Prototype and modeling of optimized ventilation

progress towards the final solution when BES finished the specified iterations. Due to flexible conflation between the two programs, the integrated coupling method has a smaller computational time compared with the fully dynamic coupling.

Fig.4.4 structure of integrated coupling

The bin coupling process provides BES the information that are pre-computed by CFD and saves in the bins for continuous energy analysis.

4.3 Prototype and modeling of optimized ventilation

With adjusting control values, ventilation system can be optimized in relation to potential energy recovery and contaminants removal by providing adequate fresh air flow rate.

Energy consumption of the whole system, finally, is capable to assess with precise and accurate prediction results compared to perfect mixing assumption.

Fig.4.5 program structure of optimized ventilation design 4.3.2 Coupling mechanism

To solve the building thermal behavior and HVAC performance in consideration of non-uniform distributions of air flow and temperature characteristics in the target office space, integration analysis of BES and CFD was carried out in this study. The framework of coupling simulation is as similar as shown in Fig.4.3. More detailed expression refers to Chapter 6. The quasi dynamic integrated simulations of BES and CFD were carried out on the basis of exchanging mutual complementary boundary conditions at each BES time step (1 hour). But the CFD case converges at each time-step and feedback the non-uniform information of the local domain to BES. A single CFD calculation was pre-carried out steadily before integrated simulation to reproduce a stable flow field in terms of convective

Heat exchanger (Type 91)

TRNBuild (Type56) Weather data

Psychometrics (Type 33 )

Meteorology (Type 109)

Pipe duct (Type 31)

Temperatures (equations)

Radiation (Type 16)

Output (Type 65) FLUENT (Type 306) PID

(Type 23)

CO₂ Demand Control

Ventilation rate (equations)

Concentrations (equations) Sky temp

(Type 69)

Pipe duct (Type 31)

Pipe duct (Type 31)

Pipe duct (Type 31) OA

RA SA

EA

Temperatures (equations)

Φ_ij Uij

Vij

Y,j

Z,k X,i CFD Part

Temperatures (equations) BES Part

and radiative environment. In the time series of coupling analysis, BES as the master program of BES-CFD integrated simulation governed the main timing in the continual process of thermal performance calculation. Alternatively, CFD was incorporated and exported non-uniform information of temperature and velocity distribution data to BES when it achieved converged solution-specified residuals and minimum number of iterations at fixed time intervals (1 hour corresponding to BES time step). There were no iterations between BES and CFD at each time-step. Instead the coupling between the two programs took place by running the CFD through a script file written by BES component. The CFD issued the results file which contained non-uniform information in the local domain, after it got convergence at each time-step. The file was returned to the BES component module as the initial boundary condition of the next time step. During each time step (1 hour in this study), steady calculations were conducted to update the data that were input to the corresponding component of BES. Before the main analysis, at least one day of a run-in period was secured to support the reasonable initial conditions of fully dynamic coupling simulation of BES and CFD. BES was judged to finish each time step when its components reached convergence by reading hourly updated boundary conditions provided by CFD.

The following process indicates the coupling mechanism of the integrated simulation:

1) A single CFD case is pre-calculated in steady-state to reproduce a stable flow field in terms of convective and radiative indoor environment.

2) Modify or update the source code of coupling controller (primitive type as Type101) based on derived information of CFD case in related to different domains.

3) Integrated simulation begin with initial boundary, FLUENT case is executed by coupling controller and read boundary conditions from script files.

4) TRNSYS keep freezing during iterations of one FLUENT case.

5) FLUENT get convergence and output non-uniform airflow properties to results files.

6) When FLUENT exits, the coupling controller read the return information and re-operate TRNSYS

7) Integrated simulation successes in alternatively updating script files and results files between two pragmas and automatically collecting results to .csv file.

4.3.3 Exchanged parameters

It has been confirmed in the previous studies that convective and radiative heat exchange between internal building surface and indoor air significantly affects energy balance of the target room. The affection can be summarized as the timing and degree of the incitation by internal surface warm the room air. Theoretically, convective heat transfer varied depending on different temperature of wall surface, in response to local airflow patterns.

Updated convective heat transfer coefficient (CHTC) and radiative heat transfer coefficient (RHTC) responded to time-dependent local domain flow conditions represent accurate prediction of heat transfer between building envelope to indoor air. Therefore, previous researches take CHTC as a significant controller for data exchanged to enhance the accuracy of coupling simulation. The choice of CHTC algorithm significantly affected the simulation prediction results of surface convection and impacted the design accuracy. Since convection in the wall vicinity varied depend on different thermal and flow boundary condition, it is necessary to decide CHTC at each time-step. Apart from exchanging the scalar in wall vicinity, this study focused on transferring average representative scalars of total occupied zone as new parameters exchange aspect for integrated simulation. Thereby, the integrated simulation can be readily applied to multi-HVAC system control.

In this study, heat flux of wall surfaces were transfer based on the Neumann thermal boundary from BES to CFD. Around the wall vicinity region of CFD model, generalized log-law wall functions was exploited to characterize momentum and temperature along with logarithmic profiles. First grid or near-to wall grid were located in turbulent region with 30<y⁺<100. Given that relatively coarse near wall treatment applied in CFD model, one-direction coupling for the wall surface thermal boundary was employed from BES to CFD. In other word, there is no CHTC feedback from CFD to BES in this integrated simulation. Instead, data exchanged were only used for controlling the thermal sensation represented by temperature, relative humidity (RH) and air quality represented by passive contaminates (CO2). Average temperature, RH and CO2 concentration of occupied zone are regarded as comprehensive reference values of local domain and feedback to BES corresponding components.

Details of parameters exchanged of the optimized ventilation with CO2 DCV is shown in Fig.4.6, Generally, TRNBuild module adopted the nodal model, which predicted the indoor environmental information with time-dependent and spatially averaged boundary condition.

In contrast, ANSYS/FLUENT discretized the mass, momentum and energy, based on FVM methods, along with the divided cell volume. Airflow properties, e.g. temperature, relative humidity, contaminant concentration as well as the thermal/environmental evaluation indices, therefore, were presented dimensionally non-uniformity.

Fig.4.6 exchanged parameters of optimized ventilation design

PID (Proport-inalIntegral-Differential) control mechanism and CO2 demand control approach were employed to maintain indoor air quality and provide thermal acceptable environment. To eliminate the effect of internal heat perturbation and provide even temperature in occupied zone, PID control was considered to adjust package air conditioner (PAC) on the basic of local domain average temperature. Supply inlet temperature of PAC and occupied zone average temperature were regarded as exchanged variables for PID applying on PAC units respectively. With the energy recovery ventilator (ERV), energy that was used for dehumidifying and conditioning air to close human requirement can be recovered. During the summer season, the ERV system cools and dehumidifies the outdoor air while humidifying and heating it in the winter season. With this device, as expected, not only an effective means of reducing energy cost can be achieved, but a visible possibility to scale down heating/cooling capacity of the whole system. In addition, CO2 demand control

Φ_ij U_ij

V_ij

Y,j X,i Z,k

PID ERV

TRNBuild

Supply T_ac

Supply T_ERVRH_ERV CO_2ERVDynamic Q Return T_ERVRH_ERVCO_2ERV Internal dynamic Heat Flux, F_surf

Occupied zone T_ave

Occupied zone CO2_ave Meteorological/

solar data Material construction

SupplyT_ERVRH_ERV Dynamic Q

Occupant density

Supply T_ac

Outdoor Air T_amb RH_ambCO_2ambdata

CO₂ Demand Control

BES Part CFD Part

approach was applied to adjust ventilation intake rate with regards to the low energy ventilation process. Under steady-state condition, ventilation rate was a function of the perfect mixing or equilibrium concentration. Generally BES can only provide single zone value refers to as perfected mixing CO2 concentration. By employing CFD, detailed and distributed CO2 concentration can be predicted and which is closer to the control target than the perfect mixing value. Therefore, ventilation effectiveness was used to build the connection between point concentrations with perfect mixing concentration based on its definition formula. In other words, constant ventilation effectiveness was assumed between each current time-step and next time-step and updated ceaselessly to limit the contaminant concentration in local domain. Supply inlet Temperature, Relative Humidity and CO2

concentration of energy recovery ventilator (ERV) were parameters undertaking the boundary condition of CFD, as for return, return outlet Temperature, Relative Humidity, CO2 concentration of ERV and occupied zone average CO2 level were feedback parameters to achieve integrated CO2 Demand Control on ERV system.

4.3.4 User defined functions in Fluent

Customized functions were also composed in this research, so-call user defined functions (UDF), which can be used to define suitable boundary conditions, material properties, and source terms in the flow domain. For specifying turbulent viscosity in solving momentum governing equations one UDF was applied. Meanwhile, UDF was written in CO2 DCV case to achieve the air circulation in PAC and to characterize constant CO2 level (keep the CO2

mass balance) inside PAC component, avoiding the default impact of OUTFLOW boundary condition.

4.3.5 TRNSYS type development

The current studies treated TRNSYS as the main driver, and called FLUENT to finish the airflow properties estimations as mentioned precisely in section 4.3.2 (Coupling mechanism). TRNSYS’s modular based on components (“Types”) also made it available to create new components by developing individually customized Types and adding them to the existing list of components.

One of the significant tasks of integrated simulation was to recompile Type101 with particular macro to update the boundary conditions for Fluent. Type 101 (DA.Arias 2006) was generally represented as the TRNSYS component that is used to create the script file for the boundary condition setting of the CFD model. Corresponding to different design model, inlet/ output nomination, controlled parameters/ purposes, and convergence requirement for each analytical case, the macro must be modified and recompiled accordingly. Since there was no always applicative simulation environment that covers all the architectural form or HAVC system or indoor environmental design with sufficient flexibility, it was appropriate to update the macro in order to fulfill user’s desire and requirements. To this purpose, understandings of computational languages, compile mechanism, theoretical fundamentals of both BES and CFD were required.

4.3.6 Building the TRNDll.dll

The following instructions give detailed information on the compiler settings used to rebuild TRNDll.dll. It is provided by TRNSYS 17 and used as reference. They explain how this project was added to the TRNSYS 17 IVFCXE2013 Solution and give all modifications to default settings.

Open the Solution in Intel Visual

Fortran: %TRNSYS17%\Compilers\IvfCXE2013\IvfCXE2013.sln

Go to File / New / Project

Under "Project Types", select "Intel ® Fortran Project"

Under "Templates", select "Dynamic-link Library"

Type in "TRNDll" for the name.

Check the "Add to solution" radio button. The Location will be adapted automatically and should say: %TRNSYS17%\Compilers\IvfCXE2013

Click OK.

Open the "Fortran Dynamic Link Library Wizard" window.

Select "DLL Settings" under "Overview", in the left frame, and choose "Empty Project" as

Click "Finish".

Create the following Project file with default

settings: %TRNSYS17%\Compilers\IvfCXE2011\TRNDll\ TRNDll.vfproj

Here ignore to introduce the following steps for Project setting and building TRNDll.dll with visual studio. (Detail procedure can be seen in TRNSYS 17). Then create a new project in to existing project and make the important liking to the TRNSYS library for rebuilding new types.

Link to TRNSYS library

Right click My Type libraries, add the TRNDll.lib file.

Location: %TRNSYS17%\Exe\TRNDll.lib.

Only when this building is succeeded and active with FORTRAN compiler, the new or updated type for integrated simulation that can be recognized by TRNSYS module.