Reducing Wind Forecasting Uncertainty with PI and ECHO<br />Using Real-time Offsite 	Meteorological Observations<br />Mart...
PI-enabled Real Time Off-site Meteorological Observations for Improved Short-term Wind Forecasting, Operating Economics an...
Topics<br />Brief introductions<br />The importance of forecasting<br />Wind power: ups and downs<br />The impact of remot...
WINData background<br />Started by Marty Wilde<br />In wind energy 1991<br />Wind energy specialists<br />From raw dirt to...
Why forecasting is important<br />
The critical role of forecasting<br />Wind penetration<br />California: <br />20% renewables by 2010<br />33% renewables b...
Working together to improve forecasts<br />Goal: Designing a program that 			results in better  forecasts<br />WINData &AW...
Bonneville Power’s balancing act<br />1,500 MW<br />Balancing Area<br />19 Wind<br />Farms<br />
BPA’s Balancing Area<br />
Scheduling wind operations is complex<br />
A series of events at BPA<br /><ul><li>Wind power  15% of peak load
Large deviations power use of Portland, OR</li></ul>Theoretical Scheduled MW<br />730 MW<br />680 MW<br />400 MW<br />625...
Consequences<br /><ul><li>Wind power is backed by another generation asset
Unintended results</li></ul>Theoretical Scheduled MW<br />Violation of:<br /><ul><li>WECC Control Standards
EPA Regulations</li></ul>0 MW<br />
Ramp Events<br />Not your friend.<br />
Ramp events captured by ERCOT (1)<br />
Ramp events captured by ERCOT (2)<br />
Ramp effects at ERCOT<br />News article from Reuters (2/27/08):<br />Sudden 1,700 to 300 MW drop in wind power<br />Increa...
Scheduling around ramp events<br />Hour ahead schedules -forecast interpretation<br />Forecasts are good at predicting tha...
WINData’s Plan<br />Introducing WINDataNOW<br />
Goal: Reduce the forecast uncertainty<br />Idea<br />Can short-term forecasts be improved by using actual upstream measure...
WINData’s technology<br />Create a modern, remote “met” sensing system<br />Combine off the shelf technologies:<br />PI Sy...
WINDataNOW! – remote met observations<br />Towers are strategically located<br />Transmit high fidelity “line of site” dat...
Tower placement<br />Prevailing <br />Wind Direction<br />
Tower placement<br />Wind Speed<br />meters / sec<br />Wind Speed<br />meters / sec<br />Wind Speed<br />meters / sec<br /...
Anticipate changes<br />Ramp up<br />Ramp down<br />Improved wind energy integration on the grid<br />More data: smoother ...
Example<br />A Site in Montana<br />
Today’s data expectations<br />(m/s)<br />Typical industry-standard data logger output<br />
The same time period, now in high fidelity<br />Actual Wind Speed (m/s)<br />
Comparison<br />Average vs. Actual Wind Speed (m/s)<br />Averages vs. higher fidelity data<br />
Using PI’s 10m Averages<br />
Using PI’s 5m Averages<br />
Zooming out a little more<br />High wind speed cut-off<br />
Increased meteorological insight<br />Re-forecast<br />Account for actual upstream observations<br />Decrease forecast err...
C-Gorge Project<br />WINData in <br />The Columbia Gorge<br />
Map of C-Gorge vs. existing sites<br />
WDN Locations vs. BPA towers<br />Goodnoe<br />7 Mile Hill<br />WDN1<br />WDN2<br />Wasco<br />
How WINDataNOW! works<br />From full-service to self-service<br />Ways of engaging WINData<br />Project management<br />De...
Subscriber architecture<br />*<br />Subscriber 1<br />Connection<br />OffsitePI System<br />WINData<br />Subscriber 2<br /...
Owner architecture<br /><ul><li>Installation by WINData
One-time setup and validation
Upcoming SlideShare
Loading in...5
×

More Reliable Wind Power Forecasting - OSIsoft Users Conference

1,226

Published on

The plan for improving wind power forecasting by using upwind meteorological observations.

Published in: Technology, Business
0 Comments
1 Like
Statistics
Notes
  • Be the first to comment

No Downloads
Views
Total Views
1,226
On Slideshare
0
From Embeds
0
Number of Embeds
1
Actions
Shares
0
Downloads
0
Comments
0
Likes
1
Embeds 0
No embeds

No notes for slide
  • In order to turn wind into a fuel source, it has to be forecast reliably and reliablyMaintain Supply > DemandWind energy adds uncertaintyWhen will the wind blow?How fast?When will it stop?How do I balance it?System operator headachesUnanticipated demandUnanticipated supply problemsParticipants do not meet commitmentsUnplanned outagesHigh spot market pricesEnronIntermittent renewables
  • Bonneville Power Authority19 wind farms interconnected (1,000 turbines)Potentially 6,000 MW online by 2009 11 years ahead of schedule
  • In the spring of 2008 BPA’s balancing area:Generated 400MW over schedule for several hoursIn August 2008 BPA’s balancing area:Generated 730 MW over schedule in 1hrGenerated 680 MW over schedule in 1hrIn September 2008 BPA’s balancing area:Generated 625 MW under schedule in 1hr
  • Wind power is “backed” by another type of utilityCombined cycle gas turbinesHydroelectric plantsEtc.Spring 2008 excess power event resulted inWECC control standard violations (fines)Hydro dam over-spill; endangering fish (EPA fines)
  • Sudden increases or decreases in wind speedIncreases usually better than decreasesSubject to seasonal variationDifficult to forecast in the short-termEffects:Operating at reduced or conservative capacityInability to participate in day and hour-ahead marketsPurchase of make-up power at high pricesDispatcher headaches
  • Another example:News article from Reuters (2/27/08):Sudden 1,700 to 300 MW drop in wind powerIncreased demand due to colder temperaturesOrdered 1,100 MW to be curtailed from interruptible customers within 10 minOther suppliers were unable to meet commitmentsHappy endingNo other customers lost powerInterruptible customers were back online 90 min later
  • Most utilities and operators own a PI SystemIs it deployed?
  • “Line of Site”: WINData’s met tower configurations are to be in line with your operations site in the predominant wind corridors.
  • More than a year of development and testingSeveral trial sites to test equipmentComparisons and validation against existing market playersNow in an industry validation phase with utility partnersWINDataNOW installation in the Columbia River Valley in OregonStarted January, 2009Upstream from large wind farm installationsPhase 1 includes installations in 2 locationsPhase 2 & 3 brings online 3 more locations
  • Transcript of "More Reliable Wind Power Forecasting - OSIsoft Users Conference"

    1. 1. Reducing Wind Forecasting Uncertainty with PI and ECHO<br />Using Real-time Offsite Meteorological Observations<br />Marty Wilde – CEO, Principal EngineerWINData, Inc.<br />Gregg Le Blanc – Flunky<br />WINData, Inc.<br /> F5Direct, LP<br />
    2. 2. PI-enabled Real Time Off-site Meteorological Observations for Improved Short-term Wind Forecasting, Operating Economics and System Reliability<br />Marty Wilde – CEO, Principal EngineerWINData, Inc.<br />Gregg Le Blanc – Flunky<br />WINData, Inc.<br /> F5Direct, LP<br />
    3. 3. Topics<br />Brief introductions<br />The importance of forecasting<br />Wind power: ups and downs<br />The impact of remote observations<br />How WINData leverages ECHO and PI<br />Future plans<br />
    4. 4. WINData background<br />Started by Marty Wilde<br />In wind energy 1991<br />Wind energy specialists<br />From raw dirt to developed wind farm<br />Over 500 meteorological tower installs<br />Specializing in high-end instrumentation and tall towers<br />Goal: <br />Advance the “state-of-the-art” by using real time measurements for model and investment validation<br />
    5. 5. Why forecasting is important<br />
    6. 6. The critical role of forecasting<br />Wind penetration<br />California: <br />20% renewables by 2010<br />33% renewables by 2020<br />Europe:<br />20% renewables by 2020<br />Already in place<br />Denmark<br />20% of demand today<br />50% by 2025<br />Spain<br />20,000 MW capacity by 2010<br />&gt; 15,000 MW in place today<br />Peaked at 40% of demand (typically ~25% of demand)<br />
    7. 7. Working together to improve forecasts<br />Goal: Designing a program that results in better forecasts<br />WINData &AWS Truewind<br />Teaming up with utilities to improve system operations<br />Better wind energy integration<br />
    8. 8. Bonneville Power’s balancing act<br />1,500 MW<br />Balancing Area<br />19 Wind<br />Farms<br />
    9. 9. BPA’s Balancing Area<br />
    10. 10. Scheduling wind operations is complex<br />
    11. 11. A series of events at BPA<br /><ul><li>Wind power  15% of peak load
    12. 12. Large deviations power use of Portland, OR</li></ul>Theoretical Scheduled MW<br />730 MW<br />680 MW<br />400 MW<br />625 MW<br />0 MW<br />
    13. 13. Consequences<br /><ul><li>Wind power is backed by another generation asset
    14. 14. Unintended results</li></ul>Theoretical Scheduled MW<br />Violation of:<br /><ul><li>WECC Control Standards
    15. 15. EPA Regulations</li></ul>0 MW<br />
    16. 16. Ramp Events<br />Not your friend.<br />
    17. 17. Ramp events captured by ERCOT (1)<br />
    18. 18. Ramp events captured by ERCOT (2)<br />
    19. 19. Ramp effects at ERCOT<br />News article from Reuters (2/27/08):<br />Sudden 1,700 to 300 MW drop in wind power<br />Increased demand due to colder temperatures<br />Ordered 1,100 MW to be curtailed from interruptible customers within 10 min<br />Other suppliers were unable to meet commitments<br />Happy ending<br />No other customers lost power<br />Interruptible customers were back online 90 min later<br />
    20. 20. Scheduling around ramp events<br />Hour ahead schedules -forecast interpretation<br />Forecasts are good at predicting that there WILL be a ramp event<br />When will it arrive?<br />Phase error<br />How significant?<br />Amplitude error<br />Forecast error tends to increase around ramp events<br />
    21. 21. WINData’s Plan<br />Introducing WINDataNOW<br />
    22. 22. Goal: Reduce the forecast uncertainty<br />Idea<br />Can short-term forecasts be improved by using actual upstream measurements from physical instruments?<br />Requirements<br />Off-site meteorological data from predominant wind directions<br />Deliver measurements with enough resolution and timeliness to be useful<br />
    23. 23. WINData’s technology<br />Create a modern, remote “met” sensing system<br />Combine off the shelf technologies:<br />PI System infrastructure <br />Sensors<br />Logger equipment<br />Focus on configuration vs. programming<br />
    24. 24. WINDataNOW! – remote met observations<br />Towers are strategically located<br />Transmit high fidelity “line of site” data<br />Use this data:<br />Collaborate with forecaster<br />Refine short-term forecasts<br />Back-cast to develop more detailed climatology<br />“Line of Site” Locations<br />Wind Farm<br />60 – 100 minutes<br />of prior notice<br />Upwind Met Tower<br />30 – 60 miles<br />upwind<br />
    25. 25. Tower placement<br />Prevailing <br />Wind Direction<br />
    26. 26. Tower placement<br />Wind Speed<br />meters / sec<br />Wind Speed<br />meters / sec<br />Wind Speed<br />meters / sec<br />Wind Speed<br />meters / sec<br />
    27. 27. Anticipate changes<br />Ramp up<br />Ramp down<br />Improved wind energy integration on the grid<br />More data: smoother operations<br />Apply “line of site” datato better understand near-term transients<br />Power Production vs.<br />Augmented forecast<br />MWh<br />Wind Speed vs. <br />Augmented forecast<br />meters / sec<br />
    28. 28. Example<br />A Site in Montana<br />
    29. 29. Today’s data expectations<br />(m/s)<br />Typical industry-standard data logger output<br />
    30. 30. The same time period, now in high fidelity<br />Actual Wind Speed (m/s)<br />
    31. 31. Comparison<br />Average vs. Actual Wind Speed (m/s)<br />Averages vs. higher fidelity data<br />
    32. 32. Using PI’s 10m Averages<br />
    33. 33. Using PI’s 5m Averages<br />
    34. 34. Zooming out a little more<br />High wind speed cut-off<br />
    35. 35. Increased meteorological insight<br />Re-forecast<br />Account for actual upstream observations<br />Decrease forecast error around ramp events<br />Advanced warning<br />Situational awareness<br />Reduce costs<br />Avoid penalties<br />Use this<br />Hour Ahead<br />To reduce this<br />
    36. 36. C-Gorge Project<br />WINData in <br />The Columbia Gorge<br />
    37. 37. Map of C-Gorge vs. existing sites<br />
    38. 38. WDN Locations vs. BPA towers<br />Goodnoe<br />7 Mile Hill<br />WDN1<br />WDN2<br />Wasco<br />
    39. 39. How WINDataNOW! works<br />From full-service to self-service<br />Ways of engaging WINData<br />Project management<br />Deployment & management of early warning met arrays<br />Enterprise-wide retrofit projects<br />“WindTelligence”<br />Analysis<br />Development of situational awareness tools and IP<br />Subscription service<br />Data from reference locations<br />Hosted central PI Server<br />Stream data to your local system<br />
    40. 40. Subscriber architecture<br />*<br />Subscriber 1<br />Connection<br />OffsitePI System<br />WINData<br />Subscriber 2<br />WINData Managed<br />Subscriber 3<br />WINData Managed<br />PI<br />SCADA<br />Your <br />Operations<br />* Not to scale<br />
    41. 41. Owner architecture<br /><ul><li>Installation by WINData
    42. 42. One-time setup and validation
    43. 43. Ownership and management by customer
    44. 44. Your own IP</li></ul>Connection<br />WINData<br />Hardware<br />Managed by you<br />PI<br />SCADA<br />Your <br />Operations<br />
    45. 45. Combination architecture<br />Other sites in the area<br />Subscriber 1<br />Connection<br />OffsitePI System<br />WINData<br />Hardware<br />Subscriber 2<br />WINData Managed<br />Subscriber 3<br />Managed by you<br />PI<br />SCADA<br />Your <br />Operations<br />
    46. 46. Acquisition architecture<br />Advantages<br />Higher resolution data<br />Rolls out quickly<br />OffsitePI System<br />Met Tower<br />Secure Connection<br />Solar Panel<br />Wireless<br />Connectivity<br />Architecture<br />Fits in place with existing PI infrastructure<br />Fault tolerant<br />Low power<br />WINData Acquisition<br />Hardware / Software<br />Data<br />Buffering<br />Data <br />Acquisition<br />Battery<br />
    47. 47. Data features<br />Focus on data confidence<br />Logger spools data if connectivity goes down<br />Supports “industry standard”<br />At least 2Hz data always streamed to PI<br />Keep the data at its original resolution<br />Tune the data with standard PI settings<br />10m averages always available via PI<br />
    48. 48. Benefits of working on OSIsoft’s platform<br />Component<br />PI Server<br />ECHO<br />New ECHO to PI Interface<br />Partner network<br />Advantage<br />Directly integrate with utilities and system operators<br />Spool data during connectivity outages<br />Speaks PINET over TCP/IP natively<br />Solution works immediately with partners like Transpara’sVisualKPI<br />
    49. 49. Demo<br />D.I.Y.<br />http://demo.transpara.com/<br />
    50. 50. Future possibilities<br />Smart Grid<br />Provide “fuel insight” to consumers<br />Encourage informed choices – demand side management<br />Better investments<br />Prospecting – High Fidelity “Bankable Data”<br />Improved climatology for new wind energy sites<br />Advanced wind farm design<br />Turbulence, FFT Analysis, Matrix correlations<br />Integrate shiny new technologies<br />Vertical profiling<br />SODAR, LIDAR, etc.<br />
    51. 51. Looking forward<br />Upstream observations can dramatically improve forecasts (Papers by Kristin Larson, 3TIER)<br />Significant interest from utilities, operators, and forecasters in short-term ramp event prediction<br />WINData wants to work with you<br />Contact Marty Wilde, Dave Roberts, or Gregg<br />
    52. 52. Special Thanks<br />OSIsoft for its support<br />Johnson City office and the embedded team for ECHO work<br />Dave Roberts, Business Development, and Sales for advocacy<br />ProSoft for its continued support<br />Transpara for use of it’s VisualKPI software<br />
    53. 53. Find us after the talk!<br />Marty Wilde<br />marty.wilde@windata-inc.com <br />Gregg Le Blanc<br />gregg.leblanc@windata-inc.com<br />gregg@f5direct.com<br />Dave Roberts – OSIsoft<br />droberts@osisoft.com<br />

    ×