BUILD SIMUL (2015) 8: 543 – 550 DOI 10.1007/s12273-015-0236-5

Modeling of active beam units with Modelica

1. Aalborg University, Danish Building Research Institute, Denmark 2. Lindab A/S, Denmark

Abstract

Keywords

This paper proposes an active beam model suitable for building energy simulations with the programming language Modelica. The model encapsulates empirical equations derived by a novel active beam terminal unit that operates with low-temperature heating and high-temperature cooling systems. Measurements from a full-scale experiment are used to compare the thermal behavior of the active beam with the one predicted by simulations. The simulation results show that the model corresponds closely with the actual operation. The model predicts the outlet water temperature of the active beam with a maximum mean absolute error of 0.18 °C. In term of maximum mean absolute percentage error, simulation results differ by 0.9%. The methodology presented is general enough to be applied for modeling other active beam units.

active beams,

Research Article

Alessandro Maccarini1 (), Göran Hultmark2, Anders Vorre2, Alireza Afshari1, Niels C. Bergsøe1

Modelica, building energy simulation, modeling

Article History Received: 10 November 2014 Revised: 4 May 2015 Accepted: 18 May 2015 © Tsinghua University Press and Springer-Verlag Berlin Heidelberg 2015

1

Introduction

E-mail: alm@sbi.aau.dk

Building Systems and Components

Energy consumption in buildings is a large share of the world’s total end-use of energy. In member states of the European Union, residential and commercial buildings require approximately 40% of the end-use of energy (IEA 2013). Heating, ventilation and air-conditioning (HVAC) systems represent the largest energy end-use both in residential and commercial buildings, accounting for almost half the energy consumed in buildings (Pérez-Lombard et al. 2008). Most of this energy is used for maintaining a room temperature of about 20–25 °C, which is close to ambient conditions and therefore requires a low content of exergy. However, in most cases HVAC systems supply energy to buildings by high quality energy sources, such as fossil fuels (high exergy systems). Extensive usage of fossil fuels causes several environmental and health issues, such as global warming, pollution and depletion of fossil natural resources. At present, several studies have been conducted on lowexergy systems for buildings (Hepbasli 2012; Balta et al. 2008; Asada and Boelman 2004; Hasan et al. 2009). By providing

heating and cooling energy at a temperature close to room temperature, low-exergy systems allow the use of low valued energy, which can be delivered by sustainable energy sources such as heat pumps, solar collectors, waste heat and energy storage (Hepbasli 2012). Therefore, the use of low-exergy systems can reduce the environmental impact of buildings, and play a crucial role towards the requirements of nearly zero-energy buildings communities. In the context of lowexergy systems, active beams represent a valuable alternative to traditional air diffuser units. Active chilled beams (ACBs) have been used for more than 20 years in Europe, mainly for cooling purpose, and increasing interests in these systems have been found in North America and Asia during the last decade. Currently, ACB systems are used for both cooling and heating of buildings and they can be simply called “active beams”. When comparing ACBs with traditional all-air variable air volume (VAV) systems, energy consumption can be reduced in various ways. First of all, ACBs decouple ventilation loads from space sensible loads, handling space cooling and outside air requirements without increasing airflow rate. Secondly, the use of chilled water at higher temperature (13 °C to 17 °C)

544

Maccarini et al. / Building Simulation / Vol. 8, No. 5

List of symbols A B, c1, c2 cp,a cp,w k L m ai m ap m ap,e m w p P

coil heat transfer area (m2) empirical coefficients specific heat of air (J/(kg·k)) specific heat of water (J/(kg·k)) coil heat transfer coefficient (W/(m2·K)) coil length (m) induced air mass flow rate (kg/s) primary air mass flow rate (kg/s) primary air mass flow rate per coil length (L/(s·m)) water mass flow rate (kg/s) pressure in the air plenum (Pa) active beam total capacity (W)

than conventional HVAC systems (4 °C to 7 °C) lead to higher efficiency of the cooling machine. Comparison reports of the energy performance of ACBs against conventional VAV systems show that the impact varies appreciably depending on the specific project and the climate location. Some studies claim that energy savings of 8%–20% can be achieved when ACBs are used (Kurt et al. 2007; Murphy 2009). Others argue that VAV systems can be more efficient than ACBs (Stein and Taylor 2013). In order to fully understand the performance of ACBs, several studied have been recently conducted. Loudermilk (2009) provided some guidelines regarding effective design of humidity levels in rooms with low primary airflow rates. Koskela et al. (2010) and Cao et al. (2010) studied the air distribution in rooms where ACB units were mounted, by using various experimental and numerical methods. Chen et al. (2014b) analyzed the heat transfer rate of four different heat exchangers in ACB application. Fong and Lee (2014) investigated the integration of ACBs in a hybrid renewable cooling system for office buildings where both the solar energy and the ground source were used. Chen et al. (2014a) proposed a hybrid dynamic model of active chilled terminal unit for real-time control and optimization applications. Livchak and Lowell (2012) showed a system of equations that can be implemented in building performance simulation (BPS) tools to describe the thermal behavior of active beam units. The usage of BPS tools represents a powerful method to analyze the performance of active beams in buildings. An exhaustive review on how active beams are modeled in existing BPS tools was conducted by Betz et al. (2012). EnergyPlus includes an empirical model developed by a manufacturer of ACBs. The model requires several inputs

Pa Pw Qmax Ta Ta,in Ta,out Tr Tw,avg Tw,in Tw,out ΔTw ε

active beam capacity provided by primary air (W) active beam capacity provided by coil (W) maximum heat transfer between two fluids primary air temperature (°C) induced air temperature entering the coil (°C) induced air temperature leaving the coil (°C) room air temperature (°C) water average temperature (°C) inlet water temperature entering the coil (°C) water temperature leaving the coil (°C) water temperature difference (°C) constant effectiveness

coefficients and it calculates the capacity of the chilled beam as a function of primary airflow rate, induction ratio, water flow rate and temperature difference between room air and water (DOE 2014). Likewise, the chilled beam component in IDA-ICE is an empirical model and it depends on two coefficients: the beam conductivity k and an exponent n. These parameters are given as functions of airflow rate or based on two user supplied performance points, at design conditions and at zero flow (EQUA Simulation AB 2013). The TRNSYS chilled beam model, like the EnergyPlus and IDA-ICE models, is an empirical model based on input coefficients. It also includes a bypass factor that regulates a portion of induced air bypassed around the coil (Betz et al. 2012). Other simulation tools such as eQuest contain no active beam models. They usually approximate active beams with induction units or fan coil objects without fan energy. It can be concluded that active beam models in BPS tools are mainly based on empirical coefficients delivered by manufacturers. The accuracy of the active beam performance is therefore significantly influenced by user inputs parameters. However, there is no study in the existing literature that compares simulation predictions of active beam models suitable for BPS tools with experimental data. In addition, no active beam model has been developed yet for Modelicabased programs. The objective of this paper is therefore to develop a model able to predict the thermal behavior of an active beam unit with Modelica and compare simulation results with experimental data. The article is organized as follow: Section 2 gives an introduction to Modelica as computational tool for building energy simulations; Section 3 describes the mathematical model developed using Dymola environment; information regarding the experimental set-up is provided in Section 4;

545

Maccarini et al. / Building Simulation / Vol. 8, No. 5

Section 5 discusses results regarding the comparison between simulations and experimental data; Section 6 concludes the article. 2

Modelica

Computer modeling and simulation is a powerful technology for calculating energy performance in buildings. For the past 50 years, a variety of BPS simulation tools have been developed and used by the building energy research community. These tools perform simulations and calculate the annual energy consumption of a building by solving a system of equations that describes the thermal behavior of envelope and HVAC systems. Climate, schedules of operation and internal loads are the boundary conditions of simulations (Trcka and Hensen 2010). Modelica, developed by the Modelica Association, is a freely available, object-oriented equation-based language for modeling of large, complex, and heterogeneous physical systems. It has been used for almost two decades especially in the design of multi-domain engineering systems such as mechatronic, automotive and aerospace applications involving mechanical, electrical, hydraulic and control subsystems. Only recently, because of the upcoming need for more complex and efficient energy systems, the use of Modelica has increased also in the building energy research community. Currently, several Modelica libraries for building components and HVAC systems exist and are continuously upgraded (Wetter et al. 2013; Van Roy et al. 2013; RWTHEBC 2014; Nytsch-Geusen et al. 2013). Moreover, the International Energy Agency has recently launched a large scale international project (IEA ECB Annex 60) with the aim to develop the new generation of computational tools for building energy systems with Modelica. Therefore, the development of active beam models suitable for Modelica-based programs represents a way to further widen the capabilities of Modelica in simulating innovative heating and cooling systems for buildings. 3 Description of the mathematical model and implementation into Modelica Schematic diagram of a general active beam unit is given in Fig. 1. It consists of a primary air plenum, a mixing chamber, a heat exchanger (coil) and several nozzles. The heat exchanger is served by a water circuit. The primary air is discharged to the mixing chamber through the nozzles. This generates a low-pressure region which induced air from the room up through the coil. The conditioned induced air is then mixed with primary air, and the mixture descent back to the space. A system of equations describing the heat

Fig. 1 Diagram of an active beam unit

transfer behavior of active beams suitable for BPS tools was given by Livchak and Lowell (2012). The total capacity of an active beam unit is the sum of capacities provided by the primary air and the water: P = Pa + Pw

(1)

The following equation calculates the capacity provided by the primary air: Pa = m ap cp,a ( Ta - Tr )

(2)

Heat transfer through the coil is described by the following system of equations under steady-state conditions and assuming no condensation on the coil surface: Pw = m w cp,w ( Tw,out - Tw,in )

(3)

Pw = kA( Tr - Tw,avg )

(4)

Pw = m ai cp,a ( Ta,in - Ta,out )

(5)

It is noted that two variables must be defined: k and m ai . For a given active beam unit, the heat transfer coefficient k is a function of several parameters including primary air and water mass flow rate, and water and air temperature. The empirical equation for k presented by Livchak and Lowell (2012) depends on six coefficients that, according to the author, can be derived from manufacturer’s capacity tests. However, a survey of performance data sheets of active beams products showed that manufacturers do not provide such coefficients. Typically, manufacturer’s data sheets only describe the capacity of the coil as a function of primary airflow rate for different levels of pressure in the plenum. Correction factors can be found to take into consideration the influence of water flow rate and water temperature difference. An alternative empirical equation to Livchak and Lowell (2012) for the heat transfer coefficient k is provided in this work. The presented equation is based on coefficients derived

546

Maccarini et al. / Building Simulation / Vol. 8, No. 5

from the performance data sheet of a specific active beam unit. However, other active beam units can be modeled by adjusting the equation with coefficients directly acquired from their respective performance data sheets. The following equation describes the coefficient k: k = Bc1c2

L A

(6)

The parameter B is a cubic polynomial function of the primary air mass flow rate per active length. Five different polynomials were developed for five different levels of pressure in the primary air plenum. Linear interpolation was assumed between the five polynomials. Table 1 shows the polynomials. Modifier c1 is a cubic polynomial function of the water temperature difference between outlet and inlet. Modifier c2 is a quartic polynomial function of the water mass flow rate. Table 2 displays c1 and c2. Length and area of the active beam are represented respectively by L and A. The induced air m ai is expressed by the following equation as function of pressure, primary air mass flow rate and primary air mass flow rate per coil length: m ai = 1.05 ´( 0.047 p + 20.043 )m ap,e-0.635 ´ m ap

(7)

The active beam model was built by using base elements of the Modelica Standard Library (MSL) and the Modelica Buildings Library (Wetter 2013). Figure 2 shows the graphic layout of the model developed with Dymola, a commercial simulation environment based on the Modelica modeling language. To generate the induction effect, air from the room is ideally moved towards the heat exchanger by the element Pipe. This component requires the amount of induced air mass flow rate calculated by the element Induction, where Eq. (7) is encapsulated. Pressure level and primary air mass

Fig. 2 Graphic layout of the active beam model developed with Dymola

flow rate are connected to this block as inputs. Water and induced air exchange heat through the component Coil, where the heat transfer model expressed by Eqs. (4) and (6) is implemented. To solve these equations three external inputs are needed: pressure level, water mass flow rate and primary air mass flow rate. Two sensors provide the value of water and air mass flow rate directly from the fluid streams while the pressure level is a design condition (depending on nozzles configuration) supplied by an external input. The temperature of water and air entering the heat exchanger is automatically detected by the inlet fluid ports, therefore no additional sensors are needed. The element Chamber mixes primary and induced air and delivers the flow to the room. Table 3 shows the base class of each element included in the model. Table 3 Base class of elements included in the model Name

Table 1 Equations for coefficient B Pressure

Equation

Library

Class

Pipe

Buildings Buildings.Fluid.Movers.BaseClasses.IdealSource

Induction

MSL

Modelica.Blocks.Interfaces.RealInput/Output

Chamber

Buildings

Buildings.Fluid.FixedResistances.SplitterFixedResistanceDpM Buildings.Fluid.HeatExchangers.ConstantEffectiveness

40 Pa

( 0.0005m ap,e - 0.0771m ap,e + 4.0232m ap,e + 9.0793 )´1.1329

60 Pa

( 0.0005m ap,e - 0.0782m ap,e + 4.0232m ap,e + 12.738 )´ 1.1329

80 Pa

( 0.0005m ap,e3 - 0.0777m ap,e2 + 4.0373m ap,e + 15.616 )´ 1.1329

Coil

Buildings

100 Pa

( 0.0005m ap,e3 - 0.0767m ap,e2 + 3.9885m ap,e + 18.528 )´1.1329

SensorAir

Buildings Buildings.Fluid.Sensors.MassFlowRate

120 Pa

( 0.0005m ap,e3 - 0.0763m ap,e2 + 3.9765m ap,e + 20.796 )´ 1.1329

SensorWater Buildings Buildings.Fluid.Sensors.MassFlowRate

3 3

2

2

Table 2 Equations for coefficient c1 and c2 Coefficient

Equation

Pressure

MSL

Modelica.Blocks.Interfaces.RealInput

WaterIn

MSL

Modelica.Fluid.Interfaces.FluidPort_a

WaterOut

MSL

Modelica.Fluid.Interfaces.FluidPort_b

InducedAir

MSL

Modelica.Fluid.Interfaces.FluidPort_a

c1

-0.0014ΔTw 3 + 0.0154ΔTw 2 - 0.0248ΔTw + 0.9428

SupplyAir

MSL

Modelica.Fluid.Interfaces.FluidPort_b

c2

 w 3 + 540.08m w 2 + 9.5737m w + 0.4903 113612m w 4 - 15884m

PrimaryAir

MSL

Modelica.Fluid.Interfaces.FluidPort_a

547

Maccarini et al. / Building Simulation / Vol. 8, No. 5

As illustrated in Table 3, the element Coil is an instance of the class Buildings.Fluid.HeatExchangers.ConstantEffectiveness. This class was originally developed for transferring heat between two fluids in the amount of: Q = Qmax ε

(8)

where ε is a constant effectiveness and Qmax is the maximum heat that can be transferred. Therefore, in order to model the heat transfer occurring in the active beam coil, the Modelica code of this class was appropriately modified, and Eq. (8) was replaced by Eqs. (4) and (6).

conditions are displayed in Table 5 for each case study. Heating and cooling loads were delivered to the test room in order to vary room air temperature and mimic situations of cooling and heating demand (Table 6). Room temperature was measured with a sensor placed at 1.7 m height in the center of the room. Table 5 Experimental parameters setting as boundary conditions

Case 1

4 Experimental setup The experimental data collection was conducted at Lindab A/S laboratories in Farum, Denmark (Fig. 3). The measurements were used in this work to compare the thermal behavior of the active beam with the one predicted by the Modelica model. A Solus active beam unit (manufactured by Lindab A/S) was mounted on a test room 3 m length, 4 m width and 2.6 m height. Technical features of the Solus active beam used for experimental validation are described in Table 4. To validate the model, four case studies were performed. The operating values of the parameters adopted as boundary

Case 2

Case 3

Case 4

Parameter

Value

Pressure

±100 Pa

Inlet water temperature

±22.1 °C

Water mass flow rate

±0.038 kg/s

Primary air temperature

±22 °C

Primary air mass flow rate

±0.03 kg/s

Pressure

±60 Pa

Inlet water temperature

±23.9 °C

Water mass flow rate

±0.038 kg/s

Primary air temperature

±25 °C

Primary air mass flow rate

±0.03 kg/s

Pressure

±120 Pa

Inlet water temperature

±22.8 °C

Water mass flow rate

±0.038 kg/s

Primary air temperature

±23.5 °C

Primary air mass flow rate

±0.03 kg/s

Pressure

±120 Pa

Inlet water temperature

±23.8 °C

Water mass flow rate

±0.038 kg/s

Primary air temperature

±23.2 °C

Primary air mass flow rate

±0.03 kg/s

Table 6 Heat loads applied to the test room. Air temperature surrounding the test room was between 18 °C and 20 °C. Therefore, when no or low thermal loads were delivered, the system mainly operates in heating mode (depending on the water temperature in the circuit)

Case 1

Fig. 3 Test room for experimental data collection Table 4 Technical features of Solus unit Feature

Value

Length

3000 mm

Width

600 mm

Height

200 mm

Weight Active length Volume of mixing chamber Heat capacity of heat exchanger

35.6 kg 2800 mm 0.04 m3 6841 J/K

Case 2

Case 3

Case 4

Time (h)

Thermal loads (W)

Operating mode

0–3

0

Heating Heating

3–16

550

16–22

350

Cooling

22–24

0

Cooling

0–3.5

2200

Cooling

3.5–10

1400

Cooling

10–15

600

Heating

15–18

0

Heating

0–4

1400

Cooling

4–12

3000

Cooling

0–3

400

Heating

3–6.5

0

Heating

6.5–9.5

250

Heating

9.5–12

400

Heating

548

The Solus active beam unit is designed to be integrated into novel low-temperature heating and high-temperature cooling systems that present an extreme low temperature difference between water and room air. Therefore, the boundary conditions related to inlet water temperature and thermal loads were chosen with the main goal to maintain a temperature difference of about 0–1.5 °C between water and room air in both heating and cooling mode. Three different values of pressure level were selected in order to validate diverse polynomials shown in Table 1. The values of water and air mass flow rate were kept constant across the four case studies and chosen according to design conditions provided by the manufacturer. Various instruments were placed into the system in order to capture experimental data. Temperatures are measured with temperature resistance sensors connected to a data logger. The accuracy is 0.2 °C. Mass flow rates are determined by pressure transmitters with an accuracy of 2%. Pressure is measured by differential pressure transducers. The accuracy is between 0.5% and 4% depending on the level of pressure. Due to high oscillations in the measurements, signals were smoothed by using DIAdem. This program calculates the floating mean for each value from the channel value and from a specified number of neighbor values. A range of 10 neighbor values was applied. Data were recorded every 180 seconds. 5 Comparison with experimental results To reproduce the experiment, the model of the active beam developed with Modelica was subjected to the same input values. These values (Table 5) were set as boundary conditions for simulations. For this purpose, a new model was built in accordance to Fig. 4. This configuration consists of two sources providing air and water flows. The operating pressure level is applied through an appropriate input signal. The model accuracy was evaluated by comparing predicted and experimental outlet water temperature. Two methods were used for comparison. Graphical techniques allowed to view the qualitative correspondence of simulated and measured data. Statistical indicators were used to quantify the goodness-of-fit to the simulated results compared to the measured data. The absolute mean error (MAE) is the mean of the absolute values of the difference between predicted and measured value. MAE has the same unit of the value evaluated. Therefore, it can be directly compared with the accuracy of the instrument. However, the relative size of the error is not always obvious. To deal with this limitation, the mean absolute percentage error (MAPE) was also calculated. MAE and MAPE are given in Eqs. (9) and (10).

Maccarini et al. / Building Simulation / Vol. 8, No. 5

Fig. 4 Dymola model for simulation N

MAE = å i =1

vi - Vi N

100 vi - Vi N å Vi i =1

(9)

N

MAPE =

(10)

In Fig. 5, water outlet temperature is compared by depicting simulated and measured data vs. time for the four case studies. Estimated results match well with monitoring data and a similar qualitatively behavior is observed. Generally, simulations slightly overestimate the measured value of water outlet temperature. Only in the Case 3 it is shown an opposite trend where simulations underestimate real operation. The opposite trend can be explained by measurement inaccuracy. Figure 5 shows also room air temperature and water inlet temperature. Heating mode occurs when room air temperature is below supply water temperature. Cooling mode occurs when room air temperature is above supply water temperature. In Fig. 6, the absolute errors are shown vs. time. This graph highlights that the highest errors are related to the Case 1 when outlet water temperature is very close to inlet water temperature. This occurs when thermal loads are extremely low during 00:00–03:00 am, at 04.00 pm (transition phase between heating and cooling mode) and during 11:00–12:00 pm. The maximum absolute error assumes a value of 0.3 °C. Considering Cases 2, 3 and 4, absolute errors are always below the uncertainty of measurements (0.2 °C). Figure 7 illustrates the quality of the simulation results by depicting simulation data vs. experimental data of outlet water temperature. These graphs show the distribution of data around the identity line: the data above this line

549

Maccarini et al. / Building Simulation / Vol. 8, No. 5

overestimates the real operation and the data below the line underestimates the real operation. The maximum absolute percentage error is related to Case 1 and it assumes a value of 1.4%. Table 7 shows the overall results as determined for the experimental data and simulation predictions. It is noticed that the model predictions are very close to the monitoring data with a mean absolute error within the uncertainty of measurements for all case studies. Fig. 6 Outlet water temperature: absolute error vs. time

Fig. 5 Outlet water temperature: comparison between simulation and experimental results for Case 1 (a), Case 2 (b), Case 3 (c) and Case 4 (d)

Fig. 7 Outlet water temperature: simulated vs. experimental for Case 1 (a), Case 2 (b), Case 3 (c) and Case 4 (d)

550

Maccarini et al. / Building Simulation / Vol. 8, No. 5

Table 7 Accuracy of the simulation results: mean absolute error (MAE) and mean absolute percentage error (MAPE) for water outlet temperature Mean absolute error (MAE) (°C)

Maximum absolute error (°C)

Mean absolute percentage error (MAPE) (%)

Maximum absolute percentage error (%)

Case 1

0.18

0.3

0.9

1.4

Case 2

0.05

0.19

0.2

0.78

Case 3

0.12

0.2

0.51

0.82

Case 4

0.04

0.13

0.19

0.56

6 Conclusion A Modelica model that can be used to predict the thermal behavior of an active beam unit was developed. The new model was based on empirical equations provided by manufacturer data sheet. In this article a comparison between experimental and simulation results was presented with respect to alternative heating and cooling mode and for different levels of pressure. The results show a similar qualitative trend for the outlet water temperature. The model predicted the outlet water temperature with a maximum mean absolute error of 0.18 °C. In terms of relative error, the outlet water temperature was predicted with a maximum mean absolute percentage error of 0.9%. The model developed in this paper is suitable for building energy simulations based on Modelica. Whole building simulation analyses can be performed by integrating the active beam model into existing Modelica libraries. The described empirical equations refer to a specific active beam unit. However, the modeling methodology is general enough that different empirical equations can be implemented to create other active beam units. Acknowledgements The authors would like to thank Miroslav Dohnal and Daniel Bochen from Lindab A/S for their valuable help given during the work. References Asada H, Boelman EC (2004). Exergy analysis of a low temperature radiant heating system. Building Service Engineering Research & Techonlogy, 25: 197–209. Balta MT, Kalinci Y, Hepbasli A (2008). Evaluating a low exergy heating system from the power plant through the heat pump to the building envelope. Energy and Buildings, 40: 1799–1804. Betz F, McNeill J, Talbert B, Thimmanna H, Repka N (2012). Issues arising from the use of chilled beams in energy models. In: Proceedings of IBPSA-USA Conference (SimBuild 2012), Wisconsis, USA, pp. 655–667.

Cao G, Sivukari M, Kurnitski J, Ruponen M, Seppanen O (2010). Particle Image Velocimetry (PIV) application in the measurement of indoor air distribution by an active chilled beam. Building and Environment, 45: 1932–1940. Chen C, Cai W, Giridharan K, Wang Y (2014a). A hybrid dynamic modeling of active chilled beam terminal unit. Applied Energy, 128: 133–143. Chen C, Cai W, Wang Y, Lin C (2014). Performance comparison of heat exchangers with different circuitry arrangements for active chilled beam applications. Energy and Buildings, 79: 164–172. DOE (2014). EnergyPlus 8.1 Input Output Reference. Department of Energy (DOE). Available at http://apps1.eere.energy.gov/buildings/ energyplus/pdfs/inputoutputreference.pdf. EQUA Simulation AB (2013). IDA Indoor Climate and Energy 4.5 User Manual. Available at http://www.equaonline.com/iceuser/ pdf/ICE45eng.pdf. Fong KF, Lee CK (2014). Investigation on hybrid system design of renewable cooling for office building in hot and humid climate. Energy and Buildings, 75: 1–9. Hasan A, Kurnitski J, Jokiranta K (2009). A combined low temperature water heating system consisting of radiators and floor heating. Energy and Buildings, 41: 470–479. Hepbasli A (2012). Low exergy (LowEx) heating and cooling systems for sustainable buildings and societies. Renewable and Sustainable Energy Reviews, 16: 73–104. IEA (2013). Transition to Sustainable Buildings: Strategies and Opportunities to 2050. International Energy Agency (IEA). Koskela H, Haggblom H, Kosonen R, Ruponen M (2010). Air distribution in office environment with asymmetric workstation layout using chilled beams. Building and Environment, 45: 1923–1931. Kurt R, Dieckmann J, Zogg R, Brodrick J (2007). Chilled beam cooling. ASHRAE Journal, 49(9): 84–86. Livchak A, Lowell C (2012). Don’t turn active beams into expensive diffusers. ASHRAE Journal, 54(4): 52–60. Loudermilk K (2009). Designing chilled beams for thermal comfort. ASHRAE Journal, 51(10): 58–64. Murphy J (2009). Understanding Chilled Beam Systems. Trane Engineers Newsletter, 38(4): 1–12. Nytsch-Geusen C, Huber J, Ljubijankic M, Rädler J (2013). Modelica BuildingSystems—eine Modellbibliothek zur Simulation komplexer energietechnischer Gebäudesysteme. Bauphysik, 35: 21–29. (in German) Pérez-Lombard L, Ortiz J, Pout C (2008). A review on buildings energy consumption information. Energy and Buildings, 40: 394–398. RWTH-EBC (2014). AixLib. A Modelica library for building performance simulations. Institute for Energy Efficient Buildings and Indoor Climate (EBC) at RWTH Aachen University. Available at https:// github.com/RWTH-EBC/AixLib. Stein J, Taylor ST (2013). VAV reheat versus active chilled beam & DOAS. ASHRAE Journal, 55(5): 18–32. Trcka M, Hensen J (2010). Overview of HVAC system simulation. Automation in Construction, 19: 93–99. Van Roy J, Verbruggen B, Driesen J (2013). Ideas for tomorrow: New tools for integrated building and district modeling. IEEE Power and Energy Magazine, 11(5): 75–81. Wetter M, Zuo W, Nouidui TS, Pang X (2013). Modelica Buildings library. Journal of Building Performance Simulation, 7: 253–270.