National Target for Renewable Energy in Electricity in 2020 is 40 % of net electricity generation

Transcription

1 Probabilistic Generation Expansion Planning under Wide- Scale Variable Renewable Energy Penetration Regular ETSAP Workshop Saturday July 9, 2011 Stanford University Presentation by K. Tigas K. Tigas, J. Mantzaris, N. Sakelaridis, C. Nakos, G. Giannakidis Backround Information Present work was carried out in the context of the implementation of the National Renewable Energy Action Plan and Energy Roadmaps 2050 National Target for Renewable Energy in Electricity in 2020 is 40 % of net electricity generation Slide 2 1

2 Methodology Overview 1 Running TIMES model, calculating electrical energy demand evolution and least cost electricity generation technologies mix under the environmental and energy policy constraints Slide 3 Methodology Overview 2 Implementing load statistics and formulating load curves for the time horizon of the models Implementing RES and CHP generation statistics and projecting the results in the time horizon of the models Slide 4 2

3 Methodology Overview 3 Running COST-P model for Probabilistic Production Costing after modifications for : incorporation of Renewables based on Residual Load Duration Curves reliability analysis based on the statistics of RES and small CHP storage requirements assesment calculating electricity generation costs taking into consideration the balancing costs due to RES penetration Slide 5 Methodology Details 1-use of TIMES model TIMES model solution is providing : Electrical energy consumption evolution Least cost electricity generation technologies mix to meet the energy demand taking into consideration : Environmental and energy policy constraints Slide 6 3

4 Methodology Details 2-use of TIMES model TIMES model solution is not providing : Accurate load calculations Generation system reliablility calculations Storage calculation Slide 7 Methodology Details 3-use of COST-P model COST-P model for Probabilistic Generation Analysis: It is being developed by the Public Power Corporation and CRES Two plant categories are considered : Dispatchable (Thermal, Dam Type Hydro, Pumped Storage) Non-dispatchable (Wind, PV, Run off River(small hydro), small CHP) Slide 8 4

5 Methodology Details 4-use of COST-P model COST-P model operation : Based on the principle of Residual Load Duration Curves Residual Load Duration Curve : Origin in the 80s They are related to the Probabilistic Production Costing methodologies for thermal and hydro Slide 9 Methodology Details 5-use of COST-P model COST-P model operation : Residual Load Duration Curves Having as input the customer load curve and the statistics of variable generation one derives the remaining load for thermal and hydro Slide 10 5

6 Methodology Details 6-Residual Load Duration Curve Load Duration Curve: Is transformed to a curve of the form : Probability (Load > existing capacity) = value Slide 11 Methodology Details 7-Residual Load Duration Curve Slide 12 6

7 Methodology Details 8-Residual Load Duration Curve Residual Load Duration Curve: How it is obtained? Based on the fact that it represents the probability : Prob (L customer > P thermal - P dthydro -P shydro -P wind -P pv -P shydro - P chp )=ε It can be formulated using the convolution formula but you have to decorrelate random variables first Slide 13 Methodology Details 9-Residual Load Duration Curve Residual Load Duration Curve: Customer Load CDF Wind Generation CDF and Slide 14 7

8 Methodology Details 10-Residual Load Duration Curve Residual Load Duration Curve: Customer Load Wind Generation CDF Slide 15 Methodology Details 11-Residual Load Duration Curve Residual Load Duration Curve: To decorrelate random variables one can use time or load zones, apply the equations in the zones and then use the equation Where p(x/y i ) is probability given that a random variable is in zone y i Slide 16 8

9 Methodology Details 12-Reserve Capacity Assesment It is related to : Reserve needed due to Residual Negative Load Reserve due to Residual Peak Load Reserve due to Residual Load Variation Reserve related to unit Ramp Rates Slide 17 Methodology Details 13-Reserve Capacity Assesment Negative load occurs due to the thermal minimum of thermal plants Triangle area corresponds to energy that will be curtailed if it is not stored or transmitted P a = P th_min L base for pumped storage : Slide 18 9

10 Methodology Details 14-Reserve Capacity Assesment The area of the triangle is the energy that will be curtailed if not stored or transmitted Storage available capacity is equal to Pa for zero curtailement To calculate storage we can resolve an equation related to the residual load duration curve: Prob{L residual <C th_min }=ε to calculate P a displaced to the left Slide 19 Methodology Details 15-Reserve Capacity Assesment For reserve related to Peak Load we use the CDF of the residual load duration curve to resolve : Prob(L residual > C available )=ε Slide 20 10

11 Methodology Details 16-Reserve Capacity Assesment Spinning Reserve required due to Residual Load Variation Maintaining a constant index of reliability on an hourly basis Per hour probability for a generator trip (dispatchable) : Per hour residual load variation (assuming that it is a Gaussian random variable) : Slide 21 Methodology Details 17-Reserve Capacity Assesment Spinning Reserve required due to Residual Load Variation The probability for Load Shedding per hour is assumed constant depending on the desired level of reliability A load shedding incident can happen : by having just a residual load variation greater than the system reserve level by having just one generator trip and an unforecasted residual load variation greater than the system reserve level by having a generator trip and an unforecasted residual load variation some time directly after a previous generator trip The probability of shedding load during a normal hour of system operation (PLSNO) is one of the terms describing PLS h and can be presented as follows : Slide 22 11

12 Methodology Details 18-Reserve Capacity Assesment Spinning Reserve required due to Residual Load Variation PLSNO h = G i=1 1 FOP i,h 1 Φ R h σ total,h G + FOP i,h i=1 G j =1 j i 1 FOP j,h 1 Φ R h Pnαfo i,h σ total,h Slide 23 Methodology Details 19-Reserve Capacity Assesment Spinning Reserve required due to Rump Rates After formulating the residual load variation curve (Gaussian or polynomial) : Total up ramp of dispatchable plants > Down ramp of residual load variation Total down ramp of dispatchable plants > Up ramp of residual load variation Total up ramp of storage plants > Up ramp of residual load variation Slide 24 12

13 Methodology Details 20-Generation Cost Calculation For each year calculation of the Long Run Average Cost of the electricity produced using the equation : Slide 25 Results 1 NREAP 2020 Residual Load Duration Curve Slide 26 13

14 Results 2 NREAP 2020 Residual Load Duration Curve Slide 27 Results 3 NREAP 2020 Storage Capacity Calculation vs Level of VRE Slide 28 14

15 Results 4 NREAP 2020 Reserve Capacity Calculation vs Level of VRE Slide 29 Results 5 NREAP 2020 Long Run Average Cost vs VRE penetration Slide 30 15

16 Results 6 NREAP 2020 Long Run Average Cost vs VRE penetration Slide 31 Results 7 NREAP 2020 Long Run Average Cost vs VRE penetration Slide 32 16

17 Results 8 NREAP 2020 Effect of Large Scale Penetration to mid merit plants Slide 33 Next Step Optimization Better combination of this methodology with TIMES Slide 34 17