INTEGRATION OF STORAGE INTO POWER SYSTEMS WITH RENEWABLE ENERGY SOURCES: CONSTRUCTION OF A STOCHASTIC SIMULATION APPROACH

Size: px

Start display at page:

Download "INTEGRATION OF STORAGE INTO POWER SYSTEMS WITH RENEWABLE ENERGY SOURCES: CONSTRUCTION OF A STOCHASTIC SIMULATION APPROACH"

Jodie Small
5 years ago
Views:

1 INTEGRTION OF STORGE INTO POWER SYSTEMS WITH RENEWBLE ENERGY SOURCES: CONSTRUCTION OF STOCHSTIC SIMULTION PPROCH presentation by George Gross University of Illinois at Champaign-Urbana PSERC Public Webinar October 16, George Gross, ll Rights Reserved. 1

2 RESERCH OBJECTIVES Construct a comprehensive simulation approach to emulate the behavior of power systems with integrated storage and renewable energy resources in a competitive environment Incorporate models for the resources and the loads that capture their salient characteristics variability and uncertainty, with spatial and temporal dependencies as well as their interactions over the grid and in the markets Demonstrate the simulation approach capabilities with a number of case studies that assess the impacts of storage and renewable resource integration on the power system variable effects 2

3 US RENEWBLE PORTFOLIO STNDRDS 29 states,+ Washington DC and 2 territories,have Renewable Portfolio Standards (8 states and 2 territories have renewable portfolio goals). Source: September

4 THE CHNGING ENVIRONMENT There is a growing worldwide interest in integrating renewable resources, as well as storage resources, into the grid to displace costly and polluting fossil-fuel-fired generation The objective of such integration is to push the creation of sustainable paths to meet the nation s energy needs and to veer it towards energy independence The context within which this progress unrolls is in the restructured, competitive electricity industry and the advances in the implementation of the Smart Grid 4

5 SLIENT CHRCTERISTICS OF SOLR ND WIND POWER OUTPUTS Highly time-dependent nature: variability characteristics intermittency effects seasonal dependence Spatially correlated Inherently uncertain and difficult to characterize analytically Limited accuracy Limited controllability / dispatchability 5

kw PV POWER OUTPUT OF 1 - MW CdTe RRY IN GERMNY 1000 900 800 700

6 kw PV POWER OUTPUT OF 1 - MW CdTe RRY IN GERMNY samples collected on a 5 minute basis 6

7 PV output (MW ) load (GW ) PV OUTPUT ND LOD Sources: ERCOT and NREL time (h) 0 7

8 MW 700 CISO DILY WIND POWER PTTERNS IN MRCH 2005 Source: CISO Source: CISO hour 8

MW 700 ONTRIO DILY WIND POWER OUTPUT 600

9 MW 700 ONTRIO DILY WIND POWER OUTPUT Source: IESO hour 9

10 wind power output (MW) MISLIGNMENT OF WIND POWER OUTPUT ND LOD load (MW) load wind power hour

11 ISSUES IN INTEGRTING WIND ND SOLR RESOURCES Misalignment of the wind power outputs with the load implies that the wind speeds are rarely adequate at times when needed to supply the loads While morning and mid-day solar power outputs are aligned with the loads, their quick decline after sunset occurs when the loads are still high The risk of spilling wind due to inadequate loads at night and the challenges of managing base-loaded unit shut down for short periods during low-load hours is a major problem 11

12 TKING DVNTGE OF INTEGRTED STORGE RESOURCES In order to take advantage of the increased flexibility imparted by the grid-integrated storage devices, we developed appropriate models, methodologies and tools The resulting approach has applications to: planning and investment analysis; policy analysis; operations; and market performance 12

13 UTILITY - SCLE STORGE CHRCTERISTICS storage unit may act as either a generating unit; or a load; or is idle neither as a load nor as a generator Storage unit operations are driven by the other resources and so storage is a highly time-dependent resource storage operations are uncertain due to its dependence on other uncertain resources Exploitation of arbitrage opportunities in the determination of storage operations is critical 13

14 UTILITY - SCLE STORGE PPLICTION MW p storage resource discharging during peak hours storage resource charging 0 during low-load hours

15 WIND/STORGE INTERCTIONS no wind resources with wind resources 15

16 THE NEED FOR NEW SIMULTION PPROCH The detailed simulation of integrated storage, wind, solar and active demand response resources requires the representation of the time dependent operational actions in line with the market results Uncertainties in the loads, wind and solar outputs and conventional unit available capacities requires their explicit representation together with their time varying nature 16

17 NEED TO EXPLICITLY REPRESENT The loads and their associated uncertainty The resources and their associated uncertainty: conventional generators utility-scale storage units renewable resources The spatial and temporal correlations among the resources at the various sites and the loads The impacts of the grid constraints The hourly day-ahead markets (DMs) 17

18 THRUST OF THE PPROCH We develop a comprehensive, computationally efficient Monte Carlo simulation approach to emulate the behavior of the power system with integrated storage and renewable energy resources We model the system load and the resources by discrete-time stochastic processes We deploy the storage scheduler to utilize arbitrage opportunities in the storage unit operations We emulate the transmission-constrained hourly dayahead markets (DMs) to determine the power system operations in a competitive environment 18

19 THRUST OF THE PPROCH We construct appropriate c.d.f. approximations to evaluate the expected system variable effects Metrics we evaluate include: nodal electricity prices (LMPs) generation by resource and revenues congestion rents CO 2 emissions LOLP and EUE system reliability indices 19

20 KEY STUDY CONTRIBUTIONS Development of a new simulation tool appropriate to address today s power industry challenges Salient features include: quantification of the power system expected variable effects economics, reliability and environmental impacts in each sub-period computationally tractable for practical systems 20

21 STUDY CONTRIBUTIONS detailed stochastic models of the time varying resources and loads allow the representation of spatial and temporal correlations storage scheduler for optimized storage operation to exploit arbitrage opportunities representation of the transmission constrained market outcomes flexibility in the representation of the market environment / policies 21

22 THE TIME DIMENSION IN SIMULTION period 1... STUDIES study period period t We decompose a multi-year study period into T non-overlapping simulation periods We specify the simulation periods in such a way that no changes in the resource mix, unit commitment, the transmission grid and the policy environment occur during each simulation period; such changes may occur in subsequent periods time period T 22

23 THE SUBPERIODS: THE SMLLEST INDECOMPOSBLE UNITS OF TIME period 1... study period period t period T time sub-period sub-period h sub-period H We introduce sub-periods to explicitly represent the time dependence of the side-by-side power system and market operations We use hourly sub-periods in the simulation and no phenomena of shorter duration are represented 23

24 THE SIMULTION PPROCH: KEY SSUMPTIONS The system is in the steady state for each subperiod and we ignore shorter duration phenomena The behavior of each market participant is independent of that of the other participants and no participant engages in strategic behavior The grid is lossless and the DC power flow conditions hold over the entire study period 24

25 PROPOSED SIMULTION PPROCH: CONCEPTUL STRUCTURE 25

26 QUNTIFICTION OF SYSTEM VRIBLE EFFECTS We use the approximations to the joint c.d.f.s of the market outcome stochastic processes to compute the expected values of the system variable effects in each sub-period In this way, we evaluate all the figures of merit of interest, including expected electricity payments, expected energy supplied by each resource, reliability indices and expected emissions 26

27 ENSURING NUMERICL TRCTBILITY We introduce various schemes to reduce the computing burden to ensure tractability Key implementational aspects: selection of a group of representative weeks variance reduction techniques: stratification, control variates and common random numbers warm-start of the linear program solver parallelization of simulation runs 27

28 TYPICL PPLICTIONS Resource planning studies Production costing issues Transmission utilization issues Environmental assessments Reliability analysis Investment analysis 28

29 CSE STUDIES 29

30 MOTIVTION We present case studies aimed at illustrating the various capabilities of the simulation approach and relevant to the integration of storage and renewable resources We perform sensitivity studies to investigate several aspects of storage integration into the grid, notably its impact in a system with deepening wind penetration, its siting, and to what extent storage and renewable resources may replace conventional generation 30

31 CSE STUDIES We present 3 sets of representative case studies: case study set I: impacts of an integrated utility-scale storage unit under a deepening wind penetration scenario case study set II: siting of 4 energy storage units and the impacts on transmission usage case study set III: substitution of conventional generation resources by a combination of storage and wind energy resources 31

32 CSE STUDY SET I: DEEPENING WIND PENETRTION The objective of this study is to perform a wind penetration sensitivity analysis and to quantify the enhanced ability to harness wind resources with the addition of a storage energy resource We evaluate the key metrics for variable effect assessment, including wholesale purchase payments, reliability indices and CO 2 emissions 32

33 THE STUDY TEST SYSTEM: MODIFIED IEEE 118-BUS SYSTEM nnual peak load: 8,090.3 MW Conventional generation resource mix: 9,714 MW 4 wind farms located in the Midwest with total nameplate capacity in multiples of 680 MW storage unit with 400 MW capacity, 5,000 MWh storage capability and 89 % round-trip efficiency Unit commitment uses a 15 % reserves margin provided by conventional units and the storage resources 33

34 SENSITIVITY CSES IN STUDY SET I case total installed wind nameplate capacity in MW base B 1,360 C 2,040 D 2,720 34

35 hourly storage charge/discharge capacity in MW load in GW CSE D: VERGE HOURLY STORGE UTILIZTION 400 load hour

36 LMP in $/MWh NODE 80 VERGE HOURLY LMPs with storage without storage base case case case B case C case D hour 36

37 LMP in $/MWh BUS 80 VERGE HOURLY LMP DURTION CURVE with storage without storage base case case case B case C case D % of hours in the week 37

38 thousand $ EXPECTED WHOLESLE PURCHSE PYMENTS 180 without storage with storage % % % % % % % % %

39 EXPECTED CO 2 EMISSIONS % % thousand metric tons +0.3 % 1.05 without storage with storage % % % % % %

40 NNUL RELIBILITY INDICES LOLP EUE 14 x 10-4 without storage with storage 0.12 without storage with storage % % % % % % % % % % % % % % % % % %

41 CSE STUDY SET II: STORGE UNIT SITING The objective of this study is to perform a sensitivity analysis on the siting of 4 storage units in the system and assess its impacts on transmission usage and on the economics at the most heavily loaded bus in the network We quantify the expected LMPs at the load center at node 59 and the total congestion rents 41

42 TEST SYSTEM OF THE STUDY: MODIFIED IEEE 118-BUS SYSTEM nnual peak load: 8,090.3 MW Conventional generation resource mix: 9,714 MW 4 wind farms located in the Midwest with total nameplate capacity 2,720 MW 4 identical utility-scale storage units, each having 200 MW capacity, 5,000 MWh storage capability and 89% round-trip efficiency Reserves margin is set at 15 % and is provided by conventional and storage resources 42

43 30 MW 12 Mvar -9 MW MW -28 MW pu pu 8 39 MW pu 24 MW 4 Mvar 17 MW Mvar 51 MW 27 Mvar MW MW MW 13 Mvar pu 2 52 MW 22 Mvar pu -9 MW 43 MW 27 Mvar 7 MW 20 MW 9 Mvar 59 MW 23 Mvar 19 MW 2 Mvar 0 MW pu pu -6 MW 85 MW 16 8 MW 3 Mvar pu MW MW MW 7 Mvar 20 MW 8 Mvar 11 MW 34 MW 16 Mvar 3 Mvar MW 1 Mvar 15 0 MW pu 19 0 MW 90 MW 3 18 MW 23 3 Mvar MW 34 Mvar pu 14 MW 8 Mvar MW 0 MW 220 MW pu 10 MW 5 Mvar pu 45 MW 25 Mvar pu MW 9 Mvar 23 MW 9 Mvar MW Mvar 0 MW 31 MW 17 Mvar -13 MW 59 MW 26 Mvar pu Mvar pu pu pu -12 MW 27 MW 11 Mvar pu MW MW 23 Mvar pu MW 7 Mvar 28 MW 1 19 MW MW 2 37 MW MW pu 16 MW 1 8 Mvar 1-59 MW 53 MW 22 Mvar 11 Mvar 74 0 MW 34 MW 37 MW 23 Mvar 68 MW 27 Mvar pu pu MW 11 Mvar 87 0 MW 85 4 MW 21 MW Mvar 11 Mvar pu MW 7 Mvar 24 MW 15 Mvar pu 33 MW Mvar 20 MW -184 MW 20 MW pu 23 MW 11 Mvar 68 MW 36 Mvar MW 0 MW MW Mvar 88 0 MW 204 MW pu MW 28 Mvar 48 MW 1 54 MW 27 Mvar MW 48 MW 607 MW 78 0 MW pu 39 MW 18 Mvar MW 26 Mvar 113 MW 32 Mvar 17 MW 4 Mvar pu pu pu MW 42 Mvar 38 MW 15 Mvar 39 MW 32 Mvar 392 MW 15 MW 9 Mvar -85 MW 12 MW 3 Mvar pu pu pu 130 MW 26 Mvar MW 31 Mvar 477 MW 65 MW 1-10 MW 17 MW 8 Mvar 12 MW 7 Mvar 0 MW pu 34 MW 93 8 Mvar 12 MW 3 Mvar 28 MW 7 Mvar 84 MW 18 Mvar pu MW 3 Mvar 30 MW 16 Mvar MW 18 Mvar 22 MW 15 Mvar MW 0 MW 63 MW 22 Mvar 252 MW pu MW 16 Mvar 40 MW MW 14 Mvar MW 0 MW 78 MW 3 Mvar 0 MW pu pu pu MW 25 Mvar pu 155 MW 113 Mvar 277 MW 113 Mvar MW 19 Mvar pu pu pu MW 6 Mvar 31 MW 26 Mvar 39 MW 3 2 MW 1 Mvar 8 MW 3 Mvar MW 16 Mvar -43 MW 25 MW 13 Mvar pu 5 Mvar -22 MW 28 MW 12 Mvar pu W STORGE SITING ON THE MODIFIED IEEE 118 BUS TEST SYSTEM W W most heavily loaded bus at node 59 W storage siting region 43

44 SENSITIVITY CSES IN STUDY SET II case base S 0 S 1 S 2 S 3 siting of the storage units no storage units at the principal load center 1 node away 2 nodes away 3 nodes away each case has 2,040 MW nameplate wind capacity 44

45 STORGE SITING REGION 0 MW 63 MW 22 Mvar -59 MW 37 MW MW 23 Mvar 0.98 pu 23 MW 11 Mvar MW MW 32 Mvar S pu 56 0 MW S 1 84 MW 18 Mvar 0.95 pu pu S Mvar 12 MW S 2 12 MW 18 MW 16 MW 3 Mvar 3 Mvar S 0 S Mvar 0.98 pu 155 MW 113 Mvar 53 MW 22 Mvar 20 MW 17 MW Mvar 11 Mvar 17 MW 8 Mvar 4 Mvar MW 113 Mvar ar r MW 47 S MW MW S pu S 0 S MW 39 MW 18 Mvar 66 S MW 1.05 pu MW 7 Mvar S 3 S 2 78 MW 3 Mvar 60 S pu S pu 160 MW MW 0.98 pu MW 39 MW 32 Mvar 2 S MW 14 Mvar 0 MW 45

46 LMP in $/MWh load in GW LMP in $/MWh NODE 59 EXPECTED HOURLY LMPs case S case S 0 case S case S case S 1 case S base case 2 case S hour 5 10 case S 3 base case number of hours in the week 46

47 congestion rents in thousand $ load in GW congestion rents in thousand $ EXPECTED HOURLY CONGESTION RENTS 60 case S case S case S case S case S 2 case S case S 3 base case case S 3 hour base case number of hours in the week 47

48 TRNSMISSION PTH CONGESTION ND ITS REENFORCEMENT 0 MW 63 MW 22 Mvar pu 37 MW -59 MW 37 MW 1 23 Mvar 48 MW 113 MW 0 MW 84 MW 23 MW 32 Mvar 18 Mvar 11 Mvar pu MW pu pu most heavily loaded bus at node Mvar 12 MW S 2 12 MW 16 MW 3 Mvar 3 Mvar S 0 S Mvar 0.98 pu 155 MW 113 Mvar 53 MW 22 Mvar 20 MW 17 MW Mvar 11 Mvar 17 MW 8 Mvar 4 Mvar MW 113 Mvar ar S 2 S 0 S 0 r pu 1 34 MW MW MW 28 MW 7 Mvar 2 x 400MW nuclear plants 391 MW 39 MW 18 Mvar MW 1.05 pu MW 3 Mvar pu 1.03 pu MW MW 0.98 pu MW 39 MW 32 Mvar MW 14 Mvar 0 MW 48

49 LMP in $/MWh PRE PTH REENFORCEMENT NODE 59 VERGE HOURLY LMPs load in GW LMP in $/MWh case S case S 0 case S case S case S 1 case S base case 2 case S hour 5 10 case S 3 base case number of hours in the week 49

50 LMP in $/MWh POST PTH REENFORCEMENT NODE 59 VERGE HOURLY LMPs load in GW LMP in $/MWh base case case S 0 case S 1 case S 2 case S base case case S 0 case S case S 1 case S hour number of hours in the week 50

51 congestion rents in thousand $ PRE PTH REENFORCEMENT VERGE HOURLY CONGESTION RENTS load in GW congestion rents in thousand $ 60 case S case S case S case S case S 2 case S case S 3 base case case S 3 hour base case number of hours in the week 51

52 congestion rents in $ POST PTH REENFORCEMENT VERGE HOURLY CONGESTION RENTS load in GW congestion rents in $ case S 2 case S 0 3,500 case S 1 3,500 3,000 base case ,000 2,500 2,500 2, ,000 case S 0 case S 2 case S 1 1,500 1, ,500 1,000 case S case S 3 hour 500 base case: no storage number of hours in the week 52

53 STUDY SET III: SUBSTITUTION OF THE CONVENTIONL RESOURCES The aim of this study is to quantify the extent, from a purely reliability perspective, wind resources can substitute for conventional generation capacity in a power system with integrated storage resources We deem storage units to be firm capacity and use them to meet the desired reserves margin s the wind resources are integrated, we decrease progressively the system reserves margin, retire conventional unit capacity and assess the impacts 53

54 THE STUDY TEST SYSTEM: MODIFIED IEEE 118-BUS SYSTEM nnual peak load: 8,090.3 MW Conventional generation resource mix: 9,714 MW 4 wind farms located in the Midwest with total nameplate capacity of 2,720 MW 4 units: each has a 100 MW capacity, 1,000 MWh storage capability and 89 % round-trip efficiency The unit commitment is performed to ensure the desired reserves margin is attained from the conventional and storage resources 54

55 SET IV SENSITIVITY CSES case reserves margin in % base (no wind, no storage resources) 15 R 0 15 R 1 14 R 2 13 R 3 12 R 4 11 R 5 10 R 6 9 R 7 8 R

56 weekly LOLP contribution WEEKLY RELIBILITY INDICES vs. weekly EUE contribution in MWh RESERVES MRGINS LOLP EUE base case 0.2 base case reserves margin in % reserves margin in % 56

57 wholesale purchase payments in thousand $ MEN HOURLY WHOLESLE PURCHSE PYMENT IMPCTS 400 base case 15% 14% % 12% 11% % 9% 8% 7% hour 57

58 KEY FINDINGS OF THE ILLUSTRTIVE STUDIES Deeper penetration of wind resources reduces DM LMPs, wholesale purchase payments, CO 2 emissions and improves system reliability Storage works in synergy with wind to drive wholesale purchase payments further down and improve system reliability Overall, CO 2 emissions are not significantly affected by the integration of a storage unit 58

59 KEY FINDINGS OF THE CSE STUDIES Storage siting significantly impacts the congestion rents and the LMPs at certain nodes In a system whose storage resources are used to substitute for conventional generation to meet the desired reserves margin requirements, large amounts of wind capacity are required to replace the retired conventional generation capacity: in the case studies presented the 2,720 MW of wind can substitute for about 300 MW of retired conventional generation capacity about 3.7 % of peak load 59

60 KEY FINDINGS OF THE CSE STUDIES bsent storage units, with all other conditions remaining unchanged, the 2,720 MW wind can replace only about 220 MW of retired conventional generation capacity about 2.7 % of peak load We attain significant reductions in wholesale purchase payments about 25 % when storage and wind resources substitute for conventional resource capacity with the same reliability level 60

61 SUMMRY OF PROJECT CONTRIBUTIONS Development of a practically-oriented approach to simulate large-scale systems over longer-term periods Comprehensive and versatile approach to quantify the impacts of the integration of storage devices into power systems with deepening penetration of renewable resources Demonstration of the capabilities of the proposed approach to a broad range of planning, investment, transmission utilization and policy analysis studies 61

62 CONCLUDING REMRKS Storage and wind resources consistently pair well together: they reduce wholesale purchase dollars and improve system reliability; storage seems to attenuate the diminishing returns trend seen with deeper wind power penetration The location of a storage unit can have large local impacts; siting requires case-by-case studies Wind resources can substitute for conventional resources to a very limited extent, even in a 62

IEEE TRANSACTIONS ON POWER SYSTEMS 1. Yannick Degeilh and George Gross, Fellow, IEEE

IEEE TRANSACTIONS ON POWER SYSTEMS 1 Stochastic Simulation of Utility-Scale Storage Resources in Power Systems With Integrated Renewable Resources Yannick Degeilh and George Gross, Fellow, IEEE Abstract