Loading...
HomeMy WebLinkAboutAlaska Wind Program Highlights Canada-USCleanEnergyDialogue 10-01-2015-WPhoto by: Cassandra Cerny, GVEA Alaska Wind Program Highlights Canada-US Clean Energy Dialogue Oct. 1, 2015 Failing to fully consider, model and design secondary loads in hybrid wind systems ensures a 15-20 point gap from expected annual energy production. 2 Secondary Heat Loads –Critical to Project Success New wind penetration classes: Impacts of curtailment: *Max wind = village demand –min diesel loading + diversion load Installed Wind Capacity (kW) Total Wind Energy Produced (kWh) Excess Electricity Net Elec kWh Net Thermal kWh Control Method Fuel Savings @ $4.5.gal Potential Benefit 300 (Hi Pen)888,180 292,307 595,873 292,307 Elec Boiler or ETS units $240,274.89 100.00% 300 (Hi Pen)888,180 292,307 595,873 0 Turbine max setpoint $206,263.73 85.84% 300 (Hi Pen)888,180 292,307 595,873 0 Non value dump load $206,263.73 85.84% 300 (Hi Pen)489,227 0 489,227 0 Curtailment $169,347.81 70.48% 300 (Hi Pen)888,180 262,731 625,449 0 15-min Batt/FW storage $216,501.58 90.11% 200 (Med Pen)592,117 107,310 484,807 107,310 Elec Boiler or ETS units $180,303.78 100.00% 200 (Med Pen)592,117 107,310 484,807 0 Turbine max setpoint $167,817.81 93.08% 200 (Med Pen)592,117 107,310 484,807 0 Non value dump load $167,817.81 93.08% 200 (Med Pen)396,716 0 396,716 0 Curtailment $137,324.77 76.16% 200 (Med Pen)592,117 90,975 501,142 0 15-min Batt/FW storage $173,472.23 96.21% Simply comparing annual heat demand with annual excess energy leads to significant error in system design. While the health clinic in this village consumes almost twice as much energy over the course of a year, the heat load is much less variable than and doesn’t coincide with the excess wind. Additional heat loads must be added to the system design to avoid significant curtailment of wind turbines. 3 Modeling of Thermal Systems Community building/load Connected to HR Loop? Current annual heating oil consumption* Thermal mass - Equiv. gals. of storage MMBTU Equiv kWh Equiv Average kW Design Day Heat Public Works-HEMF Y 19,216 2,652 743,163 84.84 Suspect boiler setpoint set above level to gain benefit from HR loop. Estimate 20% of total is unmet. Sewer Plant Y 13,695 1,890 529,639 60.46 Estimate 20% of total load is unmet School N 116,800 16,118 4,517,240 515.67 1840000 PSO N 6,348 876 245,502 28.03 100000 <BTU/Hr Health clinic N 14,219 1,962 549,925 62.78 224000 <BTU/Hr Water plant N 11,426 1,577 441,904 50.45 180000 <BTU/Hr Fire Station N 16,758 2,313 648,126 73.99 264000 <BTU/Hr Power plant Y 1,625 224 62,847 7.17 Estimate 20% of total load is unmet 0 0 0.00 0 0 0.00 Totals 200,087 27,612 7,738,346 883.37 331,107 <<Excess kWh from HOMER ^^ Poorly matched excess vs. heat load Because wind energy is variable, there are times throughout the year when there is more energy available (turquoise = excess) from the wind turbines (purple) than the current net* village electrical load (gold). 4 Detailed modeling of electric load, heat load and wind energy Thermal loads (gold) for buildings and facilities in a community can make use of this excess wind energy (turquoise) to supplement other sources (power plant heat recovery or oil-fired boiler). Reasonably well-matched excess and load: *Max wind = village demand –min diesel loading + diversion load 5 Modeling building thermal loads is easy Pull records on annual fuel deliveries for large community buildings. AKWarm estimates work too. Total annual building heat loss = total heating fuel consumed per year minus boiler inefficiency AEA can pull ASOS/AWOS data to build an hourly temperature profile for the community and calculate hourly delta T and equivalent hourly heating oil gallons or kilowatt-hours. Most village buildings do not harness significant passive solar gain, so the model can remain simple. The model doesn’t need to be exact –just good enough to compare relative loads vs. excess power. 6 Water Systems: More Complex to Model There are more factors to model, but we can still answer the question: “Will the heat loads connected to our system even have a theoretical chance of taking all the power we can give them?” We still need to know the annual fuel consumption, but should also consider as many of the following additional factors: coldest/warmest water temperature at source, storage tank size and insulation, length diameter and insulation of circ. loop pipe, washeteria dryers and DHW load. Systems with storage tanks offer a buffer to take more heat now for possible benefit later. Do buildings and water systems already connect to a waste heat loop? Does proximity to the power plant allow this? Limitations to the method and model The model is far from perfect, but attempts to simulate a typical year for thermal load and excess wind energy. Think ballpark, not section/row/seat. Understand that the diurnal thermal profile (warmest during the day and coolest at night) is greatly diminished near and above the Arctic Circle (Lambert’s Law) both during the winter and the summer. Atmospheric blocking of solar energy is also magnified at higher latitudes for longer periods of the day (Beer’s Law). An efficient passive-solar building design will deviate more from the model –but only when the sun angle is above the atmosphere-blocking level (Beer’s Law) and facing the primary windows. This may impact new schools or hospitals, but we don’t presently have any passive-solar water treatment or sewer plants built in the Arctic. Passive- solar buildings must also have unobstructed southern exposure to maximize their benefit. Very drafty buildings can consider a wind-chill-based Delta T calculation rather than straight temperature. For structures combining building heat with water heat, an estimate will need to be made on the portion of fuel attributed to each function. Diurnal profiles for water heating will be driven by both Delta T as well as time-of-day usage. The farther away a building is from the reference AWOS station, the less accurate the model. In small villages, this shouldn’t be a problem. In large metropolitan areas like Anchorage or Fairbanks where inversion layers can affect parts of the city, but not others, temperatures could be off by as much as 20 degrees F at certain times. Consider opportunities to implement heat pumps or dispatchable electric loads. 7 8 Typical Permafrost Foundation –Thermopile with Concrete Cap Bethel Avg Temperature 2013-14 warmest on record 9 Warm Permafrost is a Concern 1950 1960 1970 1980 1990 2000 2010-5 -4 -3 -2 -1 0 1 2 Temperature  Cy=0.0245x-49.8932 Sensor configuration 10 Vibration Monitoring Snapshot of Unalakleet met tower (blue hues) and wind turbine anemometers (all other colors) showing variation but with major wind speed shifts in sync. Nov. 5 & 6 very weak and weak icing signals in Nome. Met tower anemometers (green and olive) follow nacelle 1 and 2 (light and dark blue) heated anemometers. Heated anemometers still indicate light winds when cup anemometers indicate no wind, but not in the range where the wind turbine will produce power. 11 Icing Study Weak icing signal in Nome. Met tower anemometers (green and olive) follow nacelle 1 and 2 (light and dark blue) heated anemometers. Wind turbine output (orange and red) has stopped during the icing period and resumes when wind speeds increase. Unalakleet post-icing event showing met tower anemometers (blue) flat lined during the icing period and then lagging the nacelle anemometers (green) for 10 hours. Wind turbine production (red) is zero during the icing period and post-icing for 8-16 hours. 12 Icing Impact on Wind Turbine Output Nome EWT-52 turbine #1 showing overlayed power curves for non-flagged (blue) periods, met tower icing events (orange), pre-icing reduced power production (green), post-icing reduced power production (yellow) and other possible events where no met tower icing was observed (grey). Unalakleet NPS-100 turbine #5 showing overlayed power curves for non-flagged (blue) periods, met tower icing events (orange), pre-icing reduced power production (green), post- icing reduced power production (yellow). Icing (orange) is masked by other plots/curves, but is essentially flat-lined at zero power output with wind speeds of 5 m/s or less. 13 Icing Impact on Wind Turbine Output Icing Impact on Wind Turbine Output When encountered in met tower data sets, icing data should be left “as -is” rather than deleting and resynthesizing the data values. The original data accurately represents the wind conditions in the case of hoar frost and represents the expected wind turbine output in the case of hoar frost and rime ice. Tables below show the expected error in energy production forecasting using the delete/synthesize method for the study period. 14 Cold-weather evaluation to test equipment accuracy and survivability. Light detection and ranging system weighs 60 kg. Remote power module weighs 375 kg. Deployed at Delta Wind Farm – Latitude 64 deg Very limited winter performance data due to warranty troubleshooting and repair. Stable performance since reinstall in May 2015 15 SpiDAR Evaluation After several sunny days of operation following the field install at Delta Wind Farm, the unit went offline on an overcast day. Similar outages were experienced the following week. An on-board heater consumes more power than the nominal 35W to run the LIDAR system. 1-min average loads of 89W were observed with instantaneous peaks of 305W. Power pack is only usable 5 months per year. 16 SpiDAR Power Pack Capabilities SpiDAR Data Accuracy SpiDAR data were compared to the reference data sets from the met tower at 30 meters (speed) and 50 meters (speed and direction) and the EWT turbines nacelle met station (speed and direction). Trend charts of 5030 rows of data indicated 27 10-minute records that contained outlier data which were subsequently deleted. Only the most extreme outliers were screened on the initial pass to assess data quality without preconceived limits. Pre and Post outlier removal is below. 17 Performing a Pearson correlation of 10-minute time- paired data, the results showed high correlation on temperature and average wind speed data between the SpiDAR and the met tower or wind turbine anemometers. Correlation drops in the wind speed standard deviation and wind direction measurements. An analysis of the distributions of wind direction show better overlap than the Pearson correlation of step-by- step changes. 18 SpiDAR Data Accuracy RFP issued with $20k to seed development of datalogger specifically designed to meet the needs of wind resource assessment in remote Alaska. Current offerings (12-15 data channels at $1800+ per unit) targeted at large wind farm resource assessment market. Winning design proposal has 3 anemometer channels and 1 vane, on-board temperature sensor, 1-min logging interval of date & time, min, max, average and std. dev for anemometer/vane and min, max, avg for temperature. .CSV format. Data cable inputs are spring-clip, providing for fast and reliable connection in harsh weather installations. Halus Power Systems is designer, manufacturer and supplier. Unit sells for $500-$650 depending on exact configuration/options. Datalogger unit at field test site in Palmer, AK showing controller board with SD card, spring-clip connectors and water-tight seals around cable intrusions. 19 Wind Datalogger for Alaska 20 Optional Buckland photos Rich Stromberg 907-771-3053 (desk) rstromberg@aidea.org AKEnergyAuthority.org 21