PNW RTF -Rooftop Unit Working Group - RTUG – Rooftop Unit Research Project Phase 4

1. PNW RTF -Rooftop Unit Working Group- RTUG –Rooftop Unit Research Project Phase 4 November 23, 2010

2. RTUG Agenda 112310 9:30am Introductions Agenda Review; RTUG Schedule Review 9:45 Review of Results from BPA RTU Pilot Research 2009-2010 RTU gas pack characterization from annual data Model/methodology description Examination of signature predicted annual usage with 3-4 weeks of metered data Outside Air Temperature data sourcing Points to meter Sampling period Recommended metering interval of 1 minute M&V costs Points included in the basic M&V package (kW and OAT) Minimum requirements (temperature ranges/number of days) and reasons why Questions to RTUG members: What do members perceive as unknowns or gaps in this methodology and what is the process to address any gaps? How does this protocol comport with the guidelines under development for Standardized Protocols with M&V? Any other concerns small or large? Next steps: Discuss recommendations to RTF [January 2011]; schedule additional RTUG review if needed 12:15-1:00pmLunch 1:00 Status: Regional RTU Research Results Database: BPA and Web-Enabled T-Stat http://www.nwcouncil.org/energy/rtf/subcommittees/rtug/Default.htm 1:05 Energy Trust of Oregon RTU DCV M&V – Dave Robison, Stellar Processes 1:25 CO2 Sensor Research– Reid Hart, PECI 1:45 BPA request for members to join new RTF Subcommittee on Variable Capacity (VC) Heat Pumps 2:05 Review draft elements of a regional RTU Strategic Plan/Roadmap including Research/Policy/Programs/Market Transformation elements 3:30 Recap of meeting conclusions 4:00 Adjourn

3. Current RTUG Schedule • Nov 23 • Follow up discussion if needed • January 2011 RTF presentation • RTF next steps with HVAC Committee

4. BPA RTU Winter OperationPresented to:RTUGNovember 23, 2010 Presented by: Kathryn Hile, The Cadmus Group Howard Reichmuth, New Buildings Institute

5. Agenda Overview • Project summary • Proposed M&V protocol • Data chunk analysis • Outside air temperature • Simplified monitoring • Next steps

6. Annual Metering Summary • Metered 45 units for one year

7. Annual Summary (cont.) • Annual metering interval: one minute • Costs • Equipment: $1,150/RTU • Metering equipment installation and removal: $700/unit • Analysis and reporting

8. M&V Protocol – Need Your Input • Need your feedback, so we can have a high consensus recommendation to RTF by February • Will send out a draft protocol incorporating your suggestions from today • Would you support our recommendation to the RTF on this protocol for a 90% confidence fan and cooling energy savings?

9. Questions to Think About • What are unknowns or gaps in this methodology • How do we address any gaps? • How does this protocol comport with the guidelines under development for Standardized Protocols with M&V? • Any other concerns small or large?

10. Elements of the Protocol • Energy signature analysis approach (Word doc) • What points to meter • Sampling period • Number of days • Temperature ranges • Sample size • Confidence interval and the distance ratio

11. Detailed Analysis for M&V Protocol • Units selected based on consistent operation (one mode) • Gas packs • Nine units • Four on east side • Five on west side

12. What Points to Meter? 12 • Use shortest possible measurement intervals to avoid building changes • Minimum required points to meter • Outside air temperature • Power • Supply air temperature: nice, but not essential

13. Sampling Period 13 • Number of days – and why • Temperature ranges – and why • See range of temperatures for different points in the monitoring season and different date ranges • Frequency table with average daily data • Economizer and compressor and fan operation

14. Updated Energy Signature • Based on linear regression fit • Fan and A/C savings • Annualized using TMY bin data • Tested quadratic regression fit, but no significant improvement

15. Data Chunk Analysis • Basic ground rules • Statistics approach • Discussion of confidence intervals • What did we find?

16. Basic Ground Rules • Units with constant operation (no switching between fan auto/fan continuous in one RTU) • 21 and 28 days of metered data, starting every 7th day, mid-May through end of September • Extrapolate to annual using regression • Exclude all points with total daily kWh < (base load * 10%)

17. Data Chunk Statistics Approach • Applied to full annual estimate (including histogram) • 90% confidence interval for each extrapolation • Characterize temperature distributions with distance ratio: (Average distance from each daily temp to mean cooling TMY temp) (Variance among daily metering period temps) • Number of observations for 28-day monitoring period conservatively assumed to be only 21 days to account for weekend mode

18. Confidence Intervals • 90% confidence • Overlapping CIs indicate no significant difference • Distinct CIs (no overlap) indicate significant difference • Last summer’s savings loose estimate: 200 kWh/ton • Savings estimate involves two precisions treated in quadrature

19. What Did We Find? • Accurate measurements require understanding site weather and proper staging of metering • Differences between hot east side metering and mild west side metering • Measurements at the 90% confidence can be achieved • Buildings exhibit natural changes over a summer: ‘building squirm’ • Best approach involves 4 weeks pre, 4 weeks post, with 3-4 week retrofit window

20. East Side Weather Summary • Pasco weather from Pasco TMY3 • TMY3: • Hourly values for a one-year period • Derived from actual weather data 1991-2005

21. Pasco TMY3 21

22. Pasco TMY3 365 Days 22

23. Pasco TMY3 May 1 – Sept 30 23

24. East Side Results Example 1

25. Energy Signature Significant change, inadequate detailed data to diagnose.

26. East Side Results Example 2

27. Energy Signature Small but significant change, 20% minimum OSA June to 50% OSA in August.

28. East Side Findings • Best monitoring pre-late May, early June. Four-week sample • Best monitoring post-early August to end of August • Less accurate results for mid-summer, late June/July monitoring interval • Mid-summer data temperatures are above economizer range leading to inaccurate predictions

29. West Side Weather Summary • Snohomish TMY3

30. Snohomish TMY 365 Days 30

31. Snohomish TMY3 May 1 – Sept 30 31

32. West Side Results Example 1

33. Energy Signature Messy data but leads to tight confidence interval. No significant change.

34. West Side Results Example 2

35. Energy Signature Messy data. Early season heat and false cooling trigger. No significant change.

36. West Side Findings • Good results at 90% confidence, but generally not as good as east side • Pre-retrofit period should start in June • Post-retrofit period of four weeks should start in August Conclusion: Meter four weeks pre and four weeks post to achieve 90% confidence

37. Sample Size • m = sample size • = standard deviation of savings • = coefficient of variation among actual energy savings of units within program • = within the regression calculation for a given unit, the standard deviation of the distance from the actual daily energy usage to the regression line • = savings over entire year • N = # compressor on days in TMY data – we can make this determination, based on CZ • n = # compressor on days used in the analysis (should be all metered days in a given operating mode, minus weekends if off on weekends) • R = distance ratio

38. Sample Size • To get good estimates with 90% confidence, if you service… • 1,000 units, meter: • 500 units, meter: • 100 units, meter:

39. Outside Air Temperature 39 • Compared measured OAT to local weather station data for all sites • Average daily temperatures during cooling season • Examples: • Annualize using weather station data • Annualize using metered OAT data

40. OAT Comparison Snohomish 40

41. OAT Comparison Snohomish 41

42. OAT Comparison Snohomish 42

43. OAT Comparison Pasco 43

44. OAT Comparison Pasco 44

45. OAT Comparison Pasco 45

46. Simplified Monitoring • Weather station data may be key • CT current/power measurement simplifies installation? • Potential metering costs can be reduced by 80%

47. Questions to RTUG • What do members perceive as unknowns or gaps in this methodology and what is the process to address any gaps? • How does this protocol comport with the guidelines under development for Standardized Protocols with M&V? • Any other concerns small or large?

48. Next Steps in Research • Draft M&V protocol for fan and A/C savings, based on input from committee • By December 1, we will send to RTUG committee members a straw man of our recommendation • Look at building characteristics to find a better metric to predict energy savings

49. Regional RTU Research Results Database http://www.nwcouncil.org/energy/rtf/subcommittees/rtug/Default.htm • 102 RTUs 2009-10 BPA Pilot; 2wks-pre/2wks-post • 25 RTUs 12 months monitored Sept/Nov/Jan 011 • Web-enabled T-stat: 2 sites, May-Sept 2010

50. Energy Trust of OregonSmall Commercial HVAC Pilot Program Stellar Processes November 23, 2010