Shadow Flicker Monitoring 52113 to 82813 Mary Reilly

Shadow Flicker Monitoring 5/21/13 to 8/28/13 Mary Reilly Mason County Zoning and Building Director 102 E. 5 th St. , Scottville, MI 49454 mreilly@masoncounty. net (231) 757 -9272 Credits: Kirk Josvai, Flicker Observation Dr. Paul Drelles, WSCC, Model/Data Assistance

Zoning Ordinance Requirements • Mason County Zoning (Section 17. 70. 21 a): Shadow flicker on the external wall of a unpooled dwelling shall not exceed 10 hours per year (cumulative for all turbines). • Consumers Energy installed Vestas Shadow Detection System (VSDS) to curtail turbines during flicker event. 1 st system installed in N. America.

Basics: Location of Sun in Sky • Sun location is based on earths rotation (season) so flicker events are always changing – minute to minute. • The sun rises and sets at changing times.

Basics- Season and Distance from Turbine Affect Shadow Duration

VSDS components

Evening Flicker, summer

Morning flicker

Heavier in Winter Morning and Evening

Shadow Flicker Monitoring 1. ARE TURBINES PAUSING THAT ARE PROGRAMED TO? Periodic review of turbines under VSDS. 2. HOW MUCH FLICKER IS OCCURING COMPARED TO THE MODEL? Sampling of specific properties modeled to receive under 8 hours. 3. HOW LONG IS THE FLICKER SHADOW? Review of length of shadow flicker.

ARE TURBINES PAUSING THAT ARE PROGRAMED TO? Based on 16 observations: • 14 observations, turbines turned off during scheduled shut downs. • 2 observations, turbines did not shut down. • One clear day (WTG 39), • One hazy/humid day (WTG 15) • Consumers to manually override light sensors on WTG 14, 15, and others due to these and other observations.

The Data – 5400 feet from a WTG Worst Case Scenario 2 sunrise sunset Turbine ID: 51 Turbine ID: 52 Date Start Time (hrs) Duration (hrs) 1 8. 3272118 17. 299852 1 -Jan 0. 0 9. 0 0. 3 2 8. 3281061 17. 314609 2 -Jan 0. 0 9. 0 0. 3 3 8. 328303 17. 329871 3 -Jan 0. 0 9. 0 0. 3 4 8. 3278016 17. 345624 4 -Jan 0. 0 9. 0 0. 3 5 8. 3266017 17. 361852 5 -Jan 0. 0 9. 2 0. 2 6 8. 324704 17. 378537 6 -Jan 0. 0 9. 2 0. 2 7 8. 3221098 17. 395662 7 -Jan 0. 0 9. 2 0. 2 8 8. 3188209 17. 413212 8 -Jan 0. 0 9. 2 0. 2 9 8. 31484 17. 431169 9 -Jan 0. 0 10 8. 3101704 17. 449515 10 -Jan 0. 0 11 8. 3048158 17. 468234 11 -Jan 0. 0 12 8. 2987807 17. 487308 12 -Jan 0. 0 13 8. 29207 17. 506721 13 -Jan 0. 0 14 8. 2846894 17. 526454 14 -Jan 0. 0 15 8. 2766447 17. 54649 15 -Jan 0. 0 16 8. 2679426 17. 566813 16 -Jan 0. 0 17 8. 2585899 17. 587407 17 -Jan 0. 0 18 8. 248594 17. 608253 18 -Jan 0. 0 19 8. 2379628 17. 629336 19 -Jan 0. 0 20 8. 2267043 17. 65064 20 -Jan 8. 7 0. 2 0. 0 21 8. 214827 17. 672149 21 -Jan 8. 7 0. 2 0. 0 22 8. 2023398 17. 693848 22 -Jan 8. 7 0. 2 0. 0 23 8. 1892518 17. 71572 23 -Jan 8. 7 0. 2 0. 0 24 8. 1755723 17. 737752 24 -Jan 8. 7 0. 2 0. 0 25 8. 161311 17. 75993 25 -Jan 8. 7 0. 2 0. 0 26 8. 1464778 17. 782238 26 -Jan 8. 7 0. 2 0. 0 27 8. 1310826 17. 804663 27 -Jan 8. 7 0. 2 0. 0 28 8. 1151358 17. 827192 28 -Jan 8. 7 0. 2 0. 0 29 8. 0986476 17. 849813 29 -Jan 8. 7 0. 2 0. 0 30 8. 0816287 17. 872514 30 -Jan 8. 7 0. 2 0. 0 31 8. 0640896 17. 895281 31 -Jan 8. 7 0. 2 0. 0

2. HOW MUCH FLICKER IS OCCURING COMPARED TO THE MODEL? SAMPLE DATA SHOWING TURBINE, HOURS PER TURBINE and TOTAL Turbine Hours per turbine Total Note: Total flicker in model derived from multiplying “worst case” cumulative total in graphs shown by. 3 blade pass across the sun and. 4 clearness index or. 12 --. 15 was used to account for variability in climatic conditions.

2. HOW MUCH FLICKER IS OCCURING COMPARED TO THE MODEL? • EXAMPLE: 45 hours modeled under “worst case scenario” meaning always sunny, always windy, no down time for maintenance, no trees, no structures, no hills, and turbine direction oriented to create flicker. 45 hours X. 15 = 6. 75 hours per year THE QUESTION: is 6. 75 hours accurate or is the actual observed flicker something higher or lower?

Finding Potential Sites • Time of year (late spring, summer for project). • Will the turbines causing flicker be paused for another residence nearby? • Will this site receive less than 8 hours but more than 6 hours*. • Total flicker received primarily in the summer for observational period. *initial sites were selected above 6 hours, later to sites above 4 hours to be see if additional sites could be added for month of August.

Site Selection • 8 sites selected based on data. • 3 sites: turbines completely obstructed by vegetation or topography. No flicker. • 2 sites: only the tip of the blade was visible over trees such that flicker would be nearly imperceptible and highly wind direction sensitive. Little or no flicker. • 3 sites: good candidates (two AM, one PM)

Observations: Flicker appearance highly variable. • Level 0 -- modeled to receive, none exists • Level 1– receive on non-living space, in small areas with obstructions by trees, topography • Level 2 - receive in living space, at distance that diffuses light/partial diffusion by trees. • Level 3 - receive in living space, bright in appearance (less than 2000 -3000 feet to turbine), unobstructed.

Videos – Variable Appearance of Flicker

Kistler Road (Level 1/2) Site 1 - 2600’ to Turbine • Modeled to receive 7. 1 hours • Observed 8. 38 hours w/October potential • Modeled days: May 1 - August 12 (WTG 13, minor WTG 11) • Observed days: May 21 - August 8 (no flicker at end). May 21 May 1 Actual Observed Modeled Period Aug 8 Aug. 12

Observations at Kistler • Foliage gradually increases and obstructs flicker July-Aug. • Wind direction (blade direction) has a significant impact on duration of flicker and can cause no flicker for extended periods. SE/NW =no flicker. W/E=most flicker. • Morning flicker is prone to cloud banks low on the horizon that evaporate soon after dawn. • FALL observation: WTG 17 pausing for other dwelling.

Wind Direction- Variability Minimal or no Flicker Moderate or no Flicker Maximum Flicker

PM Hwy (Level 1) Site 2 - 2500 feet to turbine • Modeled to receive 6. 7 hours/year • Observed 2. 45 hours • Modeled 5 -20 to 8 -1 (WTG 17) • Observed 5 -21 to 7 -18 (ended early due to consistently viewing no flicker/obstruction). 5 -20 Actual observed days 5 -21 Modeled period 7 -18 8 -1

Observations at PM Hwy • Flicker observed on upper level of house • Briefly occurred through window. • Most often on corner of house/not inside the home. • Foliage/location of sun completely obstructed flicker toward end of observation.

Stiles Road (Level 3) Site 3 - 1600 feet to turbine • Modeled to receive 7. 7 hours • Observed 3. 95 hours in 15 observational days (total possible=80 observational days) • Turbine 40 shut down August 8 -Aug 29 (22 days) due to mechanical issues. • Model does not accurately forecast “worstcase scenario” for this property. Modeled 40 days Observed Shut Down Modeled 40 days

Observations at Stiles Rd. • Flicker here is stronger/clearer compared to other sites (Level 3). • Flicker observation clearly above modeled worst case • No trees to obstruct flicker to living room window and upstairs windows. • Recommend future observation.

3. HOW FAR IS THE FLICKER SHADOW? • Max distance observed is about 6, 000 feet. • As distance increases, light diffuses reducing appearance of shadow and potential for obstruction (trees, topo, buildings) increases. • Adjustments to model if flicker over 5400 feet done on a case-by-case basis.

CONCLUSION: ARE TURBINES PAUSING THAT ARE PROGRAMED TO? 1. Turbines pause majority of the time at expected time period. 2. Exceptions exist- VSDS and model not perfect. 3. Morning flicker – trees east of light sensors can obstruct light from reaching sensor and require manual override.

CONCLUSION: HOW MUCH FLICKER IS OCCURING COMPARED TO THE MODEL? • Cannot make broad assumptions about under estimation or over estimation of model. • In three cases*, the model under predicted worst case flicker in daily duration (event) and/or total estimated flicker. • In three cases, the model over predicted total estimated flicker when none exists due to obstructions or topography. * Data of private party included here. Two cases observed by County.

CONCLUSION: HOW MUCH FLICKER IS OCCURING COMPARED TO THE MODEL? • Turbines pause if modeled over 8 hours • Ordinance maximum 10 hours • 20% factor included for climatic variation (8 to 10 hours) • 20% factor included in probability (. 12 to. 15) • Kistler road, minutes observed (8. 5 hours) exceeds model (7. 1 hours)- not a zoning violation.

Moving Forward • Where the model under predicts “worst case scenario” it must be corrected to be valid and statistically useful. • With flicker complaints, immediate and continued response is needed (not 30 -days). • Correct model/obtain more data in order to move toward valid “sampling” at sites. • Some sites need further study- Level 3 types where impact is greatest (Stiles Road)

Questions? mreilly@masoncounty. net (231) 757 -9272