAE Aerospace Standard AS8804, 21 p.
302 SAE, 2007, Standard Test Method for Aerodynamic Acceptance of SAE AMS 1424 and SAE AMS 1428 Aircraft Deicing/Anti-icing Fluids. Document Number: AS5900, Revision Number: B, July 2007, Issuing Committee: G-12 Fluids Subcommittee.
Sætre, T., and P.K. Eian, 2006, Radiant De-Icing Gerdermoen Measurement of Temperature Rise in Various Materials Exposed for Infrared Radiation. Memo No. 1/rev3, Norconsult AS, 10 January 2006, 9 p.
Sakamoto, Y., 2000: Snow accretion on overhead wires. Philos. Trans. Roy. Soc. A, 358, 2941–2970.
Savadjiev K., and Farzaneh M.; “Modeling of Icing and Ice Shedding on Overhead Power Lines Based on Statistical Analysis of Meteorological Data”, IEEE Transactions on Power Delivery, Volume 19, No.2, Apr 2004.
Severn Bridge. (2011, February 5). Wikipedia. Retrieved May 19, 2014, from http://en. wikipedia.org/w/index.php?title=Severn_Bridge&oldid=412425962
Shin, J., Bond, T., 1993. Surface roughness due to residual ice in the use of low power deicing systems. AIAA-0031, 31st Aerospace Sciences Meeting & Exhibit, Reno, NV.
Sivas, S.L., B. Riegler, R. Thomaier, and K. Hoover, 2008, Silicone-Based Icephobic Coating for Aircraft, NuSil Technology Newsletter, 21 January 2008, Vol. 5, Issue 1—Winter 2008, 7 p.
Snow, P.G., n.d., Magnesium Chloride as a Road Deicer: A Critical Review. Burns Concrete, Inc., Idaho Falls, Idaho, 3 p.
Stein, P., 1993: Rosemount freezing rain sensor. Proceedings, Sixth International Workshop on Atmospheric Icing of Structures, Budapest, Hungary, Hungarian Electrotechnical Association, 121–124.
Sullivan, C.R., V.F. Petrenko, J.D. McCurdy, and V. Kozliouk, 2003, Breaking the ice [transmission line icing]. In Proceedings of the Thirteenth (2003) International Offshore and Polar Engineering Conference, Honolulu, Hawaii, 25–30 May 2003, pp. 49–54.
Takemoto, H., Ueno, H., Takeuchi, M., Chiba, T., Asano, Y., 2006. Evaluation of measures against snow and ice accretion on bridge members and the development of an anti-accretion lattice fence. Journal of Snow Engineering of Japan 22 (5), 357– 362.
Taylor, B., 2009, Fiber-Reinforced Polymer Honeycomb Bridge Deck Heating Evaluation, Thesis in Partial Fulfillment of a Master Degree, University of Toledo, December 2009.
303 Taylor, P. J. Verkade, K. Gopalakrishnan, K. Wadhwa,and Sunghwan Kim, 2010. “Development of an Improved Agricultural-Based Deicing Product”, Iowa State University Institute for Transportation, IHRB Project TR-581, Final Report January 2010.
The Penobscot Narrows Bridge Observatory. (2009, January 1). Penobscot Narrows Bridge Observatory: Homepage. Retrieved May 19, 2014, from http://www.maine.gov/mdot/pnbo/index.htm
Tollefsen, F., 2006, Experience from Drilling Operations in the Barents Sea, Eirik Raude. Kristiansandkonferansen 2006.
Trademarkia, 2011, www.Trademarkia.com/beet-heat-85179043.html.
TRB, 1991, Highway Deicing: Comparing Salt and Calcium Magnesium Acetate. Washington, DC: Transportation Research Board, 1991.
TRB, 1991, Highway Deicing Comparing Salt and Calcium Magnesium Acetate. Committee on the Comparative Costs of Rock Salt and Calcium Magnesium Acetate (CMA) for Highway Deicing.
Tuan, C.Y. Roca Spur Bridge: The implementation of an innovative deicing echnology, ASCE Journal of Cold Regions Engineering, 2008.
Uddevalla Bridge in Sweden. (2012, December 20). The Gazette [Montreal]. Retrieved from http://www.montrealgazette.com/Uddevalla+Bridge+Sweden/7728228/ story.html
US Coast Guard, 2008, Ice Buildup Likely Caused Lady of Grace to Sink. Press Release 25 Jan 2008, 1 p.
Vaisala, 2010, http://www.vaisala.com/en/meteorology/Pages/default.aspx. Accessed November 16, 2010.
Valley Group, 2008, http://www.cat-1.com/. Accessed November 16, 2010.
Vincentsen, L., 2006. The large fixed links in denmark - experience and developments. In: IABSE. pp. 29–35.
Vincentsen, L., Jacobsen, H., 2001. Operation and maintenance of the great belt bridge. In: IABSE.
Walsh, M.R., D.E. Garfield, J.S. Morse, K.V. Knuth, N.D. Mulherin, and G.E. Lemieux, 1993, Icing of turbine intake louvers. US Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire. Special Report 93-04, 37 p.
304 Weather Underground (2011), “METAR Tutorial,”
Wikipedia, 2013, www.en.wikipedia.org/wiki/Veterans’-Glass-City-Skyway.
Wisconsin Transportation Center, 1997, Wisconsin Transportation Bulletin No. 6: Using Salt and Sand for Winter Road Maintenance. Road Management Journal. 1 December 1997, TranSafety, Inc., [email protected].
Wood, J., Muneer, T., Kubie, J., February 2003, “Evaluation of a New Photodiode Sensor for Measuring Global and Diffuse Irradiance, and Sunshine Duration”, Journal of Solar Energy Engineering, ASME, Vol. 125, 43-48
Wyderski, M., C. Ryerson, A. Lozada-Ruiz, J. Davila, and D. Tarazano, 2003, Evaluation of Environmentally Friendly, Glycol-Free Mobile Aircraft Deicing System. 33rd International Conference on Environmental Systems (ICES), 7–10 July 2003, Vancouver, British Columbia, Canada, 10 p.
Yankielun, N., C. Ryerson, and S. Jones, 2002, Wide Area Ice Detection Using Time Domain Reflectometry. US Army Cold Regions Research and Engineering Laboratory, Hanover, New Hampshire, Technical Report ERDC/CRREL TR-02-15, 37 p.
Yankielun, N.E., and C.C. Ryerson, 2003, Device, and method of its use, for concurrent real time alerting to accumulation of material upon multiple areas of a surface. (6,608,489) (ERDC/CRREL MP-03-6085).
Yankielun, N.E., C.C. Ryerson, and A. Crocker, 2005, Clear Freshwater Ice Thickness Measurement Using Geometric Optics and Spectrometer. Cold Regions Science and Technology. Vol 43, pp. 177–186.
Zarafshan, F.; Latif-Shabgahi, G.R.; Karimi, A.(2010); "A novel weighted voting algorithm based on neural networks for fault-tolerant systems," Computer Science and Information Technology (ICCSIT), 2010 3rd IEEE International Conference on , vol.9, pp.135-139, July 9-11, 2010.
Zhi X., Ling Z., and Bai Y.; “Application of the Multi-model Ensemble Forecast in the QPF”, Intl. Conference on Informatics Science and Technology March 26-28, 2011 Nanjing, Jingsu, China
Zieve, P., Ng, J., Fiedberg, R., October 1991. Suppression of radiating harmonics Electro- Impulse Deicing (EIDI) systems. NASA STI/Recon Technical Report N 92, 19764.
305 Appendix A: Technology Matrix
This is a list and brief description of the technologies which have been reviewed. The full matrix, which includes more detail and discussion is available on the website https://extranet.dot.state.oh.us/divisions/TransSysDev/innovation/research/Ice_Project_Toledo/default.aspx. This website is open to ODOT emplyees and ODOT authorized guests.
TECHNOLOGY DECRIPTION
Chemicals and Chemical Distribution Systems
Distribution Systems
Weeping Wing Weeping Wing is a chemical aircraft ice protection system wherein a glycol-based chemical is released onto the wing surface using small orifices on the leading edge of the wing. This system provides both anti-icing (ice formation prevention) and deicing (removal of ice which has already formed).
Feltwick Anti-Icing Grate The Feltwick grate surface consists of a robust grating or tiles that wick an anti-icing fluid to the icing-prone surface from a reservoir layer located beneath.
Fixed Anti-Icing Spray Fixed Anti-Icing Spray Technology (FAST) is a class of systems marketed by several companies to spray anti-icing or Technology (FAST) deicing fluids onto walkways, roadways, bridges, and other pavement surfaces.
Zero GravityTM Third Rail Form a protective-coating barrier that prevents the buildup of ice and snow. Anti-Icer/Deicer and the Ice Free Switch.
Drip Tubing Could be as simple as running a drip, soaker irrigation hose or pipe down the top of each sheath
Chemicals
306 TECHNOLOGY DECRIPTION
Chloride deicers
Sodium Chloride As the major ingredient in edible salt, it is commonly used as a condiment and food preservative.
Calcium Chloride Calcium chloride, CaCl2, is a common salt
Magnesium Chloride Magnesium chloride as the natural mineral bischofite is also extracted (solution mining) out of ancient sea beds, for example the Zechstein seabed in NW Europe or the Permian Period seabed in the central parts of the US. Anhydrous magnesium chloride is the principal precursor to magnesium metal, which is produced on a large scale.
Potassium Chloride Potassium chloride (KCI) is a chemical compound containing both potassium and chlorine. It is considered a halide salt, which means that it contains a halogen atom and is crystalline in nature like other salts. In its pure state it is white and odorless. Impure potassium chloride varies in color from white to pink to red.
Acetate deicers
Calcium Magnesium Calcium magnesium acetate (CMA) is a relatively new deicing compound manufactured from limestone and acetic acid, Acetate and contains no salts.
Potassium Acetate Potassium acetate is the salt that forms along with water as acetic acid and potassium hydroxide are neutralized together.
Sodium Acetate Sodium acetate, (also sodium ethanoate) is the sodium salt of acetic acid
Glycols
Ethylene Glycol An organic compound widely used as an automotive antifreeze and a precursor to polymers.
Propylene Glycol Propylene glycol (PG) is currently the primary chemical used for deicing aircraft worldwide
Misc. Deicing Chemicals
307 TECHNOLOGY DECRIPTION
Sodium Formate Sodium formate, HCOONa, is the sodium salt of formic acid, HCOOH. It usually appears as a white deliquescent powder.
Urea
Agricultural-based These enhance the effectiveness of salts and are not used alone. Rather they are used in conjunction with salt to Chemicals increase the success, persistence, and reduce the amount of salt required.
Sugar-beet-based Geomelt is a trade name for a sugar beet-based deicing chemical that is used to deice roads. products
Corn-based products Caliber M1000 and NC-3000 are corn-based products designed for ice control on roads, bridges, parking lots, and sidewalks.
Alcohol-based Ice-B-Gone, also marketed as Magic Salt, consists of a sugar base stock of distilled condensed soluble (DCSs), a products slurry derived from the making of vodka and rum.
Coatings reducing ice adhesion
Rain-X Windshield Rain-X windshield treatment is intended to improve visibility in wet weather by causing rain water to bead up and Treatment reducing the adhesion strength of water droplets to glass surfaces.
NuSil Technology NuSil Technology offers a family of silicone-based coatings intended to reduce the adhesion of ice to aerodynamic surfaces and structures, such as aircraft components manufactured from aluminum or composite materials. New low adhesion coatings under development that are easily applied should be investigated.
NASA Shuttle Ice The Shuttle Ice Liberation Coating (SILC, pronounced “silk”) was developed to reduce ice formation and adhesion on Liberation Coating (SILC) the NASA Space Shuttle external fuel tank.
308 TECHNOLOGY DECRIPTION
ePaint ePaint has, or is developing, several icephobic coatings through U.S. Navy and Air Force Small Business Innovative Research (SBIR) funding. *Should be tested low adhesions strength, being considered by FAA for radars testing program may be in progress contact Tom Seliga at Volpe Center.
NanoSonic NanoSonic is developing hydrophobic, antifouling, environmentally durable coatings with a wide service temperature range and inherent anti-icing functionality. Company progress and testing should be checked.
Microphase Coatings- The coating resists abrasion, is hydrophobic causing droplets to have a large contact angle with the surface, and PhaseBreak ESL icephobic through release of an encapsulated melting point depressant that migrates to the coating surface and melts ice at the ice-coating interface. Breaks ice into small pieces.
Seashell Technology When unfrozen water droplets strike the coating, the water droplets bead into spheres and roll off the surface. Product still not released. Check with Mary Wyderski about status of SBIR progress.
Nanohomics Nanohmics is in early development of a tape that can be imprinted with a biomimetic superhydrophobic surface.
21st CenCoatings Inc WC-1 (ICE) is a modified fluoropolyurethane two-component solvent-based topcoat. Being tested by MARICE team in Norway for marine application Non-stick 2-component fluorinated polyurethane industrial and marine coating (top-coat). Applied at room temperature (spray). Non-toxic, abrasion resistant, non-stick. Extensive NRL testing.
KISS Polymers LLC KISS-COTE is made by modifying the polymerization process by adding inhibitors that halt the cross-linking process at a preselected point. Silicone-based polymer coatings, applied at room temperature, slippery, non-toxic, water-proof, non-stick, superhydrophobic.
Ross Technology- The Ross Technology NanoSH superhydrophobic coating is being developed to provide corrosion resistance, to NanoSH improve performance of boats by reducing drag, and to decrease icing on overhead transmission cables, satellite dishes, antenna towers, and aircraft.
309 TECHNOLOGY DECRIPTION
Polysiloxane(amide- A surface coating which inhibits the formation of ice upon the surface of a substrate ureide) anti-ice coating
Teflon Non-stick powder
Inertia 165
Nanaosuper- Hoowaki coating, Bill King UIUC hydrophobic
Nanaosuper- Farzaneh group in Chicoutimi hydrophobic nanomaterial coating Using a film embedded with nanofibers to prevent ice and snow from accumulating on power-grid equipment. Some films are active; they require electrical power.
Design Design changes can potentially change the amount and shape of ice accretions, and possibly control how it is released. This is also a function of the method of ice accretion. Airfoil shape of ice could be changed.
Straking Creating a helical bead on the stay surface. Potentially this could weaken the ice sheets or create turbulence that makes it more difficult for the ice sheets to adhere.
Pneumatic or electrical Pneumatic systems in which an inflatable boot covers the stay. When the boot is inflated, the ice Expulsive Deicing cracks and falls away. Or, electrical systems that use repelling forces between conductors to Systems produce the expulsive force that ejects ice from the sheath.
Deicing Boots A neoprene synthetic rubber reinforced with fabric which is inflated to remove ice from the critical control surfaces of aircraft in flight.
Electro-Impulse De- The system operates by using electromagnetic coils located behind the surface by inducing strong and sudden Icing (EIDI) magnetic forces from a high-current DC pulse through the coil.
310 TECHNOLOGY DECRIPTION
Electro-Mechanical The primary concept is to combine either an anti-icing or deicing technology with a primary low-power deicing Expulsive Deicing System technology (EMEDS) and to coordinate their operation to achieve relative degrees of ice protection and surface (EMEDS) w/ Electro condition as permitted or required by the particular application. Thermal Subsystems
Electroexpulsive The EEDS comprises two electrically conductive strips sandwiched between layers of carbon fiber or fiberglass sheet Deicing System (EEDS) material (IMS 2007). Electrical current passed through the conductors (up to 500 V at 8000–10,000 amps for 1–2 ms) cause’s magnetic fields to form in the two conductors that repulse and push the two conductors apart with an acceleration of up to 60,000 g with a cuff movement of 2 to 2.5 mm. The system is typically pulsed every 45–90 sec in an aircraft ice accretion event.
Power line ice-shedder A mechanical ice-shedding device for temporary or permanent attachment to a suspended cable, and particularly to a suspended power line. The ice-shedding device uses a motor to move at least one unbalanced weight, thereby causing a vibration of the device that is translated to the cable to which the device is attached. The vibration causes an oscillation of the cable which is sufficient to substantially shed ice that has accumulated thereon.
Internal Pressure Fill the inside of the sheath to 2 psi to cause some expansion and contraction to break up the ice
Heat General heating of the mass of the sheath to a temperature above freezing. Simple reliable and direct. Requires the most energy. Many reviewed technologies are designed for energized electric transmission lines where current flowing through the line is used directly, or in modified form, to heat the cable. The VGCS would require a heat source.
Chinook MHD Humid Chinook Mobile Heating and Deicing Corporation (Chinook MHD) has developed a technology that delivers warm, Air Deicing humid air to iced surfaces via a truck-mounted delivery head. Melts ice using sensible and latent heat.
Rockwell Collins Buddy The Buddy Start deicing nozzle is a handheld unit that uses hot air at high velocity to blow snow off aircraft surfaces Start Deicing Nozzle and to melt ice and snow from aircraft surfaces. Concept of using hot (dry) air to heat interior of cable tubes may be practical, but nozzle itself not practical unless many small nozzles paced along cable sheath.
Qfoil QFoil is an electrothermal thin-film heater technology intended for ice protection applications that allow rapid temperature rise with a low-watt density. Potentially efficient solution. Use pulse deicing for lower energy cost.
311 TECHNOLOGY DECRIPTION
ThermaWing-Kelly ThermaWing is a graphite-based thermoelectric heater that protects the leading edge of airfoils. Aerospace Thermal Potentially efficient solution. Use pulse deicing for lower energy cost. Systems
Low-Power LPED uses electrical heating elements attached to or installed within a structure requiring ice protection. Electrothermal Deicing Potentially efficient solution. Use pulse deicing for lower energy cost. (LPED)
Electrical Heat Tracing Heating wires could be installed on the sheath. Partial or complete. With careful placement may make ice pieces small enough to reduce hazard.
Variable Resistance The resistance in a cable is varied to generate heat to melt ice build-up or keep it from forming in the first place. A Cable (VRC) De-icing proprietary technology developed by the Dartmouth Ice Research Lab. System
Ice-electrolysis de-icer Variable resistance cable that increases resistance in power lines-to produce heat that melts ice on power lines or prevents new freezing.
Internal hot gas flow The strands occupy roughly 50% of the space within the sheath (See Fig. 1) so hot gases could flow up the inside of the sheath.
Internal electrical It would be possible to pull and electrical resistance cable up inside the stay. Caution: a short circuit could speedup heating corrosion of the sheath.
Solar Warming System Pave Guard, operates much like radiant heating works in a home’s floor. Tubing is installed in the bridge deck, through which a heated solution is pumped to keep the deck from freezing. The energy to heat the solution is provided by solar panels mounted near the bridge site.
Ecotemp radiant Heater mat with adhesive backing. thermal sheet
Electrically conductive Battelle Memorial Institute is developing a carbon-nanotube paint. A meeting was held with Battelle to discuss this paint technology. It has a low commercial readiness level.
Laser deicing Use a laser to remove the ice.
312 TECHNOLOGY DECRIPTION
Microwave deicing
Radiant Heating Infrared radiant heat sources.
Schaefer Ventilation The HotZone technology consists of a gas or electric infrared source with the energy focused by a unique reflective HotZone Heaters lens.
Trimac Industrial The Gas-Cat uses a gas-fueled catalytic emitter panel, although electrically powered systems are also available Systems LLC - Ice-Cat
Radiant Aviation A Radiant Aviation facility consists of an array of energy process units (EPUs), or infrared emitters, mounted on the ceiling of the tension membrane structure. A circular array to move up the stay.
Vacca Inc The Vacca system is a variable output heater that can serve as an infrared emitter.
Interface heating Applying heat directly to the interface between the ice and the stay. This eliminates the need to heat the entire mass of the stay. PETD is an electrified coating. The method uses short pulses of electricity applied directly to the ice-substrate interface Pulse electro-thermal and, therefore, only has to melt a micrometer-thin layer of ice. . This causes the ice to melt at the interface without de-icing (PETD) heating the entire mass of the stay. The potential for using this system on the VGCS has been discussed with Dr. technology Petrenko. On the VGCS, the heating would done with short pulses to different areas of the bridge. The system uses about 6kW/m2. On the VGCS, the maximum available power is 600kW. The surface area is 95,000 ft2 (9,000 m2). Thus, the VGCS stays would have to be heated in about 90 100 m2 pieces. The time to de-ice would be roughly 35 minutes. The system has been installed on the Uddevalla Bridge. (Petrenko videos) On this bridge application, batteries were used to give a surge of power. Recently, a roof with a 10,000 m2 area has been coated. PETD was developed by the Dartmouth Ice Research Lab.
High-Velocity Air, Water, or Steam
313 TECHNOLOGY DECRIPTION
High-velocity water and Using high-pressure water to cut ice from the stays. steam
AirPlus! Forced Air High-pressure air from turbine engines is used to clear snow from aircraft wings. Deicing System
Manual Deicing Breaking accumulated ice free by using a simple hand powered mechanical tool. Methods
Ice Scrapers and loosening ice from surfaces and moving the ice overboard Breaker
Ice removing tool An ice removing tool consisting of an elongated insulated handle having a yoke at the upper end with a steel grooved pulley journalled in the yoke and adapted to engage a cable having ice encrusted thereon. The pulley wheel is rolled along the cable using the handle to break up the encrusted ice on the cable thus freeing the cable of ice.
Piezoelectric A piezoelectric actuator attached to the surface breaks the bond of the ice to the stay and the ice falls off the stay.
FBS Inc The engineering goal of the technology is to guide ultrasonic energy created at a few discrete actuators located on the airfoil to locations on the airfoil where ice accretes
Creare Inc The Creare system uses a thin, electrically activated piezoelectric activator attached to the surface to be protected to break the ice-substrate adhesive bond and cause it to be ejected from the surface.
Pneumatic Systems An inflatable boot covers the stay. Air inflates the boot and cracks the ice which then falls off.
314 TECHNOLOGY DECRIPTION
Vibration and Covers
Protective Covers Sheet of material to cover the item needing protection
Covers Jetsocks
Ice Detection Any active system will require sensors. Therefore, we are tracking useful sensors.
Ice Hawk The IceHawk detects ice by analyzing the polarization of laser light reflected from surfaces. Manufacturer is attempting to develop more capable unit. Ryerson has contacts.
Ice Camera The MDA Ice Camera maps the location of ice on surfaces and its thickness.
Goodrich (Rosemount) Rosemount ice detectors sample the icing environment at the probe location. The user must determine how Icing Rate Detector representative the measurements are to other locations
Microwave Aircraft Icing System detects ice on aircraft surfaces with enough sensitivity to provide a warning before the ice accretes to a Detection System dangerous thickness. (MAIDS)
SMARTboot SMARTboot is an aircraft ice detection and protection system combining inflatable pneumatic boots and a wide-area flush-mounted ice detection system.
TAMDAR The ice detector resides within a small sensor package that protrudes from the skin of aircraft into the air stream. Should be investigated.
315 TECHNOLOGY DECRIPTION
Vaisala The in situ DRS511 sensor (Figure 102) detects roadway surface conditions by making six measurements.
Visidyne first non-contact sensing system for detecting the accumulation of ice on rotorcraft blades in flight
Pole-Ice represent ice accumulation on electrical transmission lines
Supporting Equipment Technology that can be used in conjunction with a primary technology for anti/deicing.
Robotic Climber Some type of robotic device that can climb up the stays to apply a coating or scrap off the ice in a manner that the size of the particles are of a desired size
316