Documenti di Didattica
Documenti di Professioni
Documenti di Cultura
02001 IEEE
0-7803-6659-X/01/$10.00
252
application of a sensor, provide information that assists responding thermostat. It has the same sensor and housing as
building owners when decisions are made about installing the normal thermostat. Although the time constant for this
sensors, and provide incentives to sensor manufacturers to sluggish thermostat was not experimentally determined as for
offer sensors that meet the needs of the applications. the other thermostats, it is believed to have the largest time
constant. The experimental results from this study also verify
2. Experiment Description this assumed large time constant.
The effects of the thermostat time constant on the energy 2.3. Control
usage, temperature control, and occupant comfort are
examined in this study by comparing the control A proportional integral (PI) control algorithm that resides
characteristics of four thermostats that have different time in the local controller for the test room is used to control the
constants for an interior building zone with a VAV system. zone temperature. The PI algorithm compares the thermostat
The tests are executed at the Iowa Energy Center Energy reading with the zone temperature set point. If the thermostat
Resource Station (ERS) (Price and Smith, 2000). reading is lower than the zone temperature set point, then the
PI algorithm opens the HRH valve to increase the flow rate of
2.1. Building zone and W A C system the heated water through the HRH coil and thereby heat up
the discharge air temperature to the zone. If the thermostat
The building zone is an interior building zone and is reading is higher than the zone temperature set point, then the
shown in Fig. 1. The surrounding environment for this PI algorithm partially closes the HRH valve to reduce heating
interior building zone is conditioned to be 70°F. The of the cool air (60°F in this experiment) to lower the zone air
building zone is equipped with a commercial VAV unit that temperature. The EMCS that the local controller is connected
includes a hydronic reheat (HRH) coil. Outdoor and return to records the data with a time interval of 1 min.
air is conditioned first by an air handling unit (AHU) to 60°F
and then sent to the VAV unit located in the plenum of the 2.4. Experiments
test building zone. A boiler is employed to supply heated
water to the HRH coil. The HVAC system at the ERS is During the experiments, the four thermostats are used
controlled by a commercially available energy management individually to control the VAV unit to maintain a set of
control system (EMCS). The building zone air is controlled scheduled temperature set points (Table 1). The capability of
by a local controller that is part of the EMCS. the thermostats with different time constants to track the
A comfort sensor that measures surrounding air variation of the temperature set points determines the VAV
temperature, humidity and air velocity, is placed in the energy consumption and zone comfort. The experiments are
middle of the zone to examine the occupant comfort level. repeated to check for repeatability. The experiments last for
Because in this study, the entering air flow rate and humidity six days. During the first 7 hours, the room temperature is
do not vary, only the temperature reading from the comfort controlled by the fast thermostat to stabilize the HVAC
sensor is employed as an indication of occupant comfort. system. Then, the room temperature is controlled by the four
sensors one by one, with a sequence of fast, normal, sluggish,
2.2. Thermostat and commercial thermostat. Each thermostat controls the
room for 30 hrs. During the first 10 hrs of the 30 hrs, the
Four thermostats with different time constants are installed room temperature set point is 80°F; and the set point is
on a wall in the building zone that is under test (Fig. 2). The changed to be 7OoF for 5 hrs; this 15 hrs is called a cycle.
four thermostats include a bare platinum resistance Each thermostat controls the room temperature for two
temperature detector (RTD) sensor (fast thermostat) with a cycles. The VAV damper position is fixed to be 100% open
time constant around 2.5 min; a platinum RTD with housing resulting an air flow rate of 400 CFM entering the zone. The
(normal thermostat) whose time constant ranges from 14 to heated water temperature for the HRH is around 160°F.
20 min; a platinum RTD with housing and insulation
(sluggish thermostat) whose time constant is from 30 to 40 3. Results
min; and a commercial thermostat (commercial thermostat)
whose time constant is between 20 and 25 min. The housings The experiments were executed several times from March
for the thermostats in this test are commercially available. to August 2001 to examine the effects of repeatability,
Although only one brand of housing is used in this study (Fig. housing, and energy consumption. The effect of the housing
3), other commercially available housings have similar is discussed first in Section 3.1. The repeatability are
structures. A cardboard enclosure is installed to block the reported in Section 3.2. Sections 3.3 and 3.4 compare the
radiant exchange between the fast thermostat and other occupant comfort and energy consumption when different
surfaces in the zone (Fig. 4). The fast, normal, and thermostats are in control.
commercial thermostat time constants were experimentally
verified by suddenly immersing the thermostat in a near 3.1. Housing effect
constant temperature environment. The sluggish thermostat
(Fig. 3) is employed to demonstrate the effects of a very slow Before experiments were executed, it was anticipated that
253
the effect of the housing on h e thermostat reading at steady the energy consumptions are not the same from test to test,
state condition was small. It was assumed that, although the the trend that the faster a thermostat is the less energy it
different thermostats may read differently during the transient consumes is found for each test. It is concluded that the
period, the readings would be close to each other during , experiments are repeatable.
steady state conditions. The results from a test executed in
March 2001 (Fig. 5 ) show that the thermostat readings differ 3.3. Comfort
by 1 to 2°F when the zone air temperature is at 80°F and 0.5
to 1°F when the zone air temperature is at 70°F. In Fig. 5 , the From the comfort temperature readings in Figs. 5 , 6, and
x axis is the time axis with units of min and the y axis is the 8, it is observed that the comfort temperature readings are
thermostat reading with units of OF. very close to the fast thermostat readings. The more sluggish
Each thermostat sensor was calibrated on site after this a thennostat is, the lower the reading is at 80°F and the
test, except for the commercial sensor, which is factory higher the reading is at 70°F. Hence, when the more sluggish
calibrated, to eliminate any inaccuracy caused by the sensors thermostat is controlling the VAV unit and when it reaches
themselves. From the calibration, the difference between the steady state, the zone air temperature actually has a higher
fast, normal, and sluggish thermostat sensor readings at 70°F temperature at 80°F and a lower temperature at 70°F. For
is less than 0.05"F, and the difference at 80°F is less than example, when the sluggish thermostat is in control, the
0.1"F. But the test (Fig. 6, performed in July 2001) after the comfort temperature reaches 82°F in Fig. 5 and 81°F in Figs.
thermostats were calibrated still shows that the thermostat 6 and 8, while the zone temperature set point is 80°F. And
readings differ about 0.5 to 1°F when the zone air the comfort temperature reaches 69°F in Figs. 6 and 8 when
temperature is around 70 and 80"F, respectively. the zone temperature set point is 70°F. Also, a comparison of
A housing effect test was then executed. The housing the transient periods when different thermostats are in control
effect test has the same test conditions except that the indicates that the faster thermostat brings the comfort
housings for the normal, sluggish, and commercial temperature to the desired set point faster. This means that
thermostats were removed (Fig. 4). The results (Fig. 7) from the slow responding thermostat results in the occupant being
this housing effect test show that the thermostat readings uncomfortable during both the transient and steady state
overlap each other when the housings are absent at both 70 times.
and 80°F. This test verifies that it is the housing that causes A variable, Comf, that is used to calculate the difference
the thermostat readings to differ at steady state condition. between the comfort temperature and the set point over a
It has not been verified why the housing affects the period of time that a certain thermostat is in control, and to
thermostat steady-state readings. It is suspected that the quantitatively evaluate the comfort over the period time, is
housing conducts heat into the wall, which has a lower defined as
temperature than the zone air temperature because the zone ie
surrounding temperature is around 70°F. It is also observed comf = c(Tco,f(i)-Ts(i)2 At (1)
that a thicker housing causes larger steady-state errors. i=ib
where i is the ith time interval. At is the time interval which
3.2. Repeatability is 1 min in this study. ib and ie are the beginning and ending
times for a period time that a certain thermostat is in control.
Besides the experiments in March and July 2001, another Tcod and '& are the comfort and set point temperature.
experiment was performed in August 2001. The results from Table2 supplies Comf for each test. The comfort percentage
the experiment in August 2001 are shown in Fig. 8. In this (C ) to evaluate the difference between the Comf when
test, the normal thermostat did not read correctly. Therefore, ditferent thermostats are in control is calculated using
only three thermostat readings are shown. Although all
experiments, namely March, July, and August, were executed
under the same test conditions, the small variations of the
supply air temperature from the AHU, the heated water where Comf, and Comff are the Comf values for the
temperature that enters the HRH, the zone surrounding thermostat under evaluation and the fast thermostat. Table 2
temperature, etc., are expected to produce slight differences shows that more sbw responding thermostat has larger Comf.
between the tests. The maximum C in Table 2 even reaches 8O%, which
Figures 5 , 6, and 8 show that the experiments have very indicates that t i e sluggish thermostat causes longer
similar thermostat responses. The steady state differences uncomfortable periods compared with the fast thermostat.
among thermostats are also repeatable for the first and second
cycles for each experiment. Because the thermostat sensors 3.4. Energy Consumption
were calibrated between the March and the July tests, it is not
suitable to compare the steady state differences from the In this study, only the energy consumed by the VAV HRH
March test with the later tests. The July and August tests is accounted for. The total energy is calculated from
have similar steady state differences among the thermostats. ie
The energy consumption from these tests are discussed in Et = Z P Q C ( T ~ ( ~ ) - T M ( N A ~ (3)
Section 4. Although, because of the varied test conditions, i=ib
254
where p and c are the water density and specific heat at References
160°F. Tm and & are the entering and leaving water
temperatures from the HRH coil. Q is the HRH water flow [I] Ahmed, O., Mitchell, J. W., and Klein, S. A., “Experimental
rate. T, TM, and Q are all measured variables. Validation of Thermal and Pressure Models in a Laboratory
Table 2 summarizes the total energy ansumption over Simulator,” ASHRAE Transactions, Vol. 104, Pt. 2., pp. 983 - 998,
the period when a certain thermostat is in control for the 1998.
March, July, and August tests. The energy percentage (E& to [2] Athienitis, A. K., Stylianou, M., and Shou, J., “A Methodology
evaluate the difference between the energy consumed when for Building Thermal Dynamics Studies and Control Applications,”
different thermostats are in control is calculated using ASHRAE Transactions, Vol. 96, Pt. 2, pp. 839 - 848,1990.
[3] Henderson, H., I., “Simulating Combined Thermostat, Air
(4) Conditioner, and Building Performance in a House,” ASHRAE
Transactions, Vol. 98, Pt. 1, pp. 370 - 380, 1992.
where and q-f are the total energy for the thermostat [4] Kao, J. M., and Pierce, E. T., “Sensor Errors, their effects on
under evaluation and the fast thermostat. B shows that Building Energy Consumption”, ASHRAE Journal, Vol. 25, n 12,
although different tests have different energy consumption pp. 42 45,1983.
values, all tests indicate that when the fast thermostat is in [5] Kao, J. M., “Sensor Errors”, ASHRAE Journal, Vo1.27, n 1, pp.
control, the energy consumption is the least. The more 100- 104,1985.
sluggish the thermostat is, the more energy is consumed. The [6] Lamb, G., and Tree, D. R., “Seasonal Performance of Air
maximum difference between the sluggish and fast Conditioners - An Analysis of the DOE Test Procedures: The
thermostat is 7 % of the energy. Thermostat and Measurement Errors,” Energy Conservations, U. S.
Department of Energy, Division of Industrial Energy Conservation,
4. Conclusions Report No. 2, DOE/CS/23337-2, Jan., 1981
[7] McBride, M. F., “Measurement of Residential Thermostat
The objective of this study is to compare the temperature Dynamics for Predicting Transient Performance,” ASHRA E
control, occupant comfort, and energy usage for four Transactions, Vol. 85, Pt. 1, pp. 684 - 694, 1979.
thermostats with different time constants installed on a wall [8] Nelson, L. W., “Predicting Control Performance of Residential
in an actual building zone equipped with a commercial VAV Heating System with an Analog Computer,” IEEE Transactions of
unit with a HRH coil and a DDC system. This study was Industry ApplicationsIA -10(6), pp. 73 1 - 740, 1974.
performed using experiments on a VAV system controlled by [9] Nelson, L. W., “Comments in Response to Presentation of
four thermostats with different time constants. The Nguyen and Goldschmidt (1983): ASHRAE Transactions,Vol. 89,
experimental results verify that although the thermostat time Pt. 2A,pp. 373,1983.
constant does not affect the stability of the temperature [IO] Nguyen, H. V. and Goldschmidt, V., “Modeling of a
control, a sluggish thermostat causes higher energy Residential Thermostat and the Duty Cycle of a Compressordriven
consumption, less occupant comfort, and a lack of HVAC System,” ASHRAE Transactions, Vol. 89, Pt. 2A;pp. 361 -
temperature control. The findings from this study contribute 372,1983.
to the information about the effect of the thermostat time [ 111 Price, B. A. and Smith, T. F., Description of the Iowa Energy
constant on the energy consumption and comfort condition in Center Energy Resource Station: Facility Update 111, Technical
a building zone using a VAV operation and are Report: METFS-00-001, Department of Mechanical Engineering,
experimentally demonstrated. The University of Iowa, Iowa City, Iowa, 2000.
[I21 Wen, J. and Smith, T. F., “Development and Validation of a
Dynamic Simulation Model for Thermal Response of a Building,”
accepted for presentation at the 8th International Conference on
Building Envelopes, 2001.
255
Table 2 Energy consumption and comfort level comparison.
in ft.
View A
Interior Wind01
1.0x7.0
I Thermostats
f2
256
o 600 1200 6000 6600, 7100
83
69
67
0 600 1200 1800 2400 3000 3600 4200 4800 5400 6000 6600 7300
Time,rmn
Figure 7 Resultsforexperimentwithout housing.
69
67
257