Using resource based slicing to incorporate the effects of intermittency in large scale energy models

Transcription

1 Using resource based slicing to incorporate the effects of intermittency in large scale energy models Mariliis Lehtveer IEW, Cork, Ireland

2 One week in Western Denmark (~40% of electricity from ) Mariliis Lehtveer - IEW 2016, Cork, Ireland 2

3 Principles We analyse global historic speed and solar irradiation data from European Centre for - Range Weather Forecasts (ECMWF) We separate 10 typical situations combining high, medium, low and extremely and high, medium and low solar irradiation This results in different load factors for and solar PV for each situation that can be used in large scale long term energy models Mariliis Lehtveer - IEW 2016, Cork, Ireland 3

4 Data From the European Centre for -Range Weather Forecasts (ECMWF) ERA Interim data set geographic resolution of covering the whole surface of the planet 3 hour resolution for solar irradiation 6 hour resolution for, 125m above ground 1 year of data used at a time Mariliis Lehtveer - IEW 2016, Cork, Ireland 4

5 Data filtering Sights with yearly average speed below 5 m/s excluded Population: Population mask from CIESIN Population Density Grid Future Estimates, v3 2015, 0.5 resolution Pixels with distance greater than 500km to any populated pixel excluded (10p/km2). Wind excludes pixels with population over 500 persons per km2 and below 10 persons per km2 Solar PV is placed in urban areas first Population density [1/km 2 ] Solar PV Wind onshore <10 Within 500 km Within 500 km, if speed >5m/s 10< and <500 Yes Yes, if speed >5m/s >500 Preferred No Mariliis Lehtveer - IEW 2016, Cork, Ireland 5

6 Clustering to avoid optimisation A k-means algorithm is used to cluster the pixels in each region Number of clusters and initial central points given Mariliis Lehtveer - IEW 2016, Cork, Ireland 6

7 Allocation of capacity Allocates given amount of EJs to a region Allocation to clusters by population (30%) and resource quality (70%) Allocation inside the cluster based on resource quality. A base capacity of 250 kw/km 2 for 2.5% land area for solar Solar PV efficiency 15%, facing south/north, tilt equal to latitude Wind turbine data from J.R. McLean, Equivalent Wind Power Curves, D2.4 deliverable, Mariliis Lehtveer - IEW 2016, Cork, Ireland 7

8 Mariliis Lehtveer - IEW 2016, Cork, Ireland 8

9 Resulting load factors medium high solar, low solar, medium solar, high medium high extremely Wind AFR Wind CPA Wind EUR Wind FSU Wind LAM Wind MEA Wind NAM Wind PAO Wind PAS Wind SAS Solar AFR Solar CPA Solar EUR Solar FSU Solar LAM Solar MEA Solar NAM Solar PAO Solar PAS Solar SAS Mariliis Lehtveer - IEW 2016, Cork, Ireland 9

10 Implementation in Global Energy Transitions (GET) model A cost minimizing systems engineering model of the global energy system Set up as a linear programming problem Five end use sectors: electricity, transport, feedstock, residential commercial heat and industrial process heat 10 regions Time span 100 years Daytime demand 15% higher than night Mariliis Lehtveer - IEW 2016, Cork, Ireland 10

11 New constraint Share of maximum output that must be run during the whole time period if technology is used. Aggregate part-load Biomass PP 0.35 Oil PP 0.1 Gas PP 0.1 Coal PP 0.35 LWR 0.7 FBF 0.7 MOX 0.7 Hydro PP 0.1 Mariliis Lehtveer - IEW 2016, Cork, Ireland 11

12 EJ/yr Example region without storage Curtailment 50 Solar 40 Wind FBR 10 LWR 0 medium high solar, solar, medium solar, high medium high ext. Hydrogen Hydro Mariliis Lehtveer - IEW 2016, Cork, Ireland 12

13 Storage We analysed the time connections between modelled 10 and solar resource availability situations and use that information to model electricity storage. Example: NAM, 12 hour storage From To medium high solar, low solar, medium solar, high medium high extremely medium high solar, low solar, medium solar, high medium high extremely Mariliis Lehtveer - IEW 2016, Cork, Ireland 13

14 EJ/yr Example region with storage Curtailment From storage Solar 40 Wind 30 FBR 20 LWR 10 Hydrogen medium high solar, solar, medium solar, high medium ext. high Hydro To storage Mariliis Lehtveer - IEW 2016, Cork, Ireland 14

15 Global electricity supply 450ppm target 1 slice 10 slices 10 slices with storage Mariliis Lehtveer - IEW 2016, Cork, Ireland 15

16 Sensitivity analysis Level I Level II Level III Level IV Costs in US$(2010) Wind Solar PV LWR FBR Storage 12h No storage Storage 48h No storage Storage 2w No storage Storage 2m No storage Mariliis Lehtveer - IEW 2016, Cork, Ireland 16

17 Conclusions Our results show that this approach manages to capture many features introduced by variable renewables such as need for flexible generation capacity and curtailment at high penetration levels. Adding electricity storage to the system can favour solar power but has only a minor effect on and nuclear power. Mariliis Lehtveer - IEW 2016, Cork, Ireland 17

18 Planned developments Wind and solar classes Investment rules inside the model Improved linkage to transport and heat sector Mariliis Lehtveer - IEW 2016, Cork, Ireland 18

19 Thank you for your attention! Mariliis Lehtveer, PhD Physical Resource Theory Department of Energy and Environment Chalmers University of Technology Göteborg, SWEDEN ph: Mariliis Lehtveer - IEW 2016, Cork, Ireland 19