Building energy information systems: user case studies

Jessica Granderson & Mary Ann Piette &

Girish Ghatikar

Received: 18 February 2010 /Accepted: 17 May 2010 /Published online: 16 June 2010# The Author(s) 2010. This article is published with open access at Springerlink.com

Abstract Measured energy performance data areessential to national efforts to improve buildingefficiency, as evidenced in recent benchmarkingmandates, and in a growing body of work thatindicates the value of permanent monitoring andenergy information feedback. This paper presentscase studies of energy information systems (EIS) atfour enterprises and university campuses, focusing onthe attained energy savings, and successes andchallenges in technology use and integration. EISare broadly defined as performance monitoringsoftware, data acquisition hardware, and communica-tion systems to store, analyze, and display buildingenergy information. Case investigations showed thatthe most common energy savings and instances ofwaste concerned scheduling errors, measurement andverification, and inefficient operations. Data quality iscritical to effective EIS use, and is most challenging atthe subsystem or component level, and with non-electric energy sources. Sophisticated prediction algo-rithms may not be well understood but can be appliedquite effectively, and sites with custom benchmarkmodels or metrics are more likely to perform analysesexternal to the EIS. Finally, resources and staffing

were identified as a universal challenge, indicating aneed to identify additional models of EIS use thatextend beyond exclusive in-house use, to analysisservices.

Keywords Anomaly detection . Baselining .

Benchmarking . Energy efficiency . Diagnostics .

Energy analysis . Energy information system .

Enterprise energymanagement . Performancemonitoring .Web-based energymanagementand control system

Introduction

The focus of this research is case studies of the use ofenergy information systems (EIS) in commercialenterprises and university campuses. Specifically, itaims to understand use of the technology to improveenergy performance in buildings. EIS are broadlydefined as performance monitoring software, dataacquisition hardware, and communication systemsused to store, analyze, and display building energydata. Time-series data from meters, sensors, andexternal data streams are used to perform analysessuch as baselining, load profiling, benchmarking, andbuilding level anomaly detection. At a minimum, EISprovide hourly whole-building electric data that isweb accessible, with analytical and graphical capabil-ities (Motegi and Piette 2003), as illustrated in Fig. 1.

Energy Efficiency (2011) 4:17–30DOI 10.1007/s12053-010-9084-4

J. Granderson (*) :M. A. Piette :G. GhatikarLawrence Berkeley, National Laboratory,1 Cyclotron Road,Berkeley, CA 94720, USAe-mail: JGranderson@lbl.gov

EIS may also include submeter, subsystem, orcomponent level data, and corresponding analysessuch as system efficiencies or analysis of end use, yetthese are less common.

EIS are viewed as a promising technology toreduce energy use in buildings, for a number ofreasons. There is widespread recognition that there isoften a large gap between building energy perfor-mance as designed, and measured post-occupancyenergy consumption (Diamond et al. 2006; Johnson2002; New Buildings Institute 2008). A growingbody of evidence points to the value of permanentmetering and monitoring (Piette et al. 2001b),particularly in the context of monitoring-based andcontinuous or retro-commissioning (Brown et al.2006; Mills et al. 2005; Mills and Mathew 2009).Also pointing to the value of monitoring, researchershave increasingly documented the positive behavioralimpacts of making energy consumption visible tobuilding occupants and residents (Darby 2006;Petersen et al. 2007). In addition, EIS are well alignedwith current trends toward benchmarking and perfor-mance reporting requirements.

Research questions and prior work

The case studies that are presented reveal criticalaspects of EIS usability, and the interplay betweenfeatures, diagnostics, and energy saving actions.

Research questions motivating this work include:Who uses EIS, and which features have proved mostuseful in attaining energy savings? What actions aretaken based on the information provided via an EIS?How much of a building’s low energy use or energysavings can be attributed to the use of an EIS? Whatare common challenges encountered in whole build-ing performance monitoring? What are successful,realistic EIS implementation and use models?

The body of prior work dedicated to EIS is sparsecompared to other aspects of building control anddiagnostics, yet there are several studies that meritattention. A recent study (Granderson et al. 2009)assessed the state of the technology, updating olderwork (Motegi and Piette 2003) to identify the fullspectrum of analyses and diagnostics that EISsupport, and standard features versus more sophisti-cated functionality. Two books published in 2005 and2007 contain editors’ compilations of articles thatdocument the implementation of web-based buildingcontrol and automation systems and their use forenterprise or site energy analysis (Capehart andCapehart 2005; Capehart et al. 2007). Last year,New Buildings Institute published a report thatconsiders EIS in the context of advanced meteringtechnologies (New Buildings Institute 2009). TheCalifornia Energy Commission has published EIScosts and qualitative benefits in the context ofenhanced automation (California Energy Commission

Fig. 1 Basic Energy Information System (Motegi and Piette 2003)

18 Energy Efficiency (2011) 4:17–30

2002), and Lawrence Berkeley National Laboratoryhas studied EIS and system-specific performancemonitoring and diagnostics (Motegi and Piette 2003;Motegi et al. 2003; Piette et al. 2001a, b). Finally, asubstantial body of work is dedicated to facilities’ andservice providers’ use of building automation sys-tems, or energy management and control systems.However, it tends to focus on questions concerningthe use of HVAC data for applications external to theEMCS, and on HVAC performance diagnostics andfault detection (Friedman and Piette 2001; Heinemeier1994; Webster 2005).

Case studies documenting EIS use are modest andcommonly comprise vendor-authored publicationsthat are written to publicize successful implementa-tion of a specific EIS, and to emphasize positiveaspects of the technology that the vendor wishes toadvertise. Case studies from the building energyanalysis community are more objective in theirassessments, and more varied in the content and levelof detail. Integration and installation, and the use ofEMCS data are more frequently asked about (Capehartand Capehart 2005; Capehart et al. 2007; Webster2005) than the relationship between software features,actions taken, and resulting energy savings (Motegi etal. 2003). When features actions and savings areaddressed in the literature, the overall topic is usuallynot whole-building EIS diagnostics, but rather equip-ment and system level operational diagnostics.

This paper begins with a description of themethodology used to conduct the case studies,followed by specific findings for each of the fourcases. The final two sections of the paper arededicated to research conclusions and directions forfuture work. Case study sites were identified throughlocal contacts and interviews with EIS vendors, andall reported findings are based on informationsupplied by the case study organizations at the timethat the research was conducted. Moreover, as theintent of this research is to evaluate the use of EIStechnology, inclusion in the paper does not implyendorsement of the EIS or of the participatingorganizations.

Case study methodology

Wal-Mart Stores, Inc., Sysco Corporation, the Uni-versity of California, Berkeley, the University of

California, Merced were selected for case study basedon the following criteria: users with a high level ofengagement with energy data and a role in energymanagement, aggressive savings or high-efficiencyperformance, and willingness to participate in 3-4 h ofinterviews and site visits, location permitting. Al-though each case was unique, a common set ofinformation was sought from each, including thepurpose and frequency of typical EIS use; how EISinformation is shared throughout the organization, oracross different users; challenges in implementing orlearning the technology; and quantified energy sav-ings or operational changes that resulted from use ofthe technology. The UC Merced case study wasconducted with the campus energy manager, and theUC Berkeley case included the associate director ofsustainability and engineering services, and membersof her staff groups. The Sysco case study wasinformed by the energy services provider and theemployee who is accountable for energy performanceat a Northern California warehouse site. The Wal-Mart case combined discussions with a benchmarkinganalyst from the Energy Department, the ElectricalEngineering Manager from Prototypical Design/Construction Standards, and the Senior Manger ofEnergy Systems and Technology Development.

As summarized in Table 1, the four cases that werechosen represent commercial enterprises and univer-sity campuses with a diversity of performancemonitoring technologies, commercial building types,and portfolio sizes. These cases encompass buildingsthat range from Wal-Mart and Sysco’s relativelyrepeatable warehouse and retail designs, to UCBerkeley’s legacy and historic sites, to UC Merced’svery-low energy new construction (New Buildingsand with the California Institute for Energy andEnvironment 2009). In the following, sections threethrough six, case-specific findings are presented.

Sysco (Sygma)

Between 2005 and 2006 Sysco Corporation, NorthAmerica’s largest foodservices distributor, imple-mented a 3-year corporate-wide energy efficiencyprogram with a target of 25% reduction in energyconsumption across a portfolio of 143 distributioncenters in the US and Canada. Sysco has a long-standing energy services contract with a large

Energy Efficiency (2011) 4:17–30 19

engineering firm, with whom a three-part approachwas adopted to identify and implement energy-savingchanges. First, site visits by expert refrigerationengineers and technicians were conducted for tune-ups and to identify low/no cost energy-savingmeasures. Second, the EIS was customized toaccommodate Sysco’s goals; and third, continuouscommunication and collaboration was supportedbetween corporate managers, the engineering firm,and the on-site ‘energy champions’ at each ware-house. This approach has enabled Sysco to outpace itsgoal, reaching 28% reductions before the end of theprogram period (10% from capital improvements andupgrades), with monthly savings of roughly18,000,000 kWh. While the UC Merced caserevolved around the particular constraints and powerof a densely populated, sophisticated web-EMCSplatform, the Sysco case highlights: (1) classicenterprise-wide EIS use and information sharing, (2)limited yet powerful on-site use of the EIS, (3) use ofEIS technology to ensure persistence in savings andenergy accountability.

Sysco EIS uses and challenges

Sysco uses NorthWrite’s Energy WorkSite EIS. Pulseoutputs from monitored electric utility meters arecommunicated via cellular to a central database.Sysco performs both site-specific and portfolio anal-yses on a monthly basis, using the EIS embeddedreporting capabilities. The engineering firm inputsutility billing invoices into the database and portfoliobenchmark rankings are generated as listed in Table 2.Managers coordinate monthly group reviews witheach site’s ‘energy champion’, who is accountable forenergy use. Monthly rankings are compared based ona metric called the efficiency factor, which takes into

account wet bulb temperature, the total volume offrozen and refrigerated space, total and daily energyconsumption, and weather predicted energy perfor-mance. Although not deeply understood by energychampions and managers, the efficiency factor is thebasis of site benchmarking. It is a custom definedmetric developed for Sysco’s portfolio of refrigeratedwarehouses, and preconfigured within the EIS report-ing tools. Sysco also tracks kWh/sf relative to the startof the efficiency program.

The Sysco site visit was conducted at the Stockton,CA Sygma distribution center. Stockton ranks highlyin the Sysco portfolio, and has reduced site energy useby 36% since the start of the efficiency program in2005. The energy champion makes near-exclusive useof the ‘meter monitor’ to manage his most energyintensive space. As shown in Fig. 3, this viewcontains a two-point overlay comparing the currentweek’s or day’s time series to that of the prior week,and summary of relative consumption and ambienttemperatures. The Stockton energy champion usesthis information display to implement a daily energyefficiency strategy. The existing controls do notpermit it; however, the frozen goods can toleratefluctuations in temperature between −5 and 10°F forshort periods of time, without compromising quality.In response, refrigeration setpoints are manuallyraised each morning to force the compressors to shutdown for several hours, thereby saving energy. Thelighter-colored trend in Fig. 2 reflects an instance ofthis daily strategy, whereas the darker line reflects aday in which the energy champion was on vacationand the strategy was not implemented. In thisinstance, in spite of a temperature increase, theaverage load was reduced by approximately 35%throughout the morning, relative to a day in whichstrategy was not implemented.

Table 1 Case study sites and characteristics

Case Type (m2) Controls Performance monitoring

Sysco, Stockton CASygma site

Refrigerated/dry warehouse(4.8, Stockton 8,800)

DOS-based refrigeration control NorthWrite Energy WorkSiteUtility bills

Wal-Mart, USA Retail/grocery (63 M) Novar Danfoss Emerson ComputerProcess Controls

Energy ICT EIServer Utilitybills

UC Merced, MercedCA

Campus (83,000) Automated Logic Corporation (ALC)WebCTRL

ALC WebCTRL Utility bills

UC Berkeley, BerkeleyCA

Campus (1.5 M) Barrington Some ALC, Siemens Obvius Utility bills

20 Energy Efficiency (2011) 4:17–30

Though the meter dedicated to refrigeration loadsis the primary focus of EIS use, minor energymanagement tasks are performed with all meters.Unanticipated and unexplained spikes in consumptionare identified and reviewed with equipment techni-cians, and deviations from expected profiles areinvestigated and addressed. Sysco monitors electricutility meters and has not pursued submetering, withthe result that aside from infrequent minor glitches incellular communications, data quality has not been acritical challenge.

The Stockton site visit revealed that much of theEIS functionality was unused and unexplored. Thehighly customized implementation of the EIS config-ured to meet Sysco’s needs, and the collaboration

with expert engineering service providers has resultedin a successful efficiency program based on straight-forward information, and frequent energy-based com-munication throughout the organization. Successfulmeasures implemented during the initial stages of theprogram, accountability based on monthly reporting,and an emergent corporate culture of competition,have provided significant value. At the Stockton sitethe EIS is most highly valued for its role in supportingand encouraging accountability and motivating staffso that efficiency gains might persist over time.However it is possible that additional energy savingshave gone unidentified because energy championshave not seen the value in the full set of EIScapabilities.

Fig. 2 Energy WorkSite’s ‘meter monitor’ and Stockton Sygma efficiency strategy

Table 2 Sysco metrics, benchmarks, and data sources

EIS Performance metrics Data sources Benchmarks

NorthWrite Unit-less efficiency factor Electric utility meter Portfolio rankings based on efficiency factor

Energy WorkSite Annual, monthly kWh/ksf Utility invoices Pre-program consumption [kWh/sf]

Weather feed

Energy Efficiency (2011) 4:17–30 21

Wal-Mart

Wal-Mart maintains a portfolio of 67 million ft2 ofcommercial retail space in the US, and uses EnergyICT’s EIServer to collect and monitor energy con-sumption data. Wal-Mart’s decision to implement anEIS was motivated by an overarching businessphilosophy that holds that with billion dollar utilityexpenses, energy information limited to 60- or 90-daybilling cycles is insufficient to manage energy use.Wal-Mart’s Energy Systems and Technology Devel-opment manager and building design engineersanalogize that they would never base retail decisionson 60-day-old sales data, and that energy consider-ations are just as critical.

The central themes highlighted in the Wal-Mart casecontrast markedly to those at Sysco. Rather thanintegrated EIS use throughout the enterprise to meetportfolio goals, as at Sysco, Wal-Mart is a case of EISuse by specific groups or individuals for a few keypurposes, among various departments and teams in theenterprise. Benchmarking, measurement and verification(M&V), and power procurement and demand responseare the most common uses. A group of internalsupporters champion the use of the EIS technology andmaintain a vision for how its use might be expandedthroughout the organization; regular operational analyticsare not yet widespread vertically or horizontally withinthe enterprise. In addition, Wal-Mart illustrates the factthat even the more sophisticated EIS may not satisfy allof an organization’s analytical and energy performancemonitoring needs. For uses such as M&V EIServer’sembedded functionality is well suited to user needs,while for others such as portfolio benchmarking, the EISdata is exported to third party software for analysis.

Wal-Mart EIS uses and challenges

Wal-Mart’s EIS data come from independent metersthat ‘stand alone’ from the building managementsystems. HVAC, lighting, and refrigeration electricmains are typically metered, and some stores monitorgas and water as well. Data from a subscriptionweather feed is imported into the EIS on a real-timebasis. Store and portfolio performance metrics aresummarized in Table 3.

EIServer features a custom module for M&V tasksthat has been used extensively at Wal-Mart, althoughon an ad hoc basis, to determine the effectiveness of

energy efficiency improvements. Regression analysesestablish weather-normalized baseline forecasts againstwhich actual measured consumption data are com-pared. The wholesale power procurement and demandresponse group makes considerable use of EIServer’sforecasting and normalization features, with experienceindicating that the technology is sufficiently accuratefor week-ahead predictions, and accurate to within towithin 1% for hourly time intervals.

At the individual store level, the EIS is used togauge the performance of new designs, particularly at‘High Efficiency’ supercenters that target 20-45%savings versus the typical Wal-Mart store1. However,one user does report that High Efficiency stores arebest analyzed by exporting EIS data for use inIntegrated Environmental Solutions Virtual Environ-ment models, because of the ability to run computa-tional fluid dynamics, solar thermal, and daylightingsimulation modules. Wal-Mart applies a custommodel-based approach to calculate weather and salesnormalized energy use intensities, therefore EIS dataare exported to external software for portfolio energytracking. Each month, the benchmarking analystidentifies the 20 poorest performing sites, and refersthem for further investigation at the operations andmaintenance level. In some cases, the benchmarkinganalyst explores the data for an individual store.However, she does not rely upon the EIS normaliza-tion capability, preferring to control comparisons byselecting stores from similar climates.

M&V and benchmarking activities provide twoexamples of energy savings attributable to Wal-Mart’suse of the EIS. At one site, poor store performancewas traced back to a 225 kW static lighting loadattributable to a failed dimming control module.Figure 3 shows the lighting submeter trend used toidentify this failure. As an example of M&V uses, theEIS Project Tracker module has been used to identifyseveral sites in which a failed or incorrectly installedVFD prevented actual energy savings.

Wal-Mart’s EIS challenges are largely independentof the EIS technology itself. Submetering has beendifficult because it has not been financially feasible tometer each store to the degree desired by thecorporation’s internal EIS champions. Given that theaverage supercenter contains a dozen submeters,

1 Wal-Mart Stores, Inc.–Sustainable Buildings. http://walmartstores.com/Sustainability/9124.aspx. Accessed January 2010.

22 Energy Efficiency (2011) 4:17–30

consistency in the quality of contracted installationshas also been a concern. More central to understand-ing real-world EIS use, Wal-Mart has faced difficultyintegrating regular EIS use into standard dailyactivities, particularly during the current economicdownturn. For example, supporters of the EIStechnology would like to see at a minimum, that allstaff has access to the system through web-basedexecutive reporting. Similarly, one person currentlyperforms benchmarking tasks every 30 days, whereasthe vision is to support a benchmarking group thatwould engage with the data on a daily basis.

UC Merced

Having opened in 2005, UC Merced (UCM) is thenewest University of California campus. Prior toopening, the campus made a strong commitment to

energy-efficient building design, and energy conser-vation plays a fundamental role in campus objectives(New Buildings and with the California Institute forEnergy and Environment 2009). UC Merced featuresa uniquely dense metering and monitoring infrastruc-ture, based on the campus web-EMCS. A web-EMCSis defined as an Internet-accessible system used tooperate building systems (usually HVAC) through aseries of sensors and controllers (Motegi and Piette2003). Over 10,000 control and energy monitoringpoints are currently trended at UCM’s three academicbuildings, the central plant, and smaller auxiliarybuildings. Custom benchmarks and deep monitoringcapability with a sophisticated web-EMCS are centralto the main themes embodied at the campus: (1) theuse of a web-EMCS as an EIS and (2) the web-EMCSas enabling critical information links and realizationof the campus as a living laboratory for energyefficiency analyses.

Fig. 3 Non-functional dimming module at Wal-Mart identified with EIServer, and then fixed

Table 3 Wal-Mart metrics, benchmarks, and data sources

EIS Performance metrics Data sources Benchmarks

Energy ICT Weather and sales normalized kWh/sf Building and submetered electric Portfolio rankings

EIServer Retrofit energy savings Some gas and water Pre-retrofit baseline

DR power and cost savings Subscription weather feed DR load shapes and baselines

ICT project tracking

Energy Efficiency (2011) 4:17–30 23

UC Merced web-EMCS uses

Typically, use of the WebCTRL web-EMCS at UCMerced is dominated by operational investigations.Use of the web-EMCS for traditional EIS analysesdirected at campus and whole-building energy perfor-mance has been complicated by the fact that theEMCS and monitoring instrumentation were notexplicitly commissioned or configured to track keyenergy performance metrics. The metrics used to trackenergy performance are more complicated thansimple energy use intensities (Brown 2002), and thelogic-based arithmetic in WebCTRL was not initiallyprogrammed to perform the associated calculations.WebCTRL meter data are used to track annual energyperformance; 15-min gas, electricity, hot water, andchilled water consumption are measured at thecampus level and for critical buildings, as summa-rized in Table 4. On a monthly basis, the campusenergy manager uses the web-EMCS data to deter-mine utility recharges for non-state buildings (that is,buildings that are differently financed, requiring thatthey “reimburse” the campus for utilities).

Lessons from analysis of the campus steam systemprovide an example of how the web-EMCS was usedto identify operational changes leading to energysavings. Gas trends at the central steam plant showedsignificant gas use throughout the night when thesystem was not intended to operate. At the same time,steam trends at the central plant revealed non-zerooperating pressures at night. The energy managershared the data with the superintendent, who subse-quently modified the system to true zero overnightpressure securing a 30% reduction in average dailygas consumption at the steam plant, and an estimated

$2,500 monthly savings (308 MBtu/month). In thelower portion of Fig. 4, the change to zero overnightpressure is plotted; in the upper portion, the resultingdrop in overnight gas use is shown.

In addition to providing a rich set of operationaland energy consumption data the web-EMCS has alsofacilitated the use of the campus as a livinglaboratory. To this end the web-EMCS data are usedin several ways. For example the data are used inthermodynamics course modules that the energymanager teaches. The data are also used to informstudent and faculty research efforts, and in short andlong-term research and demonstration collaborationswith the external buildings research community.

UC Merced challenges and needs

Data quality issues arise in a number of contexts atUC Merced. Networking and connectivity problemshave led to mis-communicated values that generateerrors, lock out equipment, and cause large volumesof false data and cascading false alarms. While notattributable to the capabilities of the EMCS, meteringand sensor calibration and configuration errors greatlyimpact data quality, thereby impacting the ability touse the EMCS as an EIS. With the exception of wholebuilding electric data, significant resources wererequired to manually validate the EMCS data and toquantify energy performance relative to benchmarks.

At UC Merced, the energy manager has not beenable to investigate whole-building and submeteringend-use trends to the full extent desired; and campuswide, it has taken some effort to transition fromreactive to proactive use of the data. Note that priorresearch in the use of building management systems

Table 4 UC Merced metrics, benchmarks, and data sources

Web-EMCS Annual building metrics Data sources Benchmarks

Automated Peak electric demand Building electric meter UC/CSU weather andbuilding-type normalizedenergy use intensity(Brown 2002)

Logic Total electric use Central plant electric submeters

Corporation Peak chilled water demand Central plant gas meters

WebCTRL Total gas use (incl. steam and HW) Building gas meters

Building chilled/hot water flow,and supply/return temperature

Utility bilss

24 Energy Efficiency (2011) 4:17–30

at government buildings identified similar challengesin proactivity, resources, and energy management(Webster 2005). Overall, the energy manager reportsa high level of satisfaction with WebCTRL empha-sizing that UC Merced trends extremely large vol-umes of data and that intensive monitoring needs tobe undertaken deliberately, with close attention to aspectrum of issues including wiring, system program-ming, network architecture, and hardware selection.

UC Berkeley

The University of California, Berkeley (UCB) is a140-year-old, 15.9 million ft2 campus with a widediversity of building ages, types, and sizes. Campusenergy performance has been prioritized to differingdegrees throughout the last decade, and Berkeley iscurrently experiencing a period of renewed attentionto efficiency. There is no central EIS at UCB. UCB isa contrasting case relative to the other three that isincluded to illustrate the challenges that are encoun-tered in the absence of campus-wide performance

monitoring. It also provides insights as to theinformation needs and energy management desiresof a specific energy manager, when a large ageingcampus is tasked with reducing its climate impact.Although there is no campus EIS, there is a largevolume of energy and system performance data.

UC Berkeley data uses and needs

The utility group uses utility bills and monthlymanual meter reads to manage the purchase andbilling of all campus energy. They process allinvoices, and perform accounting reviews for approx-imately 200 utility accounts, including water, electric,gas, and steam. The EMCS group at UC Berkeleyuses Broadwin’s WebAccess Project Manager toremotely oversee the campus control systems. Eachday eight person hours are dedicated to building-by-building HVAC equipment checks across 61 build-ings. Appropriate on/off status and setpoints areverified, and when problems are detected, the staffdelves further into time-series plots of relevant trenddata.

Fig. 4 Web-EMCS trends and energy saving operational change at UC Merced

Energy Efficiency (2011) 4:17–30 25

A number of campus efficiency and commission-ing interventions have implemented remotely acces-sible electric interval metering at approximately 20buildings. A 15-min interval data can be visualized,plotted, or exported using a web application. Apotentially useful tool is under development in astudent-funded research project. The Building EnergyDashboard includes monthly representations of ener-gy, water, and steam, and real-time displays of theremote access building electric meters. Figure 5shows the ‘detailed building plot’ in which thisweek’s consumption is plotted against the previousweek, with minimum maximum and average de-mand2. Student trials using such plots revealedexcessive operation of the ventilation system andover illumination in the architecture building. Basedon these observations, the ventilation schedule wasreduced by 6 h per day, and a lighting de-lamping wasconducted, resulting in a 30% reduction in totalenergy use. The two trends in Fig. 5 show thewhole-building power before and after these changeswere made. As the dashboard combines data from theutilities group with interval data that is currently usedonly on a limited basis, the application might beuseful for the campus energy management team aswell as for building occupants.

Similar to the other cases in the study, resources werecited as a challenge at UC Berkeley. In particular, theenergy manager prioritizes tracking performance at thebuilding level, and providing feedback to buildingcoordinators, EMCS and HVAC staff, and technicians.She also would like more remote-access intervalmetering, submetering beyond the whole-building level,and access-controlled public data.

Discussion

The case studies that were presented attempted toanswer questions related to energy savings and actionsattributable to EIS use, performance monitoring chal-lenges, and successful implementation models. As theassociated conclusions overlap considerably, they aregrouped into energy savings, organizational impacts andsuccess factors, and usability and analysis. Table 5summarizes actions that were taken based on building

energy data in each of the cases studied, and whereavailable, the associated energy impacts. The mostcommon actions and observations that were encoun-tered concerned incorrect implementation of scheduledloads, M&V, and inefficient or excessive operations.

Organizational impacts and success factors

The existence of data or performance monitoringsoftware does not guarantee shared knowledge oractionable information. Enterprise-wide EIS use atSysco has encouraged persistent savings, and acorporate culture of energy accountability, awareness,and competition. Similarly, extensive use and sharingof energy data at UC Merced has contributed tohighly efficient operations and energy performance,and has supported the realization of the livinglaboratory concept. On the other hand, Wal-Martand UC Berkeley are working toward more extensiveuse of data to reduce energy consumption.

Resources and staffing were a significant constraintin every case studied, and clearly impact the extent towhich energy data is successfully used to identifyenergy saving opportunities. A common view is thatEIS are primarily the domain of in-house staff andthat services are used to a minimal degree servicesduring installation and configuration. At the alternateend of the spectrum, EIS may be primarily intendedfor use by third-party energy consultants and serviceproviders. However, the general prevalence of staffingconstraints, Sysco’s successful efficiency gains, andthe number of EIS vendors that offer analyticalservices, indicate the potential for alternate modelsof successful EIS use.

Usability and analysis

Reliable high quality data are a critical aspect ofautomated analysis of building energy performance,and can have a significant impact on EIS usability. AtUC Merced, failure to commission the instrumenta-tion and web-EMCS for EIS analytics has impactedthe ability to track and diagnose building perfor-mance. More generally, usability at UC Merced isaffected by a number of challenges specific toimplementation of an intensive monitoring infrastruc-ture and the acquisition and storage of extremevolumes of trend data. The Merced case shows thatparticular attention must be paid to wiring and

2 Berkeley dashboardTM; analyze the campus metabolism.http://dashboard.berkeley.edu/building/ Accessed January 2010.

26 Energy Efficiency (2011) 4:17–30

Table 5 Summary of actions taken based on building energy information

Site Observation/action EIS data points Energy impact

UC Merced Excessive overnight gas use due tonon-zero pressure at steam boilers

Steam plant pressure, gas 30% reduction in average daily gasuse, $2,500/mo (308 MBtu/mo)avoided

Sysco Lights left on after hours atStockton Sygma

Building electricity

Sysco Multi-hour daily energy efficiencystrategy at Stockton Sygma

Building electricity, controlsystem setpointsand temperatures

35% demand reduction*Single observation

Sysco Identification of low/no cost savingsopportunities, e.g., retro-commissioningand refrigeration tune-ups

Warehouse electric meters 18% reduction in portfolio energy use36% reduction in Stockton site energy

UC Berkeley Excessive ventilation and over illuminationidentified, leading to lighting retrofit andventilation schedule change

Whole-buildingelectric meter

30% reduction in whole buildingenergy use

UC Berkeley Multi-week chiller lockout that preventedshut-down

Control system setpoints

Wal-Mart Static 225 kW load at dimming controlsubmeter

Submeter electricity $35,000/year avoided costs

Wal-Mart Failed or disconnected VFDs usedin retrofit programs

Submeter electricity Avoided zero savings at retrofit sites

Fig. 5 UC Berkeley Building Energy Dashboard, ‘detailed building plot’

Energy Efficiency (2011) 4:17–30 27

hardware integration, system programming, and net-work communications, not all of which lies wholly inthe domain of the EMCS developer. In contrast, thesystems used by Wal-Mart and Sysco are explicitlydesigned for energy analyses, and did not reportsignificant data quality issues, likely for two reasons:EIServer has embedded validation estimation errorchecking routines and data quality is usually aconcern only in cases of submetering and energysources other than electric.

The degree to which a site uses embeddedanalytical capabilities depends on the particularperformance metrics and benchmarking data that areutilized. Our cases showed that the more tailor-madethe calculations, the more likely it is that the data willbe exported for analysis in third party modeling orcomputational software. These cases indicated thatsophisticated EIS normalization and forecasting meth-ods are not universally understood across users andtechnology champions. Even so, these methods arecommonly used to great success, in a ‘black-box’manner. Finally, although EIS offer a wide range offeatures, actual use of these features can be verylimited, and it is not clear that users are always awareof how to use the capabilities of the technology togenerate energy-saving information.

Summary and future research

These case investigations have generated usefulinsights as to the value of EIS, to be extended infuture research. Three areas of key interest are: (1)features and energy savings, (2) technology defi-nitions and scalability, and (3) successful use anddeployment models. Questions concerning the mostuseful features, potentially useful but underutilizedfeatures, and energy savings attributable to EIS usemerit further attention. For example, a moreextensive set of typical actions and associatedenergy savings, and documented records of buildingconsumption before and after EIS implementationwould enable stronger conclusions on the range ofexpected savings from EIS use. In addition, typicalEIS actions and associated features can be linked toa classification of standard EIS uses such as M&Vfor retrofit support, continuous building-level anom-aly detection, or GHG emissions reporting. Closelyrelated to features and usability, there is consider-

able analytical potential in linking EIS anomalydetection methods to physical models. Today’s EISalgorithms rely purely on empirical historic perfor-mance data to detect abnormal energy consumption.However, they do not provide a means to identifyexcessive energy consumption relative to the designintent, or to realize model-predictive controlstrategies.

From a technology standpoint, definitions andscalability require further study. The UC Mercedcase illustrated some of the challenges in using anEMCS as an EIS, indicating an outstanding re-search question—how might EMCS data acquisitionand commissioning be improved so that thetechnology can serve as a robust EIS? Scalabilityis a concern that may provide insights as to whereto draw the line between EIS and related technol-ogies. In the future, it will be necessary tounderstand the tradeoffs between diagnostic capa-bilities, trend volume and number of points moni-tored, and the resulting burden on the system’sunderlying hardware and communication networks.Also related to technology definition, standardizingthe format and structure of information at the datawarehouse would encourage the development ofadvanced anomaly detection capabilities, as couldthe development of features to configure exporteddata files into formats that can be used by modelingtools such as Energy Plus or DOE-2. Standardformatting of EIS data would also facilitate thetransfer of energy information from the building tooutside entities, supporting and aligning withcurrent developments in demand side management,and the smart grid.

Finally, there remains much to learn abouteffective EIS deployment and use models withinorganizations. The Sysco case reveals a potentiallypowerful approach in which in-house use andexpert services are combined. Additional researchis needed to better understand where this approachis most useful, and to determine alternate successmodels that are appropriate to a diversity oforganizational sizes, commercial segments, andbuilding ownership models.

Acknowledgment Vendor participation was critical to thesuccess of this study, and the authors wish to acknowledge theirgenerosity and willingness to be included in this work. Thiswork was supported by the California Energy Commission and

28 Energy Efficiency (2011) 4:17–30

the California Institute for Energy and Environment underContract No. MUC-08-04.

Disclaimer This document was prepared as an account ofwork sponsored by the US Government. While this document isbelieved to contain correct information, neither the USGovernment nor any agency thereof, nor The Regents of theUniversity of California, nor any of their employees, makes anywarranty, express or implied, or assumes any legal responsibil-ity for the accuracy, completeness, or usefulness of anyinformation, apparatus, product, or process disclosed, orrepresents that its use would not infringe privately ownedrights. Reference herein to any specific commercial product,process, or service by its trade name, trademark, manufacturer,or otherwise, does not necessarily constitute or imply itsendorsement, recommendation, or favoring by the US Govern-ment or any agency thereof, or The Regents of the Universityof California. The views and opinions of authors expressedherein do not necessarily state or reflect those of the USGovernment or any agency thereof or The Regents of theUniversity of California.

Open Access This article is distributed under the terms of theCreative Commons Attribution Noncommercial License whichpermits any noncommercial use, distribution, and reproductionin any medium, provided the original author(s) and source arecredited.

References

Brown, K. (2002). Setting enhanced performance targets fora new university campus: Benchmarks vs. energystandards as a reference? American Council for anEnergy-Efficient Economy Summer Study Proceedings,4, 29–40.

Brown, K., Anderson, M., & Harris, J. (2006). Howmonitoring-based commissioning contributes to energyefficiency for commercial buildings. American Councilfor an Energy-Efficient Economy Summer Study Proceed-ings, 3, 27–40.

California Energy Commission. (2002). Money & EnergySaving Resources from the California Energy Commission.Enhanced Automation Technical Options Guidebook.California Energy Commission publication #400-02-005F.http://www.energy.ca.gov/enhancedautomation/documents/400-02-005F_TECH_OPTIONS.PDF. Accessed February2010.

Capehart, B., & Capehart, L. (Eds.). (2005). Web based energyinformation and control systems: Case studies andapplications. Lilburn, GA: Fairmont Press, Inc.

Capehart, B., Capehart, L., Allen, P., & Green, D. (Eds.).(2007). Web based enterprise energy and building auto-mation systems. Lilburn, GA: Fairmont Press, Inc.

Darby, S. (2006). The effectiveness of feedback on energyconsumption: A review for DEFRA of the literature onmetering, billing, and direct displays. EnvironmentalChange Institute, University of Oxford. http://www.eci.

ox.ac.uk/research/energy/downloads/smart-metering-report.pdf. Accessed November 2009.

Diamond, R., Optiz, M., Hicks, T., Von Neida, B., & Herrera,S. (2006). Evaluating the energy performance of the firstgeneration of LEED-certified commercial buildings.American Council for an Energy-Efficient EconomySummer Study Proceedings, 3, 41–52.

Friedman, H., & Piette, M. A. (2001). Comparison of emergingdiagnostic tools for HVAC systems. Lawrence BerkeleyNational Laboratory. Paper LBNL-47698. http://reposito-ries.cdlib.org/lbnl/LBNL-47698/. Accessed November2009.

Granderson, J., Piette, M. A., Ghatikar, G., & Price, P. (2009).Preliminary findings from an analysis of building EnergyInformation System technologies. Proceedings of theNational Conference on Building Commissioning. Law-rence Berkley National Laboratory. Paper LBNL-2224E.http://eis.lbl.gov/pubs/lbnl-2224e.pdf. Accessed February2010.

Heinemeier, K. (1994). The use of energy management andcontrol systems to manage the energy performance ofcommercial buildings. Ph.D Thesis: Department of Archi-tecture, UC Berkeley, and Energy and EnvironmentDivision, Lawrence Berkeley National Laboratory. PaperLBL-36119.

Johnson, J. (2002). Is what they want what they get?Examining field evidence for links between design intentand as-built energy performance of commercial buildings.American Council for an Energy-Efficient EconomySummer Study Proceedings, 4, 161–170.

Mills, E., & Mathew, P. (2009). Monitoring-based commission-ing: Benchmarking analysis of 24 UC/CSU/IOU projects.Report Prepared for: California Energy CommissionPublic Interest Energy Research (PIER) TechnologyDemonstration Program. Lawrence Berkeley NationalLaboratory. Paper LBNL-1972E.

Mills, E., Bourassa, N., Piette, M. A., Friedman, H., Haasl, T.,Powell, T., et al. (2005). The cost-effectiveness ofcommissioning new and existing commercial buildings:Lessons from 224 buildings. Proceedings of the NationalConference on Building Commissioning. Lawrence Ber-keley National Laboratory Report No. 56637. http://eetd.lbl.gov/emills/PUBS/PDF/NCBC_Mills_6Apr05.pdf.Accessed January 2010.

Motegi, N., & Piette, M. A. (2003). Web-based energyinformation systems for large commercial buildings.Lawrence Berkeley National Laboratory. Paper LBNL-49977. http://repositories.cdlib.org/lbnl/LBNL-49977.Accessed November 2009.

Motegi, N., Piette, M. A., Kinney, S., & Dewey, J. (2003). Casestudies of energy information systems and related tech-nology: operational practices, costs, and benefits. Proceed-ings of the International Conference for EnhancedBuilding. http://txspace.tamu.edu/handle/1969.1/5195.Accessed September 2009.

New Buildings Institute. (2008). Energy performance ofLEED for new construction buildings. Prepared for theUS Green Building Council. http://www.newbuildings.org/sites/default/files/Energy_Performance_of_LEED-NC_Buildings-Final_3-4-08b.pdf. Accessed January2010.

Energy Efficiency (2011) 4:17–30 29

New Buildings Institute. (2009). Advanced metering andenergy information systems. Prepared for the US Environ-mental Protection Agency, Grant 83378201. http://www.newbuildings.org/sites/default/files/metering-EPA%2083378201.pdf. Accessed January 2010.

New Buildings Institute, with the California Institute for Energyand Environment. (2009). Measured Performance CaseStudy: Classroom and Office Building, UC Merced. http://www.newbuildings.org/sites/default/files/Case_Study_UCM-COB.pdf. Accessed January 2010.

Petersen, J., Shunturov, V., Janda, K., Platt, G., & Weinberger,K. (2007). Dormitory residents reduce electricity con-sumption when exposed to real-time visual feedback andincentives. International Journal of Sustainability inHigher Education, 8(1), 16–33.

Piette, M. A., Kinney, S., & Friedman, H. (2001a). EMCS andtime-series energy data analysis in a large government officebuilding. Lawrence Berkeley National Laboratory. PaperLBNL-47699. http://www.osti.gov/bridge/servlets/purl/787118-Qg4Jfr/native/787118.pdf. Accessed January 2010.

Piette, M. A., Kinney, S., & Haves, P. (2001b). Analysis of aninformation monitoring and diagnostic system to improvebuilding operations. Energy and Buildings, 33(8), 783–791.

Webster, T. (2005). Trends in Energy Management Technologies -Part 5: Effectiveness of Energy Management Systems:What the experts say and case studies reveal. LawrenceBerkeley National Laboratory. Paper LBNL-57772.http://repositories.cdlib.org/lbnl/LBNL-57772. AccessedNovember 2009.

30 Energy Efficiency (2011) 4:17–30