Draft Standard Protocol for Cooling and Fan Savings from

RTUs

RTF March 1, 2011

RTF-RTUG Time LineNEEA2002-2004

RTFPhase I

2004

RTFPhase II

2006-2007

RTFPhase III

2009-2010

• Pilot Field Work• Inconclusive Savings• Identified potential deadband issues• Not ripe for Market Transformation • Lab Testing

• Limited Field Test• Honeywell Sensor Redesign• Develop Protocol to Measure Savings• Scope Next Phase

• Secondary Source Review• Frequency of Problems• Scope Next Phase

• Test Protocol to Measure Savings• Large BPA Field Test• Goal: Deemed, Deemed Calculation or MV Protocol• Scope Phase IV

Slide 2

Rooftop Unit Working Group Time Line

RTFPhase IV

2011

RTFPhase V

2012

RTFPhase VI

2013

RTFPhase VII

2014

•Analyze annual data• Propose annual savings estimation methodology• Propose M&V protocol• Elements of a regional Strategic Plan/Roadmap• New/next generation technology• Recommendation to RTF for next steps• Expand West Coast collaboration

•

•

• Embedded RTU performance monitoring/FDD specification• Test deep retrofit options• Fix DOE2/eQUEST

• Implement Roadmap elements•Web-enabled T-stat approach verified with 2-yr data analysis• Premium Ventilation benefit established• Stronger regional/national collaboration

• Identify deep retrofit options

•

• Deep RTU retrofit program

• New gen evaporative

• Adaptive controls

Slide 3

Rooftop Unit Working Group (End of) Time LineRTF

Phase XXXIII2046

• Dx RTU’s illegal• Evaporative/hybrids, VRF and ground source heat pumps required by law • Self-replicating neural networks with self-diagnosing, self-regulating, & self correcting controls on all units• Bomb proof economizers/controls• 7-10 year maintenance intervals• HVAC OEM’s required to provide performance guarantee over product life• Cherniack Retires

Slide 4

Today Review Protocol & Calculator Review issues raised & addressed at RTUG Staff Recommendation

– Resolve any RTF issues– Finalize protocol & calculator under proposed

guidelines & template– Staff to scope task to finalize protocol, test calculator,

compile documentation and conform to format

Slide 5

Regional RTU Overview Varied programs, participants, utilities and approaches

across the region

Project Goal: Protocol for cooling & fan savings usable across programs, climates and building types

Collaborative research sponsored by

– BPA, ETO, Avista, Idaho Power, RTF

– Technical analysis by Cadmus, NBI, QuEST, Stellar Processes, Ecotope & others

Slide 6

It took a village…

Slide 7

The RTUG Process 2010 7 RTUG Subcommittee meetings Prioritized research needed to develop

protocol Cadmus & NBI doing analytics Feedback from RTUG Subcommittee Lots of RTUG participation 20-30 people each

– In- and out-of-region (CA/NE) participation– Other issues taken up too: DCV, Premium

Ventilation, whole-building approach Slide 8

The DataPNW RTU 2008-09 Metered RTUs Avista = 2 pair side-by-side metered @ 12 months BPA RTU Pilot = 161 metered, 24 units @ 12 months Energy Trust of Oregon = 13 metered Idaho Power = 9 metered @ 12+ months (still going)

>> The nation’s largest, most robust, publicly available measured RTU data set

http://www.nwcouncil.org/energy/rtf/subcommittees/rtug/2009_rooftop/Default.aspx

Slide 9

Current RTU Savings Research2010-13/ Web-based T-Stats (Dreamwatts)

BPA 2010 start: 44 RTUs on 3 buildings 24-36 months DA; Cadmus analysis on first 12 months only with report March 2011

RTF discussion on data analysis for trailing 12-24 months of BPA units

Data available to upload to Regional RTU Database

RTF 2011 start: 3 heat pumps @ 24 months w/analysis

Slide 10

Results for 2010 Substantial empirical & analytical basis to

estimate savings attributable to RTU measures

– Huge progress has been made!

– Techniques are being used in impact evaluations

Developed protocol & associated calculator

Slide 11

What Did We Find? Accurate measurements require understanding the

distribution of daily temperatures and proper timing of metering

Differences between hot east side metering and mild west side metering

Measurements of average program savings at 90% confidence/10% precision can be achieved

Buildings exhibit natural changes over a summer: ‘building squirm’

Methodology will pick up any spontaneous building operations (squirm) in between pre-post- measurement periods 12

Issue: Highly variable savings

Probably not deemable UES Site-specific savings depend on as-found & program measures eg: ETO adding economizers to 3-4 ton RTUs

From BPA 2009

Slide 13

Key Elements Addressed by Cadmus & NBI & RTUG -1

Protocol issue raised: –Does a single signature accurately show annual energy use?

How issue was resolved:–Tested the signature prediction for 10+ months modeled data and 10+ months measured data

How issue is expressed in the protocol: - The protocol is based on a single signature for annualization

Slide 14

Basic Savings AlgorithmEnergy signature based on kWh/day and average daily temp– Horizontal line = fan energy– Sloped line = fan + compressor energy– Balance point = intersection of horizontal and sloped lines– Use with TMY3 data to annualize energy

Slide 15

Key Elements Addressed by Cadmus & NBI & RTUG-2

Protocol issue raised: –Does the duration of metering affect the accuracy of the prediction?

How issue was resolved:Data block analysis was done April – October in 1 week increments @ 10, 15, 21, 28 day intervals

How issue is expressed in the protocol: - The protocol uses a 4 week measurement interval as

most accurate; early/late summer intervals produce the same performance estimates within a few percent

Slide 16

Key Elements Addressed by Cadmus & NBI & RTUG-3

Protocol issue raised: - Does this method lead to biased predictions?

How issue was resolved:– Examined statistics from the data block analysis that showed hot mid-summer data can bias predictions especially on the East side

How issue is expressed in the protocol: - The protocol recommends not using data from 1st week

in July through 1st week in August

Slide 17

Key Elements Addressed by Cadmus & NBI & RTUG-4

Protocol issue raised: - Use of of site temperature data vs. recorded

hourly data from local weather station data

How issue was resolved:– Use of local hourly data was successfully tested as source of the temp data for deriving the energy signature

How issue is expressed in the protocol: - Protocol recommends use of data from local source within limits, otherwise use shielded site-based

Slide 18

Key Elements Addressed by Cadmus & NBI & RTUG-5

Protocol issue raised: – Sample size for M&V to achieve 90% confidence/10% precision

How issue was resolved:– Statistic were analyzed to determine sample sizes

1,000 RTUs: 137 500 RTUs: 129 100 RTUs: 88

How issue is expressed in the protocol: - The protocol includes a table to guide sample size

Slide 19

Key Elements Addressed by Cadmus & NBI & RTUG-6

Protocol issue raised: – Can the annualized protocol be applied to VSD on compressor/fans or to VRF technology

How issue was resolved:– No specific work has been done yet, but the approach

could and should be applied; need method to compare

How issue is expressed in the protocol: - Not currently used in the protocol

Slide 20

Key Elements Addressed by Cadmus & NBI & RTUG-7

Protocol issue raised: – How can the field M&V be accomplished at lower cost?

How issue was resolved:– Used nearby weather station data in lieu of site temp data; used CT in lieu of true power measurement;

How issue is expressed in the protocol: - Adjusted protocol to use these inputs

Slide 21

Accurate Savings Estimate – West/East Sides

Needs shortest possible measurement intervals to avoid building operation changes

Need 4 week measurement during recommended metering window to capture sufficient weekend data

Pre-measurement window 3rd wk May through 1st week July

Post-measurement window 2nd wk Aug through 4th wk Sept

Summer data usually shows no baseload in avg daily –needs to be calculated from minute data

In practice: regressions based hot weather 1st wk July through 1st wk August can bias annualized predictions due to limited cool weather data (limited or no economizer use)

Slide 22

Base Load Analysis Examined for statistical differences in

the base load calculation: calculated error did not affect sample size

Found differences due to using more power in winter: heating fan energy

23Slide 23

Related Non-Protocol Issues Not Resolved

Longevity of fan and t-stat settings

Air flow issues:

- How to treat negative savings from upgrade to minimum ventilation or ASHRAE 62.1?

- Programmatic: in field measurement/reset options – use of DCV as a proxy?

- Building code requirement

- Program benefit-cost issue

Measure life: 6th Plan = 5 years for O&M?

Slide 24

Key Elements of Calculator• RTU power by minute; could be from single calibrated CT

• RTU outside air temperature by hour from site minute data or hourly local weather station if acceptable

• RTU supply air temperature by minute; not necessary for annual usage estimate, but useful for estimating minimum percent OSA

• Annual histogram for normal average daily temperature at site from local TMY data

Slide 25

Key Element: Normalized Annual OutputCalculation of Normalized Annual Consumption

Site Temperature signature information from data tabHistogram week end week day

baseload 23.56719 22.71532balance t 54.81856 46.68124slope 1.359394 1.364004

weekly annualBin Frequency weekend weekday melded bin kWh

10 015 0 12.5 23.56719 22.71532 22.95871 020 0 17.5 23.56719 22.71532 22.95871 025 2 22.5 23.56719 22.71532 22.95871 45.9174230 4 27.5 23.56719 22.71532 22.95871 91.8348435 5 32.5 23.56719 22.71532 22.95871 114.793640 26 37.5 23.56719 22.71532 22.95871 596.926545 53 42.5 23.56719 22.71532 22.95871 1216.81250 73 47.5 23.56719 23.83212 23.75642 1734.21955 42 52.5 23.56719 30.65214 28.62787 1202.3760 45 57.5 27.21233 37.47216 34.54078 1554.33565 59 62.5 34.0093 44.29218 41.35421 2439.89870 36 67.5 40.80627 51.1122 48.16765 1734.03575 17 72.5 47.60324 57.93222 54.98108 934.678480 3 77.5 54.40021 64.75224 61.79451 185.383585 0 82.5 61.19718 71.57226 68.60795 090 0 87.5 67.99415 78.39228 75.42138 095 0 92.5 74.79112 85.2123 82.23482 0

100 0More 0 Normalized Annual Consumption, NAC >> 11,851 kWh/yr

Slide 26

Key Observations Protocol needs graphical data quality

review Protocol performs annualization &

normalization Signature will show economizer operation

changes The data should be explicitly segregated by

weekday and definable weekend Hourly load shape is available

Slide 27

Summary Protocol ready for finalization including

– Resolving any remaining issues– Similar to the 7 protocols under development– Small team with contractor with conforming protocols– Cleaning up calculator– Add ProCost module– Finishing protocol template– Documentation of supporting data & protocol testing– Other conforming details– Bundle up for adoption & publishing

Staff to scope contract approach & budgetSlide 28