RES Integration Challenges: Market and Grid Problems and Potential Solutions

1. RES Integration Challenges: Market and Grid Problems and Potential Solutions October 11, 2013, Thessaloniki, Greece Dr. Alex Papalexopoulos, CEO and Founder, ECCO International, San Francisco, CA

2. Power Market Challenges & Opportunities

3. Where is the Problem? • Load is stochastic, variable and uncertain • PV solar output is also stochastic, variable and uncertain • Supplies can also be stochastic • Need to know size, probability and duration of any shortfalls in both capacity and ramping capability • System needs flexible capacity to deal with the increased uncertainty and variability

4. Where is the Problem? The penetration of Solar PV will continue to increase as more countries adopt Renewable Portfolio Standards (RPS) and continue to enforce more stringent targets

5. The Anatomy of the “Duck”

6. Implications for the Power Market • Solar PV complicates the power market clearing process (Day-Ahead, Hour-Ahead, Real-Time) • Solar PV suffers from lack of dispatchability • Current practices treat Solar PV energy outside the market process • Solar PV puts substantial downward pressure on market clearing prices (the number of negative prices is increasing) • The transmission grid is becoming increasingly congested

7. Key Tools Available to the Power Market • Change the Power Market design rules to accommodate solar PV • Invest in flexible generation (gas fired power plants) • Implement demand response • Develop storage facilities • Curtailment of Solar PV • Improve transmission planning and expansion

8. Power Market Design Rule Changes • Develop Ancillary Services products for better balancing, better price signals, better incentives (Performance based Frequency Regulation service, Ramping products, Load Following, etc.) • Allow very large negative bids to clear the market • Develop better forecasting tools for load, solar PV, ramping requirements, etc. • Develop Intra-Hour Scheduling financially binding Markets (every 15 minutes) • Develop centralized Capacity Markets that reward flexible generation to ensure security of supply (i.e., we cannot rely on scarcity pricing)

9. Performance based Frequency Regulation • Traditional approaches typically include • a capacity payment (usually based on shadow price) • an energy payment (for the net energy injected/withdrawn in/from the system) • The new market design • a capacity payment (usually based on shadow price) • Mileage Payments adjusted for accuracy

10. Performance based Frequency Regulation • We replace the net energy payments by a mileage payment for the ACE correction provided

11. Expected Flexibility Deficiency Function • Incorporating Flexibility Requirements • Introduction of an Expected Flexibility Deficiency (EFD) function • To determine the anticipated amount of un-served energy caused by a lack of flexibility in the generating fleet • The EFD is a function of the ramp and reserve policies in any given region • The EFD is computed before executing the MIP-Based Unit Commitment • It is derived from historical system load/renewable data, as well as the forecasts associated with a given unit commitment window • Ramp and reserve policies must be defined, in order to determine the EFD

12. Expected Flexibility Deficiency Function • Sample EFD calculation: • An example of a ramp policy is that the average ramp of the system is equal to the forecasted ramp plus some constant, x [MW/min]. • An example reserve policy might be that y% of the forecasted net load is held in reserves • For these policy formulations, the EFD surface is built as a function of x and y • Note that the x and y variables are optimized within the MIP-Based Unit Commitment problem

13. Expected Flexibility Deficiency Function The EFD surface is built as a function of ramping and reserve policies. These are optimized within the MIP-based Unit Commitment Scheduling problem

14. Flexible Deficiencies • Computing Flexible Deficiencies using Historical Net Load Data

15. Day-Ahead Market Formulation:Standard Constraint Equations • Equations are used to enforce all standard equality and inequality constraints, such as: • Energy Balance (generation = load) • Unit output limits • Spinning Reserve Requirements • Regulation Reserve Requirements • Ramp rate limits (units, hydro, imports) • Unit temporal constraints (min up, min down, min run, …) • Hydro, Imports, and Pumped Hydro Energy Limits

16. Network Constraint Modelling • Optionally, the network constraints may be included in the simulation • Monte-Carlo dispatch model iterates with full power flow model (AC or DC) to enforce network constraints, including contingency constraints • Zonal model may also be used to enforce flow constraints

17. Other Constraints & Flexibility Mitigation Strategies Modelling • All relevant constraints are modeled (energy balance, over-generation, curtailment limits, capacity, UC constraints, ramping rates, hydro, imports/exports, EFD constraints, etc.) • Relative cost penalties impose flexibility mitigation strategy “loading order” • Costs will depend on specific system and applicable policies • Assuming that all renewables must be delivered is equivalent to placing an infinite penalty on curtailment and over-generation

18. Solution Methodology of the Flexibility Problem Power Flows Schedules Power Flow Power Flow Optimization Engine Power Flow Power Flow Power Flow PTDFs Loss marginal rates • Separate power flows for each time interval from the economics • Iterate with optimization engine • Execute modified Monte-Carlo simulations using minute-by-minute Solar PV data • The math here is very complicated

19. Monte Carlo Simulation Modeling • We need an Modified Monte-Carlo Simulation • Unit Outages are simulated using random draws of outages based on unit MTTF and MTTR • We need other profiles • Load profiles • Wind profiles • Solar profiles • They are all selected by Monte-Carlo draws from selected bins

20. Example Draw: High Load Weekday in August Day-Type Bins - Load Day-Type Bins - Wind Day-Type Bins - Solar Low Load High Load Low Load High Load Low Load High Load Weekends/Holidays Weekdays Jan Jan Jan Feb Feb Feb Mar Mar Mar Apr Apr Apr May May May Jun Jun Jun Jul Jul Jul Aug Aug Aug Sep Sep Sep Oct Oct Oct Nov Nov Nov Dec Dec Dec

21. Example Draw: High Load Weekday in August Within each bin, choose each (load, wind, and solar) daily profile randomly, and independent of other daily profiles Load Bin Wind Bin Solar Bin

22. Three Market Simulations • Day ahead, hour ahead and real-time markets are simulated sequentially • Load forecast inaccuracy of the day ahead market vs hour ahead is also simulated via Monte-Carlo draws • In hour ahead simulation only short start units may be committed • In real-time simulation, only units that were on-line in the HA market may be re-dispatched

23. MIP Based Flexible SCUC Results • Flexibility violations that may occur, because the penalty cost of these violations is less than the commitment of additional resources • Optimal levels of reserves and ramp-rate capability based on ramp/reserve policy in each Power Market determined by the Regulator and policy makers • Economic “pre-curtailment” of Solar PV that avoids flexibility violations and/or commitment of excessive fast-ramping generation • Requirements for flexible capacity • Optimal Procurement decisions

24. Solar PV Curtailment Could Play a Significant Role Power Flows Schedules Power Flow Power Flow Optimization Engine Power Flow Power Flow Power Flow PTDFs Loss marginal rates • Scheduled curtailment of Solar PV can help position conventional • resources to meet ramping requirements • How does the cost of curtailment compare to the cost of procuring new • flexible resources?

25. Proposed Metrics with High Solar Penetration • Resource Adequacy metrics: • LOLP, LOLE, EENS • Flexibility Deficiency metrics: • Expected Ramp Not • Served (ERNS) • Expected Regulation not • Served, etc. • Flexibility Shortage • Induced Curtailment • How does the cost of • curtailment compare to the • cost of procuring new • flexible resources?

26. Demand Response: Power Markets in Pain Price Marginal Wholesale Rate Power Demand (MW) …If we could use just 5% less power for the current hour…. No Price-Sensitive Demand -> Inefficiency, Everyone Pays For

27. Demand Response: Energy Demand Cloud Energy Demand Cloud Price Sensitive Special Programs Distributed Generation Electric Vehicles Reliability Signaled Renewable Choice Energy Storage Affinity Programs Demand Monitoring & Feedback over Internet Broadband/Cellular Individual Wireless Controllers Home/Facility Management Systems Smart Buildings, Commercial & Industrial Electric Vehicle Chargers Smart Appliances Distributed Energy & Storage TODAY FUTURE DR client is ~10kb—virtually any embedded device can run it

28. Demand Response Software in Devices INTERNET 80% of US households have broadband (as of 2011*) WiFi Router Today - Retrofit Future - Embed External Load Controller OEM Products to Seed Market Internal Load Controller

29. Storage Technologies • Storage is the game changer • Pumped Hydroelectric Storage is important but is highly site-constrained • Other technologies that have shown promise are a) Compressed Air Electric Storage (CAES), flywheels, hydrogen electrolysis • Plug-In-Electric Vehicles in Vehicle-to-Grid (V2G) mode (could serve as a major distributed storage resource) • Problem: What is the value proposition? • Develop incentives mechanisms to account for risk and reward sharing (need a regulatory framework)

30. Transmission Capacity • New transmission capacity is required • Implement technologies to permit increased utilization of the existing transmission infrastructure • Dynamic Thermal Rating • Power Flow Controls (FACTS devices) • Fault current controllers • Intelligent protection systems (adaptive relaying) • Advanced stochastic modeling and planning tools • Increased reliance on DC links

31. Conclusions • High penetration of RES creates major power market challenges • The issues involved can be viewed as a coordination problem at multiple scales in both space and time • The problems are solvable but the solutions are neither trivial nor cheap • The infrastructure upgrade costs in the legacy power system and the public’s willingness to socialize these costs could emerge as an important issue • The power market response involves solutions including a) power market design changes, b) demand response, c) Storage technologies, d) PV curtailment, e) flexible market products

32. Grid Integration Challenges & Opportunities

33. Solar PV Grid Challenges Grid Stability & Reliability Grid Stability & Reliability Load Balancing Power Systems Planning & Design CONCERN • The effects of large additions of Solar PV generation on the ac grid stability and system oscillations are not well understood • They could exacerbate pre-existing wide-area stability problems • Solar PV erodes the mechanical inertia Msec to Minutes Hours to Days Years LEARNING Milliseconds to Minutes • Develop smart inverters to mimic the intrinsically stable inertial behavior of a rotating machine • Develop Low-Voltage or Fault Ride Through Inverters • Develop plant controllers to react to grid frequency

34. Modern Solar PV Plants Need to Contribute to the Reliability to the Grid • Voltage, VAR control and/or power factor regulation • Fault Ride Through (FRT) • Real power control, ramping, and curtailment • Primary frequency regulation • Frequency droop response • Short circuit duty control

35. Automatic active power frequency control: What is Needed • Potential Issue: The frequency must be kept constant within strict limits • What is needed: • A plant controller to react to grid frequency increase by an automatic active power reduction

36. Dynamic Grid Support • Potential Issue: When a grid failure occurs many PV plants may be disconnected immediately • What is needed: • Inverters with dynamic grid support functions to act within milliseconds in such events • Devices with full LVRT or FRT behavior (Low-Voltage or Fault Ride Through) can feed reactive power into the grid during grid voltage drops

37. What Makes a PV Plant “Grid Friendly”? • Critical for Managing Grid Reliability & Stability • Potential Issue: When a grid failure occurs many PV plants may be disconnected immediately • What is needed:

38. Solar Generation is not fully dispatchable Adds variability and uncertainty . . . complicates daily dispatch Solar PV Grid Challenges Load Balancing Grid Stability & Reliability Load Balancing Power Systems Planning & Design CONCERN Msec to Minutes Hours to Days Years LEARNING Hours to Days • Integrate forecasting into daily operation • Improved operating procedures – balancing area, frequent updates, ramping support

39. Aggregation Effect Between Plants Reduces Variability Mathematically, if the short‐term output time series of N locations experiencing a similar level of variability are uncorrelated (i.e., if they vary independently from each other), the resulting variability of the ensemble should be 1/√N times that of a single location Single Location One-minute Global Irradiance (W/sq.m) 20 Bundled Locations One-minute Global Irradiance (W/sq.m) Source: Hoff et al. 2008 Spatial diversity of solar plants reduces aggregated variability… minimizing grid impact as the number of solar plants increase Source: “Implications of Wide-Area Geographic Diversity for Short-Term Variability of Solar Power”; Andrew Mills and Ryan Wiser, Lawrence Berkeley National Laboratory, September 2010

40. Grid Integration Challenges & Opportunities: Key Lessons Learned Grid Stability & Reliability Load Balancing Power Systems Planning & Design • Adopt diverse resource portfolio to increase flexibility and reduce risks • Ned flexible capacity • Integrate forecasting into daily operation • Improved operating procedures – balancing area, frequent updates, etc. • Need flexible capacity • Need grid controls and smart inverters to support reliability and grid security Years Milliseconds to Minutes Hours to Days

41. Protection • The legacy protection equipment were no designed for the presence of Solar PV (or for DG) • Solar PVs complicate protection coordination: • Reverse power flow: A fault must now be isolated not only from the substation power souce but also from the Solar PV • Fault current contribution: Until a fault is isolated, Solar PV contributed a fault current that must be modeled and managed • Relay desentitization: The presence of Solar PVs may delay of prevent the actuation of protective devices • Need transfer trip strategies to allow communication between devices

42. Islanding • Current practices do not allow power islands supported by Solar PV (very conservative policy to ensure safety, power quality, cultural issues, etc.) • Current interconnection rules for Solar PV simply require that Solar PV shall disconnect in a specific time (e.g., 10 cycles) in response to disturbances • This conflicts with the ability of the Solar PV to offer benefits to the grid if it is equipped with LVRT • Our industry needs a new set of rules, regulations and procedures to allow solar PV to support power islands • There is substantial research now in smart grids and microgrids to allow effective coordination of Solar PV with storage devices and intelligent controls so that heterogeneous power quality is achieved

43. Conclusions • High penetration of RES creates grid integration challenges • The issued involved can be viewed as a coordination problem at multiple scales in both space and time • The problems are solvable but the solutions are neither trivial nor cheap • The infrastructure upgrade costs in the legacy power system and the public’s willingness to socialize these costs could emerge as an important issue • The grid integration challenges can be resolved by deploying smart inverters that enable Voltage/VAR control , Fault Ride Through, Real Power Control, Ramping, and Curtailment, Primary Frequency Regulation, smart grid and microgrid concepts, etc.

44. Thank you for your Attention alexp@eccointl.com