predicting thermal and energy performance of mixed-mode

Research Article

Indoor/Outdoor A

irflow

and Air Q

uality

E-mail: [email protected]

Predicting thermal and energy performance of mixed-mode ventilation using an integrated simulation approach

Ali Malkawi, Bin Yan (), Yujiao Chen, Zheming Tong

Center for Green Buildings and Cities, Graduate School of Design, Harvard University, 20 Sumner Rd, Cambridge, MA 02138, USA Abstract

Mixed-mode ventilation can effectively reduce energy consumption in buildings, as well as improve

thermal comfort and productivity of occupants. This study predicts thermal and energy performance

of mixed-mode ventilation by integrating computational fluid dynamics (CFD) with energy simulation.

In the simulation of change-over mixed-mode ventilation, it is critical to determine whether outdoor

conditions are suitable for natural ventilation at each time step. This study uses CFD simulations

to search for the outdoor temperature thresholds when natural ventilation alone is adequate for

thermal comfort. The temperature thresholds for wind-driven natural ventilation are identified by

a heat balance model, in which air change rate (ACH) is explicitly computed by CFD considering

the influence of the surrounding buildings. In buoyancy-driven natural ventilation, the outdoor

temperature thresholds are obtained directly from CFD-based parametric analysis. The integrated

approach takes advantage of both the CFD algorithm and energy simulation while maintaining low

levels of complexity, enabling building designers to utilize this method for early-stage decision-

making. This paper first describes the workflow of the proposed integrated approach, followed by

two case studies, which are presented using a three-floor office building in an urban context. The

results are compared with those using an energy simulation program with built-in multizone

modules for natural ventilation. Additionally, adaptive thermal comfort models are applied in these

case studies, which shows the possibility of further reducing the electricity used for cooling.

Keywords mixed-mode ventilation,

CFD,

energy simulation,

adaptive thermal comfort model Article History Received: 9 July 2015

Revised: 11 December 2015

Accepted: 16 December 2015 © Tsinghua University Press and

Springer-Verlag Berlin Heidelberg

2016

1 Introduction

A design that enhances natural ventilation in certain climates will not only reduce energy cost, but also improve thermal comfort and air quality. However, the ability to rely solely on natural ventilation for cooling is limited by climate and thermal loads. As a result, mixed-mode ventilation becomes a promising solution that combines natural ventilation and air conditioning to achieve indoor thermal comfort (Brager et al. 2007). It is also called hybrid ventilation (Atkinson et al. 2009). Mixed-mode strategies allow buildings to be naturally ventilated during periods of the day or year when it is feasible, and to use mechanical cooling only when natural ventilation is not sufficient. A report states that mixed-mode ventilation is able to achieve 47%–79% HVAC energy savings, 0.8%–1.3% health cost savings, and 3%–18% productivity gain (CMU

2004). Another study also reports more than 40% energy savings by using mixed-mode ventilation in office buildings in a dry climate (Ezzeldin and Rees 2013).

Multizone and CFD models are widely used for predicting ventilation performance (Chen 2009). The multizone models assume uniform air temperature and neglect the momentum effect (Wang and Chen 2005). The CFD models do not employ the same assumptions and have been extensively applied, in recent years, when dealing with the challenges in natural ventilation design (Chen et al. 2007). CFD has become more and more popular in various areas with the development of computer capacity and user-friendly CFD program interfaces (Tong et al. 2012; Tong and Zhang 2015). CFD simulation could not directly predict thermal load and energy consumption. Ultimately, decision-makers would need to compare the energy consumption of different

BUILD SIMUL DOI 10.1007/s12273-016-0271-x

Malkawi et al. / Building Simulation 2

design strategies. In order to estimate the energy reduction of the design, the implementation of energy simulation is indispensable. EnergyPlus is a widely used energy simulation program with a built-in module called “Airflow Network” to simulate forced and buoyancy airflows between zones in a building, as well as between the building and the outdoors. The Airflow Network consists of a set of nodes linked by airflow components. A zone is considered as a node, and openings, such as doors and windows are considered as linkage components. The Airflow Network solves air mass balance equations to calculate the pressure of each node and airflow rate of each linkage (DOE 2014). The Airflow Network is a multizone model. The engineering reference of EnergyPlus points out that one should not expect Airflow Network to generate realistic temperature and airflow in an atrium, which is a typical design strategy for natural ventilation (DOE 2014). Moreover, surrounding buildings are not included in the airflow modeling.

A recent trend is to couple CFD and energy simulation. A significant number of coupling methods concentrate on improving predictive accuracy of surface heat convection when simulating building energy use (Zhai and Chen 2005; Beausolleil-Morrison 2002; Yi and Feng 2013; Manz and Frank 2005). The coupling has also been applied to improve natural ventilation prediction (Wang and Wong 2007, 2008, 2009; Pappas and Zhai 2008). Zhang et al. (2013) coupled EnergyPlus and FLUENT by exchanging airflow rates, air temperature, and surface heat transfer coefficients at each time step. According to their case studies, the airflow rates during natural ventilation simulated by Airflow Network module and their coupling method are very different.

There are additional efforts in coupling multizone models with CFD models (Tan and Glicksman 2005). For instance, Wang and Chen (2007) coupled a multizone network program with a CFD program in order to eliminate the assumptions in multizone models. As a case study, cross ventilation in a four-zone building is simulated using both CONTAM (a multizone model) and their coupled program. The coupled simulation exchanges pressure between the multizone model and CFD algorithm. There is around 60% difference between the airflow rate solved by CONTAM alone and by the coupled program.

The coupling methods in the studies described above have effectively improved the predictive accuracy. However, most of the methods in these studies also increase the com-plexity of simulation to some extent. Although researchers are willing to pay such a price (Chen 2009), it might not be feasible for designers and engineers to handle the complexity in their projects. Center for the Built Environment (CBE) at the University of California, Berkeley has been conducting long-term research on mixed-mode ventilation. A recent report (Gandhi et al. 2014) points out that the extensive

computing time has limited the widespread applicability of those coupling methods in practice. Some coupling methods require additional tools and code modification. The current applications of these tools are limited to research due to their programming requirements (Gandhi et al. 2014).

To further extend the applications in practice, this research intends to predict energy and thermal performance of mixed-mode ventilation in a more computationally affordable way while utilizing both CFD and energy simula-tions without code modification. The proposed integrated approach uses CFD to search for the outdoor temperature thresholds when natural ventilation alone is adequate for thermal comfort. The results are combined with energy simulation—at the end of the simulation process—to obtain the cooling energy consumption of mixed-mode strategies. In this way, the proposed approach maintains low levels of complexity so that designers and engineers are more capable of utilizing this approach for early-stage decision making.

This paper first describes the workflow of the integrated simulations. It then presents case studies to illustrate the procedures of integrating energy simulation and CFD to predict cooling electricity used for mixed-mode ventilation. Additionally, it compares the results of using energy simulation alone with a built-in multizone module for natural ventilation, to the outcomes derived from the integrated approach to illustrate differences in final results. Finally, adaptive thermal comfort models are applied in the case studies. Using higher indoor temperature setpoints of adaptive thermal comfort models can achieve more energy savings.

2 Methodology and workflow

This paper examines changeover operation—a mixed-mode operation strategy wherein a building switches between mechanical cooling and natural ventilation (Brager et al. 2007). This paper also discusses two types of natural ventilation scenarios, wind-driven and buoyancy-driven natural ventilation. Pressure difference is the main mechanism of wind-driven ventilation, while buoyancy-driven ventilation occurs as a result of temperature difference. When the temperature difference between indoor and outdoor environments is small in summer, buoyancy-driven force usually has less influence on the ventilation rate. This means that wind-driven force often dominates, especially in windy conditions (Asfour and Gadi 2007; Khan et al. 2008; Haw et al. 2012; Shen et al. 2012). Therefore, isothermal CFD simulations are conducted to simulate wind-driven natural ventilation. However, in the early design stage, designers may actually prefer to exclude wind-driven airflow when carrying out conservative design simulations for decision making, especially for designs that feature atriums and solar chimneys. Thus, this study also considers buoyancy-driven natural ventilation, without any

Malkawi et al. / Building Simulation

3

wind over the building exterior, when evaluating a new design in the early decision-making stage. Whether to perform a simulation of wind-driven or buoyancy-driven natural ventilation depends on the building design and the purpose of the simulation study. There is no existing simple guidance on selection criteria for wind-driven natural ventilation and buoyancy-driven natural ventilation from the previous studies. Some existing studies conducted either isothermal simulation for wind-driven natural ventilation or simulations for buoyancy-driven only natural ventilation depending on the purpose of their simulation studies. Jiang and Chen (2003) experimentally and numerically studied single-sided natural ventilation in buildings by buoyancy forces. The logic of the buoyancy-only simulation is that in “worst scenario” of natural ventilation, the ventilation is only driven by buoyancy forces. Liu et al. (2009) conducted simulation of buoyancy-driven natural ventilation to evaluate whether a buoyancy-only ventilation strategy is effective enough for thermal comfort in hot and humid climates. In order to evaluate the effectiveness of cross ventilation, Bangalee et al. (2012) conducted isothermal airflow simulation for wind- driven natural ventilation through multiple windows of a building. Haw et al. (2012) used isothermal CFD simulation to evaluate the performance of a wind-induced ventilation tower in hot and humid climate. Buoyancy-driven force was not included in their study because buoyancy-driven natural by itself was assumed insufficient to achieve indoor

air quality in that climate. As it is difficult to include both buoyancy and wind forces in simulations for long-term simulation, the proposed research provides methods for both types of natural ventilation. Users should select the scenario according to the building design and the purpose of their studies.

The intention of this research is to integrate CFD algorithms and energy simulations while maintaining low levels of complexity. Compared with relying solely on EnergyPlus for natural ventilation simulation, this method replaces the built-in Airflow Network module in EnergyPlus with CFD simulations. This process eliminates some of the assumptions and simplifications in airflow simulation employed by multizone models. Figure 1 describes the methods for simulating energy consumption of mixed-mode strategies using wind-driven natural ventilation (Fig. 1(a)) and buoyancy-driven natural ventilation (Fig. 1(b)). In both cases, CFD simulation is used to search for the temperature thresholds when natural ventilation alone is adequate for thermal comfort. The temperature threshold for each hour, combining other considerations, such as the humidity constraint, determines whether or not natural ventilation is adequate for that hour. Energy simulation is used to calculate cooling energy consumption without natural ventilation strategies for the entire year. Then the cooling energy consumption during natural ventilation hours is set to zero. What remains is the amount of cooling energy consumption

Fig. 1 Method for simulating cooling energy consumption using mixed-mode ventilation


needed for mixed-mode ventilation. The way of deriving temperature thresholds from CFD simulations is slightly different for these two types of natural ventilation methods.

In buoyancy-driven natural ventilation, both solar radiation and internal load are included in the CFD modeling. The highest possible outdoor temperature that still maintains indoor-thermal comfort for each solar radiation scenario is directly derived from CFD simulations. In wind-driven natural ventilation, isothermal CFD simulations are conducted to derive air change rates in different wind conditions. First, a neighborhood-scale external airflow simulation is performed to derive microclimatic wind conditions. Indoor airflow patterns are simulated using the microclimate wind conditions as boundary conditions. Meanwhile, a continuous tracer source is added to each room during the simulation in order to calculate the air change rate. When the tracer reaches equilibrium, the air change rate of a room can be determined from the amount of tracer emissions and their subsequent indoor concentration (Sherman 1990; Sherman et al. 2014):

ACH SC V

=⋅

(1)

In Eq. (1), ACH = air change rate of the room (h−1); S = tracer emission rate (g/h); C = steady state indoor tracer concentration (g/m3) simulated by CFD; and V = room volume (m3). Tracer-gas techniques are widely used methods to measure air change rates in buildings. Here, the CFD simulation mimics such experiments by adding a continuous tracer source S with a certain concentration specified by users. The space average concentration C is derived from CFD simulation, and the air change rate of each room is determined using Eq. (1). Since windows are fully open in the simulation, the air change rate is the maximum possible rate during natural ventilation. As shown in Fig. 1(a), cooling load from energy simulation is also needed to derive temperature thresholds for natural ventilation. The highest possible outdoor temperature Tout for natural ventilation in each hour can be calculated by the following equation:

( )p out in1ACH

3600c ρ V T T Q-⋅ ⋅ ⋅ ⋅ =⋅ (2)

In Eq. (2), cp = air specific heat (J/(kg·K)); ρ = air density (kg/m3); ACH = room air change rate (h−1) in that hour simulated by CFD; Tout = temperature threshold (K or °C) to be calculated; Tin = indoor-temperature setpoint (K or °C); and Q = zone sensible cooling load (W) derived from energy simulation. Ventilation load of fresh air is not included in Q. More details of the proposed approach can be found in the online version of this paper as Electronic Supplementary Material (ESM).

Computational cost is an important consideration in this

study. Since CFD simulation is computationally expensive, it is essential to reduce the required number of CFD simulations in the integrated simulations. The idea of finding the thresholds instead of running 8760 hour-by-hour simulations is critical to achieve this goal. For wind-driven natural ventilation, thermal simulation is completely separated from CFD airflow simulation, which decreases the input dimensions and thus reduces the number of required CFD simulations.

In our study, FloVENT software (Mentor Graphics 2014) is used for the CFD simulation of buoyancy-driven natural ventilation, while FlowDesigner software (MI Research 2014) is employed for the CFD simulation of wind-driven natural ventilation. These two CFD programs both specifically target simulations for buildings and have different advantages. FloVENT is capable of integrating solar irradiance into CFD simulation and its parametric study function is useful for threshold search. The parametric study function allows users to perform a series of simulations automatically where one or more parameters are varied. FlowDesigner has a “nesting” function that can automatically use simulation results from a neighborhood-scale as boundary conditions for indoor airflow simulation. In addition, FlowDesigner can take complex geometries directly from the computer-aided 3D design program, for example, Rhinoceros (Robert McNeel & Associates 2015). EnergyPlus is used for energy simulation.

3 Case study

This section presents two case studies to illustrate the process of predicting the cooling energy consumption of mixed- mode ventilation. The building used for the case studies is an existing three-floor structure with a basement, located in Cambridge, Massachusetts. At one time, the building was a residential house but it has since been converted into an office building that accommodates forty researchers. Window air-conditioners are used for indoor cooling with a setpoint of 26.7 °C and 60% relative humidity according to the Predicted Mean Vote Model (PMV) in ASHRAE Standard 55 (ASHRAE 2013a). There is no mechanical fresh air supply. During mechanical cooling and heating, fresh air is simply introduced through infiltration. According to ASHRAE Standard 62.1 (ASHRAE 2013b), 0.82 ACH will satisfy the ventilation requirement. Since the building envelopes are not well sealed, the infiltration rate is set at 1 ACH in the simulation. As a simple example, it is assumed that the building is perpetually occupied with the cooling setpoint remaining the same across time (24/7). More building parameters are listed in Table 1.

In a recent retrofit project, a new design proposes to change the interior of the building to an open-office


5

Table 1 Building parameters

Total area (m2) 365

Occupant density (person/m2) 0.1096

Lighting and equipment (W/m2) 16.5

Wall U-value (W/(m2·K)) 0.23

Roof U-value (W/(m2·K)) 0.17

Window U-value (W/(m2·K)) 2.7

Window solar heat gain coefficient 0.7

Infiltration rate (ACH) 1

EER (energy efficiency ratio) of window air-conditioners

9 (equivalent COP=2.6)

Indoor cooling setpoint (°C) 26.7

Indoor humidity upper limit 60%

concept. Therefore, this case study simulates cooling energy consumption of both the existing and proposed new design. The results are compared with the findings generated by Airflow Network, the built-in module of EnergyPlus for mixed-mode ventilation. The adaptive thermal comfort model has been applied to the case study to examine additional energy reduction.

3.1 Wind-driven natural ventilation

Figure 2 shows the existing second-floor layout and a cross section of the building. The rooms and floors are completely separated from one another. The temperature difference inside these small rooms is most likely negligible due to the insignificance of the buoyancy-driven effect especially in windy conditions. Therefore, the simulation only considers the wind-driven natural ventilation in mixed-mode strategies for the existing design. The simulation results are detailed and presented for the room at the southwest corner on the second floor.

3.1.1 Simulation for wind-driven natural ventilation

As a first step, the 3D model of surrounding buildings is exported from Google Earth and imported into FlowDesigner to perform external airflow simulations at neighborhood- scales. The size of this computational domain is approximately 1000 m × 800 m × 65 m. It is meshed with approximately two million structured grids. A grid-independency study is conducted to ensure that results are independent of the mesh resolution (Fig. 3(a) and Table 2). A fully developed wind profile is applied at the inlet, with a turbulent intensity of I =10%. A sensitivity study on the inlet turbulence is conducted by varying I, from 5% to 20%, as shown in Fig. 3(b). The difference in velocity profiles is almost indistinguishable as the flow field at complex urban environment is dominated by large-scale motion generated by scales comparable to the size of buildings and street canyons (Xie and Castro 2006). Logarithmic wall function is applied to the near-wall region due to the impracticality to resolve all viscous sublayers in such a large domain (Launder and Spalding 1974). Symmetry boundary condition is applied on top of the domain as slip walls with zero shear, and open boundary condition is specified at the flow outlet. The standard k–epsilon turbulence model is chosen for computational stability and its affordable isothermal flow over buildings (Neofytou et al. 2006; Wang et al. 2013). After the neighborhood-scale airflow simulation is done, the built-in “nesting” module in FlowDesigner allows users to select a smaller domain to continue to the building scale simulation. In this case study, the size of the building-scale domain is 20 m × 20 m × 30 m and it is meshed with one million structured grids. The flow variables including velocity components, turbulent kinetic energy, and dissipation rate on the boundaries of the building scale domain are automatically transferred and interpolated from the neighborhood-scale simulation.

Fig. 2 Floor plan and section of the case study for wind-driven natural ventilation simulation


Table 2 A sensitivity study on mesh resolution

Grid number 1 million 2 million 4 million

ACH 17 29 30

Simulation scenarios, in terms of wind speed and wind

direction, also need to be decided. Figure 4 shows the wind speed distribution when outdoor air temperature is between 15 and 26.7 °C. Within this outdoor temperature range, it is possible to naturally ventilate the buildings. In natural ventilation mode, outdoor temperature must be lower than the indoor setpoint, which is 26.7 °C in this study. Otherwise, it is impossible to maintain indoor temperature below the cooling setpoint when there is heat gain from lights, plug- ins, occupants, and solar radiation. A lower limit of outdoor temperature is also set in this analysis. There are two reasons for this lower temperature limit. First, when it is cold outside, occupants are less likely to keep windows open (Rijal et al. 2007; Dutton and Shao 2010) as that might cause undesirable draft. Second, when outdoor temperature is low, there might not be any cooling load for external load dominant buildings. Even when there is cooling load, allowing in a small amount of outdoor air through windows can easily offset the load. Since this scenario is not the interest of this study, this paper sets the lower limit for outdoor temperature as 15 °C. According to Fig. 4, the wind speed is below 9 m/s during more than 95% of the hours that are possible for natural ventilation. Therefore, eight wind directions (45° apart) and five wind speeds (1, 3, 5, 7 and 9 m/s at the height of 10 m) are selected as simulation scenarios. The wind profiles for different heights are taken into account in the CFD simulation.

Figure 5 shows a view of the simulated external airflow distribution when the wind is coming from the west at the

speed of 9 m/s, and indicates that the microclimate wind speed is significantly reduced in an urban environment. The red rectangle specifies the location of the investigated

Fig. 4 Normalized histogram of wind speed during hours that are possible for natural ventilation

Fig. 5 Neighborhood airflow simulation results

Fig. 3 Sensitivity study on mesh resolution (a) and inflow turbulence level (b)


7

house, while the airflow conditions including velocity com-ponents, turbulent kinetic energy, and dissipation rate inside of the red rectangle are used as boundary conditions when simulating indoor airflow.

Air change rates of eight wind directions at wind speeds of 1, 3, 5, 7 and 9 m/s are simulated. Figure 6(a) shows results from an indoor airflow simulation for one wind condition. Figure 6(b) shows part of the air change rate outcomes. Using interpolation, the air change rates of different wind directions at speeds ranging from 1 through 9 m/s can be derived. Wind speeds of 9 m/s or greater are considered to have the same air change rate. When the wind arrives from the west at the speed of 9 m/s, the investigated room reaches 42 h−1. In this condition, the indoor airflow speed is still well below 0.5 m/s, as shown in Fig. 6(a). Therefore, drafts that cause discomfort will not be of concern in this case study (Cândido et al. 2008; Evans 1980).

The next step is to calculate the air change rate of each hour based on wind conditions in weather data and the simulated air change rates from CFD. Meanwhile, energy simulation outputs hourly cooling load. Then the outdoor temperature threshold of each hour can be calculated using Eq. (2). Natural ventilation is feasible if the actual outdoor temperature of that hour is lower than the temperature threshold and higher than the lower limit. Additionally, the weather conditions must satisfy other constraints such as humidity. If there is cooling consumption in that hour, it is reset to zero. This enables the determination of the number of hours that natural ventilation alone is adequate for thermal comfort, as well as the cooling energy consumption using mixed-mode strategies. In this case study, the humidity constraint is that outdoor dew-point temperature must be below 18 °C. Details regarding indoor temperature and humidity setpoints are discussed in a later section.

As shown in Table 3, the annual cooling energy con-sumption using mixed-mode ventilation for the investigated room is 23.5 kWh/m2. Meanwhile, there are 3258 hours

wherein outdoor temperature is above 15 °C, which indicates how much cooling time the building might need. Among these 3258 hours, outdoor air temperature is below the indoor temperature setpoint with acceptable humidity for 2363 hours. These hours are possible for natural ventilation. Define weather natural ventilation index as the number of hours in which outdoor temperature and humidity are possible for natural ventilation divided by the total cooling hours. For instance, a cool and dry climate will yield a high index. In this study, the weather natural ventilation index is 73%, according to the Boston-Logan International Airport TMY3 weather data. This index can be easily derived from weather data without any simulation. It is recommended to check this index at the beginning, to decide whether natural ventilation is applicable in a certain climate.

There are 1925 hours out of 2363 hours in which the outdoor air temperature is lower than the temperature thresholds for natural ventilation derived from CFD and energy simulations. The building’s natural ventilation index takes the actual number of hours in which natural ventilation is adequate for thermal comfort in the investigated zone/ building and divides that figure by the total number of hours that outdoor temperature and humidity are possible for natural ventilation. This index considers building cooling load, envelope design, and weather conditions, suggesting how much natural ventilation potential has been exploited for a certain zone/building. A high percentage indicates that there is not much room left to further improve the natural ventilation design. In this case study, for the investigated room, the building natural ventilation index is 81%, which means that up to 19% more natural ventilation hours is theoretically possible to achieve.

3.1.2 Comparison of using EnergyPlus and the integrated approach when simulating wind-driven natural ventilation

In EnergyPlus, Airflow Network could not fully take surrounding buildings into consideration. The local wind

Fig. 6 Airflow simulation results


Table 3 Natural ventilation hours and energy reduction

Number of hours

Tout>15 °C 3258

15 °C < Tout <Tin,set, Tout,dew < 18 °C 2363

15 °C < Tout <Tout,threshold, Tout,dew < 18 °C 1925

Weather natural ventilation index 73% (2363 / 3258 = 73%)

Building natural ventilation index 81% (1925 / 2363 = 81%)

Annual cooling energy consumption 23.5 kWh/m2

speed at a certain height, is extrapolated from the wind speed measured at a meteorological station using coefficients that depend on the roughness characteristics of the surrounding terrain (DOE 2014):

metmet

metmet

α α

zδ zz δ

V V= ( ) ( ) (3)

where z = height above ground (m), Vz = wind speed at height z (m/s), α = wind speed profile exponent at the site, δ = wind speed profile boundary layer thickness at the site (m), zmet = height above ground of the wind speed measured at the meteorological station, Vmet = wind speed measured at the meteorological station (m/s), αmet = wind speed profile exponent at the meteorological station and δmet = wind speed profile boundary layer thickness at the meteorological station (m). The wind speed profile coefficients α, δ, αmet, and δmet depend on the roughness characteristics of the surrounding terrain. The terrain in this case study is set as urban and the actual surrounding buildings are not taken into account in the model, meaning that the accuracy of the natural ventilation simulation is compromised.

In this comparison study, the air change rate of each room due to wind-driven airflow is simulated by the EnergyPlus Airflow Network module and CFD program FlowDesigner. In EnergyPlus, the terrain is set to be urban. The air change rate of each room comes directly from the EnergyPlus simulation. The results of the investigated room in the case study are compared in Fig. 7. As shown in Fig. 7(a), for north, east, south and west wind directions, the air change rates simulated by EnergyPlus are significantly larger than those generated by CFD. In addition, as shown in Fig. 7(b), their patterns are also different. According to the EnergyPlus simulation, the highest ventilation rate occurs at the window orientations in the south and west wind directions. According to CFD simulations, the highest ventilation rate occurs from the southwest direction.

In terms of cooling consumption of mixed-mode ventilation, if humidity is not considered, the output of the integrated simulations is 18.2 kWh/m2, while the EnergyPlus simulation result is 14.5 kWh/m2 for the investigated room. The difference is noticeable.

Fig. 7 Air change rate of the investigated room simulated by EnergyPlus and FlowDesigner

3.2 Buoyancy-driven natural ventilation

Figure 8 shows the proposed design for a recent project. In the new design, the partitions are removed to create an open-office floor plan. The roof has been redesigned. A solar chimney and an atrium are added to enhance buoyancy- driven natural ventilation. In the early decision-making stage, wind-driven airflow is excluded in the simulation in order to carry out a conservative design evaluation.

3.2.1 Simulation for buoyancy-driven natural ventilation

To monitor the temperature distribution in the house in the CFD model, thirty-three temperature sensors (monitored points) are placed in the occupied area, as shown in Fig. 8. If the temperatures at these monitored points are below the indoor temperature setpoint, thermal comfort is achieved. Mean temperature is not used as an indicator because the indoor temperature is not uniform and only occupied area is of concern. Indoor temperature setpoint is 26.7 °C and the humidity constraint is that the outdoor dew-point temperature must be below 18 °C. The goal of CFD simulation is to find temperature thresholds for different scenarios when natural ventilation alone is adequate for thermal comfort. In the buoyancy-driven simulation, the size of the domain is roughly 15 m × 17 m × 14.5 m. Open boundary condition is applied to all external boundaries except the ground. Solar radiation is included in the model and internal heat sources are specified at each floor inside the house. Because the internal load is fixed for different solar radiations, there is a highest possible outdoor temperature below which natural


9

ventilation can be used to achieve the indoor temperature setpoint. When outdoor temperature is higher than this threshold, mechanical air-conditioning must be used. The range of solar radiation, from zero to maximum solar radiation, can be acquired from weather data file.

A search for the temperature thresholds uses the parametric study function in FloVENT. Parametric studies are conducted for different solar radiation intensities from 0 to 1000 W/m2, with 1000 W/m2 of solar radiation being the highest in the weather file. Figure 9 shows the simulation results of the temperature thresholds in dots for this case study. The outdoor threshold temperature is 24.3 °C when there is no solar radiation and 23.1 °C when the solar radiation intensity is 1000 W/m2. Using interpolation, temperature thresholds for all solar radiation intensities can be derived. As shown in Fig. 9, any combination of outdoor temperature and solar radiation intensity that falls in the blue area is adequate for buoyancy-driven natural ventilation. Figure 10 shows the airflows in the atrium. The highest air speed is around 0.7 m/s, which occurs at the openings on the roof.

Fig. 9 Suitable weather conditions for natural ventilation–simulation results

An energy simulation is then conducted to derive the cooling energy consumption without natural ventilation. Energy consumption of mixed-mode ventilation is derived by deducting the energy consumption during the natural ventilation hours. In this case study, the electricity consump-tion for cooling is 25.1 kWh/m2 per year without natural ventilation. Taking the buoyancy-driven natural ventilation into consideration, there are 2086 hours during which natural ventilation is adequate for thermal comfort. The electricity consumption for cooling reduces to 14.0 kWh/m2 per year using mixed-mode ventilation. The building natural ventilation index is 88%. Even according to this conservative estimate, the proposed design is effective in terms of natural ventilation.

3.2.2 Comparison of using EnergyPlus and the integrated approach when simulating buoyancy-driven natural ventilation

The Airflow Network module in EnergyPlus is a multizone model, which assumes that air temperature is uniform in a zone and that air momentum effects are neglected. These assumptions might cause inaccuracy in the conditions of strong buoyancy or strong momentum (Wang and Chen 2008). Because of the atrium, using a multizone model might not be adequate for the proposed design. Using EnergyPlus with its built-in Airflow Network model and hybrid natural ventilation control configuration, the cooling electricity is 6.7 kWh/m2 per year. The EnergyPlus simulation does not consider the humidity constraint. The corresponding outcome of the EnergyPlus simulation combined with the FloVENT simulation is 9.9 kWh/m2 per year, which results in a 32% difference in terms of cooling electricity.

3.3 Adaptive thermal comfort model

In the case study above, we assume a fixed indoor temperature

Fig. 8 Proposed design and temperature sensor location in buoyancy-driven natural ventilation simulation


setpoint of 26.7 °C according to the Predicted Mean Vote Model in ASHRAE Standard 55 (ASHRAE 2013a). Tem-perature of 26.7 °C with relative humidity of 86% is still within the comfort zone when local air speed is 0.2 m/s. If the local air speed reaches 0.3 m/s, relative humidity could be as high as 100% at 26.7 °C. Therefore, in natural ventilated buildings, the acceptable humidity range also depends on local air speed. In this method, the humidity requirement is not simulated in the same way as the temperature requirement in CFD. Instead, a humidity constraint is set as an additional weather criterion. In this study, the humidity constraint is set as an outdoor dew-point temperature that must be lower than 18 °C. With a large ventilation rate, this condition will most likely ensure thermal comfort even when there is latent load. Designers and engineers can use different humidity constraints for their projects if necessary.

Instead of a fixed setpoint, the upper limit of indoor temperature during natural ventilation can be determined according to the adaptive model in ASHRAE 55 (ASHRAE 2013a). This adaptive model established by de Dear and Brager (1998; 2002) has been widely used. The model uses mean monthly outdoor temperature to determine indoor- temperature setpoints. For each calendar month, there is one indoor-temperature setpoint. This study uses the 80% acceptable indoor temperature upper limit as indoor cooling setpoint during natural ventilation, as shown in Fig. 11. The adaptive thermal comfort only applies to natural ventilation. When the building uses mechanical cooling, the temperature setpoint is still fixed at 26.7 °C.

Fig. 11 Indoor cooling setpoint for natural ventilation based on adaptive thermal comfort model

In order to use the adaptive thermal comfort model in the integrated simulations, only one more energy simulation is needed to estimate the cooling load with new indoor temperature setpoints. There is no modification on the CFD side. In this case study, as the indoor cooling setpoints for natural ventilation from June to September are higher than the fixed 26.7 °C, cooling load in these months decreases. As a result of applying adaptive thermal comfort model, the number of natural ventilation hours increases and the cooling energy consumption of mixed-mode ventilation is further reduced from June to September, especially in July, as shown in Fig. 12. In conclusion, more energy savings potential can be achieved by using the adaptive thermal comfort model. The proposed approach provides the flexibility of incorporating different thermal comfort models.

Fig. 10 CFD simulation results, airflows in the atrium (solar radiation intensity =1000 W/m2 and outdoor temperature = 23.1 °C)


11

4 Discussion and conclusion

This paper predicts the energy and thermal performance of mixed-mode ventilation by integrating CFD and energy simulations in a computationally affordable way. In this study, CFD simulation and energy simulation are performed independently and their results are combined to derive the number of natural ventilation hours and the amount of cooling energy consumption for mixed-mode ventilation. It takes approximately 30–40 CFD simulations for a one-year study. With parametric study function, the simulations can be automatically conducted in parallel. Any CFD and energy simulation programs can be used without code modification. No other integration platform is required. The implementation of integrating the results of EnergyPlus and CFD simulations can be realized in any existing tools even if users do not have coding experience. This study also investigates the impact of adaptive thermal comfort model on the further energy reduction of mixed-mode ventilation, which demonstrates the flexibility of incorporating different thermal comfort models and other constraints using integrated simulations.

Two case studies are discussed in detail for buoyancy- driven and wind-driven natural ventilation simulations. Weather and building natural-ventilation indexes are developed to assist with early decision-making in natural ventilation design. Simulation results of using EnergyPlus alone and using EnergyPlus combined with CFD are compared. For buoyancy-driven mixed-mode ventilation, there are noticeable differences in cooling electricity. One reason is that the Airflow Network module in EnergyPlus could not provide realistic results for large spaces, such as atriums. For wind- driven natural ventilation, the air change rates simulated by the Airflow Network module in EnergyPlus are significantly higher than those simulated by CFD. The Airflow Network module is unable to capture the complex fluid dynamics due to the presence of surrounding buildings, which accounts for the difference of air change rate predictions. The comparison conducted in this research assumes that CFD is more

comprehensive than multizone models due to the fact that it is able to provide more realistic airflow simulation. Further validation against experimental data is needed to evaluate the accuracy of these two approaches.

The approach of this research is to replace the built-in Airflow Network module in widely used energy simulation tool EnergyPlus with more comprehensive airflow simulation algorithm CFD in an easy-to-implement way. The proposed approach will enable building designers and engineers to inform early-stage decision making. This study demonstrates the process of using integrated CFD and energy simulations to provide more comprehensive analysis when large spaces and surrounding buildings need to be included in the simulations. The validation of this methodology should be further considered in future work while the trade-off between the complexity and accuracy should also be studied. The uncertainty of separating wind-driven and buoyancy-driven natural ventilation can also be subject to future research.

Electronic Supplementary Material (ESM): supplementary material is available in the online version of this article at http://dx.doi.org/10.1007/s12273-016-0271-x.

References

ASHRAE (2013a). ASHRAE Standard 55-2013. Thermal Environmental Conditions for Human Occupancy. Atlanta: American Society of Heating, Refrigerating and Air-Conditioning Engineers.

ASHRAE (2013b). ASHRAE Standard 62.1-2013. Ventilation for Acceptable Indoor Air Quality. Atlanta, American Society of Heating, Refrigerating and Air-Conditioning Engineers.

Asfour OS, Gadi MB (2007). A comparison between CFD and network models for predicting wind-driven ventilation in buildings. Building and Environment, 42: 4079–4085.

Atkinson J, Chartier Y, Pessoa-Silva C, Jensen P, Li Y, Seto WH (2009). Natural Ventilation for Infection Control in Health-care Settings—WHO Guidelines. Geneva: World Health Organization.

Bangalee MZI, Lin SY, Miau JJ (2012). Wind driven natural ventilation through multiple windows of a building: A computational approach. Energy and Buildings, 45: 317–325.

Fig. 12 Cooling energy consumption of mixed-mode using modified adaptive thermal comfort model or fixed setpoint


Beausoleil-Morrison I (2002). The adaptive conflation of computational fluid dynamics with whole-building thermal simulation. Energy and Buildings, 34: 857–8871.

Brager G, Borgeson S, Lee YS (2007). Summary Report: Control Strategies for Mixed-mode Buildings. Berkeley: Centre for the Built Environment, University of California.

Cândido C, de Dear R, Lamberts R, Bittencourt L (2008). Natural ventilation and thermal comfort: Air movement acceptability inside naturally ventilated buildings in Brazilian hot humid zone. In: Proceedings of Air Conditioning and the Low Carbon Cooling Challenge Conference, Windsor, UK.

Chen Q (2009). Ventilation performance prediction for buildings: A method overview and recent applications. Building and Environment, 44: 848–858.

Chen Q, Glicsman L, Lin J, Scott A (2007). Sustainable urban housing in China. Journal of Harbin Institute of Technology (New Series), 14s: 6–9.

CMU (2004). Guidelines for High Performance Buildings. NSF/IUCRC Center for Building Performance and Diagnostics at Carnegie Mellon University, Advanced Building Systems Integration Consortium.

de Dear RJ, Brager GS (1998). Developing an adaptive model of thermal comfort and preference. ASHRAE Transactions, 104(1): 145–167.

de Dear, RJ, Brager, GS (2002). Thermal comfort in naturally ventilated buildings: Revisions to ASHRAE Standard 55. Energy and Buildings, 34: 549–561.

DOE (2014). EnergyPlus engineering reference: The reference to EnergyPlus calculations. US Department of Energy.

Dutton S, Shao L (2010). Window opening behaviour in a naturally ventilated school. In: Proceedings of SimBuild, New York, USA, pp.260–268.

Evans M (1980). Housing, Climate and Comfort. New York: John Wiley and Sons.

Ezzeldin S, Rees SJ (2013). The potential for office buildings with mixed-mode ventilation and low energy cooling systems in arid climates. Energy and Buildings, 65: 368–381.

Gandhi P, Brager G, Dutton S (2014). Mixed mode simulation tools. Internal report, Center for the Built Environment (CBE).

Haw LC, Saadatian O, Sulaiman MY, Mat S, Sopian K (2012). Empirical study of a wind-induced natural ventilation tower under hot and humid climatic conditions. Energy and Buildings, 52: 28–38.

Jiang Y, Chen Q (2003). Buoyancy-driven single-sided natural ventilation in buildings with large openings. International Journal of Heat and Mass Transfer, 46: 973–988.

Khan N, Su Y, Riffat SB (2008). A review on wind driven ventilation techniques. Energy and Buildings, 40: 1586–1604.

Launder BE, Spalding DB (1974). The numerical computation of turbulent flows. Computer Methods in Applied Mechanics and Engineering, 3: 269–289.

Liu PC, Lin HT, Chou JH (2009). Evaluation of buoyancy-driven ventilation in atrium buildings using computational fluid dynamics and reduced-scale air model. Building and Environment, 44: 1970–1979.

Manz H, Frank T (2005). Thermal simulation of buildings with double-skin facades. Energy and Buildings, 37: 1114–1121.

Mentor Graphics (2014). FloVENT. Available at http://www.mentor.com/ products/mechanical/flovent. Accessed 31 Jul 2014.

MI Research (2014). FlowDesigner. Available at http://www.mi-research. com/MI_Research/FlowDesigner.html. Accessed 31 Jul 2014.

Neofytou P, Venetsanos AG, Vlachogiannis D, Bartzis JG, Scaperdas A (2006). CFD simulations of the wind environment around an airport terminal building. Environmental Modelling & Software, 21: 520–524.

Pappas A, Zhai Z (2008). Numerical investigation on thermal per-formance and correlations of double skin facade with buoyancy- driven airflow. Energy and Buildings, 40: 466–475.

Rijal HB, Tuohy P, Humphreys MA, Nicol JF, Samuel A, Clarke J (2007). Using results from field surveys to predict the effect of open windows on thermal comfort and energy use in buildings. Energy and Buildings, 39: 823–836.

Robert McNeel & Associates (2015). Rhinoceros. Available at https:// www.rhino3d.com. Accessed 22 Jun 2015

Shen X, Zhang G, Bjerg B (2012). Comparison of different methods for estimating ventilation rates through wind driven ventilated buildings. Energy and Buildings, 54: 297–306.

Sherman MH (1990). Tracer-gas techniques for measuring ventilation in a single zone. Building and Environment, 25: 365–374.

Sherman MH, Walker IS, Lunden MM (2014). Uncertainties in air exchange using continuous-injection, long-term sampling tracer-gas methods. International Journal of Ventilation, 13: 13–27.

Tan G, Glicksman L (2005). Application of integrating multi-zone model with CFD simulation to natural ventilation prediction. Energy and Buildings, 37: 1049–1057.

Tong Z, Wang YJ, Patel M, Kinney P, Chrillrud S, Zhang KM (2012). Modeling spatial variations of black carbon particles in an urban highway-building environment. Environmental Science & Technology, 46: 312–319.

Tong Z, Zhang KM (2015). The near-source impacts of diesel backup generators in urban environments. Atmospheric Environment, 109: 262–271.

Wang L, Chen Q (2005). On solution characteristics of coupling of multizone and CFD programs in building air distribution simulation. In: Proceedings of 9th International IBPSA Building Simulation Conference, Montreal, Canada.

Wang L, Chen Q (2007). Theoretical and numerical studies of coupling multizone and CFD models for building air distribution simulations. Indoor Air, 17: 348–361.

Wang L, Chen Q (2008). Evaluation of some assumptions used in multizone airflow network models. Building and Environment, 43: 1671–1677.

Wang L, Wong NH (2007). The impacts of ventilation strategies and facade on indoor thermal environment for naturally ventilated residential buildings in Singapore. Building and Environment, 42: 4006–4015.

Wang L, Wong NH (2008). Coupled simulations for naturally ventilated residential buildings. Automation in Construction, 17: 386–398.

Wang L, Wong NH (2009). Coupled simulations for naturally ventilated rooms between building simulation (BS) and computational fluid dynamics (CFD) for better prediction of indoor thermal environment. Building and Environment, 44: 95–112.

Wang YJ, Nguyen MT, Steffens JT, Tong Z, Wang Y, Hopke PK, Zhang KM (2013). Modeling multi-scale aerosol dynamics and micro- environmental air quality near a large highway intersection using the CTAG model. Science of the Total Environment, 443: 375–386.

Xie Z, Castro IP (2006). LES and RANS for turbulent flow over arrays of wall-mounted obstacles. Flow, Turbulence and Combustion, 76: 291–312.

Yi YK, Feng N (2013). Dynamic integration between building energy simulation (BES) and computational fluid dynamics (CFD) simulation for building exterior surface. Building Simulation, 6: 297–308.

Zhai ZJ, Chen QY (2005). Performance of coupled building energy and CFD simulations. Energy and Buildings, 37: 333–344.

Zhang R, Lam KP, Yao SC, Zhang Y (2013). Coupled EnergyPlus and computational fluid dynamics simulation for natural ventilation. Building and Environment, 68: 100–113.

predicting thermal and energy performance of mixed-mode

Documents