1、2011 ASHRAE 755ABSTRACTThis paper provides an overview of the pilot phase of afield study to determine the feasibility of a method of directlymeasuring the waste of water and energy caused by currenthot-water distribution systems (HWDS) in California resi-dences using wireless sensor network technol
2、ogies. Theexperience gained in the pilot phase study using wirelesssensor networks demonstrates that it is clearly feasible to usethis technology for measuring water and gas flows andtemperatures.The goal was to demonstrate a method to reliably collectwater flow and temperature data from every indoo
3、r hot waterend-use point and at the water heater in one-second intervalswhen water was flowing. The overall success of the pilotphase study indicates that this technique can work. However,the pilot phase study did reveal shortcomings in many areas.The recommendations in this paper address those shor
4、tcom-ings and provide ways to improve the outcomes of any follow-on field study. The projects tasks were to test and evaluatethe proposed hardware, installation protocols, data collec-tion, and processing procedures. The techniques developed inthis project provide a way to accurately measure tempera
5、tureand flow of indoor water use events at one-second resolution.The technologies used in this pilot phase study are viable foruse in a larger field study to determine the energy and waterefficiency of hot-water distribution systems in Californiahomes. The lessons learned from this experience will i
6、mproveprocedures, programming, and wireless sensor networkspecifications. INTRODUCTIONCalifornias energy and water resources are both at apremium, and the states economic and environmental vitalitydepend on the efficient use of these resources. Heating wateris one of the most energy-consumptive acti
7、vities in a house-hold, accounting for about 40% of California residential natu-ral gas consumption. In terms of water use, water heatingsystem designs often require users to run the water for a timebefore it achieves the desired temperature, wasting water in theprocess. The purpose of this project
8、was to conduct a pilot study todetermine the feasibility of a method to directly measure thewaste of water and energy caused by current hot-water distri-bution systems in California residences using wireless sensornetwork technologies. Monitoring of hot-water end uses inresidential buildings has bee
9、n done before (Lowenstein andHiller 1996, 1998; Tiller and Henze 2004). However, thosestudies did not investigate the efficiency of the HWDS. Thisproject explored a methodology to determine the efficiency ofthe HWDS.This project sought to gain experience with the fieldmeasurement process and collect
10、 actual data to understand theautomated collection and processing protocol necessary tomeasure the waste of water and energy caused by current resi-dential hot-water distribution systems.Monitoring was successfully completed on three housesfor a total of 22 days. Activities included equipment instal
11、la-tion and monitoring at the water heater and hot-water end uses.APPROACHA wireless sensor network was developed to measureflow and temperature of water at the trunk (water heater) andPilot Phase of a Field Study to Determine Waste of Water and Energy in Residential Hot-Water Distribution SystemsJa
12、mes D. Lutz, PE Peter Biermayer Derek A. KingMember ASHRAE Associate Member ASHRAEJames D. Lutz is a research associate supervisor and Peter Biermayer is a staff research associate in the Environmental Energy TechnologiesDivision, Lawrence Berkeley National Laboratory, Berkeley, CA. Derek A. King is
13、 an engineer at 4D Imaging, Berkeley, California.LV-11-0092011. American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions, Volume 117, Part 1. For personal use only. Additional reproduction, distribution, or transmission in eit
14、her print or digital form is not permitted without ASHRAES prior written permission.756 ASHRAE Transactionstwigs (individual end-use points) of a residential HWDS. Datawere collected at one-second intervals while water was flow-ing. The points in the HWDS that were monitored were theinlet and outlet
15、 of the water heater and several hot-water enduses in single-family residences. The collected data wereminimally processed on site and sent to a central server by cellmodem for further processing. At the central data processingsite, the one-second interval field data were analyzed andaggregated into
16、 summary data about individual hot-waterdraws.All measurements were taken at points that are capable ofbeing isolated by shutoff valves, such as sink faucets andclothes washers. This strategy allowed relatively simpleinstallation. The appropriate shutoff valve was closed, thedownstream plumbing was
17、disconnected, the flowmeter andthermistor mountings were installed, and the plumbing wasreconnected. Similar flow and temperature monitoring equip-ment was applied to the gas line supplying the water heater tomeasure energy use by the water heater.The temperature and flow of water was measured into
18、andout of the water heater and at the clothes washer, dishwasher,showerhead, and kitchen sink faucet. The temperature andflow of gas to the water heater was also measured. Figure 1shows the points at which temperatures and flows weremeasured.Monitoring EquipmentWater flow was measured with an inline
19、 turbine meter.The flowmeter is manufactured for use in tankless water heat-ers. It is small enough to be installed without trouble in mostlocations. The standard pickup for the flowmeter is a Hall sensor,which detects a magnetic pulse each revolution from a magnetmounted on the turbine blades. Anot
20、her type of pickup, aWiegand sensor, was available from the manufacturer. Unlikea Hall sensor, this type of sensor does not need external power.The Wiegand sensor pickoff worked well with the equipmentused in this study. No modifications to the data acquisitionsoftware or hardware were necessary to
21、use the Wiegandsensor. Since energy management is an issue for the wirelesssensor network, the Wiegand sensors were used to conserveenergy and extend battery lifetime.The output of the flowmeters is rated at 515 pulses perliter (1950 pulses per gallon) at flow rates above 2 L/min(0.5 gpm), with a fu
22、ll scale linearity of 1% over a range from1.0 to 30 L/min (0.26 to 7.9 gpm) (SIKA USA, Inc. 2006).Since many of residential hot water flows at end uses are belowthe lower rate, the flowmeters were calibrated in the labora-tory. Below about 0.5 L/min (0.13 gpm) the flowmeter doesnot register any flow
23、. Figure 2 is a sample calibration curve forone of the flowmeters. At a 2 liter per minute flow rate, the0.002 liter per pulse (515 pulses per liter) is equivalent to 1%accuracy. Unfortunately, the accuracy decreases dramaticallyat lower flow rates. Temperature was measured with a thermistor probein
24、serted into the water flow. Thermistors are ceramic semicon-ductors whose resistance drops nonlinearly as temperaturesrise. The thermistors used for this project were rated at 10,000ohm () resistance at 25C (77F), with a tolerance of 0.2C(0C to 70C) (0.4F 32F to 158F) (QTI 2005). Althoughthe configu
25、ration of the probes and the thermistor specifica-tions were chosen to maximize temporal sensitivity, measure-ments were not taken during the project to determine the timeconstant (Linkous et al. 2007).For this project, temperature was determined by measur-ing the resistance of the thermistor and ap
26、plying the followingequation:whereT = temperature, Ka = 0.001116b = 0.0002330c = 0.0000003723d = 0.00000009906Rtherm = resistance of thermistor (Ohms)This raises the possibility of higher flows due to coinci-dent uses of hot water at different end-use points. Duringconcurrent hot-water draws at diff
27、erent end uses, the flow ofwater at the water heater will be the sum of these flows.Because of the potentially higher flow rates at the water heater,a different flowmeter was used at that location. The flowmeters used at the inlet and outlet at the waterheater were a rotor design with a single jet.
28、They had a ratedrange of 0.8 to 38 L/min (0.3 to 10 gpm). The accuracy of theseflowmeters was listed as 1% full scale by the manufacturer. Figure 1 Measurement points.1T- ab Rtherm()cRtherm()ln2dRtherm()ln3+ln+=2011 ASHRAE 757The output of these flowmeters was a square-wave pulsedvoltage, at 87.2 pu
29、lses per liter (330 pulses per gallon). Sincethe output of these flowmeters was also in pulses, the identicaldata acquisition system was used to read the output of thesemeters.An auxiliary gas meter was installed on the gas line to thewater heater to record gas use by the water heater. A magneticpul
30、se generator attached to the gas meter generated 17.7 pulsesper liter (500 pulses per cubic foot). Gas consumption duringgas draw events was recorded at one second intervals. Thepilot light to ignite the main burner of water heaters burnscontinuously. The gas use by the pilot light was recorded assi
31、ngle pulses at approximately 20-second intervals. For acomplete knowledge of the energy consumption of the waterheater, the heating value of the gas is needed. In this project theenergy use of the water heater was not calculated.In addition to collecting pulses from the gas flowmeter,the temperature
32、 of the surface of the gas pipe was also moni-tored. This temperature was used as a proxy for the ambient airtemperature near the water heater when gas was not flowing tothe burner. Ambient air temperature does not change veryrapidly. This was sufficient to get a good record of the ambienttemperatur
33、e.Wireless Data Acquisition SystemThe data acquired by the sensors was read and transmittedto a local on-site computer by a wireless sensor network. Theindividual units of the wireless sensor network, also calledmotes, did minimal data processing and sent the data to a basemote at an on-site dedicat
34、ed computer. More processing of thedata was done at the on-site computer. The data were sent viacell modem to a server at a central location once a day. A sche-matic of the data collection and transmission scheme is shownin Figure 3.The mote platforms used an embedded operating systemdeveloped for s
35、pecifically for wireless sensor networks (Levis2006). The radio signal from the motes followed the IEEE802.15.4 standard at 2.4 GHz (IEEE 2006).The circuit board for the motes was approximately 32 by65 mm (1.27 by 2.58 in.) with a USB port at one end. The motes assembled data in five-second interval
36、s intopackets during flow. During periods of no flow, packets weresent at approximately 20-minute intervals. The data packetscontain identification fields for the computer and the mote, atimestamp, five seconds of flow information, five seconds oftemperature, and some clock synchronization flags. Da
37、ta acquisition software using wireless sensor networkshas problems standard hard-wired data acquisition systems donot have. Among the problems faced in developing the soft-ware for the motes were: signal interference (making sure thatconcurrent signals from different motes did not result in dataloss
38、), time synchronization (making sure that all the moteswere registering the correct time and applying corrections tothe timestamp in the data packets if the mote was not timesynchronized), and node reboots (making sure the motesrebooted automatically if the software running on them froze).A base mot
39、e for the wireless sensor network wasattached to a USB port on the on-site computer. The on-sitecomputer was located near the water heater in all cases. Theon-site computer unloaded the data packets from the basemote and performed some preliminary data processing.Packets were sorted into files by mo
40、te and duplicate packetswere eliminated. Retroactive time synchronization wasapplied if necessary. Figure 2 Calibration curve for flowmeter 1A.758 ASHRAE TransactionsThe data were sent once per day via cell modem to a serverat the central site. The scheme to process the data and transmitit to the ce
41、ntral server included several automated scripts andbatch files.Monitored HousesMonitoring was done on four houses during the pilotperiod. The houses, all single-family detached homes builtfrom 1970s through 1990s, were of volunteers. The waterheater for each house is in the garage, as is standard fo
42、r homesof this vintage in California. All houses are slab-on-grade. Ashort description of each house, along with the number of resi-dents, is shown in Table 1.The data collection at the house in Moraga was notsuccessful. No transmission was received from the mote in theshower. The L-shape of the hou
43、se meant that transmissionfrom that mote would have gone through two layers of exteriorsiding. It is probable this prevented signal transmission. Therewere problems with getting data from other motes at this siteas well.Collected DataThe collected data included flow rates and temperaturesfrom hot-wa
44、ter end-use points at one-second intervals duringdraws. The hot-water use points monitored in the pilot studywere kitchen sinks, showers, dishwashers, and clothes wash-ers. Water flows into and out of the water heater weremeasured. Water temperature was measured at each end-usepoint and at the inlet
45、 and outlet of the water heater. Data wereonly collected for those periods when draws were occurring.Table 2 lists the points where measurements were taken inpilot phase houses and the type of sensor.Monitoring was successfully completed for a total of 22full days at three houses. Due to the develop
46、mental nature ofthe data acquisition system, data from some days were notusable. The successful monitoring days and dates are listed inTable 3.Data were collected for 9416 water draw events. Thesewater draw events include measurements of flow into and outof the water heater as well as at the end-use
47、 points. Because ofthe standing pilot light, the flowmeter on the gas line was send-ing data packets much more frequently. This resulted in77,260 additional data packets being sent that were not asso-ciated with water draws.Data ProcessingThe raw measurement data collected by the motes anduploaded t
48、o the server were processed sequentially by severalcomputer scripts. The processing was done for one day at atime. At this point, the data were in multiple separate files foreach mote. First, the data files sent over from each mote wereconcatenated into one data file per mote. Because of the uncer-t
49、ain nature of the wireless data transmission, occasionallyFigure 3 Data collection and transmission scheme.2011 ASHRAE 759multiple data packets would be received. Duplicate data pack-ets were removed at this stage.Next, the information from the data packets wasconverted into a simpler format for analysis. In this format,each record contained the data from one second: the time-stamp, one flow measurement, and one temperature measure-ment.After this, the flow data from the motes were translatedfrom pulses to volumes. This was based on the calibrationsthat were done in
