Demand Reduction in the Forward Capacity Market


Published on

Published in: Technology, Business
  • Be the first to comment

  • Be the first to like this

No Downloads
Total views
On SlideShare
From Embeds
Number of Embeds
Embeds 0
No embeds

No notes for slide
  • I’m Jennifer Chiodo and want to point out the lead author on this paper, Kathryn Parlin is presenting another paper in this session and she asked me if I would present on the Vermont Demand Reduction Verification Effort.I got involved in Jan 2008 when I was asked to help plan implementation of Vt MV PlanAfter reading the ISO MMVDR and Market Rule 1, I entered the first meeting thinking – this is great! We are going to install permanent meters on our EE projects, have real time monitoring and be able to KNOW how much energy we are saving.I quickly learned that this was not the case and that or task was to figure out how to meter efficiency in compliance with ISO standards without permanent metering.
  • Independent System Operators (ISOs) in NE and PJM territory are holding auctions to procure future power capacity.They mechanism allows for demand resources resulting from energy efficiency programs to be bid into the system as future supply capacity.Most programs in NE participate.Participation of DSM programs in the PJM RPM was not as extensive as that in NE when last reviewedOther ISOs are looking at these models
  • Reliability is crucial for the credibility of efficiency as a demand side resource.Prior evals had focused on energyPrograms have traditionally used customer specific peak demand reduction estimates.Generation is knowable, easily measured and is operated by knowledgeable market actors whose business is primarily to generate and sell power. (Changing a little with grid interconnected small scale renewable generation, but these are not yet bidding in as capacity and they are individually metered)Energy Efficiency – as I noted at the beginning, even with smart grid we are pretty far from having real time metering of energy savings.The projects are diverse, individually small and abundant. How many people here live in NE? How many of the NE residents have CFLS in home and/or super T8s in your office? These and thousands of people like us are the operators.Presents the overarching questions – how do we ensure reliability?
  • ISO NE issued a guide for the participants bidding in demand resources (addressed DR too) called the Manual for Measurement and Verification of Demand Reduction Value from Demand ResourcesParticipants required by the manual to file and have an approved M&V Plan prior to bidding into the marketVt filed M&V Plans were approved and they bid into the marketThe Vermont Department of public service was identified in the contract as the required independent third party verifier or compliance with the filed M&V Plan.DPS contracted with WHEC and members of the team for the first verification included Lexicon, GDS and Cx Associates. ERS has since joined the team.During the rest of this presentation I’ll be sharing with you the process the team used including sampling, M&V, QC and findings of realization rates
  • The FCM M&V for Vermont represented a fundamental shift from past verifications which had focused on kWh and were done using engineering review of the program paper and electronic files.The new order was to focus on demand during recently defined summer and winter peak hoursThere were also required changes from previously agreed upon baseline assumptions such as the understanding that in Vermont small new commercial construction is typically below code. To comply with ISO – the code is the baseline for all NC even if local construction may not have a code baseline. In addition, prescriptive programs that have a standard demand reduction value for lights going into NC must be analyzed using an LPD baseline.First round covered two yearsSecond round was just evaluated, don’t have numbers, but Kathryn mightRR from Round 1 began being applied to June 2010’s savings Vermont began receiving payments for capacity in 2008
  • Used prior program participation to determine sampling strategy.Began with an astronomical number of stratification variables and ended up with these.HVAC and associates/Lighting and controls/Other includes refrigeration non-hvac motors, process, envelop, compressed air, etc.Retrofit had to be separated out because of the need for pre and post metering, where that is typically not required for MOP and is unavailable for NCSize – large was set based on prior years and was ultimately lower than we would have liked.Realtime sampling was a key component of EVTs filed M&V PlanConcept was to have an ongoing sampling strategy where projects would be pulled as they came into the program and pre metering could be installed on retrofit projects by the M&V team where metering was not planned by the program.Straight lighting (with no controls) was determined to typically not require pre/post since the schedule was unlikely to change. Therefore for retrofit lighting controls was included in “other” to ensure pre-metering.
  • The team encountered numerous challenges with real time sampling.The sample frame was continuously changing so being confident of our sampling rate and total quantity in the varying strata was a challenge. A separate tracking and reporting mechanism had to be created in which PMs entered likelihood of projects, end uses and estimated kW – this proved very challenging for PMs. Project schanged dramatically from the initial input and could not be resampled.Timing – many projects completed before even showing up in the database. Others were sampled and had meters installed but did not move forward or were completed in subsequent years.Customers – one large customer was contacted as the provider had not anticipated metering the project. The customer was unwilling to have the project metered but meter data not be used in the measure analysis which would have created bias since the program had not planned to meter.7 of 23 were included. Total retrofit # XX, a couple were metered and had installs in subsequent year and will also be used.Filed a plan change – real time sampling no longer in use.
  • WHEC did the sampling and data management.Provided project files and M&V Planning templates. Each project received M&V plansLarge projects were to be metered by EVT and the team was intended to collaborate with EVT on M&V Planning.We used all IPMVP options and had full customer cooperation with the process. Only customer issues were timing related and trying to get meters in for peak periods.
  • In planning for verification we needed to look at the summer and winter peak kW and determine criticallity of metering timing. For snowmaking – needed to meter in winter, for HVAC efficiency needed to meter in summer. Other measures were less obvious. Schools tend to have complex schedules, as shown here.For schools we tried to capture the three or 4 determined operating modes:Shut down, partial occupancy, (teachers) summer sessions and school in session. Metering typically occurs in early to mid August and goes into Sept. to capture the full range of occupancy patterns.Detailed schedules are essential for establishing reliable estimates of the average kW reduction over the performance hours.Performance hours are June – Aug 1-5 weekday nonholidayandDec-Jan 5-7 weekday non-holidayThe team typically used licensed electricians to install metering. We often used people referred by the owner as their familiarity with the electrical distribution in the buildings was very beneficial and when issues occurred the owner’s resource was there to help.
  • RLW issued a study of what metering equipment and approaches came closest to metering the requirements in the M-MVDR.Our team typically used DENT Elite Pros, Flukes and Hobos for the workWe used NOAA weather data from around the state and TMY3 data for normalization.As shown on these graphs we found that it was essential to screen for operating schedule within the performance hours prior to running regressions. These two graphs are for a rooftop air handler on a NC project during the performance hours. We are looking at energy use vs temperature during the performance hours in the first slide and seeing no correlation. On the second slide we see why – the equipment shuts off before the building closes. The final regression on this equipment was very good and showed strong temp dependence.Needed to make sure team understood difference between taking out zeros for scheduled off vs taking out zeros where equipment was cycling off.
  • Review the overall findings. This is for C&I custom measures only.Issues included the change in definition of peak period occurred during the period that we were reviewing. Some values used by the programs did not yet have better data so no adjustment was made by the implementers.First two columns are by type – see hovering around 70%Next three are by size: smalls were not high performers while in past they did very well in paper review. This may happen because standardized approaches are often applied to analysis of small projects which may not capture the nuances of operation.Large projects received more pre/post metering and most of the lighting by kW occurred in medium and large projects.Summer hvac was off because the peak demand value was known not to be optimal but no new value existed at that time. Has since been updated to reflect the KEMA study.Lighting is pretty knowable, especially during the evening hours when variability in commercial use decreases as does use.Overall a little better than 70%
  • Here we can see how the RRs are clustered around unity but have a wide distribution. The zeros are typically for projects affected by the economic downturn in which equipment was removed or in some cases plants were closed.
  • The evaluation exceeded the required confidence/precision targetsThis was a fairly expensive exercise, but that level of investment is required for ensuring reliability of the resource.We invested a lot of time in project planning and needed the flexibility to change things as the evaluation proceeded overall and at the project level.We found that the real time sampling model could not be effectively implemented with the available resources.From a policy perspective, we see this effort contributing to future IRP efforts as well as serving as a more targeted resource when looking at using demand reductions to defer T&D capacity investments.
  • Demand Reduction in the Forward Capacity Market

    1. 1. Demand Reduction in the Forward Capacity Market: Expectations & Reality <br />Kathryn Parlin, West Hill Energy & Computing<br />Jennifer L. Chiodo, Cx Associates<br />1<br />
    2. 2. Overview<br />ISO NE FCM<br />Vermont’s Custom Commercial Portfolio<br />Sampling<br />M&V<br />Findings<br />Recommendations<br />2<br />
    3. 3. Demand Markets<br />What?<br />ISO capacity procurement includes demand savings from efficiency programs<br />Where?<br />ISO NE Forward Capacity Market (FCM)<br />PJM Reliability Pricing Model (RPM)<br />3<br />
    4. 4. Grid Reliability<br />Energy Efficiency<br />Generation<br />Extensively Metered<br />Concentrated<br />Real time data<br />Accurate<br />Dispatchable<br />Known operating schedule<br />Operators – few, known and knowledgeable<br />Not Metered<br />Aggregated<br />Widely dispersed<br />Diverse<br />Not individually metered<br />Varying schedule and output<br />Operators – many, unknown and unaware of project interaction with market<br />4<br />
    5. 5. Process<br />ISO M-MVDR<br />VT M&V Plan<br />VT FCM Bid<br />DPS Verification Contract<br />Sampling<br />Project level M&V<br />Analysis and Report Reviews<br />Annual Realization Rates Established<br />5<br />
    6. 6. VT Portfolio Verification<br />Demand Savings Calculation<br />Paper verification of kWh IPMVP methods<br />Baseline changes<br />Round 1: January 2007 – December 2008<br />Round 2: 2009<br />Round 1 RR applies to savings from June 2010 <br />Capacity payments began in 2008<br />6<br />
    7. 7. Round 1 Sampling<br />Stratification Variables<br />End use: HVAC, Lighting, Other<br />Type: Retrofit, MOP/NC<br />Size: Large, Medium, Small<br />“Real Time” Measure Sampling<br />Retrofit projects<br />No metering planned by program<br />Evaluation team pre/post metering<br />7<br />
    8. 8. Real Time Sampling Redux<br />Dynamic Sample Frame<br />Size threshold established before population known<br />Project tracking and reporting issues<br />Timing – projects completed prior to sampling<br />Customer issues<br />7 of 23 sampled projects completed<br />8<br />
    9. 9. Project Verification<br />M&V Plans developed for each project<br />All IPMVP Options used<br />Option A (Partially measured) typically for lighting <br />Option B (Metered equipment) HVAC, refrigeration, compressed air, etc.<br />Option C (Whole facility) 15 minute interval data, large  kW, consistent measure performance<br />Option D (Calibrated Modeling) New Construction<br />Excellent customer participation<br />9<br />
    10. 10. Project Approach<br />Scheduling<br />10<br /><ul><li>Safety</li></li></ul><li>Project M&V<br />Metering Equipment<br />Seasonality of Weather<br />Temperature<br />Daylight<br />Regressions/Schedules<br />11<br />
    11. 11. TYPE<br />SIZE<br />END USE<br />12<br />
    12. 12. 13<br />
    13. 13. Conclusions<br />+ 80/10 Confidence/Precision<br />Level of rigor essential for ensuring reliability<br />High degree of planning and flexibility required<br />Real time sampling infeasible<br />Benefits have broad potential<br />IRP<br />T&D Constraints <br />14<br />