STUDIES OF CONTROL STRATEGIES FOR THE ... - CiteSeerX
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007
STUDIES OF CONTROL STRATEGIES FOR THE CONCORDIA SOLAR HOUSE José A. Candanedo1, Brendan O’Neill1, Sasa Pantic1, and Andreas Athienitis1 1 Department of Building, Civil and Environmental Engineering, Concordia University Room EV16.117, 1455 Maisonneuve W., Montréal, Québec, H3G 1M8 Tel. (514)-848-2424, ext. 7080, e-mail: [email protected] Type of Paper: Refereed
ABSTRACT Efficient control strategies are essential for the effective operation of a building, and must be considered a fundamental part of the design. This paper presents studies of control strategies implemented at the Concordia Solar House. The investigation is focused on the control of three important subsystems and their integration: a) the fan driving the flow in the BIPV-thermal system; b) the mechanical system connecting a water-to-water heat pump to the HVAC system (including the water storage tanks); and c) the motorised interior roller blinds on the south facing façade.
INTRODUCTION Since May 2006, a team from Concordia University has been conducting research activities at the Concordia Solar House (Figure 1).
Proper regulation of the fan speed is necessary for adequate heat transfer while avoiding excessive power consumption by the fan. 2. The mechanical heat recovery and HVAC systems of the Solar House were originally designed to switch between several heating and cooling modes as needed, and to store the energy recovered in the water tanks. However, the implementation of the control system was not completed during the construction of the house. 3. The large windows in the south façade of the Solar House were designed to maximise solar heat gains and to allow daylighting. However, in order to avoid overheating and to minimise heat losses to the exterior, well-designed control algorithms are required to control the roller blinds covering these windows. The control algorithms for these subsystems are based both on experimental data (such as solar radiation, temperature and flow measurements) and on simulations carried out in Mathcad. The control system of the individual components of the house will be used as building blocks of a more comprehensive control strategy, using the solar house as a case study of predictive control based on weather forecasts.
CONTROL OF BIPV/T AND HVAC SYSTEMS Figure 1. Concordia Solar House in its present location. This house, developed by students for the 2005 Solar Decathlon competition, has been undergoing several modifications in order to convert it into a permanent research facility. This paper describes the current stage of development, and preliminary results, of control strategies for the systems of the house. Three subsystems of the house are of particular interest: 1. The Building Integrated Photovoltaic-Thermal System (BIPV/T system) installed on the roof has the dual purpose of generating electricity and recovering the excess heat to be used as a source for a heat pump or to preheat domestic hot water.
A variable speed fan located in the rear of the solar house drives the flow underneath the PV panels. As seen in Figure 2, this air flow can either be directed through an air-to-water heat exchanger or rejected directly to the exterior (in which case the only purpose of the air flow is to cool down the PV panels). At full fan speed, the flow rate is about 165 Collecting Duct
PV
nels @ PV pa oor Outd Air
15°
A
@ ls ne pa HVAC System
° 30
Heat Exchanger
A
Fan
Figure 2. Simplified schematic of BIPV/T System. 1
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 30 S=1000 W/m2 S=800 W/m2
25
S=600 W/m2 S=400 W/m2 S=200 W/m2
20
Tout-Tin (°C)
L/s (348 cfm) when the heat exchanger is bypassed. In this case, the average velocity is about 0.45 m/s in the 30° slope section and about 0.55 m/s in the 15° slope section. When passing through the heat exchanger the flow rate is reduced to 120 L/s (253 cfm), corresponding to 0.33 m/s (at the 30° section), and 0.40 m/s (at the 15° slope section). Different flow rates can be obtained by modifying the fan electrical frequency, and consequently its speed (see Figure 3).
15
10
5
0 0
20
40
200
80
100
120
140
160
Air Flow Rate (L/s)
HE Bypassed HE Used
180 160
Figure 4. Simulated air temperature rise for different solar radiation conditions vs air flow rates (Tinlet = 13.3 °C, Vwind = 1.5 m/s).
y = 3.16x - 10.5 2 R = 0.9922
140 PV Air Flow (L/s)
60
120 100 80
y = 2.5253x - 11.25 2 R = 0.9908
60 40 20 0 10
20
30
40
50
60
70
Fan Frequency (Hz)
Figure 3. Air flow in BIPV/T cavity vs. fan speed. The outlet temperatures are a strong function of the solar radiation conditions, wind speed and the air velocity in the cavity, as shown in the results of the simulation displayed in Figure 4. The variation of the fan speed offers the possibility of controlling the outlet temperature of the air in the BIPV/T cavity. The HVAC system of the Solar House permits the use of the hot air from the BIPV/T as a source for the heat pump (as shown schematically in Figure 5). Automatic controls for the valves and dampers of the
mechanical system have recently been implemented. The work at the Solar House thus far has concentrated on the analysis, upgrading and operation of the individual components of the HVAC and BIPV/T system. The study and control of the passive response of the home, independent of the HVAC system, has also been an important factor in working towards devising an integrated control strategy that will allow for the operation of all of the home’s components as a complete system.
PREDICTIVE CONTROL OF SOLAR HEAT GAINS Theoretical Derivation The windows located on the south wall of the living room are covered by large interior roller blinds. A relay circuit has been installed to allow the integration of the blind position control into the main Data Acquisition and Control Unit. Commands can
Q Solar Collector Heat Exchanger
Heat Pump Hot Tank
Cold Tank
Solar Storage Tank
AHU
Figure 5. HVAC System of the Solar House. The coloured lines indicate the water flow for the heating mode using the BIPV/T air. 2
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 be issued remotely from a desktop computer, manually or automatically according to a control algorithm.
Given that the house is well insulated, the value of the transfer admittance ( YT (ω ) ) approaches zero, and the effect of the sol-air temperature can be neglected.
The first control strategy, implemented during the winter months, consisted of opening the roller blinds when the solar irradiance exceeded 100 W/m2 on the south oriented, 30° roof slope, where the pyranometer is installed
Thus, the effect of the solar heat gains on the indoor temperature swing, in the frequency domain, is given by:
Figure 6 shows representative results obtained following this strategy, without the operation of any HVAC system, in late February 2007.
(1)
25 20
Passive Thermal Response
1200
15 1000
800
W/m2
Temperature (°C)
10 5 0
⎞⎛ ⎟⎜ 1 ⎟⎜ ⎟⎜ U o ⎟⎜ 1 + U S ⎠⎝
⎞ ⎟ ⎟ ⎟ ⎟ ⎠
The effect of the exterior temperature is given by:
1400
Int_Temp Ext_Temp Irradiation on 30° tilt, due south surface
⎛ ⎜ G (ω ) T1 (ω ) = ⎜ U SU o ⎜ ⎜ Y (ω ) + U + U S o ⎝
600
-5
⎛ 1 1 + ⎜ YS (ω ) U S T2 (ω ) = ⎜ 1 1 ⎜ 1 ⎜ Y (ω ) + U + U S o ⎝ S
⎞ ⎟ ⎟ T (ω ) ⎟ o ⎟ ⎠
(2)
400
-10 200
-15 Time
-20 24/Feb
0
25/Feb
26/Feb
27/Feb
28/Feb
Figure 6. Interior and exterior temperatures and global solar irradiance on a 30° due south slope, from February 24th through 27th, 2007.
Uo TR ( )
-YT ( )Teo ( )
YS ( )
case if there is very little infiltration and doors and windows have high R-values, then T1(ω ) is the dominant factor and: TR (ω ) ≈
G( ) US
The resulting room air temperature, TR (ω ) , is given by adding T1(ω ) + T2 (ω ) . If Uo is very small in comparison with US and YS (ω ) , which would be the
Teo( )
A simplified mathematical model of the Solar House, based on thermal networks and wall admittance models (Athienitis et al., 1990; Athienitis, 1994; Athienitis and Santamouris, 2002; Davies, 2004) is shown in Figure 7. The meaning of the symbols is described in the nomenclature section. The indoor temperature can be calculated by considering the effect of each of the three forcing functions (namely the solar heat gains, exterior temperature and sol-air temperature) independently, and by applying the superposition principle.
(3)
For non-negligible infiltration or forced ventilation, the effect of exterior temperature has to be considered. Equation 3 could be modified as follows: TR (ω ) ≈ k1
Figure 7. Simplified model of the Solar House. The inspection of several days of data suggested that the indoor temperature swing is largely due to the daily cycle of solar radiation. Certainly, the exterior temperature variation has an effect on the indoor temperature; however, due to the levels of insulation of the Solar House, it may be assumed that passive solar heat gains are the most important factor in the daily indoor temperature oscillation.
G (ω ) Y (ω )
G (ω ) + k2To(ω ) Y (ω )
(4)
Where k1 and k2 are complex correction factors, with magnitudes between 0 and 1, which could be determined experimentally for a particular situation. Solar House Case In the case of the Solar House, temperature and solar radiation data from a very clear day were chosen (February 26th, 2007), when the blinds were fully open. Figures 8 and 9 display the solar radiation and temperature curves (passive response) used to calculate the self-admittance of the interior surfaces of the house. Assuming that the exterior temperature has negligible influence on the interior temperature swing, Fourier analyses were performed for each of the curves, and Equation 5 was derived for the self-admittance: Y (ω ) ≈
S (ω ) Awindowsτ TR (ω )
(5)
With this value, it is possible to estimate the indoor temperature swing from an estimate of the solar heat
3
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 gains based on the sky conditions and the position of the blinds. This scheme can be used to control, to 1200
τ eq = rτ open + (1 − r )τ closed
1200
Shading effect of nearby building
1000 800 S it W 2 m
Shading effect of trees
window without the blind (estimated to be about 65%) and τ closed is the effective transmittance of the window including the blind (set to 40% to account
400 200 0
0
(6)
In Equation 6, r is the blind position as a percentage of fully open, τ open is the transmittance of the
600
0
equivalent transmittance was defined for each of these positions as shown in Equation 6.
6
12
0
18
24
tit
Table 1. Temperature swings for different solar radiation conditions and blind positions (°C).
24
Condition Overcast P. Cloudy P. Sunny Clear Sunny
hr
Figure 8. Solar radiation on the south façade for February 26th, 2007 (including snow reflection)
0% 2.4 4.7 7.1 9.4
25% 2.7 5.4 8.2 10.9
Blind Position 50% 75% 3.1 3.5 6.2 6.9 9.3 10.4 12.3 13.8
100% 3.8 7.6 11.5 15.3
25
25
for the heat absorbed by the blinds and transmitted through smaller, non-shaded windows).
20
Table 1 shows the temperature swings expected with the blind positions and the solar radiation conditions.
15
IntTempit Celsius
10 5
0
0
0
6
12
0
18
tit
24 24
hr
Figure 9. Indoor air temperature, Feb. 26th, 2007. some extent, the indoor temperature.
With the temperature swing table, a control algorithm was designed (see Figure 11) to prevent overheating in the house by measuring the indoor temperature at early morning (where the minimum temperature is expected), and adjusting the blind position in order to avoid exceeding a temperature threshold, while maximising solar heat gains. This algorithm should work provided there are no drastic changes in exterior air temperature.
Algorithm Implementation Four sky conditions, with their corresponding solar irradiance curves, were defined arbitrarily: Clear sunny, partly sunny, partly cloudy, and overcast (see Figure 10). These categories were decided to be consistent with a typical weather forecast report, when accurate solar radiation data is not available. 1000
1000
Clear sunny
800 Sc it
Partly sunny
Sps it 600 Spc it
Partly cloudy
400
Scs it
Overcast
200 0
0
0 0
6
12 tit
18
24 24
hr
Figure 10. Solar irradiance curves defined for the control algorithm. Five blind positions were defined: fully open (100%), 75%, 50%, 25% and fully closed (0%). An Figure 11. Control algorithm for the roller blinds. 4
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007
As a first attempt to use weather predictions for the blind control algorithm, the sky conditions from the weather forecast as reported on the Environment Canada website were introduced manually on March 28th for the subsequent three days. The conditions used as inputs were: 29th, “clear sunny”; 30th, “partly sunny”; and 31st, “clear sunny”. It was decided to set the weather conditions as “partly sunny” from April 1st onwards.
RESULTS AND DISCUSSION
because the solar radiation curves were modelled for the conditions of the previous month. Figure 13 shows a revised solar irradiance curve for a clear sunny day, modeling the shading effect of trees by means of a random function. Figure 14 compares the temperature rise predicted by the admittance model using the revised curve and the actual temperature rise recorded on March 29th, 2007. The agreement between the two curves is good. It is also worth pointing out that the main purpose of the algorithm, to avoid overheating, was achieved.
The results of the blind control algorithm for the days following March 29th are presented in Figure 12 and Table 2. The moving hourly average of solar radiation is presented to make the interpretation of results clearer. The measurements of temperature swings from March 30th, 31st and even April 1st are in good agreement with the predicted values. This is because the weather forecast accurately predicted the sky conditions for these days. 1400
1100
Scs it
Outdoor T.
550
W 2
m
0
28
Indoor T.
1100
0
18 per. Mov. Avg. (Solar Radiation) 24
0
2
4
6
8
10
0
1200
2
Solar Radiation (W/m )
16 800
12
14
16
18
20
22
24
t it
20 1000
12
24
hr
Figure 13. Improved irradiance model for a clear day in late March.
°C 8
600
Measured Temperature 4
Modelled T Int
Solar Radiation
10
1200
400 0
8
1000
200
6 0 31/Mar
01/Apr
02/Apr
Figure 12. Recorded indoor temperature, exterior temperature and moving hourly average of solar radiation for March 29th through April 6th, 2007. Table 2. Comparison of predicted and measured temperature swings. March 29th-April 6th, 2007. Mar Day Predicted Condition Blind Position Expect Temp. Swing Swing Assessment
800
-8 30/Mar
29th Clear sunny 25% 10.87 11 Correct
April
30th 31st 1st Partly Sunny Clear Sunny Partly Sunny 75% 25% 75% 10.36 10.87 10.36 9.8 10 8.3 Correct Correct Correct
Additional sky condition categories and their associated solar irradiance curves could be defined. These curves should also be a function of the time of the year. However, in order to be used appropriately, the weather forecast data used as input should be more detailed and accurate. The use of numerical weather forecast data, downloaded periodically from the internet is planned in collaboration with Environment Canada. The curves in Figure 10, used in the control algorithm, predicted that the minimum temperature occurs about 6:30 a.m. It was observed that the actual minimum values occurred at about 8:00 a.m., partly
T (°C)
29/Mar
4 600 2 400 0 200
-2
-4 06:00
Solar Radiation (W/m 2)
-4
0 08:00
10:00
12:00
14:00
16:00
18:00
20:00
22:00
00:00
Local Standard Time (Official Time - 1hr)
Figure 14. Predicted (green) vs measured (red) temperature rise for March 29th, 2007. As a final remark, it is important to mention that a large portion of the thermal mass of the house is not exposed to solar heat gains, and therefore, the temperature swings are more extreme than would be advisable for a passive solar home.
PERFORMANCE OF PV ARRAYS Current-voltage curves (I-V curves) are commonly used for evaluating the performance of arrays of photovoltaic panels under different conditions. An IV curve is obtained by applying different loads to the terminals of an array (from open circuit to short circuit) and measuring current and voltage. This can be done manually, or more often, by means of an
5
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 electronic curve tracer, which performs the load variation automatically. I-V curves have been taken from the Solar House
A similar pattern (performance below rated values) has been observed in the other arrays. Figure 17 shows a typical curve for Array 1. The maximum power in this case was 1378 W. 30
3
3-1
3-2
3-3
3-4
3-5
4-1
4-2
3-6
3-7
3-8
3-9
3-10
4-3
4-4
25
4 2-1
2-2
2-3
2-4
2-5
4-5
4-6
2-6
2-7
2-8
2-9
2-10
4-7
4-8
15 Current
2
20
10
5
1-1
1
1-2
1-3
1-4
1-5
0
5-1
0
10
20
30
40
50
60
70
80
90
5 1-6
1-7
1-8
1-9
1-10
-5
5-2
Voltage
Figure 17. IV curve, Array 1. Fan off.
As expected, it has been observed that when the PV arrays tend to perform below the rated values when their temperature exceeds standard test conditions. The PV panels installed at the Concordia Solar House have a large temperature coefficient (according to the manufacturer, -0.5% of rated power per °C above STC). Since the temperatures of the PV panels were near 60 °C, reductions of 17.5% or more were expected. Figure 16 shows curves corresponding to Array 2 (on the 30° slope), taken near solar noon in spring, at slightly more than 1000 W/m2 on the roof slope. Since this array has 10 PV panels, each of them rated at 175 Wp, the expected maximum power was close to 1750 W. For the first test (with the fan off), it was found that the maximum power was 1371 W (78.3% of its rated capacity). After turning on the BIPV/T fan, and letting it run for about half an hour, the maximum power was 1462 W (83.5% of its rated capacity), under nearly identical conditions of solar irradiance. This increase can be attributed to the cooling effect of the fan on the PV panels. This effect seems to be more evident in Array 2 (perhaps the one experiencing most the effects of the air flow). 30
25 Array 2, No fan Array 2, Fan operating
Current (A)
20
15
10
5
0 0
10
20
30
40 50 Voltage (V)
60
70
80
90
Figure 18 shows an IV curve corresponding to Array 5, the smallest one, with only two panels. The maximum power measured was 270 W (77% of rated value). 6
5
4 Current (A)
Figure 15. Configuration of the PV arrays. arrays employing a commercial Array Tester system (ARRAY TESTER 550, EETS, United Kingdom). Some of the curves obtained are presented below.
3
2
1
0 0
10
20
30
40
50
60
70
80
90
Voltage (V)
Figure 18. IV curve, Array 5. Fan off.
CONCLUSIONS This paper has presented an investigation of control strategies for the BIPV/T, HVAC system and motorised roller blinds of the Concordia Solar House. The algorithm used for the control of the roller blinds of the south facing façade is presented. The results obtained with this algorithm show the possibility of using the concept of thermal admittance and weather forecasts for predictive control of the indoor temperature. The influence of temperature on the performance of the PV panels has been confirmed. A properly designed BIPV/T system can help mitigate this effect. Lessons learned from the Solar House about the design and operation of solar homes have been implemented in Alstonvale Net Zero House (Candanedo et al., 2007), a building designed for the EQuilibrium Housing competition organised by CMHC.
Figure 16. IV curves, Array 2. Fan on and off.
6
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 More information about the Concordia Solar House is available in the works by Pantic (2007), Moussa (2006), and Pasini (2006).
ACKNOWLEDGEMENTS This work was supported by Natural Resources Canada through the Innovative Research Initiative and the Technology and Innovation Program as part of the Climate Change Plan for Canada, and by NSERC through the Solar Buildings Research Network. The collaboration of our colleagues from the Concordia Solar Laboratory has been instrumental in the efforts performed at the Solar House.
NOMENCLATURE
the SESCI/SBRN Joint Conference, Calgary, June 2007. Davies, M.G. 2004. Building Heat Transfer. John Wiley and Sons. Hoboken, N.J., USA. Moussa, R. 2006. The 2005 Canadian Solar Decathlon House Revisited. Report presented to the BCEE Department, Concordia University Pantic, S. 2007. Energy Analysis of Photovoltaic Thermal System Integrated with Roof and HVAC System. Master’s Thesis. Concordia University, Montréal, Canada. Pasini, M.A. 2006. Modeling and Design of an Independent Solar House. Master’s Thesis. Concordia University, Montréal, Canada.
Awindows: Area of the south façade windows (m2) G: Solar heat gains incident on the internal surfaces (W/m2) Temperature of the room air (°C) TR: S (ω ) : Solar irradiance on the windows in the frequency domain (W/m2) To: Exterior temperature (°C) Sol-air temperature (°C) Teo: Conductance between the room air and the U o: exterior. It includes the conductance through the windows, doors, and the effect of infiltration (W/K) Conductance between the surfaces having US: self-admittance (walls and floor) and the room air (W/K) Y: Self admittance of walls and floor (W/K) Transfer admittance (W/K) Y T: τ : Transmittance of the windows τ eq : Equivalent transmittance of the windows τ open :
Transmittance of the windows without the
blinds τ closed : Transmittance of the windows including the blinds ω: Angular frequency (rad/s)
REFERENCES Athienitis, A.K.. 1994. Buiding Thermal Analysis, Mathcad Electronic Book. Athienitis, A.K., Stylianou, M., Shou, J. 1990. A Methodology for Building Thermal Dynamics Studies and Control Applications. ASHRAE Transactions, Vol. 96, Part 2, pp. 839-848. Athienitis, A.K., Santamouris, M. 2002. Thermal Analysis and Design of Passive Solar Buildings. James & James. London, UK. Candanedo, J.A., Pogharian, S., Athienitis, A.K. 2007. Design and Simulation of a Net Zero Energy Home in Montréal. To be presented at
7
STUDIES OF CONTROL STRATEGIES FOR THE CONCORDIA SOLAR HOUSE José A. Candanedo1, Brendan O’Neill1, Sasa Pantic1, and Andreas Athienitis1 1 Department of Building, Civil and Environmental Engineering, Concordia University Room EV16.117, 1455 Maisonneuve W., Montréal, Québec, H3G 1M8 Tel. (514)-848-2424, ext. 7080, e-mail: [email protected] Type of Paper: Refereed
ABSTRACT Efficient control strategies are essential for the effective operation of a building, and must be considered a fundamental part of the design. This paper presents studies of control strategies implemented at the Concordia Solar House. The investigation is focused on the control of three important subsystems and their integration: a) the fan driving the flow in the BIPV-thermal system; b) the mechanical system connecting a water-to-water heat pump to the HVAC system (including the water storage tanks); and c) the motorised interior roller blinds on the south facing façade.
INTRODUCTION Since May 2006, a team from Concordia University has been conducting research activities at the Concordia Solar House (Figure 1).
Proper regulation of the fan speed is necessary for adequate heat transfer while avoiding excessive power consumption by the fan. 2. The mechanical heat recovery and HVAC systems of the Solar House were originally designed to switch between several heating and cooling modes as needed, and to store the energy recovered in the water tanks. However, the implementation of the control system was not completed during the construction of the house. 3. The large windows in the south façade of the Solar House were designed to maximise solar heat gains and to allow daylighting. However, in order to avoid overheating and to minimise heat losses to the exterior, well-designed control algorithms are required to control the roller blinds covering these windows. The control algorithms for these subsystems are based both on experimental data (such as solar radiation, temperature and flow measurements) and on simulations carried out in Mathcad. The control system of the individual components of the house will be used as building blocks of a more comprehensive control strategy, using the solar house as a case study of predictive control based on weather forecasts.
CONTROL OF BIPV/T AND HVAC SYSTEMS Figure 1. Concordia Solar House in its present location. This house, developed by students for the 2005 Solar Decathlon competition, has been undergoing several modifications in order to convert it into a permanent research facility. This paper describes the current stage of development, and preliminary results, of control strategies for the systems of the house. Three subsystems of the house are of particular interest: 1. The Building Integrated Photovoltaic-Thermal System (BIPV/T system) installed on the roof has the dual purpose of generating electricity and recovering the excess heat to be used as a source for a heat pump or to preheat domestic hot water.
A variable speed fan located in the rear of the solar house drives the flow underneath the PV panels. As seen in Figure 2, this air flow can either be directed through an air-to-water heat exchanger or rejected directly to the exterior (in which case the only purpose of the air flow is to cool down the PV panels). At full fan speed, the flow rate is about 165 Collecting Duct
PV
nels @ PV pa oor Outd Air
15°
A
@ ls ne pa HVAC System
° 30
Heat Exchanger
A
Fan
Figure 2. Simplified schematic of BIPV/T System. 1
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 30 S=1000 W/m2 S=800 W/m2
25
S=600 W/m2 S=400 W/m2 S=200 W/m2
20
Tout-Tin (°C)
L/s (348 cfm) when the heat exchanger is bypassed. In this case, the average velocity is about 0.45 m/s in the 30° slope section and about 0.55 m/s in the 15° slope section. When passing through the heat exchanger the flow rate is reduced to 120 L/s (253 cfm), corresponding to 0.33 m/s (at the 30° section), and 0.40 m/s (at the 15° slope section). Different flow rates can be obtained by modifying the fan electrical frequency, and consequently its speed (see Figure 3).
15
10
5
0 0
20
40
200
80
100
120
140
160
Air Flow Rate (L/s)
HE Bypassed HE Used
180 160
Figure 4. Simulated air temperature rise for different solar radiation conditions vs air flow rates (Tinlet = 13.3 °C, Vwind = 1.5 m/s).
y = 3.16x - 10.5 2 R = 0.9922
140 PV Air Flow (L/s)
60
120 100 80
y = 2.5253x - 11.25 2 R = 0.9908
60 40 20 0 10
20
30
40
50
60
70
Fan Frequency (Hz)
Figure 3. Air flow in BIPV/T cavity vs. fan speed. The outlet temperatures are a strong function of the solar radiation conditions, wind speed and the air velocity in the cavity, as shown in the results of the simulation displayed in Figure 4. The variation of the fan speed offers the possibility of controlling the outlet temperature of the air in the BIPV/T cavity. The HVAC system of the Solar House permits the use of the hot air from the BIPV/T as a source for the heat pump (as shown schematically in Figure 5). Automatic controls for the valves and dampers of the
mechanical system have recently been implemented. The work at the Solar House thus far has concentrated on the analysis, upgrading and operation of the individual components of the HVAC and BIPV/T system. The study and control of the passive response of the home, independent of the HVAC system, has also been an important factor in working towards devising an integrated control strategy that will allow for the operation of all of the home’s components as a complete system.
PREDICTIVE CONTROL OF SOLAR HEAT GAINS Theoretical Derivation The windows located on the south wall of the living room are covered by large interior roller blinds. A relay circuit has been installed to allow the integration of the blind position control into the main Data Acquisition and Control Unit. Commands can
Q Solar Collector Heat Exchanger
Heat Pump Hot Tank
Cold Tank
Solar Storage Tank
AHU
Figure 5. HVAC System of the Solar House. The coloured lines indicate the water flow for the heating mode using the BIPV/T air. 2
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 be issued remotely from a desktop computer, manually or automatically according to a control algorithm.
Given that the house is well insulated, the value of the transfer admittance ( YT (ω ) ) approaches zero, and the effect of the sol-air temperature can be neglected.
The first control strategy, implemented during the winter months, consisted of opening the roller blinds when the solar irradiance exceeded 100 W/m2 on the south oriented, 30° roof slope, where the pyranometer is installed
Thus, the effect of the solar heat gains on the indoor temperature swing, in the frequency domain, is given by:
Figure 6 shows representative results obtained following this strategy, without the operation of any HVAC system, in late February 2007.
(1)
25 20
Passive Thermal Response
1200
15 1000
800
W/m2
Temperature (°C)
10 5 0
⎞⎛ ⎟⎜ 1 ⎟⎜ ⎟⎜ U o ⎟⎜ 1 + U S ⎠⎝
⎞ ⎟ ⎟ ⎟ ⎟ ⎠
The effect of the exterior temperature is given by:
1400
Int_Temp Ext_Temp Irradiation on 30° tilt, due south surface
⎛ ⎜ G (ω ) T1 (ω ) = ⎜ U SU o ⎜ ⎜ Y (ω ) + U + U S o ⎝
600
-5
⎛ 1 1 + ⎜ YS (ω ) U S T2 (ω ) = ⎜ 1 1 ⎜ 1 ⎜ Y (ω ) + U + U S o ⎝ S
⎞ ⎟ ⎟ T (ω ) ⎟ o ⎟ ⎠
(2)
400
-10 200
-15 Time
-20 24/Feb
0
25/Feb
26/Feb
27/Feb
28/Feb
Figure 6. Interior and exterior temperatures and global solar irradiance on a 30° due south slope, from February 24th through 27th, 2007.
Uo TR ( )
-YT ( )Teo ( )
YS ( )
case if there is very little infiltration and doors and windows have high R-values, then T1(ω ) is the dominant factor and: TR (ω ) ≈
G( ) US
The resulting room air temperature, TR (ω ) , is given by adding T1(ω ) + T2 (ω ) . If Uo is very small in comparison with US and YS (ω ) , which would be the
Teo( )
A simplified mathematical model of the Solar House, based on thermal networks and wall admittance models (Athienitis et al., 1990; Athienitis, 1994; Athienitis and Santamouris, 2002; Davies, 2004) is shown in Figure 7. The meaning of the symbols is described in the nomenclature section. The indoor temperature can be calculated by considering the effect of each of the three forcing functions (namely the solar heat gains, exterior temperature and sol-air temperature) independently, and by applying the superposition principle.
(3)
For non-negligible infiltration or forced ventilation, the effect of exterior temperature has to be considered. Equation 3 could be modified as follows: TR (ω ) ≈ k1
Figure 7. Simplified model of the Solar House. The inspection of several days of data suggested that the indoor temperature swing is largely due to the daily cycle of solar radiation. Certainly, the exterior temperature variation has an effect on the indoor temperature; however, due to the levels of insulation of the Solar House, it may be assumed that passive solar heat gains are the most important factor in the daily indoor temperature oscillation.
G (ω ) Y (ω )
G (ω ) + k2To(ω ) Y (ω )
(4)
Where k1 and k2 are complex correction factors, with magnitudes between 0 and 1, which could be determined experimentally for a particular situation. Solar House Case In the case of the Solar House, temperature and solar radiation data from a very clear day were chosen (February 26th, 2007), when the blinds were fully open. Figures 8 and 9 display the solar radiation and temperature curves (passive response) used to calculate the self-admittance of the interior surfaces of the house. Assuming that the exterior temperature has negligible influence on the interior temperature swing, Fourier analyses were performed for each of the curves, and Equation 5 was derived for the self-admittance: Y (ω ) ≈
S (ω ) Awindowsτ TR (ω )
(5)
With this value, it is possible to estimate the indoor temperature swing from an estimate of the solar heat
3
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 gains based on the sky conditions and the position of the blinds. This scheme can be used to control, to 1200
τ eq = rτ open + (1 − r )τ closed
1200
Shading effect of nearby building
1000 800 S it W 2 m
Shading effect of trees
window without the blind (estimated to be about 65%) and τ closed is the effective transmittance of the window including the blind (set to 40% to account
400 200 0
0
(6)
In Equation 6, r is the blind position as a percentage of fully open, τ open is the transmittance of the
600
0
equivalent transmittance was defined for each of these positions as shown in Equation 6.
6
12
0
18
24
tit
Table 1. Temperature swings for different solar radiation conditions and blind positions (°C).
24
Condition Overcast P. Cloudy P. Sunny Clear Sunny
hr
Figure 8. Solar radiation on the south façade for February 26th, 2007 (including snow reflection)
0% 2.4 4.7 7.1 9.4
25% 2.7 5.4 8.2 10.9
Blind Position 50% 75% 3.1 3.5 6.2 6.9 9.3 10.4 12.3 13.8
100% 3.8 7.6 11.5 15.3
25
25
for the heat absorbed by the blinds and transmitted through smaller, non-shaded windows).
20
Table 1 shows the temperature swings expected with the blind positions and the solar radiation conditions.
15
IntTempit Celsius
10 5
0
0
0
6
12
0
18
tit
24 24
hr
Figure 9. Indoor air temperature, Feb. 26th, 2007. some extent, the indoor temperature.
With the temperature swing table, a control algorithm was designed (see Figure 11) to prevent overheating in the house by measuring the indoor temperature at early morning (where the minimum temperature is expected), and adjusting the blind position in order to avoid exceeding a temperature threshold, while maximising solar heat gains. This algorithm should work provided there are no drastic changes in exterior air temperature.
Algorithm Implementation Four sky conditions, with their corresponding solar irradiance curves, were defined arbitrarily: Clear sunny, partly sunny, partly cloudy, and overcast (see Figure 10). These categories were decided to be consistent with a typical weather forecast report, when accurate solar radiation data is not available. 1000
1000
Clear sunny
800 Sc it
Partly sunny
Sps it 600 Spc it
Partly cloudy
400
Scs it
Overcast
200 0
0
0 0
6
12 tit
18
24 24
hr
Figure 10. Solar irradiance curves defined for the control algorithm. Five blind positions were defined: fully open (100%), 75%, 50%, 25% and fully closed (0%). An Figure 11. Control algorithm for the roller blinds. 4
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007
As a first attempt to use weather predictions for the blind control algorithm, the sky conditions from the weather forecast as reported on the Environment Canada website were introduced manually on March 28th for the subsequent three days. The conditions used as inputs were: 29th, “clear sunny”; 30th, “partly sunny”; and 31st, “clear sunny”. It was decided to set the weather conditions as “partly sunny” from April 1st onwards.
RESULTS AND DISCUSSION
because the solar radiation curves were modelled for the conditions of the previous month. Figure 13 shows a revised solar irradiance curve for a clear sunny day, modeling the shading effect of trees by means of a random function. Figure 14 compares the temperature rise predicted by the admittance model using the revised curve and the actual temperature rise recorded on March 29th, 2007. The agreement between the two curves is good. It is also worth pointing out that the main purpose of the algorithm, to avoid overheating, was achieved.
The results of the blind control algorithm for the days following March 29th are presented in Figure 12 and Table 2. The moving hourly average of solar radiation is presented to make the interpretation of results clearer. The measurements of temperature swings from March 30th, 31st and even April 1st are in good agreement with the predicted values. This is because the weather forecast accurately predicted the sky conditions for these days. 1400
1100
Scs it
Outdoor T.
550
W 2
m
0
28
Indoor T.
1100
0
18 per. Mov. Avg. (Solar Radiation) 24
0
2
4
6
8
10
0
1200
2
Solar Radiation (W/m )
16 800
12
14
16
18
20
22
24
t it
20 1000
12
24
hr
Figure 13. Improved irradiance model for a clear day in late March.
°C 8
600
Measured Temperature 4
Modelled T Int
Solar Radiation
10
1200
400 0
8
1000
200
6 0 31/Mar
01/Apr
02/Apr
Figure 12. Recorded indoor temperature, exterior temperature and moving hourly average of solar radiation for March 29th through April 6th, 2007. Table 2. Comparison of predicted and measured temperature swings. March 29th-April 6th, 2007. Mar Day Predicted Condition Blind Position Expect Temp. Swing Swing Assessment
800
-8 30/Mar
29th Clear sunny 25% 10.87 11 Correct
April
30th 31st 1st Partly Sunny Clear Sunny Partly Sunny 75% 25% 75% 10.36 10.87 10.36 9.8 10 8.3 Correct Correct Correct
Additional sky condition categories and their associated solar irradiance curves could be defined. These curves should also be a function of the time of the year. However, in order to be used appropriately, the weather forecast data used as input should be more detailed and accurate. The use of numerical weather forecast data, downloaded periodically from the internet is planned in collaboration with Environment Canada. The curves in Figure 10, used in the control algorithm, predicted that the minimum temperature occurs about 6:30 a.m. It was observed that the actual minimum values occurred at about 8:00 a.m., partly
T (°C)
29/Mar
4 600 2 400 0 200
-2
-4 06:00
Solar Radiation (W/m 2)
-4
0 08:00
10:00
12:00
14:00
16:00
18:00
20:00
22:00
00:00
Local Standard Time (Official Time - 1hr)
Figure 14. Predicted (green) vs measured (red) temperature rise for March 29th, 2007. As a final remark, it is important to mention that a large portion of the thermal mass of the house is not exposed to solar heat gains, and therefore, the temperature swings are more extreme than would be advisable for a passive solar home.
PERFORMANCE OF PV ARRAYS Current-voltage curves (I-V curves) are commonly used for evaluating the performance of arrays of photovoltaic panels under different conditions. An IV curve is obtained by applying different loads to the terminals of an array (from open circuit to short circuit) and measuring current and voltage. This can be done manually, or more often, by means of an
5
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 electronic curve tracer, which performs the load variation automatically. I-V curves have been taken from the Solar House
A similar pattern (performance below rated values) has been observed in the other arrays. Figure 17 shows a typical curve for Array 1. The maximum power in this case was 1378 W. 30
3
3-1
3-2
3-3
3-4
3-5
4-1
4-2
3-6
3-7
3-8
3-9
3-10
4-3
4-4
25
4 2-1
2-2
2-3
2-4
2-5
4-5
4-6
2-6
2-7
2-8
2-9
2-10
4-7
4-8
15 Current
2
20
10
5
1-1
1
1-2
1-3
1-4
1-5
0
5-1
0
10
20
30
40
50
60
70
80
90
5 1-6
1-7
1-8
1-9
1-10
-5
5-2
Voltage
Figure 17. IV curve, Array 1. Fan off.
As expected, it has been observed that when the PV arrays tend to perform below the rated values when their temperature exceeds standard test conditions. The PV panels installed at the Concordia Solar House have a large temperature coefficient (according to the manufacturer, -0.5% of rated power per °C above STC). Since the temperatures of the PV panels were near 60 °C, reductions of 17.5% or more were expected. Figure 16 shows curves corresponding to Array 2 (on the 30° slope), taken near solar noon in spring, at slightly more than 1000 W/m2 on the roof slope. Since this array has 10 PV panels, each of them rated at 175 Wp, the expected maximum power was close to 1750 W. For the first test (with the fan off), it was found that the maximum power was 1371 W (78.3% of its rated capacity). After turning on the BIPV/T fan, and letting it run for about half an hour, the maximum power was 1462 W (83.5% of its rated capacity), under nearly identical conditions of solar irradiance. This increase can be attributed to the cooling effect of the fan on the PV panels. This effect seems to be more evident in Array 2 (perhaps the one experiencing most the effects of the air flow). 30
25 Array 2, No fan Array 2, Fan operating
Current (A)
20
15
10
5
0 0
10
20
30
40 50 Voltage (V)
60
70
80
90
Figure 18 shows an IV curve corresponding to Array 5, the smallest one, with only two panels. The maximum power measured was 270 W (77% of rated value). 6
5
4 Current (A)
Figure 15. Configuration of the PV arrays. arrays employing a commercial Array Tester system (ARRAY TESTER 550, EETS, United Kingdom). Some of the curves obtained are presented below.
3
2
1
0 0
10
20
30
40
50
60
70
80
90
Voltage (V)
Figure 18. IV curve, Array 5. Fan off.
CONCLUSIONS This paper has presented an investigation of control strategies for the BIPV/T, HVAC system and motorised roller blinds of the Concordia Solar House. The algorithm used for the control of the roller blinds of the south facing façade is presented. The results obtained with this algorithm show the possibility of using the concept of thermal admittance and weather forecasts for predictive control of the indoor temperature. The influence of temperature on the performance of the PV panels has been confirmed. A properly designed BIPV/T system can help mitigate this effect. Lessons learned from the Solar House about the design and operation of solar homes have been implemented in Alstonvale Net Zero House (Candanedo et al., 2007), a building designed for the EQuilibrium Housing competition organised by CMHC.
Figure 16. IV curves, Array 2. Fan on and off.
6
2nd Canadian Solar Buildings Conference Calgary, June 10 – 14, 2007 More information about the Concordia Solar House is available in the works by Pantic (2007), Moussa (2006), and Pasini (2006).
ACKNOWLEDGEMENTS This work was supported by Natural Resources Canada through the Innovative Research Initiative and the Technology and Innovation Program as part of the Climate Change Plan for Canada, and by NSERC through the Solar Buildings Research Network. The collaboration of our colleagues from the Concordia Solar Laboratory has been instrumental in the efforts performed at the Solar House.
NOMENCLATURE
the SESCI/SBRN Joint Conference, Calgary, June 2007. Davies, M.G. 2004. Building Heat Transfer. John Wiley and Sons. Hoboken, N.J., USA. Moussa, R. 2006. The 2005 Canadian Solar Decathlon House Revisited. Report presented to the BCEE Department, Concordia University Pantic, S. 2007. Energy Analysis of Photovoltaic Thermal System Integrated with Roof and HVAC System. Master’s Thesis. Concordia University, Montréal, Canada. Pasini, M.A. 2006. Modeling and Design of an Independent Solar House. Master’s Thesis. Concordia University, Montréal, Canada.
Awindows: Area of the south façade windows (m2) G: Solar heat gains incident on the internal surfaces (W/m2) Temperature of the room air (°C) TR: S (ω ) : Solar irradiance on the windows in the frequency domain (W/m2) To: Exterior temperature (°C) Sol-air temperature (°C) Teo: Conductance between the room air and the U o: exterior. It includes the conductance through the windows, doors, and the effect of infiltration (W/K) Conductance between the surfaces having US: self-admittance (walls and floor) and the room air (W/K) Y: Self admittance of walls and floor (W/K) Transfer admittance (W/K) Y T: τ : Transmittance of the windows τ eq : Equivalent transmittance of the windows τ open :
Transmittance of the windows without the
blinds τ closed : Transmittance of the windows including the blinds ω: Angular frequency (rad/s)
REFERENCES Athienitis, A.K.. 1994. Buiding Thermal Analysis, Mathcad Electronic Book. Athienitis, A.K., Stylianou, M., Shou, J. 1990. A Methodology for Building Thermal Dynamics Studies and Control Applications. ASHRAE Transactions, Vol. 96, Part 2, pp. 839-848. Athienitis, A.K., Santamouris, M. 2002. Thermal Analysis and Design of Passive Solar Buildings. James & James. London, UK. Candanedo, J.A., Pogharian, S., Athienitis, A.K. 2007. Design and Simulation of a Net Zero Energy Home in Montréal. To be presented at
7
