1、260 2010 ASHRAEABSTRACTThis is the second paper in a four-part series reporting onthe test and evaluation of typical carbon-dioxide sensors usedin building HVAC applications. Fifteen models of NDIRHVAC-grade CO2sensors were tested and evaluated to deter-mine the accuracy, linearity, repeatability, a
2、nd hysteresis ofeach sensor. This paper describes the performance of thesensors and provides a comparison with the manufacturersspecifications. The sensors were tested at 40% relative humid-ity, 73oF (22.8oC) temperature, 14.70 psia (101.35 kPa) pres-sure, and at five different CO2concentrations (40
3、0, 750, 1100,1450, and 1800 ppm). The test results showed a wide variationin sensor performance among the various manufacturers andin some cases a wide variation among sensors of the samemodel.In all, 45 sensors were evaluated: three from each of the15 models. Among the 15 models tested, eight model
4、s have asingle-lamp, single-wavelength configuration, four modelshave a dual-lamp, single-wavelength configuration, and threemodels have a single-lamp, dual-wavelength configuration.INTRODUCTIONThis is part two of a four-part series of papers reporting onthe test and evaluation of typical CO2sensors
5、 used in buildingHVAC systems. In this study, fifteen models of NDIR (non-dispersive infrared) HVAC-grade CO2sensors were tested andevaluated. In all, 45 sensors (three from each model) wereevaluated to determine the sensor accuracy, linearity, repeat-ability, and hysteresis. The results are compare
6、d with themanufacturers specifications. The experimental procedureused to test and evaluate the sensors was described in Part 1(Shrestha and Maxwell 2009) of this paper. Among the 15models tested, eight models have a single-lamp, single-wavelength configuration, four models have a dual-lamp,single-w
7、avelength configuration, and three models have asingle-lamp, dual-wavelength configuration. All single-lamp,single-wavelength sensors and one single-lamp, dual-wavelength sensor incorporate an “automatic baseline adjust-ment” algorithm in the sensors electronics package.This paper presents an overvi
8、ew of the past studiesperformed by researchers to evaluate the performance of CO2sensors used in HVAC application. Further, a brief discussionon CO2sensor specifications and experimental test procedures(detailed in Part 1 of this paper) is provided. In addition, thepaper presents test and evaluation
9、 results, including a compar-ison of the performance of various CO2sensors.PREVIOUS STUDIESIn the past, limited studies have been done to investigatethe performance of HVAC-grade CO2sensors using a con-trolled environment. Fahlen et al. (1992) evaluated the perfor-mance of two CO2sensors, one photo-
10、acoustic type and oneinfrared spectroscopy type, in lab tests and long term fieldtests. The lab tests included performance and environmentaltests. The authors conclude that the deviation between actualconcentration and the sensors reading are normally wellwithin 50 ppm at a concentration level of 10
11、00 ppm. How-ever, at a concentration of 2000 ppm the test results showed adeviation of up to 300 ppm. The output of one sensorincreased dramatically during environmental testing. Thissensor failed to return to its normal value.Fisk et al. (2006) conducted a pilot study that evaluatedthe in-situ accu
12、racy of 44 NDIR CO2sensors located in ninecommercial buildings. The evaluation was performed eitherAn Experimental Evaluation of HVAC-Grade Carbon-Dioxide Sensors Part 2: Performance Test ResultsSom S. Shrestha, PhD Gregory M. Maxwell, PhDStudent Member ASHRAE Member ASHRAESom S. Shrestha is a R the
13、refore, thisoutput provides a reference signal used for sensor compensa-tion. Filter 2 (Figure 1C) passes 4.26 m IR radiation which isused for the CO2concentration measurement.The configuration shown in Figure 1D uses a silicone-based electronically-tunable Fabry-Perot interferometer infront of the
14、detector. This solid-state device provides an elec-tronic method of switching between the band pass filters, thusallowing for sensor compensation using a single filter-detectorpackage.CO2SENSOR SPECIFICATIONSCO2sensor manufacturers provide detailed specificationsfor their products. This information
15、is available from thecompanys website and/or literature packaged with the prod-uct. Table 1 summarizes some of the product information forthe models evaluated in this study. (Specific product names arenot used in this paper, rather the sensors are referred to as S1through S15.) The table indicates t
16、he sensor configuration, themanufacturer-specified accuracy, linearity, and repeatability.The accuracy statements for some sensor models include thepressure and temperature conditions for which the accuracystatement applies. For other sensor models, these conditionsare not explicitly stated. Three m
17、anufacturers state the nonlin-earity of their sensors, and five manufacturers state the repeat-ability of their sensors. None of the manufacturers specifyhumidity levels as part of the accuracy statement nor explicitlystate the hysteresis of their sensors.Sensors S1 through S8 are single-lamp, singl
18、e-wavelengthsensor, sensors S9 through S12 are dual-lamp, single-wave-length sensors, and sensors S13 through S15 are single-lamp,dual-wavelength sensors. Of the three single-lamp, dual-wave-length sensors, sensor model S14 incorporates an “automatic1.The term “beam” is used by sensor manufacturers;
19、 how-ever, this leads to confusion on the technology employed inthe sensor configuration. A single lamp can produce mul-tiple “beams”, so when the term “dual beam” is used, it isnot clear if one or two lamps are employed.Figure 1 NDIR sensor configurations. 2010, American Society of Heating, Refrige
20、rating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2010, Vol. 116, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission. 262 ASHRAE T
21、ransactionsbaseline adjustment” algorithm. Many sensor models provideoptions for the output signal produced. In this study, all sensorswere tested using an output signal of 0 to 10 VDC. Sensormodel S11 only provides an output signal of 4 to 20 mA. Forthis sensor, a precision resistor was used to con
22、vert the signalinto VDC. Measurement range of all sensors is 0 to 2000 ppm.EXPERIMENTAL TEST PROCEDUREThe CO2sensors were tested using a test chamber specif-ically designed and fabricated for the performance evaluation.Technical details of the test chamber and instrumentation aredescribed in Part 1
23、(Shrestha and Maxwell 2009) of this paper.All tests were conducted while maintaining the chamber rel-ative humidity, temperature, and pressure at 40%, 73F(22.8C) and 14.70 psia (101.35 kPa), respectively. Estab-lished testing procedures, including requirements for steady-state conditions, are descri
24、bed in Part 1 (Shrestha and Maxwell2009). The accuracy and linearity of the sensors were evalu-ated using the first forward and reverse measurements at CO2concentrations of 400, 750, 1100, 1450, and 1800 ppm (cor-responding to data points 1 to 9 in Figure 2). Repeatability wasevaluated using the two
25、 forward measurements at CO2con-centrations of 750, 1100, and 1450 ppm (corresponding to datapoints 2 to 4 and 10 to 12 in Figure 2). Hysteresis was evalu-ated using the first forward measurement and the reverse mea-surement at CO2concentrations of 750, 1100, and 1450 ppm(corresponding to data point
26、s 2 to 4 and 6 to 8 in Figure 2).ACCURACY TEST RESULTSThe accuracy test results are presented in terms of thedeviation of the measured CO2concentration by a sensor fromthe actual CO2concentration in the test chamber (i.e., deviation= measured CO2concentration actual CO2concentration).Deviation is ca
27、lculated for each sensor at each test condition.The mean deviation for a given sensor at a given condition isthe average deviation of the first forward measurement and thereverse measurement. As mentioned above, three sensors fromeach of fifteen models were tested. The letters A, B and C areused to
28、distinguish between each sensor of a given model.Data plots are used to illustrate the accuracy of the CO2sensors while numerical results are provided in tables. Specif-ically, a plot that compares the mean deviation and actual CO2concentration for a single sensor model was created. Eachaccuracy plo
29、t also illustrates the manufacturer-specified accu-racy. The accuracy statement appears as a set of dotted lines onthe plot. For most sensor models, the accuracy band increasesat higher CO2levels; therefore, the dotted lines are seen todiverge as the CO2level increases.Before discussing the results
30、for each sensor, it should bepointed out that for some sensors the maximum output voltage(nominally 10 VDC) was reached when the sensors wereexposed to the higher level CO2concentration test conditions.This is seen in Figure 3 which shows the output voltage fromsensor S2B (model S2, sensor group B)
31、as the CO2con-centration in the test chamber increases. As seen in the fig-ure, when the CO2concentration reaches approximately1700 ppm, the output voltage of the sensor is 10 VDC. Sincethis is the maximum voltage produced by the sensor, furtherincreases in CO2concentration do not produce any change
32、 inthe output voltage. For sensor S2B, the output reading wouldindicate 2000 ppm for any CO2concentration in excess of1700 ppm. The sensor is said to be saturated. For all sensormodels tested, the operating range of each sensor is 0 to 2000ppm. Saturated data points were not considered for analysis.
33、Analysis of Deviations for Each Sensor ModelAnalysis of deviations for each sensor model is presentedin this section. All sensors were tested under “as received”conditions from the suppliers. No calibration was performedon the sensors prior to testing. Sensors were powered up andallowed to stabilize
34、 in the laboratory environment before theywere tested.Single-lamp, single-wavelength sensors: Figure 4 showsthe accuracy test results for the sensors that use the single-lamp, single-wavelength configuration. These sensors use amanufacturer-specific proprietary algorithm to periodicallyFigure 2 Exam
35、ple data set to illustrate numbering schemeused to identify data points. Figure 3 Saturated reading of S2B sensor at 1800 ppm. 2010, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2010, Vol. 116, Part 1. For personal
36、 use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission. ASHRAE Transactions 263adjust the baseline CO2 concentration according to the mini-mum CO2concentration observed over a time period. It isassum
37、ed that the minimum CO2concentration observed corre-sponds to the outdoor CO2concentration. A value of 400 ppmis commonly used as the value for the outdoor CO2concen-tration level to which the sensor output is set when the auto-matic baseline adjustment is made.Prior to placing the sensors in the te
38、st chamber for per-formance testing, all sensors were operating in the laboratoryenvironment where the average environmental conditions(relative humidity, temperature, and barometric pressure)were, 14.7%, 76F (24.2C), and 14.3 psia (98.6 kPa), re-spectively. The minimum CO2concentration in the labor
39、a-tory varies on a daily basis according to the variation in theambient CO2concentration. Figure 5 shows the CO2con-centration in the laboratory for two weeks prior to the per-formance test.Operating in the laboratory environment for an extendedperiod of time allowed the sensors with automatic basel
40、ineadjustment algorithm to “adjust” their baseline; therefore,when these sensors are placed under test conditions andcompared to “actual” CO2 concentration levels, there is somebias in the measured value reported by the sensor. Because thisbias is unknown, no compensation was made when showingthe te
41、st results. It is therefore up to the reader to interpret themeaningfulness of the comparison of the sensor accuracy tothe accuracy statement for each sensor that uses an “automaticbaseline adjustment” algorithm.Table 2 shows the numerical values of the deviations foreach sensor at each CO2level. Da
42、ta points corresponding tosaturated sensor output are highlighted in the table.Sensor S1The deviations of the measured CO2concentration fromthe actual CO2concentration for Model S1 sensors arepresented graphically in Figure 4A. The manufacturer-statedaccuracy for this sensor model is 1% of measureme
43、nt range+5% of measured value. The deviation of all three sensorsTable 1. Manufacturer-Specified Accuracy, Nonlinearity, and Repeatability for CO2SensorsSensor Model Configuration AccuracyManufacturer-SpecifiedNonlinearityManufacturer-Specified RepeatabilityS1 Single-lamp, single-wavelength1% of mea
44、surement range + 5%of measured valueNA NAS2 Single-lamp, single-wavelength 100 ppm or 7% whichever is greater NA NAS3 Single-lamp, single-wavelength50 ppm or + 3% of reading at 77F (25C) at standard pressure1500 ppm readings at standard pres-sure 760 mm Hg and 77F (25C)NA 8 ppmS10 Dual-lamp, single-
45、wavelength 50 ppm + 2% of measured value NA NAS11 Dual-lamp, single-wavelength 100 ppm + 3% of reading NA NAS12 Dual-lamp, single-wavelength (50 ppm + 2% of measure value) at 68F (20C)NA NAS13 Single-lamp, dual-wavelength (30 ppm + 2.0% of reading) at 68F (20C) 1% FS NAS14 Single-lamp, dual-waveleng
46、th50 ppm or 5% whichever is greater (7% for levels over 1500 ppm) at 60F to 90F (15C to 32C)NA NAS15 Single-lamp, dual-wavelength (30 ppm + 2% of reading) at 77F (25C) 1% FS NANotes: NA indicates that the information was not available in the manufacturers product literature. Full scale (FS) is 2000
47、ppm for all sensors. The nominal operating temperaturerange is 32 to 122F (0 to 50C) with 0% to 95% RH. 2010, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2010, Vol. 116, Part 1. For personal use only. Additional r
48、eproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission. 264 ASHRAE TransactionsFigure 4 Comparison of deviation from actual CO2 concentration of single-lamp, single-wavelength sensors.Figure 5 CO2concentration in the labor
49、atory for two weeks prior to commencing the performance test. 2010, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2010, Vol. 116, Part 1. For personal use only. Additional reproduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission. ASHRAE Transactions 265shifts upward when the actual CO2concentration wasincreased. However, the slope of sensors A and C are a littlehigher than
