Capacity value of intermittent generators Preliminary findings - PowerPoint PPT Presentation

Capacity value of intermittent generators Preliminary findings Market Regulations May 2018 1

Outline 1. Introduction 2. Capacity value assessment methods 3. Current method 4. Current issues in the SWIS Page 2

1.Introduction Page 3

Background • Capacity value: the contribution a capacity makes to system adequacy • Relevant Level Methodology – RLM • contribution of variable generation to system adequacy in the SWIS • The ERA is currently required to review the method every three years • IMO last reviewed the RLM in 2014 Page 4

Capacity valu lue outcomes • Significant change in the RLM • Previous method: average output of IGs • Change in method transitioned over 3 years • Current method: average output during high-risk periods Use of average output Current risk-based method 600 60% proportion of installed capacity (%) 500 50% 400 40% Capacity credits Capacity (MW) 300 30% 200 20% 100 10% 0 0% 2005-06 2006-07 2007-08 2008-09 2009-10 2010-11 2011-12 2012-13 2013-14 2014-15 2015-16 2016-17 2017-18 2018-19 2019-20 Installed Capacity Certified Capacity Accredited capacity as share of installed capacity Page 5

Basis is of capacity valu luatio ion • Effective load carry rying capability (ELCC): the amount of incremental load that a resource can serve without a change in the system reliability • ELCC considers: • probabilistic nature of generation output • random forced outages • Correlation between system random variables 6

Basis is of capacity valu luatio ion 3500 LOLP = 10% 3000 Available firm capacity 2500 2,400 MW Load (MW) 2000 1500 1000 500 0 100 200 300 400 500 600 700 800 900 1000 Interval 7

Capacity valu lue of fir irm capacity - example • Reliability target: LOLP=10% • Additional generation: 100 MW installed capacity (firm) 3500 LOLP = 10% Effective Load Carrying Capability (ELCC) = 100 MW 3000 Capacity value = 100% of the installed capacity 2500 Load (MW) 2,300 MW 2000 1500 1000 500 0 100 200 300 400 500 600 700 800 900 1000 Interval Load (MW) Net load 8

Addition of random capacity (low penetration) • Generator: Normally distributed output, 𝑛 = 100 𝑁𝑋, 𝑡 = 50 𝑁𝑋 • Assume: generator output is independent of load distribution 3500 𝐹𝑀𝐷𝐷 ≈100 𝑁𝑋 3000 ELCC is close to mean output of 2500 the generator Load (MW) 2000 1500 1000 500 0 100 200 300 400 500 600 700 800 900 1000 Interval Load (MW) Net load 9

Addition of random capacity (high penetration) • Generator: Normally distributed output, 𝑛 = 1000 𝑁𝑋, 𝑡 = 500 𝑁𝑋 • Assume: generator output is independent of load distribution 3500 𝐹𝑀𝐷𝐷 ≈ 470 𝑁𝑋 3000 ELCC is close to 47% of mean 2500 output of the generator 2000 1500 Load (MW) 1000 500 0 0 100 200 300 400 500 600 700 800 900 1000 -500 -1000 -1500 Interval 10 Load (MW) Net load

The effect of correlation (ext xtreme example) • Assume the generator with 100 MW mean output and 50 MW std. dev. is not available during extreme demand periods (above 2,200 MW) 𝐹𝑀𝐷𝐷 ≈ 0 𝑁𝑋 3500 3000 2500 Load (MW) 2000 1500 1000 500 0 100 200 300 400 500 600 700 800 900 1000 Interval Load (MW) Net load 11

2017 WEM distribution 3750 3500 Load Duration Curve 3250 Net load duration curve Net Load 3000 2750 Demand (MW) 2500 2250 2000 1750 1500 1250 1000 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% 55% 60% 65% 70% 75% 80% 85% 90% 95% 100% % Time demand exceeded 12

2017 WEM distribution 3750 Load Duration Curve Net load duration curve Net Load 3500 Demand (MW) 3250 3000 2750 0.0% 0.5% 1.0% 1.5% 2.0% 2.5% % Time demand exceeded 13

Win ind capacity valu lue in in other jurisdictions Source: Milligan et al. 2017, Capacity value assessments of wind power Page 14

2. Capacity valu lue assessment methods Page 15

Assessment methods in in practice • Two approaches for ELCC calculation: • Fundamental analysis (reliability model) • Approximation method: to approximate the outcomes of fundamental analysis • Data required for calculation • Coincident data during high LOLP/peak intervals: • Output of intermittent generators • Output of conventional generators • System load 16

Fundamental l analy lysis (ELCC) Decreased LOLE=1.5 day in load Base System Base System ten years -300 MW LOLE=1 day in ten years + + Increased additional additional Base System Base System resource resource load 1000 MW 1000 MW +100 MW LOLE=0.8 day in ten years ELCC = 400 MW 17

ELCC calc lculation challenges • Historical data is usually not sufficient (for rare events in the system) • Eg. In the SWIS (between 2006 and 2012) we never experienced a peak load above the one in ten year peak forecast • We need a model to forecast how IGs perform during extreme demand/high-risk periods 18

3. Current method in the SWIS IS Page 19

Current method • Approximation method to estimate (individual) ELCCs • Mean output during peak LSG (net load) intervals • Less: • K factor (define) • To account for variability of IGs • Previously was 0.003 (international experience) but in 2014, Sapere estimated it for the SWIS (set to 0.000) • U-Factor • To account for the (negative) correlation of IGs with load during high-risk periods 20

Capacity valu luation in in other jurisdictions • Approximation methods • average output of IGs • Time-based approaches: specified (peak/high-risk) intervals • Risk-based approaches: when the system is under the highest reliability risk • Fundamental analysis: • Mid-continent ISO (MISO) • System-wide ELCC calculation (wind resources) • Deterministic allocation of ELCC to individual IGs (based on historical performance) 21

Jurisdiction Reliability criteria Method PJM 1 in 10 year LOLE Approximate Time-based Mean output during peak periods SWISS Hybrid: Approximate • 1 in 10 year peak demand LOLE Risk-based • <0.002% USE Adjusted mean output during peak net load (LSG) NYISO 1 in 10 year LOLE Approximate Time-based Mean output during peak intervals ISO-NE 1 in 10 year LOLE Approximate Time-based, also allows for intervals with system- wide shortages California ISO 1 in 10 year LOLE Approximate Time-based Mean capacity during peak intervals (70% exceedance factor) MISO 1 in 10 year LOLE Fundamental analysis Calculation of system-wide ELCC Allocation of ELCC to individual wind farms based on historical data 22

Does fundamental l analy lysis provide an exact capacity valu lue number? 2018 penetration of wind (~14%) 45% For 2018-19, MISO uses the average of 11 points ELCC=15.2% ELCC (%) 2017 20XX 2008 5% 2006 23 Wind penetration (% of peak load)

Fundamental l vs approximation methods • Fundamental analysis entails building a reliability model: • Can a Plexos model be ready in time? • Constraints on the use of data collected (eg SRMC data) to use for purposes other than 2.16 • Do the results of such analysis provide a significantly different estimate of ELCC (than approx. methods)? • Fundamental analysis is more complex and less transparent • Approximation methods: • Relatively simple • More transparent • However, underlying assumptions may no longer be valid 24

4. Current issues in the SWIS Page 25

Valuation of capacity in a security constrained network • The PUO’s consultation paper: • the valuation of capacity in a security constrained network design • Resources to receive capacity credits subject to network constraints • Current RLM does not consider capacity constraints • Timing of PUO’s review: • Capacity valuation method review after outcomes of network access review • PUO is exploring design of different mechanisms to provide for system adequacy and security Page 26

Collgar’s rule change proposal • Collgar : use of mean output at peak LSG periods is discriminatory • Does not reflect the contribution of IGs to peak demand periods • AEMO argued that contribution towards high-risk periods is more relevant (noting the increased penetration of IGs) • Some (including the PUO) supported Collgar’s argument • Others noted the upcoming review of the capacity valuation method by the ERA 27

Collgar’s rule change proposal… • Hybrid reliability criteria defined in the market rules • With increased penetration of IGs the likelihood of energy shortfall during not highest peak periods increases • If most of energy shortfall events happen during highest peak periods: • Use of peak LSG and peak demand interval would provide similar results (in theory) • If energy shortfall events and highest peak do not coincide: • Peak LSG (net-load) can be relevant for the calculation of ELCC 28

Technology differences • Emergence of behind the meter technologies. • Differences in operational characteristics (solar, wind) • Battery storage installed with intermittent generators • Battery combined with intermittent capacity : firm capacity • How to value such capacity? • MISO uses a system wide ELCC and allocates that to individual IGs based on historical performance • In the SWIS, ELCC is calculated individually (with a common adj. factors) 29