1、924 2009 ASHRAEABSTRACT The software CONTAM was used to create a calibrated multi-zone model to replicate in-field tracer gas decay measurements of a new 2-story, 2600 ft2(240 m2), single-family house in Sacramento, CA under different whole-house dilution ventilation scenarios. The model incorporate
2、d measured values of ventilation system airflow rate, building enclosure leakage, fan-forced mixing between floor levels, indoor and outdoor temperature, and wind speed and direc-tion. The enclosure leakage distribution was adjusted to tune the model to the measured tracer gas concentration data. Th
3、e calibrated model was then used to compare different ventila-tion systems under identical outdoor conditions over a one-day period. Zones that received more ventilation air had faster concentration decay rates compared to zones that received less ventilation air. Results showed that ventilation sys
4、tems that delivered air to all zones, either by a dedicated duct system or by incorporation of the central forced-air space conditioning system, had more uniform ventilation air distribution.INTRODUCTIONThis paper describes the creation of a calibrated computer model for residential ventilation syst
5、ems and the use of the calibrated model to extend the results obtained in previous field testing. The model calibration process used test data from tracer gas testing of residential ventilation systems in a new single-family house near Sacramento, California. Hendron (2007) detailed the tracer gas t
6、esting and conclusions. The work described in this paper was performed in order to eval-uate ventilation systems that were not present in the houses tested by Hendron and to provide the capability to extend the results of field testing in one location under one set of envi-ronmental conditions to ma
7、ny locations under many sets of environmental conditions.DESCRIPTION OF HOUSEThis work concentrates on one of the two houses tested by Hendron (2007). The house is two-story, approximately 2600 ft2(240 m2), with four bedrooms and three bathrooms. The first floor consists of one bedroom, one bathroom
8、, a laun-dry room, the living room area, and a kitchen and dining room. The second floor consists of the master bedroom and bath-room, two additional bedrooms, an additional bathroom, and a small common area at the top of the stairway which over-looks the living room below. Figure 1 contains a drawi
9、ng of the floor plan of the house.Figure 1 House floor plan.A Calibrated Multi-Zone Airflow Model for Extension of Ventilation System Tracer Gas TestingAaron Townsend, PE Armin Rudd Joseph Lstiburek, PhD, PEngAssociate Member ASHRAE Member ASHRAE Fellow ASHRAEAaron Townsend is an associate and Armin
10、 Rudd and Joseph Lstiburek are principals with Building Science Corporation, Somerville, MA.LO-09-088 2009, American Society of Heating, Refrigerating and Air-Conditioning Engineers, Inc. (www.ashrae.org). Published in ASHRAE Transactions 2009, vol. 115, part 2. For personal use only. Additional rep
11、roduction, distribution, or transmission in either print or digital form is not permitted without ASHRAEs prior written permission.ASHRAE Transactions 925DESCRIPTION OF PREVIOUS TRACER GAS TESTINGHendron (2007) describes the tracer gas testing in detail. In total, seventeen ventilation tests were pe
12、rformed on the house using tracer gas decay methods. Table 1 lists the tracer gas tests performed. For each test, the house was first brought to a well-mixed state with uniform tracer gas concentration in all zones by running the central air handler (AHU) and auxil-iary mixing fans. The test was ini
13、tiated by deactivating the mixing systems and activating the ventilation system as appro-priate for the test, and leaving the house in that state for a period of 2 to 14 hours. Three ventilation systems were tested. The first ventilation system tested was the central-fan-inte-grated supply (CFIS) ve
14、ntilation system, which consists of an outside-air duct to the return side of the AHU and a controller that operates the AHU on a minimum duty cycle. The outside-air duct contains a damper that remains closed except when the CFIS system is activated. The duty cycle of the AHU and CFIS system varied
15、from test to test. This ventilation system was operated at different ventilation rates using a variable-speed fan installed inline with the outside air duct, as described in Table 1. The second and third ventilation systems were upgraded exhaust fans located in the laundry room and master bedroom, r
16、espectively. The exhaust fans were tested only at 100% of the ASHRAE Standard 62.2-2003 (ASHRAE 2003) (referred to elsewhere in this paper simply as 62.2) ventilation rate, and were tested with and without simultaneous operation of the AHU for mixing. In addition to the ventilation tests, natural in
17、filtration and air handler bump (natural infiltration with the AHU running) tests were also conducted. During the tracer gas tests, the bedroom doors were either open or closed. The house was built with transfer grills, which are passive openings above the doorways that allow a return air path when
18、the bedroom doors are closed. The transfer grills were also either open or closed (taped over) during the tracer gas tests. The doors to the bathrooms and laundry room were always open. All exterior doors and windows were always closed.DESCRIPTION OF MODELING SOFTWARECONTAM is a multi-zone air flow
19、network modeling software developed by the National Institute of Standards and Technology (Walton 2005; Emmerich 2003; Emmerich 2001). It is commonly used in ventilation research to model build-ings, ventilation systems, and contaminants in indoor and outdoor air (Emmerich 1995; Persily 1998). In CO
20、NTAM, the user specifies attributes of the buildings zones, air flow path-ways between zones (such as leaks or fans and ducts), contam-inant sources and sinks, and other relevant inputs. The software performs the simulation and the results are available for visualization or export.TESTING OF SUBSTIT
21、UTE HOUSEAt the time of the work described in Hendron (2007), an enclosure air leakage test was performed with a blower door (ASTM 2003), but no further diagnostics were performed on the house enclosure or interior demising walls as further work was not planned at the time. Later, when the decision
22、was made to create a calibrated computer model, much more detailed information about the enclosure and interior airflow paths was needed in order to provide a reasonable starting point for the calibration process. The original house was no longer available for testing, so another house of the same f
23、loor plan was tested instead. While two houses of the same floor plan can certainly have different leakage characteristics, these Table 1. Tracer Gas TestsTest Number DescriptionCFIS Tests With Mixing (All have AHU 20 min off/10 min on)1Doors Closed, Transfer Grills Open, 95% of the 62.2 Ventilation
24、 Rate*2Doors Closed, Transfer Grills Open, 60% of the 62.2 Ventilation Rate3Doors Closed, Transfer Grills Open, 33% of the 62.2 Ventilation Rate4Doors Closed, Transfer Grills Closed, 60% of the 62.2 Ventilation RateLaundry Exhaust Tests With Mixing (All at 100% of the 62.2 ventilation rate)5Doors Cl
25、osed, Transfer Grills Open, AHU 20 min off/10 min on6Doors Closed, Transfer Grills Open, AHU 25 min off/5 min on7Doors Closed, Transfer Grills Closed, AHU 25 min off/5 min onLaundry Exhaust Tests Without Mixing (All at 100% of the 62.2 ventilation rate)8 Doors Open, Transfer Grills Open9 Doors Close
26、d, Transfer Grills Open10 Doors Closed, Transfer Grills ClosedMaster Bathroom Exhaust Tests With Mixing (All at 100% of the 62.2 ventilation rate)11Doors Closed, Transfer Grills Open, AHU 25 min off/5 min onMaster Bathroom Exhaust Tests Without Mixing (All at 100% of the 62.2 ventilation rate)12 Doo
27、rs Closed, Transfer Grills Open13 Doors Closed, Transfer Grills ClosedNatural Infiltration Tests (No ventilation or AHU operation)14 Doors Open, Transfer Grills OpenAir Handler Bump Tests (No ventilation, AHU on)15 Doors Open, Transfer Grills Open16 Doors Closed, Transfer Grills Open17 Doors Closed,
28、 Transfer Grills Closed*Test 1 was 95% instead of 100% of the 62.2 ventilation rate due to hardware limita-tions.926 ASHRAE Transactionstwo houses were built within a few months of each other, by the same builder and likely the same subcontractors, and the overall enclosure leakage testing results w
29、ere similar. The original house had a leakage rate of 1346 cfm (635 L/s) at 50 Pascals (0.2 in. of water) pressure difference across the enclo-sure (CFM50), and the substitute house had 1608 CFM50 (759 L/s at 50 Pa). The substitute house was slightly larger due to an option that added two additional
30、 bedrooms and an addi-tional bathroom; after subtracting the leakage in the additional bedrooms, the substitute house was 1411 CFM50 (666 L/s at 50 Pa). As the substitute house was simply a starting point for calibrating the model, differences between the houses were of minor consequence and were re
31、medied during the calibration process.Air leakage characterization on the substitute house was performed to quantify both house-to-exterior and room-to-room leakage characteristics. The testing also included tests of zone pressures and central forced-air system airflow to each room. The testing proc
32、edure was able to quantify the leakage characteristics of each room to the exterior and to neighboring zones, but no attempt was made to identify the specific loca-tions of leakage within each room. Further details of the test-ing at the substitute house are included in the appendix.MODELING PROCEDU
33、REThe goal of the modeling was to produce a set of inputs for the house enclosure and zone-to-zone leakage pathways that, when simulated with CONTAM, would produce the same results as the tracer gas tests when the ventilation systems were operated in the same manner as each of the tracer gas tests.A
34、s a starting condition, leakage values calculated from the leakage testing in the substitute house were used for the exterior enclosure and the interior partition walls. Because the actual leakage locations within each room were not deter-mined by the testing, leakage within each room was initially
35、distributed proportional to the wall and ceiling area. Wall leak-age was broken into leakage for each wall orientation and into five vertical locations on each wall, with equal vertical sepa-ration between the locations. Each leakage location on a wall had the same leakage coefficient and exponent.
36、Initial test runs with simplified models showed the vertical spacing chosen (5 leaks per wall, equally spaced on a 9 ft (2.7 m) wall) approx-imated diffuse wall leakage, while still maintaining a manage-able number of leakage elements in the model. The temperature in each room and the outdoor temper
37、ature and wind speed had been recorded during the tracer gas testing, and were used as inputs to the model. Wind direction was not recorded during the tracer gas testing, so meteorological data from the nearest airport (Auburn, CA, approximately 10 miles (16 km) away) was obtained and the wind direc
38、tion data was used as an input to the model. Drawings and specifications for the AHU and duct system were obtained from the subcontrac-tor, which were used to create a full duct and AHU model. The AHU and all ductwork in this house are located within condi-tioned space, greatly simplifying the need
39、to characterize duct leakage. For each test simulated, a schedule was created that controlled the ventilation systems, AHU operation, and trans-fer grill and bedroom door status to replicate operation as performed in the tracer gas tests. Results from the model were compared to the tracer gas data a
40、nd the leakage inputs were modified via trial-and-error to decrease the error between the model output and the tested data. No formal method was used to obtain a minimized error function, only visual comparison of the measured and simulated tracer gas decay curves, so there is no reason to assume th
41、at the final inputs represent a unique or optimized solution.During the initial comparisons of measured and simulated data, it became clear that the most difficult tests to replicate were the tests with large differences in tracer gas decay rates between the different rooms. Stated differently, it i
42、s easier to replicate the decay rate in a well-mixed house (which might be approximated as a single well-mixed zone) than it is to repli-cate the decay rates of six interconnected zones. Conse-quently, a single test was used for the calibration, and the remaining tests were used after the calibratio
43、n was complete in order to evaluate the results. The test used to calibrate the model was test 9, which utilized the continuously-operating laundry room exhaust fan as the ventilation system, did not have mixing via the AHU, and had the bedroom doors closed and the transfer grills open. Test 9 was s
44、elected as the calibra-tion test because it was a long test (14 hours) without mixing, there were substantial differences in the tracer gas concentra-tions, and it had an interesting change in tracer gas decay rate during the test due to a temperature change in two of the zones.Figure 2 shows the tr
45、acer gas decay curves from the initial simulation of test 9 using the enclosure leakage values calcu-lated from the substitute house. The results are not unreason-able, yet clearly there is room for improvement. Figure 3 shows the results of the final simulation of test 9, which were deemed to agree
46、 with the measured data sufficiently to cease further trial and revision of the model.STATISTICAL METHOD FOR EVALUATION OF MODELING RESULTSResults were evaluated statistically using ASTM D5157-97 Standard Guide for Statistical Evaluation of Indoor Air Quality Models (ASTM 2008). ASTM D5157 has three
47、 crite-ria relevant to evaluating the results of this work. The first criterion is that the data used for the evaluation should be inde-pendent from the data used to develop the model. All of the test results with the exception of test 9 meet this criterion, as they were not used to calibrate the mo
48、del. The second criterion consists of a set of quantitative parameters related to the agree-ment between the predicted (modeled) and observed (measured) data sets. These parameters are:1. Correlation coefficient, r, between the predicted and observed data sets. r ranges from -1 to +1, with -1 indi-c
49、ating an inverse relationship, 0 indicating no relation-ship, and +1 indicating a strong relationship between the ASHRAE Transactions 927two data sets. D5157 suggests that r values greater than 0.9 generally indicate adequate model performance with respect to correlation coefficient.2. Best-fit line of regression between the predicted and observed data sets. Regression of perfectly-matched data sets would have a slope, m, of 1.0 and an intercept, b, of 0.0. D5157 suggests that an a slope of 0.75 to 1.25 and an intercept less than 25% of the mean value of the observed data set (thus 0.9 0.7
