Measurement and Verification for Residential DR Training

Transcription

1 Measurement and Verification for Residential DR Training July 20, 2015

2 Disclaimer PJM has made all efforts possible to accurately document all information in this presentation. The information seen here does not supersede the PJM Operating Agreement or the PJM Tariff both of which can be found by accessing: For additional detailed information on any of the topics discussed, please refer to the appropriate PJM manual which can be found by accessing: The new rules have not yet been approved by FERC and are not yet in effect. All manual/tariff language subject to change based on FERC ruling. 2

3 Agenda Overview Sampling Switch operability Data validation and conversion Customer churn Documentation Transition mechanism Registration and market participation 3

4 New Manual/Tariff language PJM Issue Tracking Includes Manual/Tariff language, stakeholder presentations Manual 19 has most of the information FERC filing Docket No. ER Filed 6/4/2015 Expected FERC ruling 8/3/2015 4

5 Reason for changes Current M&V methods based on legacy programs from 20 years ago New methods will make residential M&V more consistent with C&I customers M&V will reflect actual performance during events instead of average performance over previous seasons 5

6 Definitions Measurement and Verification (M&V): General term that includes compliance for Load Management (LM) and Synchronized Reserves (SR), and energy settlements for LM and economic markets. Legacy Direct Load Control (LDLC): Formerly Direct Load Control (DLC). Refers to M&V method in elrs. Direct Load Control (DLC): New tariff definition. Refers to direct load control technology. Ability fo CSP to directly curtail end-use device at end-use customer without intervention from end-use customer. Population: Registered customers. Sample: Customers selected from the registered population of non-interval metered customers who have interval meters installed for the purpose of M&V for all registered customers. 6

7 Overview of changes Deemed Savings Report and load research studies not valid for M&V Starting with DY16/17 Non-interval metered DLC not eligible for CBL/compliance method Statistical sampling: new M&V method Residential non-interval metered customers Real interval meter data for all participants Sample data from same time period as event Must use GLD/FSL for compliance 7

8 Interval metered customers No change in status quo for M&V and meter data collection Actual hourly meter data for all customers is used Not eligible to participate in sampling Some Manual/Tariff changes do apply to interval metered customers, relating to churn, documentation, etc. 8

9 Non-interval metered customers M&V done via real-time sample Random and representative sample of customers with interval meters Sample is in real-time i.e. data from sample is from same time period as event Hourly data from sample is scaled to population After data is scaled to population, processes are same as interval metered customers 9

10 Agenda Overview Sampling Switch operability Data validation and conversion Customer churn Documentation Transition mechanism Registration and market participation 10

11 Statistical sampling - qualifications Residential customers only Curtailment via DLC technology Locations can be accurately sampled to reflect population Must otherwise qualify for relevant market Eligible markets Load Management (LM) Emergency/Pre-Emergency No interval meters recording at 1 hour or better Economic Energy (DA/RT) No interval meters recording at 1 hour or better Synchronized Reserves (SR) No interval meters recording at 1 minute or better 11

12 Sample requirements Separate sample required for: LM Economic Energy SR EDC CSP End-use device/grouping Curtailment algorithm Switch vintage (if substantial variation) Registration or Dispatch group End-use device/grouping Curtailment algorithm Switch vintage (if substantial variation SR subzone Registration or Dispatch group End-use device/grouping Curtailment algorithm Switch vintage (if substantial variation 12

13 Sample must be designed to be representative of population Maximum error of 10% at 90% confidence Sample design If we draw sample same way, 90% of samples draw will give a result within +/- 10% of true population value 13

14 Sample stratification To ensure representative sample Control device size in 2 groups roughly at median e.g. median AC size is 3.1 kw, stratification by AC size < 3.1 kw and > 3.1 kw Based on sum of device sizes at EDC account level Geographic Stratification For large or highly variable geographic areas e.g. AEP wide program would likely require geographic stratification, RECO probably not CSP may propose alternate stratification to reduce variance, subject to PJM approval 14

15 Status quo for CSP owned meters EDC account meter level i.e. entire premise is metered Metering requirements shall meet: EDC requirements for accuracy, or Sample meters Max error of 2% over full range of meter equipment (including PT/CT). For PDR this includes PDR error plus EDC meter error and ANSI compliance See Manual 11 Section 10 Interval Meter Equipment and Load Data Requirements. Required documentation: Meter quality assurance plan, meter qualification form 15

16 Sample size determination First year only: Option to use 150 customers for sample without variance study, and use existing sample. Sample size to satisfy 90% confidence, < 10% error Sample size determined in variance study At least 75 randomly selected participants LM/Energy: 4 weeks of contiguous hourly meter data SR: 2 weeks of contiguous minute meter data Data collection during season that end use device is curtailed E.g. June September for ACs CSP may propose alternate sample size determination technique but must be approved by PJM 16

17 Variance study calculations 1 17

18 Variance study calculations 2 18

19 Variance study calculations III 19

20 Sample recalibration Recalibrated annually Sample size recalculated using data from customers in sample, and same calculations as variance study Proportion of each stratum in the sample must be within +/- 1 sample of population proportion e.g. Sample size = 150 customers Population proportion stratum A= 20% Stratum A should be 30 customers does not need to be recalibrated if customers Replacements if necessary must be randomly selected, maintain strata integrity, etc. If population is expanded in non-random manner, sample must be expanded appropriately 20

21 Agenda Overview Sampling Switch operability Data validation and conversion Customer churn Documentation Transition mechanism Registration and market participation 21

22 Switch operability Switch operability study no longer used Two way switch communication One-way switch communication 22

23 Switch operability 1 way communication CSP cannot accurately determine if each switch successfully cycled based on CSP instruction Operability value is implicit in sample Must report sample data from all switches, even if inoperable Cannot repair failed switches until: Repair faulty switches in population OR Reselect entire sample Includes any system/device that would cause end-use device not to reduce load properly in the population Metering and metering communication Can be fixed in sample Includes only systems/devices that would not affect load reduction in population Component that is related to both metering and switching cannot be repaired Switch failures in sample must be reported to PJM within 2 business days 23

24 Switch operability 2 way communication CSP receives verification from switch that it successfully cycled based on CSP instruction Performance factor for each event based on actual population operability: F = number of switches in population that successfully cycled/total number of switches Inoperable switch in sample Sample size > requirement: do not report load data from inoperable switch Sample size < requirement: must report load data from switch Can repair faulty switch in sample or population at any time 24

25 Agenda Overview Sampling Switch operability Data validation and conversion Customer churn Documentation Transition mechanism Registration and market participation 25

26 Data validation and estimation Data validation and estimation should use NAESB Validating, Editing, and Estimating (VEE) Protocol. Should be used for 1-minute SR data as well as hourly data for LM/Energy. Use same rules as NAESB VEE but substitute 1- minute interval where it says 1-hour Sample meter data is disqualified from being reported for an event if 5 minutes or more are missing or faulty from 1-min meter data for a single event 2 hours or more are missing or faulty from hourly meter data for a single event 26

27 Missing meter data 1 If meter data cannot be reported from a sample meter for an event then: One-way switch: report PLC value for any sample meters that don t have valid data, regardless of whether there are sufficient meters to meet the minimum sample requirement Ex: Min sample size: 150 total sample meters: 155 sample meters with missing or faulty data: 7 submit data from 148 function meters, and PLC values for remaining meters 27

28 Missing meter data 2 If meter data cannot be reported from a sample meter for an event then: Two-way switch: If sufficient valid meter data to meet minimum sample size, do not need to report any data from faulty/missing meters. If faulty/missing meters mean that sample no longer meets minimum sample size then must report PLC values for faulty meters. Ex: Min sample size: 150 total sample meters: 155 sample meters with missing or faulty data: 7 submit data from 148 function meters, and PLC values for 2 meters to meet 150 sample size. No data required for remaining 5 meters. 28

29 Sample data population data Note: sample data must be from same time interval being settled XX ii,tt meter reading for customer i during interval t, after VEE is applied B is the set of EDC accounts in sample to be included in estimation MM ss is the sample size (number of EDC accounts in B) MM cc is the population of cycled customers F is the operability factor one-way switch: F = 1 Two-way switch: F = number of switches in population that successfully cycled/total number of switches Meter data value to be submitted to PJM for interval t: YY tt = FF MM cc MM ss XX ii,tt ii BB 29

30 Agenda Overview Sampling Switch operability Data validation and conversion Customer churn Documentation Transition mechanism Registration and market participation 30

31 Customer churn Applies to both interval and non-interval metered residential mass market customers Customers may be replaced if: Customer moves from their premises Customer terminates contract with CSP CSP to maintain list of all replaced customers and furnish to PJM within 2 business days of request Include date of replacement, reason for replacement, new/old EDC account, PLC, zip code 31

32 Customer churn 2 Load Management Registration CSP may not add/remove customers other than replacement CSP must maintain number of registered customers every day of product availability period. If number of customers falls below, CSP to report to PJM within 2 business days and subject to resource deficiency charges if applicable. Interval metered LM: replacement customers selected must maintain PLC and load drop in aggregate Non-interval metered LM: replacement customers must be randomly selected to maintain integrity of strata, and if applicable, PLC and load drop (in aggregate). 32

33 Customer churn 3 Economic energy & SR CSP can add/remove customers at any time Maintain documentation of which customers are added/removed (date, EDC account, PLC, zip code) Update number of customers on location in elrs. Must be accurate before each offer. List of offered customers must be finalized at time of offer. List of offered/cycled customers must be maintained for each offer and event for 2 years. Number of offered customers cannot exceed number of customers on location. Interval metered: no restrictions on replacement customers since actual meter data is submitted Non-interval metered: customers can only be added/removed if can be done such that sample remains representative of population. CSP must maintain documentation indicting this. If CSP offers partial list of customers to market, then offered customers must be randomly assigned from pool of all registered customers so sample remains representative. 33

34 Agenda Overview Sampling Switch operability Data validation and conversion Customer churn Documentation Transition mechanism Registration and market participation 34

35 Documentation To be submitted to PJM before registration M&V plan (before May 1) Meter quality assurance and qualification documentation List of all registered customers: EDC account, PLC, zip code (before June 1) Pseudo-EDC account permitted where legal/regulatory conditions prohibit sharing To be maintained by CSP for 2 years and submitted to PJM on request within 2 business days. List of any replacement customers List of customers cycled during an event Individual meter data (sample or actual) from event/test 35

36 Documentation: measurement & verification plan M&V plan must be submitted and approved by PJM before registration is submitted. Updated and resubmitted annually to continue participation in PJM markets Content: Details on variance study How was sample size determined How was sample selected and stratified Meter qualification Meter quality assurance Data validation and error correction protocol How will sample meter data be converted to population meter data 36

37 Agenda Overview Sampling Switch operability Data validation and conversion Customer churn Documentation Transition mechanism Registration and market participation 37

38 Transition mechanism 1 MW that were committed before FERC approval of new rules can be relieved via transition mechanism if: Cannot comply with statistical sampling requirement Estimated reduction using meter sample less than load research study or Deemed Savings Applicable to MW already cleared for Delivery Years 2016/17, 2017/18, 2018/19 For any auction for which CSP is using transition mechanism, CSP is not permitted to sell MWs in any modeled LDA or sub-lsa for which non-viable DR MWs are declared Bilateral transactions are subject to review and referral to MMU and/or FERC OE. Bilateral transactions should not be executed for non-viable MW for financial gain. PJM to reduce CSP capacity commitment in erpm by product/zone and prorate amount of non-viable MW as necessary if CSP cleared MWs in multiple auctions. CSP responsible to register sufficient DR to cover final capacity commitment or received daily deficiency charge PJM to only pay capacity revenue based on final CSP capacity commitment 38

39 Notify PJM: in writing via at Transition mechanism 2 UCAP that cannot be delivered due to sampling requirements by enduse customer site and/or detailed analysis Notification 7 days prior to posting of planning parameters for IA DY2016/17 3 rd IA DY2017/18 1 st, 2 nd or 3 rd IA DY2018/19 1 st, 2 nd or 3 rd IA IA planning parameter posting date can be found here: 39

40 Agenda Overview Sampling Switch operability Data validation and conversion Customer churn Documentation Transition mechanism Registration and market participation 40

41 Registration and market participation For details on using elrs, emkt, erpm and general market participation, please see the following: Economic Demand Resource in Energy Market Economic Demand Resource in Ancillary Service Market Economic Demand Response Offer Administration in emkt System Emergency DR (Load Management) in Capacity Market Additional information is on the PJM Training Material Page 41

42 Create location in elrs EDC account: use pseduo account number to represent multiple customers. elrs: creating locations # of Sites Multiplier: number of customers that are being registered with PJM. This should be based on EDC customers accounts, not on the number of AC s or water heaters being controlled. elrs has a limit for site multiplier of 99,999 so if you are registering more accounts, must use multiple locations. kw Reduction (per site): the average per site reduction, at the customer meter level (i.e. not including losses). This number is determined by CSP based on expected reduction. Total load reduction: elrs will automatically calculate this for you. You should expect to see (Total Per Site) X (# of Sites Multiplier). This should reflect the total quantity of load reduction for all registered customers, at the customer meter level (not including losses). Peak Load Contribution: This should be the sum off all PLCs for all registered customers. Actual individual PLCs should be used instead of class averages when available. Address: Pseudo address. Zip-code: should be used for all residential mass market registrations (sub-zonal or not). If the same customers will be used for LM and economic registrations, the same elrs location must be used. 42

43 elrs: LM Registrations 1 Load Management Measurement method: should be Firm Service Level (FSL) or Guaranteed Load Drop (GLD) Only. In nearly all cases will be FSL. Firm Service Level (FSL) Emergency or Pre-Emergency DR (Load Management) achieved by a customer reducing its load to a pre-determined level upon the notification from the CSP s market operations center. The customer must be able to reduce load below the pre-determined level which must be lower than the amount of capacity reserve for the customer as represented by the peak load contribution ( PLC ). 43

44 elrs: LM Registrations 2 FSL = PLC (Firm Load Level X Losses) Adjusted to include system losses This represents the load value that customers will reduce to Aggregate value for all customers on registration 44

45 elrs: Economic Registrations Economic If participating in DA market, correct LSE is required. If not, may set LSE = EDC for residential mass market only. CBL Method must qualify for CBL via RRMSE test 45

46 All meter data submitted should be aggregated to population level Meter data submission Sum of meter data for all locations for interval metered customers Sample data scaled to population for non-interval metered customers Templates for meter data submission located here: Economic Energy May require 60 days of aggregate meter data depending on CBL LM Compliance FSL 24 hours of data LM Energy 24 hours of data (unless using economic CBL) SR 10 min before to 10 min after full event 46