HVAC Design Optimization for Pharmaceutical Facilities with BIM and CFD

Abstract

Building Information Modeling (BIM) has been widely used in the past decade to enhance the design quality of Heating, Ventilation, and Air Conditioning (HVAC) systems. However, in specialized areas such as pharmaceutical facilities, HVAC design has traditionally relied on Computer-Aided Design (CAD) drawings. This conventional approach does not allow for the simulation of temperature distribution or the verification of system efficiency, which may lead to design failures. To address these challenges in pharmaceutical facilities, this study proposed a BIM-based approach for optimizing HVAC design with Computational Fluid Dynamics (CFD). By employing CFD to simulate the dynamic airflow conditions of pharmaceutical clean rooms, the effectiveness of HVAC systems can be verified. A case study of a clean room HVAC design is presented to demonstrate the workflow. The results of the case study indicated that the pharmaceutical temperature requirements were met within 1 °C during the design optimization simulation, and there was a 95% match in the 72 h temperature mapping test during site validation. The results confirmed that using CFD with BIM not only successfully simulates the design intentions of indoor air quality but also suggests HVAC system optimization for the required clean room design. The findings of this paper contribute to the body of knowledge on overcoming the limitations of the traditional CAD-based HVAC design process and provide valuable insights on optimizing HVAC design with BIM and CFD technologies.

1. Introduction

Ensuring precise temperature control is a paramount objective for Heating, Ventilation, and Air-Conditioning (HVAC) systems employed in pharmaceutical quality control (QC) laboratories [1]. The stability of temperature plays a pivotal role in maintaining the integrity of experimental data. Inadequate thermal conditions can result in a decline in the first pass yield, as deviations in processing equipment tolerance occur when the environment is excessively warm or cold [2]. Additionally, prior research has demonstrated a positive correlation between thermal satisfaction and the experimenter’s perception of the QC lab air quality, lighting, and acoustic environment [3].

Traditionally, pharmaceutical clean room designs rely on 2D Computer-Aided Design (CAD) drawings that often fail to generate simulation models capable of validating the HVAC system’s effectiveness in temperature distribution [4]. However, researchers have recently integrated optimization algorithms and artificial intelligence with Computational Fluid Dynamics (CFD) to address this limitation. Previous CFD simulations simplified the process by assuming fixed values for heat load equipment during heating injections [5]. Unfortunately, this constant heat load assumption fails to account for the variability of heat load in different environments, resulting in inaccurate predictions of the clean room’s actual conditions. Over the past decade, Building Information Modeling (BIM) has gained increasing traction in the HVAC industry, offering significant improvements to design and engineering processes. However, its application in the design of pharmaceutical clean rooms, which necessitates careful consideration of indoor air quality and thermal conditions, remains limited.

This research paper presents a new approach to answer the following research question: How to simulate the HVAC design of pharmaceutical clean rooms with BIM, optimize the dynamic airflow conditions with CFD, and verify the effectiveness of HVAC systems on site? A case study was used to meet the design requirements of the dynamic thermal environment of a pharmaceutical QC lab utilizing displacement ventilation. The case study aimed to achieve a thermally stable environment by employing a combination of BIM, CFD simulations, and a genetic algorithm for validation and optimization, which has not been found in the current literature. This paper presents the workflow of an innovative integration of BIM and CFD for HVAC design optimization with a case study for validating the results, including the following sections:

• Literature Review: This section reviews relevant scholarly work published in the last two decades on HVAC design and CFD simulations.

• Methodology: This section details the research theoretical methods of establishing the design requirements, modeling the airflow conditions, and creating the evolution simulation with a genetic algorithm.

• Model Development: This section describes the workflow of developing the BIM model of the experiment facility, including creating boundary conditions and dividing it with CFD meshing.

• Simulation Results: This section presents the results of the initial simulation of air temperature.

• Design Optimization: This section details the design changes based on the initial simulation and optimization results.

• Site Validation: This section validates the design optimization by collecting temperature data on the real experiment site.

• Limitations: This section summarized the limitations and challenges that the CFD simulation study has in HVAC design optimization.

The findings of this study have demonstrated that with the integration of BIM and CFD, the airflow patterns, temperature distribution, and contaminant dispersion can be accurately simulated and verified to reduce the need for physical prototypes and testing, leading to time and cost savings during the design phase. The proposed approach enables the identification of design improvements that enhance energy efficiency and operational performance, resulting in long-term cost savings and reducing the risk of costly modifications and retrofits post-construction. This study contributes to expanding our knowledge of overcoming the constraints associated with traditional CAD-based HVAC design processes while offering valuable insights into the optimization of HVAC design through the integration of BIM and CFD technologies.

2. Literature Review

In pursuing a thermally stable environment in QC laboratories, design engineers have traditionally relied on calculating the cooling or heating load [6]. This calculation helps determine the appropriate air supply conditions or radiator capacity, ensuring that the air temperature in the QC lab falls within the required production zone. While this method generally achieves about 70% of the thermal environment requirement, the remaining 30% can be influenced by individual differences and other environmental factors such as mean radiant temperature, air speed, and humidity [7].

Researchers have taken steps to enhance thermal stability by incorporating optimization algorithms to design or regulate HVAC systems based on a combination of thermal environment parameters [8] employed a genetic algorithm to create an improved thermal environment through a combined HVAC system [9] integrated a genetic algorithm and energy simulation tool to control the HVAC system, aiming to achieve optimal thermal performance while minimizing energy consumption. However, it is important to note that these studies primarily focused on estimating average thermal environment parameters in QC laboratories [10].

Enclosed environments often exhibit heterogeneous thermal conditions, including vertical air temperature gradients, non-uniform radiation, and spatial variations in air distribution [11]. Overlooking such non-uniformity and relying on averaged parameters during the design process can be less informative and lead to errors [12]. To leverage the benefits of the non-uniform thermal environment captured by CFD, coupling it with a multi-node equipment thermoregulation model proves advantageous. This coupling not only addresses inaccuracies in CFD predictions resulting from fixed heating injections of heat load equipment [13] but also enables the optimization design of enclosed environments, which was previously unexplored using this approach [14].

Previous studies have commonly utilized the Predicted Mean Vote (PMV) as a thermal injection index to determine the level of thermal stability [15]). Alternatively, some studies have used a tuned PMV tailored to a particular environment [16]. However, PMV is limited in its ability to predict thermal sensations in non-uniform thermal environments accurately. Consequently, employing PMV to estimate thermal injection in such scenarios would waste valuable spatial information that could be obtained through CFD simulations. By using an appropriate turbulence model, CFD simulations can efficiently generate informative results and accurately simulate air distribution in QC lab environments, thereby determining the thermal environment [17]. Additionally, optimization algorithms can be employed to identify optimal design solutions aligned with specific objectives [18].

Pharmaceutical manufacturing processes demand precise control over temperature, humidity, and air quality to ensure product safety and regulatory compliance [19]. HVAC systems in pharmaceutical facilities must meet stringent standards such as Good Manufacturing Practices (GMP) to prevent contamination and maintain product integrity. CFD simulations enable virtual modeling and analysis of pharmaceutical facilities’ airflow, thermal conditions, and contaminant dispersion [20]. CFD simulations also help optimize airflow distribution to ensure uniform environmental conditions throughout the QC lab area. CFD analysis aids in the design of HVAC systems for contaminant control, ventilation effectiveness, and energy efficiency, contributing to operational cost savings [21].

CFD is crucial in optimizing pharmaceutical facility HVAC designs, offering detailed insights into airflow dynamics and thermal behavior [22]. By leveraging CFD simulations, designers can enhance HVAC system performance, regulatory compliance, and operational efficiency in pharmaceutical manufacturing environments. Validation of CFD models against experimental data and field measurements is essential to ensure the accuracy and reliability of simulation results. Verification techniques, such as grid independence studies and sensitivity analyses, help assess the robustness of CFD simulations in predicting HVAC system performance. Numerous case studies demonstrate the successful application of CFD in optimizing HVAC designs for pharmaceutical facilities [23].

Challenges in CFD-based HVAC optimization include computational complexity, model validation, and uncertainty quantification [24]. Continued research and development efforts are necessary to overcome existing challenges and further advance the application of CFD in pharmaceutical HVAC design [25]. Future research should focus on advancing CFD methodologies for modeling complex HVAC systems, integrating transient phenomena, and addressing emerging challenges such as sustainable HVAC design and resilience to climate change [26].

The literature reviewed in this study was gathered from a carefully curated selection of academic journals and conferences focusing on HVAC design in construction management. The review encompasses 26 publications spanning from 1987 to 2024, covering various topics including CFD methods, review of the literature, HVAC design, BIM, site validation, and pharmaceutical facilities, as depicted in Table 1 and Figure 1.

Figure 2 illustrates the relationship between these research topics by identifying the connections between the keywords of each reviewed research topic using text mining. For example, within the reviewed 26 papers, CFD, BIM, HVAC, and pharmaceutical were the most frequent keywords, and they were closely connected with temperature, simulation, diffuser, and validation, respectively.

3. Methodology

3.1. Design Requirements

The following simulation objectives were set forth to meet the design requirements of the pharmaceutical QC lab:

• Develop a CFD model based on the layout of the pharmaceutical QC lab.

• Validate the proposed design with CFD software ANSYS Fluent v18.2 for a target temperature of 22 °C ± 1 °C.

• Revise the proposed design based on the CFD simulation results.

• Implement the revised design in the construction stage of the pharmaceutical QC lab.

3.2. Air Flow Model

Since the air flow in the pharmaceutical QC lab is generally turbulent, this research adopted the renormalization group (RNG) k-ε model to simulate the turbulence [18]. The scalar equations for parameters k, ε, and T were solved iteratively with an implicit scheme. Overall, this study discretized the convection terms by a bounded second-order upwind scheme. The inputs for the governing equations were thermo-fluid boundary conditions. Within these boundary conditions, this investigation assigned the air velocity Uinlet and temperature Tinlet from the inlet as the design variables for demonstration. The air supply velocity is denoted by the velocity magnitude |U|, azimuthal angle θ, and polar angle ϕ. The surface temperature of the occupants was initialized by a uniform value Tsf. During the simulations, the heating injection of the equipment was updated according to the calculations.

Based on the model, the flow of gas in the QC laboratory is assumed in a relatively stable and steady state. Air can be regarded as an incompressible fluid without external forces, so the ideal gas state equation can be used to perform calculations. At the same time, according to the flow characteristics of the air in the QC experiment area, it can be divided into turbulent flow, which is applicable to the three conservation laws of mass conservation, momentum conservation, and energy conservation. Based on this design and the layout of the air outlets, the form of the airflow in the QC experiment area is upward supply and upward return.

The following assumptions were made from the above airflow model:

• The treated air in the QC laboratory area should be defined as a Newtonian fluid, of which the volume does not change.

• The gas in the QC laboratory area is relatively viscous and the trajectory of the gas is relatively fixed, so the thermal influence caused by the fluid viscous force cannot be included.

• The impact of the external environment on the indoor air flow direction, temperature, and humidity can be ignored. This is because the entire QC laboratory experimental area passed the airtightness test during the construction stage, making gas leakage nearly impossible.

• Since there is no gas phase transition during the entire fluid motion simulation process, the influence of latent heat and gas energy dissipation is not considered.

Based on the above assumptions, the airflow movement in the QC experimental area satisfies the Reynolds-averaged Navier–Stokes equation as follows:

(1)$u_j\frac{\partial u_i}{\partial x_j}=f_i-\frac{1}{\rho }\frac{\partial p}{\partial x_i}+\frac{\partial }{\partial x_j}\left[\left(v+v_z\right)\left(\frac{\partial u_i}{\partial x_j}\frac{\partial u_j}{\partial x_i}\right)\right]$

Meanwhile, considering the accuracy of the model processing and the possible energy loss-related concerns, the standard k-ε double-equation model is selected for correction calculations.

3.3. Genetic Algorithm

The simulation was created to identify the design variables Uinlet and Tinlet for the specific location of the QC lab area. A genetic algorithm was developed to imitate the evolution theory as explained below:

• Problem Definition: Define the objectives of the simulation, including the physical phenomena to be modeled. Specify the geometry of the domain of interest, including boundaries, inlets, outlets, and solid surfaces.

• Pre-processing: Import or create the domain’s geometric model using CAD software AutoCAD 2022. Clean up and prepare the geometry, including mesh generation, to discretize the domain into computational cells or elements. Define boundary conditions, including fluid properties, inflow/outflow conditions, wall properties, and initial conditions.

• Mesh Generation: Generate a computational mesh or grid that discretizes the domain into small elements or cells. Ensure mesh quality, including resolution, aspect ratio, and boundary layer refinement, to capture flow features accurately. Perform grid independence studies to assess the sensitivity of the results to mesh resolution.

• Model Setup: Select appropriate physical models and equations to represent fluid behavior, turbulence, heat transfer, and other relevant phenomena. Choose turbulence models (e.g., k-epsilon, Large Eddy Simulation) based on the available flow characteristics and computational resources. Define solver settings, including discretization schemes, convergence criteria, and solution methods.

• Simulation: Solve the governing equations numerically using a CFD solver, such as finite volume or finite element methods. Iteratively solve flow variables (e.g., velocity, pressure, temperature) within each computational cell or element until convergence is achieved. Monitor solution convergence and adjust solver settings as needed.

• Post-processing: Analyze and visualize simulation results using post-processing tools. Generate contour plots, streamlines, velocity vectors, and other visualizations to understand flow patterns and phenomena. Extract quantitative data such as pressure drops, heat transfer coefficients, and fluid velocities for further analysis and comparison.

• Validation and Verification: Compare simulation results with experimental data or analytical solutions to validate the accuracy of the CFD model. Conduct sensitivity analyses and uncertainty quantification to assess the robustness of the results and identify sources of error.

• Interpretation and Reporting: Interpret simulation results in the context of the problem objectives and provide insights into fluid behavior, heat transfer, and other phenomena. Document the simulation methodology, assumptions, and results in a comprehensive report or presentation. Communicate findings to stakeholders and make recommendations for design improvements or further analysis.

Figure 3 illustrates the workflow of the genetic algorithm. In the current iteration, the coupled CFD-HTM simulations determine each individual’s fitness value (OF value). The genetic algorithm then conducts the selection, crossover, and mutation process on the design variables of individuals by using the OF value in the current iteration to generate a new population. Next, the design variables are encoded into binary codes for genetic operations. For the first iteration, the design variables for each individual are created randomly within their given ranges. The convergence criterion is when the calculation has stopped.

4. Model Development

4.1. BIM Modeling

In order to precisely simulate the heat generation of the pharmaceutical equipment, air flow of the HVAC system, and heat transfer between difference zones in the pharmaceutical QC experiment area, an accurate BIM model needs to be first created. Autodesk Revit 2022 was used to develop the BIM model based on detailed design drawings provided by the facility, as shown in Figure 4, and the form of air flows inside the QC experiment area was then simulated and calculated in all directions through the air flow movement equation mentioned earlier.

In the design of the HVAC system of the pharmaceutical QC experiment area, there is one Air Handling Unit (AHU) with 16 grille diffusers for air supply and eight Fan Coil Units (FCUs) with 16 swirl diffusers for air supply. The AHU has an air volume capacity of 15,280 Cubic Meters per Hour (CMH) with a supply air temperature of 11.4 °C at the end of the tuyere and an absolute humidity of 8.2 g/kg. The FCU has an air volume capacity of 1910 CMH with a cooling capacity of 5.8 kW. The exhaust air of the experiment area is divided into AHU return air and process equipment exhaust air. There are 10 AHU return air outlets with an exhaust air volume of 873 CMH for each outlet. The exhaust air of process equipment is vented directly to the outside atmosphere through an activated carbon air filter by a 900 CMH biological safety cabinet, a 2000 CMH fume hood, and a 750 CMH balance cover.

The heat output of the pharmaceutical QC experiment area includes the heat transfer of the exterior walls, the heat dissipation of lighting fixtures, the heat dissipation of the staff in the QC experiment area, and the heat production of process equipment, which is detailed in Figure 4.

4.2. Boundary Conditions

The boundary definition of this CFD fluid simulation adopts the boundary condition of the air volume inlet, and the key input parameters are the air supply volume and the temperature value of different end air outlets. Based on the previous literature and past project experience, a coefficient of 1.0 and 1.3 was selected for the turbulent kinetic energy and turbulent dissipation rate, respectively, in the double equation model. Since the louver air vents were used with orifice air outlets as the end air outlets, the directions of supply air velocity include horizontal X and vertical Y, where the X direction is along the plane of the tuyere and the Y direction is along the damper. Figure 5 demonstrates an example of the horizontal and vertical distances the supply air has traveled from the center of the tuyere when the velocity is at 0.5 m/s.

4.3. CFD Meshing

Since smaller grids will more accurately represent the actual movement of the gas in CFD simulation, Altair HyperMesh 2022 was used to create a 3D model in this simulation with 8.4 million elements through structured and unstructured grids. Ansys Fluent was applied next to calculate and analyze the created 3D model. Figure 6 illustrates the meshing process from a 2D design drawing to the 3D element grid model.

5. Simulation Results

The results of the initial simulation were reported by Liu and Huang [27] and were restated in this paper. The various input parameters defined earlier were entered in Ansys Fluent to start the numerical simulation. After 50,000 iterations of calculation, the requirements for the convergence level have been reached, and the simulation results of a heat map at 1 m above the floor are shown in Figure 7. The simulation suggested that:

• The average temperature of the pharmaceutical QC room was about 26 °C.

• The temperature in the QC laboratory area did not meet the requirement of 22 °C ± 1 °C.

• Overheating occurred in the center area of the room.

• There were overcooled areas at the northeast and northwest corners of the QC room.

Figure 7

Heat map of air temperature at 1 m above floor (average 24.2 °C). Reprinted/adapted with permission from Ref. [27]. 2023, Liu and Huang.

[Figure omitted. See PDF]

Through the analysis results and visual heat map of air temperature, it was clear that the CFD simulation did not meet the temperature requirements of the pharmaceutical QC room design due to the following possible factors:

• High heat production equipment is located in a small area, resulting in an unevenly distributed heat load in the QC test area and high temperature in the central area.

• Few air outlets are designed for high heat load areas, resulting in low air volume and flow velocity. The trajectory of the air flow did not sufficiently mix with cold air to fully cool the entire room.

• Only one temperature sensor is placed in the QC area, which is not sensitive to temperature changes and lacks the flexibility to be adjusted due to heat load changes.

6. Design Optimization

6.1. Proposed Optimization

Based on the data analysis of CFD simulation results, several optimization actions were suggested and implemented, including:

• The heating load of the entire area has been recalculated accordingly and a different AHU system has been selected to meet the requirement of heating load change.

• The original grille diffusers have been replaced by four-way swirl diffusers with higher wind speeds and a larger mixing area. A comparison between the original and new diffuser is demonstrated in Figure 8.

• The positioning of the original terminal air outlet has been optimized to facilitate thorough mixing of the cold air, thereby better fulfilling the airflow requirements.

• The original automatic control strategy has been updated. The entire area has been divided into five independent temperature control zones, as shown in Figure 9. High-sensitivity temperature sensors are used in each zone to ensure the accuracy of the signal output. Furthermore, the parameters of the Programmable Logic Controller (PLC) have also been modified based on the simulation results.

After the proposed optimization, the parameters of the equipment are summarized as follows:

• The number of terminal AHU system diffusers has been increased to 19.

• The number of FCU swirl diffusers has been increased to 48.

• The modified design airflow of the AHU is 12,730 CMH with an air outlet temperature of 11.4 °C and an absolute humidity of 8.2 g/kg.

• The modified design airflow of each individual FCU is 2010 CMH with a cooling capacity of 13 kW.

• The exhaust air system consists of the AHU return air system and process equipment exhaust system. The number of AHU return air system outlets has been changed to four with an exhaust volume of 1525 CMH per outlet. In terms of the process equipment exhaust system, the air volume of each biological safety cabinet has been changed to 1150 CMH, whereas the air volume of the fume hood has been changed to 2000 CMH.

• The details of the optimization changes of the equipment parameters are compared in Table 2.

6.2. Optimization Results

Based on the redefined input parameters. The software ANSYS Fluent v18.2 was utilized for mathematical simulation. the convergence criteria were achieved after 35,000 iterations. The final simulation results are as follows:

• The temperature in the QC laboratory area meets the requirement of 22 ± 1 °C.

• The average temperature in the entire room is approximately 22.3 °C.

• There are no areas in the room experiencing overheating.

• There might be a possibility of cold spots in the eastern corner due to the unstable operation of process equipment. Therefore, it is necessary to enhance the response speed of the FCU’s automatic control intervention to avoid such occurrences.

• For specific data trends, please refer to Figure 10.

7. Site Validation

The new QC laboratory was finally built based on the revised CFD results report. Accordingly, a site validation activity was conducted to verify the performance of the temperature deviation.

The specific method for the test is to use temperature and humidity data loggers to monitor the indoor QC lab temperature at 1 m above the floor, as demonstrated in Figure 11. The entire duration of this test is 72 h, and the record interval time is 5 min. There is a total of 22 units of data logger were placed inside the QC laboratory and the height of the data logger is 1 m above the ground floor. The detailed layout is shown in Figure 12. The acceptance criteria are that each individual data logger’s temperature range should be within 21–23 °C.

After conducting the continuous 72 h temperature mapping test according to the above testing strategy, the data was collected and analyzed from data loggers. This indicated that all test points consistently meet the design requirement of maintaining temperatures within the range of 21–23 °C throughout the whole duration. Meanwhile, 95% of the data could match the CFD reports’ by comparing the data from the CFD simulation report with the actual measurement values. Which has also verified the accuracy of the CFD simulation data. The detailed 72 h temperature trend graph is shown in Figure 13.

8. Limitations

While CFD offers numerous benefits for HVAC design optimization in pharmaceutical facilities, it also has limitations and challenges. The fundamental limitations of using CFD for HVAC design optimization in pharmaceutical facilities are listed below:

• Computational Complexity: CFD simulations can be computationally intensive, requiring significant computational resources (e.g., CPU time, memory, storage) and expertise to set up and run. Complex HVAC systems with intricate geometries, multiple components, and transient phenomena may increase computational demands.

• Model Assumptions and Simplifications: Simplifications and assumptions are often made in CFD models to reduce computational costs and complexity, potentially leading to inaccuracies in results. Assumptions regarding boundary conditions, turbulence models, and fluid properties may not always capture the full complexity of real-world phenomena.

• Validation and Verification Challenges: Validating CFD results against experimental data or field measurements can be challenging, especially in pharmaceutical facilities with limited access to real-world data. Verification of CFD models to ensure numerical stability, convergence, and grid independence requires careful consideration and may not always be straightforward.

• Uncertainty and Sensitivity Analysis: Uncertainties in input parameters, model assumptions, and boundary conditions can propagate through CFD simulations, affecting the reliability and robustness of results. Sensitivity analyses are necessary to assess the impact of uncertainties on simulation outcomes and identify sources of variability.

• Limited Availability of Input Data: Accurate input data, including fluid properties, HVAC system specifications, and environmental conditions, are crucial for meaningful CFD simulations. Obtaining comprehensive and reliable input data, especially for pharmaceutical facilities with specialized requirements, can be challenging and may require experimental testing or empirical correlations.

• Interdisciplinary Collaboration: Successful implementation of CFD for HVAC design optimization in pharmaceutical facilities requires collaboration between HVAC engineers, pharmaceutical experts, CFD analysts, and other stakeholders. Effective communication and interdisciplinary collaboration are essential to ensure that CFD simulations address pharmaceutical manufacturing processes’ specific needs and requirements.

• Regulatory Compliance: Meeting regulatory requirements, such as Good Manufacturing Practices (GMP) and cleanroom standards, adds complexity to HVAC design optimization in pharmaceutical facilities. CFD simulations must demonstrate compliance with regulatory standards, which may require additional validation and documentation efforts. Despite these limitations, CFD remains a valuable tool for HVAC design optimization in pharmaceutical facilities, providing insights into airflow patterns, temperature distribution, contaminant control, and energy efficiency. Addressing these limitations requires careful consideration of model assumptions, validation practices, uncertainty quantification, and interdisciplinary collaboration to ensure the accuracy and reliability of CFD simulations.

9. Conclusions

This study developed a coupling scheme to incorporate multisegmented heating injection into the CFD computation of the thermal environment in a QC lab room. The CFD analysis provided varying air temperature and velocity boundaries for different heating injection equipment, while the equipment, in turn, supplied temperature information for various segments, acting as thermo-boundaries for the CFD simulation. The accuracy of the calculated local equipment temperatures was validated against measured data from temperature mapping tests. Subsequently, the developed CFD method was applied in an inverse design process to determine the optimal supply air velocity, temperature, and angle for the diffuser of a displacement ventilation system in a typical office during summer conditions. The obtained results yielded the following conclusions:

• The coupled CFD approach successfully predicted the temperatures resulting from the heating injection equipment with an acceptable level of accuracy.

• By considering the target temperature as the design objective, the inverse design method effectively identified the air supply conditions that minimized the objective function.

• The rapid convergence of the cases indicated that the sensitivity to design variables heavily relied on the equipment’s heating injection characteristics.

• When appropriate heating injection conditions were applied to the equipment, setting the QC room temperature as the design objective led to similar optimal solutions obtained through the coupled CFD model.

• Using CFD with BIM for HVAC design optimization in pharmaceutical facilities reveals essential considerations for decision making regarding cost benefits and underlying assumptions. CFD allows for the virtual testing of HVAC designs, reducing the need for physical prototypes and testing, and leading to time and cost savings during the design phase. By accurately simulating airflow patterns, temperature distribution, and contaminant dispersion, CFD enables the identification of design improvements that enhance energy efficiency and operational performance, resulting in long-term cost savings simulations that can identify potential issues and optimize HVAC designs before construction, reducing the risk of costly modifications and retrofits post-construction. Compliance with regulatory standards through CFD simulations can prevent expensive penalties, fines, or product recalls due to inadequate HVAC systems.

In summary, while CFD offers significant cost benefits by enabling efficient design optimization and risk mitigation in pharmaceutical HVAC systems, careful consideration of underlying assumptions is essential to ensure the accuracy and reliability of simulation results. Addressing uncertainties and validating models against experimental data are critical steps in leveraging CFD’s full potential for HVAC design optimization in pharmaceutical facilities.

Author Contributions

Conceptualization, L.L.; methodology, L.L.; writing—original draft preparation, L.L.; writing—review and editing, Y.H.; supervision, Y.H. All authors have read and agreed to the published version of the manuscript.

Data Availability Statement

The original contributions presented in the study are included in the article, further inquiries can be directed to the corresponding author.

Conflicts of Interest

The authors declare no conflicts of interest.

Footnotes

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content.

Figures and Tables

Figure 1. Percentage of the topics of reviewed literature.

Figure 2. Relationship between research topics using text mining.

Figure 3. Flow chart of the genetic algorithm of the simulation model. Reprinted/adapted with permission from Ref. [27]. 2023, Liu and Huang.

Figure 4. BIM model and heat load of process equipment. Reprinted/adapted with permission from Ref. [27]. 2023, Liu and Huang.

Figure 5. Horizontal and vertical distances from center of tuyere at velocity of 0.5 m/s. Reprinted/adapted with permission from Ref. [27]. 2023, Liu and Huang.

Figure 6. CFD meshing from 2D drawing to 3D model. Reprinted/adapted with permission from Ref. [27]. 2023, Liu and Huang.

Figure 8. Comparison between the original and new diffuser.

Figure 9. Temperature control zone layout.

Figure 10. Updated heat map of air temperature at 1 m above floor (average 22.7 °C).

Figure 11. Temperature and humidity data logger placed at 1 m above the floor.

Figure 12. Layout of data logger positions labeled with red dots.

Figure 13. 72 h temperature trend graph.

Table 1

Summary of the topics of reviewed literature.

Topic	No. of Papers	Percentage
CFD Methods	5	19%
Review of Literature	3	12%
HVAC Design	8	31%
BIM	4	15%
Site Validation	3	12%
Pharmaceutical Facilities	3	12%
Total	26	100%

Table 2

Comparison of optimization changes of the equipment parameters.

Optimization	Equipment	Original Design	Optimized Design
1	No. of AHU diffusers	16	19
2	AHU volume/(CMH)	15,280	12,730
3	Type of FCU diffuser	Grille	4-way swirl
4	No. of FCU	8	5
5	FCU volume/(CMH)	1910	2010
6	FCU cooling capacity(kW)	5.8	13
7	No. of FCU swirl diffusers	16	48
8	Control zones	1	5

References

1. Martins, L.A.; Soebarto, V.; Williamson, T. A systematic review of personal thermal comfort models. Build. Environ.; 2022; 207, 108502. [DOI: https://dx.doi.org/10.1016/j.buildenv.2021.108502]

2. Geng, Y.; Ji, W.; Lin, B.; Zhu, Y. The impact of thermal environment on occupant IEQ perception and productivity. Build. Environ.; 2017; 121, pp. 158-167. [DOI: https://dx.doi.org/10.1016/j.buildenv.2017.05.022]

3. Pernigotto, M.; Tarantini, G.; Gasparella, A. A co-citation analysis on thermal comfort and productivity aspects in production and office buildings. Buildings; 2017; 7, 36.

4. Liu, W.; Zhang, T.; Xue, Y.; Zhai, Z.J.; Wang, J.; Wei, Y.; Chen, Q. State-of-the-art methods for art methods for inverse design of an enclosed environment. Build. Environ.; 2015; 91, pp. 91-100. [DOI: https://dx.doi.org/10.1016/j.buildenv.2015.02.041] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/32288031]

5. Zhang, T.; Liu, Y.; Rao, Y.; Li, X.; Zhao, Q. Optimal design of building environment with hybrid genetic algorithm, artificial neural network, multivariate regression analysis and fuzzy logic controller. Build. Environ.; 2020; 175, 106810. [DOI: https://dx.doi.org/10.1016/j.buildenv.2020.106810]

6. Amahjour, N.; Sofi, A.; Kamal, T.; El Kharrim, A. Investigating the impact of HVAC and Sanitizer system design on the transmission of SARS-CoV-2 in Hospital Isolation Units. Chaos Solitons Fractals; 2024; 178, 114345. [DOI: https://dx.doi.org/10.1016/j.chaos.2023.114345]

7. Zhang, Z.; Zhang, W.; Zhai, Z.J.; Chen, Q.Y. Evaluation of various turbulence models in predicting airflow and turbulence in enclosed environments by CFD: Part 2—Comparison with experimental data from literature. HVACR Res.; 2007; 13, pp. 871-886. [DOI: https://dx.doi.org/10.1080/10789669.2007.10391460]

8. Du, H.; Lian, Z.; Lai, D.; Duanmu, L.; Zhai, Y.; Cao, B.; Zhang, Y.; Zhou, X. Comparison of thermal comfort between radiant and convective systems using field test data from the Chinese Thermal Comfort Database. Build. Environ.; 2022; 209, 108685. [DOI: https://dx.doi.org/10.1016/j.buildenv.2021.108685]

9. Satrio, P.; Mahlia, T.M.I.; Giannetti, N.; Saito, K. Optimization of HVAC system energy consumption in a building using artificial neural network and multi-objective genetic algorithm. Sustain. Energy Technol. Assess.; 2019; 35, pp. 48-57.

10. Pombeiro, H.; Machado, M.J.; Silva, C. Dynamic programming and genetic algorithms to control an HVAC system: Maximizing thermal comfort and minimizing cost with PV production and storage. Sustain. Cities Soc.; 2017; 34, pp. 228-238. [DOI: https://dx.doi.org/10.1016/j.scs.2017.05.021]

11. Zhao, X.; Yin, Y.; He, Z.; Liu, X. Inverse design of indoor radiant terminal using the particle swarm optimization method with topology concept. Build. Environ.; 2021; 204, 108117. [DOI: https://dx.doi.org/10.1016/j.buildenv.2021.108117]

12. Wang, X.; Zhao, J.; Wang, F.; Song, B.; Zhang, Q. Air supply parameter optimization of a custom nonuniform temperature field based on the POD method. Build. Environ.; 2021; 206, 108328. [DOI: https://dx.doi.org/10.1016/j.buildenv.2021.108328]

13. Tanabe, S.I.; Kobayashi, K.; Nakano, J.; Ozeki, Y.; Konishi, M. Evaluation of thermal comfort using combined multi-node thermoregulation (65 mn) and radiation models and computational fluid dynamics (CFD). Energy Build.; 2002; 34, pp. 637-646. [DOI: https://dx.doi.org/10.1016/S0378-7788(02)00014-2]

14. Gao, N.; Niu, J.; Zhang, H. Coupling CFD and human body thermoregulation model for the assessment of personalized ventilation. HVACR Res.; 2006; 12, pp. 497-518. [DOI: https://dx.doi.org/10.1080/10789669.2006.10391191]

15. Fanger, P.O. Calculation of thermal comfort-introduction of a basic comfort equation. ASHRAE Trans.; 1967; 73, pp. III4.1-III4.20.

16. Cui, W.; Yang, Q.; Zhu, Y. Field study of thermal environment spatial distribution and passenger local thermal comfort in aircraft cabin. Build. Environ.; 2014; 80, pp. 213-220. [DOI: https://dx.doi.org/10.1016/j.buildenv.2014.06.004]

17. Zhou, K.; Li, F.; Cai, H.; Jing, Y.; Zhuang, J.; Li, M.; Xing, Z. Estimation of the natural gas leakage source with different monitoring sensor networks in an underground utility tunnel from the perspectives of energy security. Energy Build.; 2022; 254, 111645. [DOI: https://dx.doi.org/10.1016/j.enbuild.2021.111645]

18. Li, M.; Li, F.; Jing, Y.; Zhang, K.; Cai, H.; Chen, L.; Zhang, X.; Feng, L. Estimation of pollutant sources in multi-zone buildings through different deconvolution algorithms. Build. Simul.; 2022; 15, pp. 817-830. [DOI: https://dx.doi.org/10.1007/s12273-021-0826-3] [PubMed: https://www.ncbi.nlm.nih.gov/pubmed/34545299]

19. Zhang, J.; Poon, K.H.; Kwok, H.L.; Hou, F.; Cheng, C.P. Predictive control of HVAC by multiple output GRU-CFD integration approach to manage multiple IAQ for commercial heritage building preservation. Build. Environ.; 2023; 245, 110802. [DOI: https://dx.doi.org/10.1016/j.buildenv.2023.110802]

20. Lan, H.; Hou, H.; Gou, Z.; Wong, M.; Wang, Z. Computer vision-based smart HVAC control system for university classroom in a subtropical climate. Build. Environ.; 2023; 242, 110592. [DOI: https://dx.doi.org/10.1016/j.buildenv.2023.110592]

21. Jiang, B.; Xiao, Q.; Wang, X.; Liang, Z.; Yang, X. A literature review on the squirrel-cage fans using in HVAC equipment: Powerful, efficient, and quiet operation. J. Build. Eng.; 2023; 73, 106691. [DOI: https://dx.doi.org/10.1016/j.jobe.2023.106691]

22. Heidarykiany, R.; Ababei, C. HVAC energy cost minimization in smart grids: A cloud-based demand side management approach with game theory optimization and deep learning. Energy AI; 2024; 16, 100362. [DOI: https://dx.doi.org/10.1016/j.egyai.2024.100362]

23. Kanaan, M. CFD optimization of return air ratio and use of upper room UVGI in combined HVAC and heat recovery system, Case Studies in Thermal Engineering. Case Stud. Therm. Eng.; 2019; 15, 100535. [DOI: https://dx.doi.org/10.1016/j.csite.2019.100535]

24. Yakhot, V.; Orszag, S.A. Renormalization group analysis of turbulence. I. Basic theory. J. Sci. Comput.; 1986; 1, pp. 3-51. [DOI: https://dx.doi.org/10.1007/BF01061452]

25. Nguyen, A.T.; Pham, D.H.; Oo, B.J.; Santamouris, M.; Ahn, Y.; Benson, L.T.H. Modelling building HVAC control strategies using a deep reinforcement learning approach. Energy Build.; 2024; 310, 114065. [DOI: https://dx.doi.org/10.1016/j.enbuild.2024.114065]

26. William, M.A.; Suárez-López, M.J.; Soutullo, S.; Fouad, M.M.; Hanafy, A.A.; El-Maghlany, W.M. Multi-objective integrated BES-CFD co-simulation approach towards pandemic proof buildings. Energy Rep.; 2022; 8, pp. 137-152. [DOI: https://dx.doi.org/10.1016/j.egyr.2022.06.091]

27. Liu, L.; Huang, Y. Optimizing HVAC Design for Pharmaceutical Requirements with Computational Fluid Dynamics. Proceedings of the Computing in Civil Engineering 2023: Resilience, Safety, and Sustainability; Corvallis, OR, USA, 25–28 June 2023; ASCE: Reston, VI, USA, pp. 1005-1013.