Quantification of energy losses caused by blade icing and the development of an Icing Loss Climatology Using SCADA data from Scandinavian wind farms Staffan Lindahl Winterwind 201 1 SAFER, SMARTER, GREENER
Blade icing loss factor [%] Annual energy loss due to icing [%] Individual turbine annual icing loss [%] Winterwind 2014 What we did 100 90 80 70 60 0 40 What we found 3. R² = 0.4498 3.0 2. 2.0 1. 1.0 0. 0.0 0 0 100 10 200 Hub elevation [m] relative an arbitrary datum Region 1 Region 2 Region 3 16.0 14.0 12.0 10.0 8.0 6.0 4.0 2.0 0.0 0 Effective hub height [m] 900 700 m 720 m 740 m 760 m 780 m 800 m Conclusions we drew SCADA data are great for quantifying energy loss caused by blade icing Losses in Scandinavia vary greatly, from close to 0% to more than 10% of annual energy production. Evidence of correlation between elevation and icing loss. Linear over small elevation range. Polynomial over large range? Potential scope for developing empirical relationship between icing losses and elevation 30 Jul Aug Sep Oct Nov Dec Jan Feb Mar Apr May Jun Season 2
Contents Review of operational data considered Re-cap on loss calculation method Specific investigations undertaken Influence of control strategy Inter-annual variability Importance of elevation update from Winterwind 2014 Conclusions 3
Data included Data from 30 wind turbines (+200) 18 Wind Farms (+8) Reasonable geographical coverage 10+ projects in Sweden < Projects in Norway < Projects in Finland Excludes projects where icing loss is managed manually Includes projects where: Turbines that shut down when controller detects icing Turbines that remain operational during blade icing events 4
Energy loss quantification Define Base-line power curves based on data for Normal operation only; The energy loss is defined by the Actual less the Expected production; An energy loss value is calculated for each each 10-minute record. Results in a database of Actual Power, Expected Power and an icing event log, for each turbine and each 10-minute record. Power deficit
Annual Icing loss - always shut down when iced [%] Influence of control strategy Question: What s the potential benefit of keeping turbines operational during icing events? Method: Simulate energy losses which would have been incurred during icing for projects which remain operational during icing events Compare actual to simulated losses Assumptions about sensitivity of controller ice detection required. Impact on loads not considered 2 20 1 10 0 0 10 1 20 Annual Icing loss - always operate when iced [%] 6
Annual Icing Loss [%] Icing loss error as percentage of annual energy Jan-08 Apr-08 Jul-08 Oct-08 Jan-09 Apr-09 Jul-09 Oct-09 Jan-10 Apr-10 Jul-10 Oct-10 Jan-11 Apr-11 Jul-11 Oct-11 Jan-12 Apr-12 Jul-12 Oct-12 Jan-13 Apr-13 Jul-13 Oct-13 Jan-14 Apr-14 Jul-14 Oct-14 Icing energy loss factor [%] Monthly production contribution [%] Inter-annual variability Question: How much do icing losses vary from year to year? Derive monthly icing loss for each wind farm Calculate annual icing loss (July to June) based on nominal production profile. Only projects with very long operating periods useful Inter-annual variability (IAV) defined by the coefficient of variation High mean loss coincides with low variability Low mean loss coincides with high variability Very long datasets required to accurately determine long-term mean losses 100 1 10 90 80 70 60 0 Time 0 0 2 0 7 100 Icing IAV [%] 6 4 3 2 1 12 11 10 9 8 7 6 Assumed production profile Annual rolling average (trailing) Annual mean (July to June) Monthly icing energy loss factor Mean 1%; IAV 30% Mean 3%; IAV 7% 0 0 10 1 Measurment period [years] 7
Annual Annual energy energy loss loss due due to icing to Icing [%] [%] Importance of elevation Questions: How do icing losses vary with altitude? Individual turbine mean annual losses calculated Correlation of loss vs. effective hub-height Strong relationship between loss and elevation throughout Sweden; Coastal Norway and Finland do not follow trend of Swedish sites, although data-sets are small. 20 20 1 1 10 10 Sweden Norway Region 1 Finland Region 2 Region 3 0 0 0 20 00 70 1000 0 20 00 70 1000 Effective Effective hub height hub-height [m] elevation 8
Icing loss climatology Relationship of annual icing loss and elevation used to define icing climatology Represents loss for projects were turbines remain operational through icing events. Geographical coverage limited by: Data availability Reliability of loss / elevation relation General experience of factors driving icing: cloud base elevation, Arctic / Siberian weather systems, Atlantic / Gulf stream effect. Uncertainty in loss estimate needs to be recognised High variability inevitable leads to high uncertainty Confidence elevated due to good length of datasets (6+ years) in combination with geographical diversity. 9
Conclusions Increased confidence in using elevation as a proxy for icing loss in Sweden Deemed sufficient to create an icing loss climatology covering most of Sweden. Initial results suggest the relationship is not applicable to Finland and coastal Norway. More data required to patch gaps in Sweden, and understand Norwegian and Finish conditions. Great potential for reducing energy loss by keeping turbines operational through icing conditions, rather than shutting down. Relative benefit diminishes with increasing icing loss. Impact on loads not considered here. Inter-annual variability in icing losses is very high. Measurement period in excess of years required to reduce loss prediction error below 2% of AEP. 10
End Staffan Lindahl, DNV GL Energy Staffan.Lindahl@dnvgl.com +44 117 972 9761 www.dnvgl.com SAFER, SMARTER, GREENER 11