Autonomous HVAC CFD Precision - Case Studies and Insights!
Thursday, October 19, 2023
Autonomous HVAC CFD Precision - Case Studies and Insights!
Karan Beeshm
Blog Author - Karan Beeshm
Written by Karan Beeshm
5 Minutes Reading
5 Minutes Reading
CFD analysis comprises several stages, including geometry model creation, meshing, and solving for the given physics. In various industries, this analysis is handled either by an in-house CFD team or outsourced to consultancy firms. However, the process demands High-Performance Computing (HPC) machines, licensed and maintained software, resulting in significant costs.
Autonomous HVAC CFD revolutionizes this analysis, particularly for indoor rooms, by automating the entire process. Simply possessing a virtual design model, which can also be created within the app, allows users to obtain results with just a few clicks. This substantially reduces the designer's effort.
Autonomous HVAC CFD presents distinct advantages over other industry tools
  • Elimination of In-house HPC Setup : The app performs simulations on a cloud platform, eliminating the need for an in-house HPC arrangement.
  • Automated CFD Process : No CFD expertise is required, as the app handles all CFD processes automatically.
  • Results Accessibility : Users can access results through a web browser.
In summary, the app takes care of all concerns related to conducting CFD analysis. In line with simulationHub's commitment to empowering designers, the process has been streamlined and automated to yield the best possible results quickly. The extensive CFD expertise of simulationHub has contributed to this effort, ensuring top-notch results for users.
While autonomy is invaluable, reliability and accuracy are equally crucial. Therefore, to address this, a "Validation study" is essential to confirm result accuracy.
This blog highlights three specific cases that showcase the precision of Autonomous HVAC CFD:
Case Study 1: Comfortina Benchmark Case
In this first case study, we delve into the Comfortina Benchmark Case, a design by Nielsen et al. This study is centered around the investigation of indoor environments. Professor Nielsen from Aalborg University led this research, collaborating with scholars from the University of Tokyo and Keio University.
Case Study 2: Full-Scale Experimental Measurement of CO2
Our second case study delves into the full-scale experimental measurement of CO2, conducted by G Pei et al. This study involves measurements taken within an environmental chamber designed to replicate a typical one-occupant office setting. The experiment was conducted at the Pennsylvania State University.
Case Study 3: CCTech Office Conference Room
The third case study explores our CCTech Office Conference Room. This study focuses on the enhancement of Indoor Air Quality (IAQ) within their office conference room. This research was conducted at CCTech by our inhouse IOT team and provides insights into optimizing the workspace environment.
Each of these case studies offers unique insights and contributes to the validation of the accuracy of Autonomous HVAC CFD. Through these analyses, we aim to demonstrate the capability and reliability of our methodology.
This blog is divided into several parts, each focusing on a specific aspect of the CFD study:
Case Description
In this section, we provide an overview of the cases under investigation. This includes a description of the scenarios, environments, and objectives of the study.
Case Study 1
In this first case study, a manikin is positioned within a rectangular wind tunnel measuring 2.44 m x 2.46 m x 1.2 m. The air flows into the domain from one side at a temperature of 22°C and an average velocity of 0.2 m/s. This incoming air subsequently exits the domain through two circular outlets located on the opposite surface. The manikin is seated and oriented towards the unidirectional inlet air. Moreover, the manikin's placement is symmetrical with respect to the z-plane, featuring a heat flux of 38 W.
Wind tunnel used by Professor Nielsen for benchmarking the case
Fig 1: (a) Major dimensions of Case Study 1 (b) Wind tunnel used by Professor Nielsen for benchmarking the case
Case Study 2
The second case study involves the placement of a manikin, a computer, a monitor, and two lights within an environmental chamber of dimensions 4.27 m x 4.27 m x 3 m. The chamber's air intake occurs at a temperature of 18°C with an inflow rate of 0.0385 m^3/s. This inflow is channeled through a low-momentum diffuser measuring 1.215 m x 0.615 m, positioned at floor level. The air subsequently exits the chamber through four circular outlets located at the ceiling level. Notably, to emulate CO2 emissions through breathing, continuous CO2 release is set at a flow rate of 0.026 m^3/hr via the manikin's nose. The heat load within the chamber originates from the manikin, computer, monitor, and lights, with respective values of 91 W, 108 W, 26 W, and 95 W.
The experimental configuration utilized for CO2 emission in the present study
Fig 2: (a) Major dimensions of the Case Study 2 (b) The experimental configuration utilized for CO2 emission in the present study.
Case Study 3
The third case examines the arrangement of 3 manikins within an office space spanning 20 ft. x 28 ft. Air circulation is facilitated by a centrally located cassette air conditioner with a 1.5-ton capacity. The room is entirely enclosed by walls that remain shielded from direct sunlight, and though it lacks windows, a dual door system is situated on one side. To mitigate noise infiltration from the neighboring bullpen workspace, the gaps around the glass doors have been outfitted with acoustic seals. Moreover, each manikin contributes to the scenario with a heat flux of 71 W. These manikins additionally emit CO2 at a rate of 40,000 ppm, expelling it at a velocity of 0.88 m/s through their nasal passages.
CCTech Conference room
Fig 3: Major dimensions of the Case Study 3 (b) CCTech Conference room.
Geometry and Mesh
Here, we delve into the details of the geometry models used in the simulations. We discuss how these models were created and processed for accurate representation. Additionally, we explore the meshing process, which is crucial for obtaining reliable results.
Inlet, Outlet and Heat load surface Case Study
Fig 4: Showcases Inlet, Outlet and Heat load surface for (a) Case Study 1 (b) Case Study 2 (c) Case Study 3.
The meshing process was executed according to the application's standards. The mesh was refined around significant geometries, including the manikin, light, and computer. This refinement adequately captured surface intricacies and fulfilled the requirements of the chosen flow and turbulence model.
Inflation layer and mesh resolution
Fig 5: Case Study 1 (a) Surface Mesh (b) Inflation layer and mesh resolution around manikin. Case Study 2 (c) Depicts Mesh resolution around object such as monitor. Also mesh refinement near the nose of the thermal manikin to capture plume and jet. Case Study 3 (d) Region wise refinement of mesh to capture diffusion from inlet (e) To capture thermal stratification effectively, additional refinement has been introduced in proximity to objects like the light source. This refinement is aimed at enhancing the accuracy of the simulation by ensuring that thermal patterns are accurately represented.
Boundary Conditions and Numerical Setup
This segment covers the setup of boundary conditions and the numerical parameters employed in the simulations. Precise boundary conditions are essential to mirror real-world scenarios, and the numerical setup ensures the accuracy of the computational methods.
Boundary conditions Inlet Objects (manikin, computer, etc) Outlet Wall
Velocity FixedValue or Velocity profile No-slip Zero gradient No-slip
Pressure FixedFlux FixedFlux FixedValue FixedFlux
Temperature FixedValue FixedFlux Zero gradient FixedFlux
Turbulence In accordance with the velocity values, length scale, and turbulent intensity Wall function for ⍵ and 1E-10 for k Zero gradient Wall function for ⍵ and 1E-10 for k
The numerical schemes were established to align with the application's strategy. A residual convergence criterion of 10^-6 was selected, and simultaneous monitoring of outlet temperatures was conducted. The turbulence was modeled using the SST k-omega two-equation model, and radiation was simulated using the Discrete Ordinates model.
A mesh independence study was conducted. Various mesh sizes and the application's meshing strategy were scrutinized. The subsequent analysis confirmed that the current meshing strategy employed by the application was sufficiently accurate. However, in order to capture the intricate geometry presented in Case Study 1, additional refinements were introduced.
Results and Discussion
The results obtained from the simulations take center stage in this section. We present the outcomes of the Autonomous HVAC CFD analysis, discussing key findings and insights derived from the data. Additionally, we engage in a comprehensive discussion of the results, drawing connections to the initial objectives and the broader context of the study.
CFD data (represented by lines) extracted from the simulation and experimental data case study
Fig 6 CFD data (represented by lines) extracted from the simulation and experimental data (depicted as dots): (a) Case Study 1 (b) Case Study 2 (3) Case Study 3.
Plots for Case Study 1
Velocity profile
Fig 7: Velocity profile.
In the case of plane x=0.19 m, the velocity results obtained from the CFD analysis align closely with the experimental findings. However, for planes x=1.69 m and x=2.19 m, the velocity values derived from the CFD analysis exhibit trends that mirror the experimental velocity values.
It's worth noting that the velocity at z=0 on plane x=2.19 m in the CFD analysis slightly exceeds the values recorded in the experimental measurements.
Across all considered points on plane x=2.19 m, at their respective z values, the mean relative deviation stands at 4.47%. Simultaneously, the standard deviation is measured at 18.29% when compared to the experimental velocity values.
These discrepancies in results can be attributed to several factors. Notably, differences in geometry between the experimental setup and the CFD model, variations in the inlet velocity profile, and the distribution of heat flux on the manikin contribute to the observed deviations.
Temperature distribution on manikin surface
Fig 8: Case Study 1 (a) Temperature distribution on manikin surface (b) Temperature contour plot on z= 0 m (c) Velocity contour plot on z= 0 m.
Figure 8 (a) presents the temperature distribution across the manikin's surface. Notably, in areas where the air velocity is low, convective heat transfer is diminished. Consequently, this leads to elevated temperatures on the manikin's surface, particularly evident in the back region. This heightened temperature is a direct result of reduced velocity in the wake region, as illustrated in Figure 8 (c).
Plots for Case Study 2
Temperature profile
Fig 9: (a) Temperature profile (b) CO2 concentration profile.
In Figure 9 (a), the temperature trend follows an increase with height, in accordance with the fundamental principle of displacement ventilation. Upon comparison, it's evident that while the simulated temperatures don't perfectly match the measurements, the disparity remains minimal, with discrepancies as low as 0.89°C. Despite this, the simulated temperature profile consistently mirrors the observed trend displayed in the measured data.
Figure 9 (b) highlights the stratification of CO2 concentration within the displacement-ventilated room. The simulation results above the breathing plane (1-1.1 m) deviate slightly from the actual measurements. However It is important to acknowledge that below the breathing plane the simulation outcomes don't exactly match the measurements. This can primarily be attributed to the simplified geometry of the manikin used in the simulation. Nonetheless, it's encouraging to note that both the simulation and measurement outcomes exhibit a comparable trend in the vertical CO2 concentration profile
Temperature contour
Fig 10: Case Study 2 (a) Temperature contour (b) CO2 concentration contour.
Sensor Temperature in C CO2 concentration in PPM Relative Humidity in %
Physical AHC Data % Accuracy Physical AHC Data % Accuracy Physical AHC Data % Accuracy
1 26.5 26.1 98.49 1198 1226 97.66 66.2 66.7 99.24
2 25.9 25.7 99.23 1115 1216 90.94 65.5 66.5 98.47
3 27 26.4 97.78 1192 1238 96.14 68.7 66.9 97.38
Supply 22.4 22.4 100 1190 1190 100 76.4 76.4 100
CO2 concentration contour
Fig 11: Case Study 3 (a) Temperature contour (b) CO2 concentration contour (c) Relative humidity contour (d) Velocity contour.
The CFD simulation results closely match the experimental data, particularly within the occupant breathing zone. However, the recommended CO2 concentration of 1000 ppm in the conference room is not achieved. Therefore, CFD analysis can be instrumental in identifying indoor air quality issues during the design phase, well before commissioning, thereby preventing discomfort, health complaints, and the expenses associated with design modifications after installation.
To summarize, the results demonstrate that the app's methodology is capable of accurately predicting the indoor environmental flow physics to a high degree of accuracy.
Get your FREE trial
Wish to explore Autonomous HVAC CFD to optimize your HVAC design for IAQ and thermal? Sign-up now and get free credits worth $500 for 90 days. You can explore the product and evaluate HVAC designs for Indoor Air Quality and estimate thermal comfort upto space area of 5000 square feet.
Blog Author -Karan Beeshm
Karan Beeshm
Karan Beeshm is a CFD support Engineer at Centre for Computational Technologies Private Limited (CCTech), Pune. He loves to work in fields physics and mathematics. Skilled in OpenFOAM, Fluent, C, MATLAB, CAD Modelling. He has completed his M.Tech in Thermal and Fluids Engineering from (Dr. BATU), Lonere, Raigad. His areas of interest are Heat Transfer, Fluid Mechanics, Computational Fluid Dynamics, Numerical Methods, Operation Research modeling. Driving and traveling, playing cricket and chess are his hobbies and he likes to explore historical places.
Blog Author -Karan Beeshm
Karan Beeshm
Karan currently serves as a Member of the Technical Staff at the Centre for Computational Technologies Private Limited. Within the organization, he demonstrates a high level of enthusiasm for OpenFoam Development and system dynamics modeling. His professional interests lie primarily in computational and data science applied to advanced energy systems. Karan obtained his undergraduate degree in Mechanical Engineering from Ramaiah Institute of Technology.