Evaluation of the control performance of hydronic radiant heating systems based on the emulation using hardware-in-the-loop simulation

Building and Environment 46 (2011) 2012e2022

Contents lists available at ScienceDirect

Building and Environment journal homepage: www.elsevier.com/locate/buildenv

Evaluation of the control performance of hydronic radiant heating systems based on the emulation using hardware-in-the-loop simulation Kyu Nam Rhee a, Myoung Souk Yeo b, *, Kwang Woo Kim b a b

Department of Architecture, Graduate School of Seoul National University, Republic of Korea Department of Architecture, College of Eng., Seoul National University, Republic of Korea

a r t i c l e i n f o

a b s t r a c t

Article history: Received 10 December 2010 Received in revised form 26 March 2011 Accepted 9 April 2011

This study presents an emulation method to evaluate the control performance of a hydronic radiant heating system. Since heat output in the system is dependent on the pressure loss and flow rate in the hydronic network, the interaction between thermal and hydronic models needs to be considered in the evaluation of the control performance. For this reason, many studies apply an integrated simulation to the evaluation; however, the analysis of the hydronic network sometimes leads to unreliable results due to the improper initial values for algebraic loops or the lack of modeling information on the hydronic components. In order to deal with this problem, this study suggests an emulation method, where the hydronic network is replaced by real hardware and the building physics is analyzed by a simulation. In the emulation, the pressure loss and flow rate in the hydronic network were represented by replacing the real pipe with equivalent hydraulic resistance. In addition, by using real control systems that connect the hydronic network and building simulation, the interaction between building physics and hydronic network could be considered in the evaluation. Based on the proposed emulation method, the performance of several control strategies was evaluated in terms of the accuracy and the rise time. The result shows that the individual control needs to be combined with hydronic balancing for more accurate control. Hydronic control devices such as a flow limit valve and a pressure differential control valve also proved to be helpful to the improvement of the control performance. Ó 2011 Elsevier Ltd. All rights reserved.

Keywords: Hydronic radiant heating system Emulator Emulation Hydronic balancing Control performance Hardware-in-the-loop simulation

1. Introduction The hydronic radiant heating system is widely used in residential buildings due to its high thermal comfort level, energy saving potential, and clean and quiet operation [1e3]. As it uses hot water as a heating medium, it is controlled by modulating the temperature and/or flow rate of hot water, or the heat flux from the heat source [4]. Previous studies show that it is practical to apply water flow rate control combined with water temperature control to the hydronic radiant heating system for multi-zone buildings [5,6]. Thus, the condition of pressure loss and flow rate in a hydronic network should be considered in the control performance analysis of a hydronic radiant heating system. In addition, it should be noted that the control performance is considerably affected by hydronic components such as manifolds, embedded pipe, circulation pump, balancing valves, control valves and so on. Therefore, many studies argue that the control performance should be analyzed with an integrated simulation [7e11] or co* Corresponding author. E-mail address: [email protected] (M.S. Yeo). 0360-1323/$ e see front matter Ó 2011 Elsevier Ltd. All rights reserved. doi:10.1016/j.buildenv.2011.04.012

simulation [12] that interconnects the different physical domains. This approach makes it possible to consider the interaction between building-side physics (room air temperature, heat output from the floor surface, and so on) and system-side physics (water flow rate, water temperature, pressure, and so on) in a building with hydronic networks. These studies investigated the trend of building simulations and laid an emphasis on the necessity of the integrated or coupled simulation. It is thought that these studies also presented the direction for the simulation of combined heat and fluid flow in a building and system context. However, there have been few studies on the integrated simulation method to analyze the coupled hydronic-thermal physics in hydronic radiant heating systems. Gamberi et al. [13] developed a hydraulic-thermal approach to simulate the dynamic behavior of water flow rates, water temperatures, and room air temperatures in a building with radiator heating system. Thermal quantities of a heating system were determined by applying the result of a hydronic system simulation. Although the effects of the hydronic network on thermal output (water and air temperature) were taken into account, the change of flow rate or pressure in the hydronic network due to the on/off of the control valve, which is caused by the change of room air

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

temperature, was not considered. Xu et al. [14] developed an integrated model for simulating the thermal and hydronic behavior of space heating systems with radiators in a multi-family building. By utilizing a Thermostatic Radiator Valve (TRV) model as a bridge between two parts, the interaction between the thermal and hydronic model was simulated more efficiently. Thus, hydronic interactions between consumers, or the variations of flow rate and pressure in the hydronic network due to the TRV on/off, were considered in the simulation. However, the simulation was developed under the assumption that pressure difference at the building entrance is constant and equivalent to the constant pump head. In addition, the TRV model was represented with a characteristic curve, based on the specific condition of EN 215 (European standard on thermostatic radiator valves). As a result, the effects of the inherent characteristics of the hydronic components on the performance were not considered in the simulation. On the other hand, Franco et al. [15] adopted real hydronic components such as pipe, pumps and mixing valves in order to examine the influence of network layout and piping arrangement on the stability of the thermal control system. However, building models were simply replaced by shell-tube heat exchangers to simulate heating or cooling load, making it difficult to analyze detailed building physics such as room air temperature, floor surface temperature, heat output from the floor surface and so on. Yang et al. [16] proposed a Simulink-based simulation for analyzing the thermal performance of hydronic radiant heating systems. Thermal models, including a floor heating module, were developed based on dynamic heat balance equations and a hydronic network was represented with a steady-state equation to calculate water flow rates. The thermal and hydronic model run in a sequential manner, in which thermal model receives flow rates from the hydronic model, while the hydronic model receives on/off status from the thermal model. Nonetheless, there is still the problem of analyzing the impact of system hardware on the hydronic network and thermal performance. Through the review of the previous studies, it can be found that hydronic networks are numerically analyzed and coupled with thermal models in most thermal-hydronic integrated simulations. In this approach, algebraic loops will be created when hydronic networks are solved by an iterative numerical method because pressure loss has a nonlinear relation with flow rate in the network. This iterative procedure may slow down simulation speed under certain conditions and prevent the simulation from converging if an initial condition is not given plausibly [17e19]. As Zupancic et al. [20,21] argued, the simulation can be risky by an algebraic loop. This problem may result in the decrease of the robustness of the simulation. In addition, the lack of information on hydronic components can increase the uncertainty of the simulation [22]. The characteristics of hydronic components, for example, a pump curve, friction coefficient of internal flow, pressure drop across a valve under various flow rates, and so on, are difficult to model because most manufacturers tend to provide simple specifications rather than detailed performance data [23]. Hence, this study aims at establishing a method to improve thermal-hydronic integrated simulation by mitigating the uncertainty of the hydronic network. To do this, hydronic networks were physically represented in the simulation in order to avoid algebraic loops. In other words, real hydronic components were included in the simulation so that hydronic characteristics could be considered explicitly in the analysis of the control performance of the hydronic radiant heating system. This approach can be regarded as an emulation, especially Hardware-In-the-Loop-Simulation (HILS) in that some part of the simulation is replaced by real hardware [24e27]. In this study, an emulation method appropriate for the analysis of the hydronic radiant heating system was suggested.

2013

Based on the emulation method, the control performance according to control strategy was evaluated and compared. 2. Emulator development From the viewpoint of a hydronic system, a hydronic radiant heating system can be schematized as described in Fig. 1. The flow rate and pressure loss (head) of the system are solved by searching for the intersection of a pump curve and system curve [28]. In general, both curves can be approximated as quadratic equations and the solution can be obtained by solving the following equations together.

Dhpump ¼ aQT2 þ bQT þ c ½kPa

(1)

Dhsystem ¼ pQT2 ½kPa

(2)

where Dhpump ¼ pump head [kPa] Dhsystem ¼ pressure difference between supply and return junction [kPa] a, b, c ¼ coefficients of pump curve [e] QT ¼ total flow rate [m3/s] p ¼ coefficient of system curve [e] Since the pump and the pressure difference between supply and return junction is equal at the pump operating point, assuming Equations (1) and (2) is equal yields the total flow rate and pressure difference between the supply and return junction. The total flow rate is then distributed to each circuit according to the hydraulic resistance of each circuit. It should be noted that coefficients for the pump curve, a, b, and c, are constant, but the coefficient of the system curve, p, is dependent on the flow rate in the network when a fixed speed pump is applied to the variable flow system where the system demand varies with heating loads. Especially, the pressure distribution and flow rate are influenced by the on/off operation of the control valve, which supplies hot water to each thermal zone. Thus, Equation (2) needs to be modified each time the heating demands of the thermal zones are changed. As discussed earlier, the equation for the hydronic network contains an algebraic loop due to the non-linear relation between

Fig. 1. Conceptual diagram of a hydronic network.

2014

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

Fig. 2. Concept of an emulation for a hydronic radiant heating system.

pressure loss and flow rate, which is solved by a numerical method such as the NewtoneRaphson Method [13]. The reliability of this solution is highly dependent on the plausible initial condition. Although the initial condition is provided appropriately, the condition may not be valid if the equation is modified due to the change of heating demands. In addition, the reliability of the result is also affected by the input parameters of hydronic components such as control valves, balancing valves, distribution manifold, and circulation pump and so on. However, considerable effort is required to

obtain enough information on the parameters because the information is usually limited to a technical specification, which is measured under the specific condition. For example, the flow rate through a balancing valve cannot be obtained from the flow coefficient (Kv) data if the pressure drop across the valve differs from the standard condition. In order to evaluate the control performance of the hydronic radiant heating system, the interaction between the building thermal zone and hydronic network should be taken into account. Thermal-hydronic integrated simulation can be an alternative to include the interaction in the evaluation. However, this simulation method needs to be improved by mitigating an ‘initial condition’ problem of the hydronic network and enhancing the reality of hydronic components. To do this, this study adopted an emulation method where the hydronic network was replaced by real hardware, while the thermal zones of a building were represented by numerical simulation. The concept of the emulation is illustrated in Fig. 2. A distribution manifold, a circulation pump, a heat source (gas-fired boiler), balancing valves and control valves were used to constitute a hydronic network. For the purpose of compact and flexible configuration, an embedded pipe of the hydronic radiant heating system was replaced by a balancing valve, which has the equivalent hydraulic resistance as the original embedded pipe. Plate heat exchangers were used to simulate heat output from the floor surface of the hydronic radiant heating system. On the other hand, building simulation was developed by applying the heat balance method and programmed with MATLAB/Simulink [29]. The hydronic network and building simulation were then interconnected by a data acquisition and control device. The developed emulator is shown in Fig. 3 and information on the components is summarized in Table 1. At every time step of the simulation, the data acquisition and control device acquires measured data (e.g. water temperature, water flow rate) from the hydronic network and transfers the data to the computer simulation. Based on these measured data, the simulation calculates room air temperature, heating demand, and so on, and control signals are then transferred to valve actuators or a heat source in the emulator. An emulation is conducted by repeating the abovementioned process of measurement, data transfer, simulation, and control signal transfer.

Fig. 3. The developed emulator for hardware-in-the-loop simulation.

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

2015

Table 1 Information on the components in the emulator. Components Hardware

Panel

Distribution system

Control system Heat source

Interface

Sensor

DAQ device

Software

Item

Specification

Balancing valve Heat exchanger

Kv: w45, opening rate 0e100% (1% per step) Capacity: 11,667W (10,000 kcal/h) Primary: 80  C/70  C/secondary: 50  C/60  C DN20, with actuators 5 circuits, with shut off & balancing valve 20 A 5e250 kPa (0.05e2.5 bar)

2 way valve Manifold Flow limit valve Pressure differential control valve Circulation pump Actuator Valve controller Gas fired boiler Electric heater Chiller Water bath Flow meter Flow meter Pressure gauge Temperature sensor DAQ board D/A converter A/D converter Relay controller Simulation program DAQ program Operating system

3. Emulation method 3.1. Representation of hydronic network Previous studies indicate that a hydronic radiant heating system for a multi-zone building can be operated efficiently by controlling hot water flow rates [6,30]. Thus, hot water flow rates to each room should be known to evaluate the control performance of the system. As flow rates are determined by hydraulic resistance or pressure loss in the piping, it is essential to represent pressure loss in the hydronic network when evaluating the control performance based on the emulation method. Pressure loss in a hydraulic circuit is dependent on pipe length, diameter, fittings, piping layout, as described in Equation (3).


 f $L X 8 þ K $ 2 4 $Q 2 ½kPa D gp D

Dh ¼ 10$

(3)

where Δh: pressure loss [kPa] f: Darcy-Weisbach friction factor [e] L: pipe length [m] D: pipe internal diameter [m] K: minor loss coefficient [e] Q: flow rate [m3/s]

Flowrate: 5.6  104 m3/s (33 lpm), Head: 260 kPa (26 mAq) On/off type (24 V) Valve and boiler control relays þ transformer Capacity: 29,167 W (25,000 kcal/h) Capacity: 12,000 W Capacity: 52,500 W (45,000 kcal/h) 0.6 m3 (600 L) Volumetric type, 0e1.25  104 m3/s (0e7.5 lpm) Turbine type, with flow indicator Bourdon type T-type thermocouple NI SCXI-1600 NI SCXI-1124 þ NI SCXI-1325 NI SCXI-1102 þ NI SCXI-1303 NI SCXI-1160 þ NI SCXI-1324 MATLAB 2009a LabVIEW 8.5 Window XP

In this study, pressure loss in a hydraulic circuit was represented using the pressure drop across a balancing valve installed in the emulator. According to the definition of valve flow coefficient, the relation between pressure drop and flow rate of a balancing valve can be expressed by the following equation.

Kv ¼ 0:087Q

sffiffiffiffiffiffi sg

Dp

½e

(4)

where Kv : valve flow coefficient [e]. Q : flow rate [m3/s] sg: specific gravity of hot water [e]. Δp: pressure drop [Pa] Thus, the pressure loss can be represented with a balancing valve by using the following equation.

Dp ¼

a2 Kv2

Q 2 ½kPa

(5)

where a ¼ 0.087 It should be noted that the pressure loss in the real circuit can be approximated as the sum of the pressure drop across a balancing valve and the pipe section of the emulator. The pressure drop across

Table 2 Information on the panel to be represented by the emulator. Pipe layout

Variables

Quantity

Pipe length [m] Number of 90 bends Number of 180 bends Control valve Kv (when fully opened) Balancing valve Kv (when fully opened) Heating load [W] Design flow rate [m3/s ] Supply water temperature [ C] Pressure loss in pipe section [kPa] Pressure loss due to minor loss [kPa] Sum of pressure loss [kPa]

74.3 2 15 133 45 1040 2.5  105 (1.5 lpm) 60 7.21 (0.721 mAq) 1.13 (0.113 mAq) 8.34 (0.834 mAq)

2016

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

Fig. 4. Represented pressure loss by the emulator.

the pipe section is composed of the pressure drop in the polybutylene (PB) pipe, fittings, a heat exchanger, a flow meter and so on. Designating this pressure drop as Δh0,n, the pressure loss of the n-th circuit in a multi-zone building can be represented by the following equation.


a2 fn Ln X 8 10$ þ Kn $ 2 4 $Qn2 ¼ Dh0;n þ 2 Qn2 Dn Kvn g p Dn

(6)

The left-hand side of Equation (6) refers to the pressure loss of the real circuit, the right-hand side refers to the represented pressure loss, which includes the pressure loss in the pipe section (first term) and a balancing valve (second term). As the emulator is composed of a relatively short length of PB pipe and a small number of fittings and pipe bends as shown in Fig. 2 and Fig. 3, a balancing valve was used to account for most of the pressure loss in representing a real circuit. In other words, the Kv value of the balancing valve was adjusted so that Equation (6) can be satisfied. To do this, the opening rate of the balancing valve was determined according to the pressure loss of the real circuit. In order to examine the representation method, the pressure loss of a hydronic circuit, as described in Table 2, was represented using the emulator. The expected pressure loss of the circuit is 8.34 kPa (0.834 mAq) when the design flow rate is 2.5  105m3/s (1.5 lpm). Under this condition, the Kv value to satisfy Equation (6) was calculated as approximately 10. Based on the flow coefficient chart from the manufacturer, the opening rate of the balancing valve was then adjusted to attain the Kv value. With this configuration, the measured pressure loss of the represented circuit in the emulator was 8.50 kPa (0.85 mAq), which means that the pressure loss of the real circuit can be represented successfully by the emulator.

Fig. 6. Speed difference between the simulation and real hardware. (a) when the speed difference is not reconciled. (b) when the speed difference is reconciled.

In addition, in a variable flow system such as usual hydronic radiant heating system, the flow rate of a circuit is affected by the on/off status of other zones because the valve on/off changes pressure distribution in the hydronic network. To examine whether the pressure loss is represented well under the various on/off conditions of other zones, pressure loss was measured by applying all of the on/off combinations that can be expected in a building with 4 thermal zones. The result of the measurement is shown in Fig. 4. A solid line refers to the expected pressure loss in a real circuit (reference pressure loss), while the circle refers to the measured pressure loss represented by the emulator. As depicted in the figure, it can be seen that the pressure loss is represented well enough under the variation of the flow rate. 3.2. Connecting control system The emulator in this study was equipped with real control systems such as control valves and a valve controller. Control valves were actuated by a valve controller, which receives on/off status from building simulation and transfers control signals to control valves and a heat source. Control systems were linked to building simulation using LabVIEW data acquisition and control devices (NI

Fig. 5. Data transfer in the process of the developed emulation.

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022 Table 3 Control strategies for the evaluation. Strategy Strategy Strategy Strategy Strategy Strategy Strategy

1 2 3 4 5 6

Control reference

Flow rate balancing

Hydronic control device

Central Central Individual Individual Individual Individual

No balancing Balancing No balancing Balancing Balancing Balancing

None None None None Flow limit valve Pressure differential control valve

SCXI-1600, -1124, 1103, 1160). For the purpose of efficient data exchange, Simulink-based building simulation models were transformed to the LabVIEW-based code using the LabVIEW Control Design and Simulation Toolkit. Fig. 5 shows an emulation process from the viewpoint of data exchange between the control system and building simulation model. Control systems affect the hydronic radiant heating system by changing the hot water flow rate, which is a major parameter of building simulation. In general, the flow rate is maintained steadily except when the control valve is operating (opening or closing). In this steady state, the flow rate measured from the emulator can be used directly for the simulation. However, the flow rate increases or decreases gradually until the control valve is fully opened or closed because the valve has its own opening/closing time. During the opening/closing time of the valve, the direct use of measured flow rate for the simulation can degrade the accuracy of the analysis because the simulation proceeds much faster than the actuation of the real control valve. For example, if the valve has 30 seconds of opening/closing time and the simulation runs 10 times faster than in real world, the simulation will consider the opening/closing time as 300 seconds. As a result, the building will be analyzed with much less flow rate (for opening operation) or much more flow rate (for closing operation) than the real condition, as illustrated in Fig. 6(a). Moreover,

2017

the valve may repeat opening or closing continually if the on/off cycle analyzed by the simulation is shorter than the actual opening/ closing time of the valve. Thus, how the opening/closing operation of the valve is to be dealt with in the simulation needs to be considered. In general, the hydronic radiant heating system has so much thermal mass that the effect of the system operation for a short time span can be attenuated in the thermal analysis. In addition, compared with the overall time of the system operation, the opening/closing time of a general control valve is so short that the effect of flow rate change during the valve operation can be negligible. Therefore, this study adopted an emulation method in which the simulation pauses when the valve starts to operate and the simulation restarts when the valve finishes operating, as illustrated in Fig. 6(b). It is expected that this method can reduce the time required for the real time simulation as well as reconcile the speed difference between the simulation and real hardware. To realize this concept, the status of the control valve was checked at every time step. If the status differs from that of the previous time step, data acquisition was postponed in order to pause the simulation. After the opening/closing time of the valve, data acquisition was resumed to restart the simulation. This strategy was programmed using the LabVIEW Control Design and Simulation Toolkit. 4. Evaluation of control performance 4.1. Control strategies Although there are many control strategies for hydronic radiant heating systems, it is more practical to apply flow rate control using on/off valves due to the advantages in installation cost, maintenance and operation. This control strategy can be classified into central and individual control according to the reference point by which the supply of hot water is controlled. Each control strategy can be combined with hydronic balancing which can mitigate the imbalance

Fig. 7. Horizontal section of the residential building to be evaluated.

2018

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

Table 4 System information of the evaluated building. Heat source

Type capacity

Hydronic network Design flow rate [m3/s] Pipe length [m] Pressure loss [kPa]

Gas-fired boiler 30,000W

Head [kPa] flow rate [m3/s]

130 (13 mAq) 1.7  104(10.4 lpm)

Room1

Room2

Room3

Room4

2.2  105 (1.30 lpm) 83 2.20 (0.220 mAq)

3.0  105 (1.80 lpm) 85 2.52 (0.252 mAq)

4.2  105 (2.50 lpm) 93 3.25 (0.325 mAq)

2.4  105 (1.44 lpm) 86 2.65 (0.265 mAq)

of flow rate due to the difference in hydraulic resistance of each circuit [31]. In addition, the control performance can be degraded by an excessive flow rate due to the increased pressure difference between the inlet and outlet in the hydronic circuit, especially under part load

conditions. To prevent this problem, a control strategy can be complemented with the hydronic control device such as a flow limit valve (FLV) and a pressure differential control valve (PDCV), which are designed to prevent excessive flow rate.

Fig. 8. Emulation results for control performance of each control strategy. (a) Strategy 1: central þ no balancing. (b) Strategy 2: central þ balancing. (c) Strategy 3: individual þ no balancing. (d) Strategy 4: individual þ balancing. (e) Strategy 5: individual þ balancing þ FLV. (f) Strategy 6: individual þ balancing þ PDCV.

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

2019

In order to consider the above-mentioned control approach, 6 control strategies were derived as described in Table 3 and the control performance of each strategy was evaluated using the emulation. The evaluation was carried out for a conventional residential building in Seoul, Korea, which is composed of 4 thermal zones and hydronic circuits as depicted in Fig. 7. Information on heating load, design flow rate, hydronic network and heat source is summarized in Table 4. 4.2. Performance index In this study, two indices for control performance were analyzed: accuracy and rise time. Firstly, since the primary objective of control is to maintain the controlled variable (e.g. room air temperature) at set-point as closely as possible, the control accuracy was selected as the first performance index. This was quantified with an averaged room air temperature with regard to each room. Secondly, the control needs to ensure that a controlled variable approaches the set-point as fast as possible. In order to evaluate this performance, the rise time of room air temperature was selected as the second performance index and analyzed using emulation results. In general, the rise time is defined as the time required for a signal to change from a specified low value to a specified high value. In this study, rise time was defined as the time required for room air temperature to change from initial temperature (0  C) to set-point temperature (20  C). 4.3. Emulation results

Fig. 9. Total flow rate according to heating demand. (a) Strategy 4. (b) Strategy 5. (c) Strategy 6.

Fig. 8 shows the time-variation of flow rate and room air temperature with 6 control strategies. For Strategies 1 and 2, a room thermostat was assumed to be installed in Room 3 (living room) in order to realize a central control, which is still the most common control strategy for residential buildings in Korea. It can be found that air temperatures of all rooms swing simultaneously because all control valves are operated by the signal from Room 3. In the case of Strategy 1, the hydraulic resistance of each room was not balanced, therefore each room is provided with a similar flow rate regardless of the required flow rate. Since Room 3 has the largest heating load and pressure loss in the hydronic circuit, Rooms 1, 2 and 4 are provided with more flow rate than necessary; however, Room 3 can suffer from a relatively insufficient flow rate. As a result, Room 3 is the slowest to approach set-point temperature, while the room air temperatures of other rooms exceed the set-point temperature. On the other hand, a central control with hydronic balancing (Strategy 2) proved to mitigate the uneven heating because each room was provided with the required flow rate by hydronic balancing, as plotted in Fig. 8(b). However, there is still an imbalance of air temperatures between each room because every room is controlled by the air temperature of Room 3. Fig. 8(c) shows the result of the emulation in which an individual control is applied but hydronic balancing is not conducted (Strategy 3). Compared with Strategies 1 and 2, Strategy 3 resulted in the prevention of the overheating problem due to unnecessary heating. However, room air temperatures rise at different speeds because each room is not provided with the proper flow rate due to an unbalanced hydronic network. As a result, this control strategy may cause the uneven heating problem, especially in the start-up period. Strategy 4, an individual control combined with hydronic balancing, can result in the entire building being maintained around the set-temperature as plotted in Fig. 8(d). In addition, the temperature imbalance in the start-up period could be prevented by balanced flow rates. Hence, in order to improve the control performance of the hydronic radiant heating system, an individual control needs to be complemented with hydronic balancing. Emulation results also indicate that hydronic control devices can affect the temperature control performance as well as the hydronic

2020

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

Fig. 10. Comparison of the control accuracy.

control performance as shown in Fig. 8(e) and (f). The application of Strategy 5, where an FLV was added to Strategy 4, resulted in a similar pattern of room air temperatures as that of Strategy 4. However, compared with Strategy 4, the maximum total flow rate was reduced from 18.6105 m3/s (11.2 lpm) to 11.7105 m3/s (7.0 lpm) as described in Fig. 9(a) and (b). Even though the reduced flow rate meant that each room approached set-point temperature at a slower rate, the throttling range of room air temperatures was reduced and thus the control was more accurate. In the case of Strategy 6, where a PDCV was added to Strategy 4, the flow rate was varied almost proportionally to the number of heated rooms, or heating demand, as described in Fig. 9(c). Since a PDCV absorbs the increased pressure difference, which results in the excessive flow rate, it can supply proper flow rate even under part load conditions where only one room is heated. 5. Discussion For the evaluation of a thermal-hydronic coupled system such as a hydronic radiant heating system, it is required to model the hydronic components such as a balancing valve, a flow limit valve, a pressure differential control valve. When modeling these components, it is common to be provided with the limited information, for

example, a flow rate passing through the valve under the fixed pressure condition. Thus it is difficult to evaluate the impact of such hydronic components on hydronic network and thermal performance under dynamic conditions with existing simulation methods. By applying the emulation suggested by this study, it could be found that the emulation can evaluate the effect of hydronic components on the hydronic network (flow rates and pressure) and the control performance under the dynamic conditions, which is difficult to be evaluated by a pure simulation. In addition, the emulation method made it possible to observe the operation of a valve controller, control valves, a heat source, which is actuated according to the change of the room air temperature. Thus it was also possible to examine the operation of system hardware while conducting the emulation. Based on the emulation results, the control performance of a hydronic radiant heating system can be evaluated more quantitatively. From the viewpoint of performance indices introduced in Chapter 4, the control performances according to control strategies can be compared as follows. Fig. 10 shows the control accuracy which is quantified with an average room air temperature. It can be found that the central control (Strategies 1 and 2) leads to overheating and thus its accuracy is deteriorated. If a thermostat is installed in a room with a smaller heating load than the living room (e.g. Room 1 in Fig. 7), the accuracy will also be deteriorated due to an under-heating problem. Although the accuracy of the central control can be improved considerably by applying hydronic balancing (Strategy 2), more improvement cannot be expected than the individual control (Strategy 3 to 6). It should be noted that hydronic balancing also contributes to the improvement of the accuracy of the individual control (compare Strategies 3 and 4). The control can be more accurate by applying a hydronic control device such as an FLV or a PDCV; however, the extent of the improvement is not significant (Strategies 5 and 6). In terms of the rise time, control strategies without hydronic balancing (Strategies 1 and 3) caused Room 3, the hydraulically unfavoured room, to approach set-point temperature much later than other rooms as described in Fig. 11. This is because Room 3 is provided with a smaller flow rate than required but other rooms are provided with surplus flow rates due to the unbalanced hydronic network. On the other hand, the application of hydronic balancing (Strategies 2 and 4) reduces the rise time of the hydraulically unfavoured room. However, since a limited total flow rate is shared by each circuit in a hydronic network, flow rates to other rooms will decrease and the rise time of other rooms will consequently

Fig. 11. Comparison of the rise time.

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

increase. Thus, in terms of the rise time, hydronic balancing has an advantage in that it prevents a hydraulically unfavoured room from being heated too slowly but it is not advantageous in that the overall rise time of a building may increase slightly. Even though strategies 5 and 6 were very beneficial for the accuracy, the rise times of all rooms were increased because the total flow rate was reduced by an FLV or a PDVC. Based on the quantitative comparison using emulation results, some strategies to improve the control performance of a hydronic radiant heating system can be suggested as follows. Even when a central control is adopted due to the low cost and simple installation, hydronic balancing should be complemented to prevent the control performance from being deteriorated. If all performance indices are taken into account, the individual control with hydronic balancing (Strategy 4) can be the most appropriate alternative in the practical application. Although the application of hydronic control components (Strategies 5 and 6) can improve the accuracy of Strategy 4, the improvement is not considerable. However, these strategies (Strategies 5 and 6) can provide hydraulic or mechanical advantages such as low circulation power, the prevention of cavitation problems owing to the decreased flow rate and pressure difference under part load conditions, which can be schematized as Fig. 12. The performance improvement in this aspect needs to be explored further by additional studies. For example, a method to evaluate the overall energy performance including pump circulation power can be proposed by measuring the fuel consumption and electrical power directly from the emulator. In addition, by measuring the pressure drop across a balancing valve installed in

Fig. 12. Schematic diagram of the effect by a hydronic control strategy. (a) Decreased pressure difference (b) Decreased flow rate.

2021

the emulator, a control strategy or valve structure that is beneficial to the prevention of the cavitation can be suggested. As the accuracy is further enhanced, the performance with regard to the rise time tends to decline further. Thus, the importance of the performance index can be weighted by user’s preferences and different control strategies can be selected according to the preferences. Furthermore, a combined control strategy can be considered in order to optimize the control performance. For example, if the optimized strategy aims at enhancing the accuracy while reducing the rise time of a room, the individual control combined with dynamic balancing can be suggested [32]. With this strategy, the control valve is fully opened in order to supply maximum flow rate in the start-up period. If the room air temperature approaches setpoint temperature, the full-opened valve is returned to its initial position for the hydronic balancing. Lastly, the optimum control strategy can be established by analyzing the energy consumed by the heat source as well as the aforementioned control performance. Further studies are needed to determine the optimum strategy by analyzing the characteristics of the heat source, which is connected with building models by way of the emulation. 6. Conclusions In this study, an emulation method for a hydronic radiant heating system was developed using the concept of Hardware-In-the-Loop Simulation (HILS), where the hydronic network was replaced by real hardware and the building physics was analyzed by a Simulinkbased simulation. In order to represent the pressure loss and flow rate, which are essential factors for the control performance of a hydronic radiant heating system, the pipe section was replaced by a balancing valve, which has the same hydraulic resistance as the real hydronic circuit. In addition, a strategy to connect real control systems and virtual building models was suggested. With the developed emulation method, a thermal-hydraulic integrated problem in a hydronic radiant heating system could be solved successfully. It was possible to analyze the influence of hot water flow rate on the control of room air temperature, and vice versa. In addition, pressure loss and flow rate in the hydronic network could be analyzed without numerical simulation, which can result in an unreliable solution when the plausible initial condition is not given. Especially, the emulation made it possible to examine the impact of hydronic control devices such as a flow limit valve (FLV) and a pressure differential control valve (PDCV), which are difficult to model due to the lack of input parameters. The developed emulation was applied to the evaluation of control performance for a hydronic radiant heating system. Several control strategies were evaluated using performance indices of the accuracy and the rise time. Emulation results show that the central control results in the overheating problem and thus the accuracy can be highly deteriorated. However, the central control combined with hydronic balancing reduces the room air temperature differences and consequently improves the accuracy. Although the individual control proved to provide an accurate control, temperature differences between each room can increase when the hydronic balancing is not applied. Hence, it can be concluded that the individual control should be combined with hydronic balancing for more accurate control. It was also observed that hydronic control devices such as an FLV or a PDCV can improve the accuracy; however, the improvement is not considerable and the rise times of entire rooms tend to be increased. Since the devices are expected to have more advantages in the enhancement of hydraulic performance, rather than temperature control performance, some issues such as circulation power or cavitation problems need to be explored by further studies. In addition,

2022

K.N. Rhee et al. / Building and Environment 46 (2011) 2012e2022

the operation and energy consumption of the heat source should be considered in order to establish the optimum control strategy. By using the developed emulation method, further studies will be conducted to determine the optimum strategy that can enhance energy saving potential as well as control performance. Acknowledgement This work was supported by the Korea Science and Engineering Foundation (KOSEF) grant funded by the Korea government (MEST) (No. 2010-0013451s). References [1] Olesen B. Radiant floor heating in theory and practice. ASHRAE Journal 2002; 44(7):19e24. [2] Chen Q. Comfort and energy consumption analysis in buildings with radiant panels. Energy and Buildings 1990;14:287e97. [3] Laouadi A. Development of a radiant heating and cooling model for building energy simulation software. Building and Environment 2004;39:421e31. [4] Leigh, S. An experimental approach for evaluating control strategies of hydronic radiant floor heating systems. PhD thesis. Ann Arbor: The University of Michigan, 1991. [5] Gibbs D. Control of multizone hydronic radiant floor heating systems. ASHRAE Transactions 1994;100(1):1003e9. [6] Ryu S, Lim J, Yeo M, Kim K. A study on the control methods for radiant floor heating and cooling System in residential building. ASHRAE Transactions 2004;110(2):106e16. [7] Augenbroe G. Trends in building simulation. Building and Environment 2002; 37:891e902. [8] Citherleta S, Clarke J, Hand J. Integration in building physics simulation. Energy and Buildings 2001;33:451e61. [9] Felsmann C, Knabe G, Kremonke A, Perschk A. Application of simulation in design and operation of refurbished buildings and heating systems. Proceedings of IBPSA 1997;2:149e56. [10] Hensen J. Towards an integral approach of building and HVAC system. Energy and Buildings 1993;19:297e302. [11] Tr cka M, Hensen J. Overview of HAVC system simulation. Automation in Construction 2010;19(2):93e9. [12] Tr cka M, Hensen J, Wetter M. Co-simulation for performance prediction of integrated building and HVAC systems e an analysis of solution characteristics using a two-body system. Simulation Modelling and Practice and Theory 2010;18:957e70. [13] Gamberi M, Manzini R, Regattieri A. Simulink simulator for building hydronic heating systems using the NewtoneRaphson algorithm. Energy and Buildings 2009;41:848e55.

[14] Xu B, Fu L, Di H. Dynamic simulation of space heating systems with radiators controlled by TRVs in buildings. Energy and Buildings 2008;40:1755e64. [15] Franco W, Sen M, Yang K, McClain R. Dynamics of thermal-hydraulic network control strategies. Experimental Heat Transfer 2004;17:161e79. [16] Yang, K., Jung, H., Rhee, K., Ryu, S., Yeo, M., and Kim, K.A Study on the thermal performance simulation integrated with hydronic analysis in radiant floor heating System, Proceedings of The First International Conference on Building Energy and Environment 2008; 2060 e 2067 [17] Jreijiry D, Husaunndee A, Inard C, Villenave J. Control of ventilation in buildings using SIMBAD building and HVAC toolbox. Proceedings of IBPSA 2003;2:591e8. [18] Dabney J, Harman T. Mastering Simulink. Prentice Hall; 2003. 293e295. [19] Internet website http://www.mathworks.com/access/helpdesk/help/toolbox/ physmod/powersys/ug/f1-7758.html. [20] Zupan ci c B, Sodja A. Object oriented modelling of variable envelope properties in buildings. WSEAS Transactions on Systems and Control 2008;12(3): 1046e56. [21] Zupan ci c, B. and Sodja, A. Thermal conditions in buildings - advances of acausal OO modeling in comparison with traditional approach, Proceedings of the 7th WSEAS International Conference on System Science and Simulation in Engineering (ICOSSSE’08) 2008: 173 e 178. [22] Wetter MA. Modelica-based model library for building energy and control systems. Proceedings of IBPSA 2009;2:652e9. [23] Clarke J. Energy simulation in building design. 2nd ed. Butterworth-Heinemann; 2001. [24] Palla, S., Srivastava, A., and Schulz, N. Hardware in the loop test for relay model validation, Proceedings of the IEEE Electric Ship Technologies Symposium, 2007: 449 e 454 [25] Lahrech R, Gruber P, Riederer P, Tessier P, Visier J. Development of a testing method for control HVAC systems by emulation. Energy and Buildings 2002; 34:909e16. [26] da Silva P, Knabe G. LabHouse: system simulation and emulation within boiler development. Building Service Engineering Research and Technology 2003; 24(4):281e7. [27] Peitsman H, Wang S, Kaerki S, Park C, Haves P. The reproducibility of tests on energy management and control systems using building emulators. ASHRAE Transaction 1994;100(1):1455e64. [28] Fu Y, Wu K, Cai Y. Stability analysis of closed-loop water system. Journal of Thermal Science 2006;15(2):152e8. [29] Kalagasidis A, Weitzmann P, Nielsen T, Peuhkuri R, Hagentoft C, Rode C. The international building physics toolbox in Simulink. Energy and Buildings 2007;39:665e74. [30] Cho S, Zaheer-uddin M. An experimental study of multiple parameter switching control for radiant floor heating systems. Energy 1999;24:433e44. [31] Ryu S, Rhee K, Yeo M, Kim K. Strategies for flow rate balancing in radiant floor heating systems. Building Research and Information 2008;36(6): 625e37. [32] Rhee K, Ryu S, Yeo M, Kim K. Simulation study on hydronic balancing to improve individual room control for radiant floor heating system. Building Services Engineering Research and Technology 2010;31(1):57e73.