Power System Reliability: Operating Reserves From Responsive Load

1. Power System Reliability:Operating Reserves From Responsive Load Brendan Kirby Oak Ridge National Laboratory 865-576-1768 kirbybj@ornl.gov

2. DOE’s Interests • Bulk power system reliability • Reducing energy costs • Increase energy efficiency The National Labs have no commercial interests in any power markets or load response technologies

3. ORNL and Ancillary Services • 1985 – Took a 2000 MW DOE load from firm power to real-time markets. Dropped average power prices from $45 to $17/MWH. Both the load and the utilities won. • 1995 – Identification and Definition of Unbundled Electric Generation and Transmission Services, preceded FERC NOPR for 888 by 1 month • 1998 – NERC Policy 10 • 2001 – NERC spinning reserve requirement is technology neutral • 2003 – Blackout field investigation • 2004 – NERC/FERC Reliability Readiness Reviews and FERC reliability technical staff • Continuous working with large and small load response projects at WECC, CEC, CAISO, LIPA, TVA, WAPA, Alberta, AEP, CSW, NIPSCO, NEDRI, PJM, EPRI, EEI, BPA, etc.

4. Factors to Keep In Mind • Reliability “rules” were developed as guidelines • Guidelines were developed around incumbent technologies • It is not necessary for all loads to be responsive • Different loads will want to respond differently • Society and the ISO should actively draw in as many reliability resources as practical

5. Why Do We Think There Is A Good Match Between Some Responsive Loads And Contingency Reserves? • What are the power system’s reliability needs? • How do these match to load response capabilities? • Examples of contingency reserve responsive load projects • Examples of existing responsive load programs that relate to contingency reserves • Obstacles and concerns

6. Conceptually, Responsive Load Could Greatly Enhance Bulk Power System Reliability • Good match between some load’s inherent response capability and power system’s physical reliability requirements • Preconceptions about what response is easiest for loads to provide are often wrong • Advances in communications and control technology make it possible to exploit load response • Market and reliability rules need to be reexamined • Our goal should be to enhance reliability using the best performing, most cost-effective resources available, not to preserve the existing rules • Achieving this goal requires changes to existing rules • Responsive load is the largest underutilized reliability asset in North America

7. A Series of Reserves Respond To Contingency Events A contingency is the sudden, unexpected loss of generation or transmission Ideally, we want the spinning reserves restored and ready to respond again at this point Exact timing requirements and reserve names vary from region to region

8. Spinning Reserve Is Typically Deployed for 10 Minutes or Less Service definitions often require reserves to be capable of 2 hr sustained response but the power system’s physical requirements are much shorter

9. Contingency Reserve Prices Show Value • Faster (not longer duration) services typically command higher prices with spinning reserves typically being twice as expensive as non-spinning or supplemental reserves • Responsive load is available when price (and value to the power system) is highest The power system does not fundamentally value reserves in “square blocks”

10. Four Types of Demand Response • Energy Efficiency programs reduce energy consumption and usually reduce peak demand • Price Response programs reduce consumption during high priced hours and sometimes shift consumption from day to night (real time pricing or time of use) • Peak Shaving programs require more load reduction during peak hours and focus on reducing peaks every high-load day • Reliability Response (contingency response) requires the fastest, shortest duration response. Response is only required during power system “events”

11. Responsive Loads: Better Matched To Spinning Reserves Than Peak Reduction • Better for the load: shorter, less frequent disruption (less storage is required) • Better for the power system: faster response, more reliable, more redundancy, distributed, better use of generation • Better for other loads: reduced energy and ancillary service prices – not all loads need respond • Better for society: reduced need for generation and transmission • Historically, loads have not been allowed to provide spinning reserves, the fastest, highest price contingency reserves • Advances in communications and control now make it technically possible • Power system reliability needs now make it desirable • Spinning reserve is capacity Complements energy management, peak shaving, and price response

12. Both Large and Small Loads Are Attractive Spinning Reserve ProvidersCurrent Projects Include: • CDWR – large pumping loads • Xcel – large pumping loads • Carrier Comfort Choice – Residential and small commercial thermostat response • DigiLog – Packaged, through-the-wall air conditioning response; commercial & residential

13. Spinning Reserves From Large Pumping Loads • This demonstration with the CA Department of Water Resources consist of controlling water utility pumps so that they respond to frequency deviations and system operator commands for contingency events • Variable speed drives may help increase operational flexibility and motor lifetimes

14. CDWR Manages Pumping Loads To Reduce Energy Costs Spinning Reserve Opportunities and Revenues Are Still Substantial

15. CDWR Pumps Could Supply Significant Spinning Reserve • 62% of total California spinning reserve requirements • 100% of California spinning reserve requirements for 3292 hours in 2002

16. Xcel Energy DemonstrationCabin Creek – 2x124MW Pumps At night, when XCEL is pumping up the pond, and they already have their efficient combined cycle plants running at full load, they must run low efficiency peakers as spin. Turning off the pumps would be more reliable, and would save natural gas. Jim Williamson Xcel Energy 303-808-1849 An immediate opportunity • Pumps currently supply operating reserves • Xcel wants to supply spin from pump load • Load removed in <1 minute

17. WECC Related Activities • Stability analysis shows benefits of spin from load • CMOPS forms committee to design summer 2005 large pump demonstration • CEC funded CAISO/SCE residential/commercial spinning reserve and feeder relief demonstration for summer 2005 • MORC approved Pacificorp request for non-spin from load

18. Spinning Reserve From Residential and Small Commercial Thermostats • Existing Carrier ComfortChoice technology for peak reduction • Faster than generation for spinning reserve • Spinning reserve capability ~3x peak reduction • Significant monitoring in place

19. Communicationsand Control • Designed for multi-hour peak reduction • Deployment signal <90 seconds • Verification delayed to protect paging system • Grouping by location, type, or any other criteria • Customer override allowed for peak shaving, not for spinning reserve • Control can be duty cycle, set point, or turn off • Monitors temperature, run time, communications • Customer remote monitoring and control web interface

20. Can Provide Spinning Reserve While Providing Peak Reduction

21. Substantial Installed Capacity That Could Be Tested Today

22. Spinning Reserves From Commercial Loads • Supervisory controllers in hospitality industry can: • Save energy and can reduce electric demand • Reduce space heating/cooling in rooms that are unoccupied • Shift electric loads during peak periods for short time intervals • Provide the capability for utilities to satisfy spinning reserve • Easy retrofit – Fast deployment • Developed by Digi-Log Technologies; tested for energy savings by ORNL • Spinning reserve is an easy add-on • Modified to operate by pager signal from utility • Demo with LIPA and possibly others • Will work with many technologies

23. DigiLog Motel Energy Management System Provides Spinning Reserve • 80 room Howard Johnson motel • 1 minute revenue metering • Pager deployed spinning reserve • 34kW, 36% load drop in 1 minute

24. Loads May Have The Capability To Provide Regulation • Regulation is the most valuable ancillary service and the most difficult to provide • Most intrusive • continuous minute-to-minute movements • Greatest communications and control requirements • If some loads are willing and able to provide regulation we should expect others to be good at supplying contingency reserves • Energy storage technologies (flywheels) may be ideal regulation suppliers as well

25. Many Loads Appear To Be Capable of Providing Regulation • Air liquefaction 1,000MW • Induction & ladle metallurgy furnaces 1,000MW • Gas & water pumping with variable speed motor drives • Electrolysis: >14,000MW • Aluminum 6,500MW • Chlor-alkali 4,500MW • Potassium hydroxide 1,000MW • Magnesium, sodium chlorate, copper • Always address regulation on an aggregate system basis CPS 1&2 are the metrics for minute-to-minute control area balancing Regulation is the most expensive ancillary service

26. Initial Testing of Load Supplying Regulation Should Begin in December • The same net energy is consumed – the same production is met • Two response options are shown – blue may be better for both the load and the power system than red

27. Demand Response Program Examples • This list is not exhaustive • The focus is on programs relationship to reliability and operating reserves • Georgia Power’s price responsive programs are not included, for example

28. NYISO Responsive Load Programs • Emergency • Emergency Demand Response Program & ICAP Special Case Resource Program • Day ahead advisory, 2 hour notice • 1577 MW total • 1277 MW responsive load, 300 MW distributed customer owned generation • Mostly large industrial loads • 3-4 deployments per year • 3-12 hours per event, must be capable of 4 hours of response • Economic • Day-Ahead Demand Response Program (DADRP) • 377 MW registered but 10 MW typically deployed

29. Spinning Reserve, ICAP, &Co-Optimization • ICAP requirement/payment up to 130% of peak load • Block response requirements • Not tied to system hourly need • Do accommodate generator limits (minimum load, emissions, etc.) • ICAP generator must bid into energy and AS markets • Co-Optimization of energy and AS markets • Results in cost shift from AS to ICAP • AS requirements tied to generator capability, not reliability requirements

30. PJM Responsive Load Programs Load resources in PJM must be: >100kW, available 9am-10pm (4-7hrs/event typical), and achieve full response in 1 hour • Emergency • 659 MW • $40/MWh average payment • Economic • 724 MW • $46/MWh average payment • PJM’s long-term goal is to make market rules technology neutral

31. ERCOT Responsive Load Programs • 4000 MW of historic interruptible and curtailable load • Three categories under new system total ~800 MW • Voluntary load response • 15 min energy market • Qualified balancing up load • Capacity & energy but only paid for response • Load acting as a resource • Capacity and energy • Underfrequency relays for Responsive Reserve • <2 minute response • Does not count as spinning reserve - yet • Regulation not yet accommodated

32. Florida Power and Light Responsive Load Programs • Historic 1980’s programs • Customers elect which loads to include • AC (4 hrs 50% duty cycle) or (3 hrs continuous) • Electric heat (4 hrs 50% duty cycle) • Pool pumps (4 hrs continuous) • Water heaters (4 hrs continuous) • 1000 MW peak reduction • 2000 MW emergency response • 2 sec – 1 minute response • Does not qualify as spinning reserve Ed Malemezian 772-286-9831

33. Florida Power and Light Responsive Load Programs • 800,000 out of of 4.4 million customers • Two way communications • Fast broadcast, slow response • Counts requests, operations, effective operations • Used to verify all operations • Now periodic checks for equipment failure • Starting to do AMR • Incentive was ~$85/yr now ~$42/yr • Response can be location specific • Operator tools to forecast and tailor response

34. Other Responsive Load Programs • Cinergy: 600 MW commercial and industrial • PEPCO: 150,000 residential & >300 commercial customers • …

35. Interesting Complexities & Concerns • Observability – metering, communications, forecasting • Statistical response • Forecasting • Override • Stability • Reliability rules and market structure

36. Metering and Communications Requirements • Givens: • Payment must be tied to actual response • Deployment signals have to be fast • One SCADA monitoring system currently performs three functions • Continuous readiness monitoring • Real-time event monitoring • Performance monitoring • How much monitoring is required? • Large aggregations of small, independent resources (statistical response) may not need the individual real-time monitoring that large, individual resources require (deterministic response) • Redundancy may be better than observability. • A 5% error in total load forecast can be a problem. A 5% error in reserve response may not be. Especially if the error is usually in the direction of needed response. FP&L has sophisticated response forecasting tools • Performance monitoring can be slower • What information does the system operator really require in real-time?

37. Communications Requirements Are Asymmetric(This is a Big Benefit) • System-to-load communications are typically broadcast • Resource need – MW of response desired • Price • Deployment – respond Now! • Load-to-system communications are typically individual • Capabilities and price offer • Performance monitoring – conceptually can be slower • Aggregator may help

38. Statistical Response May Be Better Than Monitored Response Aggregation of many (smaller) individually lower reliability resources still provides higher guaranteed response than fewer (larger) individually higher reliability resources

39. Common Concerns: Overrides – A Benefit and a Problem, But Less of a Problem for Spinning Reserve Some technologies block emergency response overrides

40. Common Concern: Stability Will Suffer With Reduced Inertia – Here It Improves WECC Analysis: Loss of Palo Verde 1 • Case A is the normal base case simulation. • Case F, normal governors but 300 MW of spin has been removed and one Helms pump trips at 59.95 – A better freq. response than the normal base case. Analysis done by Donald Davies, WECC

41. Service Definitions Are Critical • Most generators do not care if they run for 30 minutes or 8 hours • May have minimum run times, minimum loads, minimum off times, ramping constraints, emissions limits, startup/shutdown costs, etc. • A load may be able to respond for 10-30 minutes but not 2 hours (rising cost curve) • Can re-arm immediately if not used frequently – an excellent match to reliability requirements but not to energy markets • Response capability matches spinning contingency reserve much better than demand relief

42. Generation System Reliability Market Rules Responsive Load Market Rules Do Not Align Perfectly With Reliability Needs Market and reliability rules naturally tend to be aligned with the capabilities of the dominant supply technology Minimum load Minimum run time 10 minute ramp Minimum off time Start time Inaccurate response …

43. Capacity Market Interactions • A Conceptual Discussion • A viable power system must have contingency reserves • If long-term reserve payments do not cover the capital cost of a CT (the lowest cost generation) then some reserve costs are being covered elsewhere • Market interactions should be examined to assure they are doing what you want Spinning reserve is part of the real-time capacity requirement Supplying spinning reserve frees other generation to serve load Example: If a system requires an additional 50 MW to meet peak requirements and a new CT is economically viable but a cheaper contingency reserve resource is not there is likely a flaw in one of the market rule sets

44. Co-Optimization • Co-Optimization optimizes specific supply technologies (generation) • Co-Optimization does not optimize the reliability functions being procured • Spinning reserve • Non-spinning reserve • Supplemental reserve • Co-Optimization assumes that any resource that can provide spinning reserve can provide non-spinning reserve and supplemental reserve and energy • Dispatches reserves for energy when energy prices are high

45. Sample Economics • Difficult to get good public cost numbers • FP&L • $150/customer installed • $120/kW demand reduction • $60/kW contingency response • LIPA Carrier Comfort Choice • $515 per residential and $545 per commercial customer • $487/kW demand reduction • $162/kW contingency response • Compares favorably with $500/kW combustion turbine

46. Needed Rule Changes • NPCC definition of 10 Minute Synchronized Reserve • Technology neutral • Recognize value of fast response • Recognize the value of reserve restoration • Markets • Do not force reserve providers into the energy market • Payment for Capacity • Value the resource with respect its availability coincident with system need & loss of load probability • Make sure, in whatever fashion, that the entire reliability service and value proposition is addressed

47. Conclusions • Communications and control technology now makes spinning reserve from load possible • There are advantages to the responding loads, the power system, other customers, and society • Spinning reserve is a better match for some loads than peak reduction • Rules are important

48. Conclusions (continued) • If a technology does not serve the physical reliability needs of the power system we do not want it • If a technology does provide reliability benefits but does not meet current market or reliability rules we may want to reexamine the rules