Benefits of on-site real-time solar forecasting with CloudCAM

Transcription

1 Benefits of on-site real-time solar forecasting with CloudCAM Colin Bonner, Fulcrum3D Sodar Cloud Monitoring Solar Forecasting Environmental Monitoring 10 October 2016 All Energy 2016 Page 1

2 Monitoring for renewable energy Compact-beam Sodar CloudCAM solar forecasting High speed data loggers Flightdeck data centre low cost, flexible, reliable monitoring solution innovative camerabased cloud monitoring system unique design allows sampling rate up to 200Hz data delivery, analysis & asset management performance verified by industry experts optimised for the wind energy industry cloud detection and cloud statistics short term cloud prediction (up to 30 min) noise and weather monitoring high speed solar monitoring wind monitoring online fault reports and error detection 24hr secure access web server Fulcrum3D and 3 rd party products Page 2

3 The problem: solar variability solar power output is can be variable due to clouds output can drop or rise by ~80% in a matter of a seconds major problem as PV penetration increases on microgrids or weakgrids. current solution are: storage to provide support during a cloud events, or excessive spinning reserve in case of cloud event limit penetration Page 3

4 The solution: CloudCAM real-time cloud detection & solar forecasting integrate with SCADA via Modbus completely standalone system optional emote access via Flightdeck Focusing on grid stability / ramprate limiting and spinning reserve control. Opportunity for LCOE, bidding etc Page 4

5 CloudCAM images and detection Clearsky: no cloud detected Note camera flare, dirt etc removed Cloud: robust cloud detected and identified Site specific horizon mask removes impacts of local vegetation etc Page 5

6 All Sky image Cloud detection Page 6

7 Spinning reserve management Reduce spinning reserve to save fuel if we know there are no clouds in area. Smaller diesel for less time: Operational maintenance saving Direct fuel savings For example, switching from a 640kW set to a 320kW set would save around 12L/h, or reduce spill of PV by reducing the minimum diesel loading constraint, e.g. switching down from a 640kW at 40% minimum loading to a 320kW set at 80% loading would save -- as an extremely rough estimate -- 20L/h for 6h/d for 250d/y at 1$/L = 30000$/y * ROI of ~1 year for 500~1000kW micro grids without getting too fancy. * Rough estimate based on NT Power Water Corp tenders Page 7

8 Curtail PV to increase yield? Batteries and inverters are cool but they still obey the first law of thermodynamics PV IN System out Battery Power Page 8

9 Curtail PV to increase yield? Batteries and inverters are cool but they still obey the first law of thermodynamics PV IN System out Battery Power Page 9

10 Curtail PV to increase yield? Batteries and inverters are cool but they still obey the first law of thermodynamics. Pre-emptive ramp down PV IN System out Battery Power Page 10

11 Curtail PV to increase yield? Reduction in output from pre-emptive curtailment is less than the power used to charge batteries as charger, battery chemistry and inverter inefficiencies are removed PV IN System out Battery Power System out (CloudCAM) Page 11

12 Benefits of solar forecasting Forecasting make sense for low-energy high-power storage systems Efficiency driver: Use of solar forecasting can increases system yield by reducing use of lossy elements (batteries, chargers, inverters, quiescent fly-wheel load) Financial drivers: CAPEX: Reduces size of battery required ( current thought is ~25% ) OPEX: Reduces cycling through battery to extend their life Spinning reserve management direct maintenance and fuel savings Cost of CloudCAM is in the noise for utility scale solar projects Page 12

13 CloudCAM project timeline Started development 2012, first system in field 2013, ARENA funding in 2014, commercialisation 2016 Deployment at 6 sites to date CloudCAM in real-time control system at three sites (4 th end Oct 2016) Strong pipeline ahead Page 13

14 Case study: TKLN TKLN sites are three high penetration sites on NT Power Water Corp microgrids. Sites are owned and operated by Epuron. Operational since 2013 with up to 90% instantaneous PV penetration, designed for 30% energy. AC coupled batteries for ramp-rate limiting during cloud events. Battery designed life of 10 years (GEL Lead Acid) Greater than expected degradation after ~4 years Retrofitted CloudCAM without modifying PLC or previously commissioned systems. CloudCAM in direct control of PV array Ran CloudCAM in passive mode for before switching live Page 14

15 Case study: TKLN TKLN sites are three high penetration sites on NT Power Water Corp microgrids. Sites are owned and operated by Epuron. Page 15

16 Case study: TKLN PWC Set point Tripower Tripower PLC Modbus SMA Cluster Controller Tripower Tripower Batteries Battery Inverters Metering point Page 16

17 Case study: TKLN PWC Set point Tripower Tripower PLC Modbus CloudCAM Modbus SMA Cluster Controller Tripower Tripower Batteries Battery Inverters Metering point Page 17

18 CloudCAM installation at Ti Tree, NT Page 18

19 Case study: TKLN Cloud detected; predictive ramp-down commences; cloud clears; ramp up and resume normal operation Cloud detected; predictive ramp-down occurs until solar output reaches safe level Cloud clears; ramp up and resume normal operation Setpoint (Max kw) PV Output (kw) System Output (kw) Battery output (kw) Possible solar Output (kw) safe level pre-set Note different ramp-up and ramp-down rates are allowed for in this design. Page 19

20 Case study: TKLN Significant fluctuations avoided; minor loss of production; elimination of penalties Setpoint (Max kw) PV Output (kw) System Output (kw) Battery output (kw) Possible solar Output (kw) Elimination of battery cycling and battery capacity required. Page 20

21 Case study: TKLN Significant fluctuations avoided; minor loss of production; elimination of penalties Setpoint (Max kw) PV Output (kw) System Output (kw) Battery output (kw) Possible solar Output (kw) Dramatic reduction in battery cycling; dramatic reduction in battery capacity required (max 50kW battery to cover a 250kW potential step change) Page 21

22 Case study: TKLN Significant reduction in penalties Reduced battery use by 90% Yield (normalised to irradiance) increased 4~5% * * Daniel Gilbert March 2016 RAPS conference Page 22

23 Case study: Karratha Airport ARENA funded system with storage and solar forecasting on Horizon network Power topology similar to TKLN, however, CloudCAM supplied as black box to integrator and PLC controls PV. Commissioning Q Page 23

24 ARENA knowledge sharing 1 Hz pyranometer data (K&Z SMP11) from test sites: Requires free registration for access to static archived data. Terms and conditions on website. We ll be adding more data with a typical delay of a month or so. Visit F3D at stand 2850 Page 24

25 Level 11, 75 Miller St North Sydney NSW 2060 AUSTRALIA (2) October 2016 Page 25