Abstract:Computational Fluid Dynamics (CFD) is a well-established tool to study fluid dynamics and particle movement, while Artificial Neural Network (ANN) models offer machine learning capabilities to accelerate indoor airflow predictions, but they still maintain a reasonable level of accuracy for prediction purposes. This study pioneers the integration of Deep Neural Network (DNN) models into indoor airflow dynamics, aiming to provide an accurate and accelerated prediction efficiency. The objective is to train two DNN models (classical and modified DNN models) to capture the complex relationships between ventilation rate, airflow patterns, and particle dispersion characteristics within buildings. Using a dataset generated from CFD simulations encompassing various air change rates, the trained modified DNN model significantly enhances prediction efficiency in term of the computational cost by 67 % reduction of CFD computational time (1 h to 20 min) while also resulting in very similar accuracy compared to the CFD outputs. The R2 values of classical and modified DNN models (plane 1) at air flow rate equals to 4 ach are 0.6867 and 0.9567 in term of the DPM distribution, respectively. The similar pattern is observed as the accuracy of modified DNN is higher than the classical DNN for other air flow rates in terms of the DPM and velocity distributions. Accordingly, the number of prediction errors is significantly decreased as the model alters from the classical DNN to modified DNN model. The significance of this research lies in its potential to enhance the efficiency of assessing particle dispersion, allowing for the more efficient design of targeted ventilation strategies and indoor air quality control measures tailored to diverse pollutant sources emitted from humans. Integrating DNN and CFD in assessing particle dispersion characteristics is promising for improving the understanding of indoor air dynamics and facilitating data-driven decision-making for ensuring healthier and safer indoor environments.

Development of self-adaptive low-dimension ventilation models using OpenFOAM: Towards the application of AI based on CFD data

Implementation and visualization of artificial intelligent ventilation control system using fast prediction models and limited monitoring data

Possible User-Dependent CFD Predictions of Transitional Flow in Building Ventilation

Deep learning to develop zero-equation based turbulence model for CFD simulations of the built environment

An open-source and experimentally guided CFD strategy for predicting air distribution in data centers with air-cooling

Improving indoor air flow and temperature prediction with local measurements based on CFD-EnKF data assimilation

Rapid prediction of indoor airflow field using operator neural network with small dataset

A Method Toward Real-Time CFD Modeling for Natural Ventilation

Convolutional Neural Networks-Based Surrogate Model for Fast Computational Fluid Dynamics Simulations of Indoor Airflow Distribution

Particle dispersion for indoor air quality control considering air change approach: A novel accelerated CFD-DNN prediction

Data-driven prediction of indoor airflow distribution in naturally ventilated residential buildings using combined CFD simulation and machine learning (ML) approach

Simulation-aided development of a compact local ventilation unit with the use of CFD analysis

RANS turbulence model development using CFD-driven machine learning

Reducing Spatial Discretization Error on Coarse CFD Simulations Using an OpenFOAM-Embedded Deep Learning Framework

A CFD-based surrogate model for predicting flow parameters in a ventilated room using sensor readings

Fast prediction of indoor airflow distribution inspired by synthetic image generation artificial intelligence

Fast Prediction and Optimization of Building Wind Environment Using CFD and Deep Learning Method

Non-intrusive Reduced Order Model of Urban Airflow with Dynamic Boundary Conditions

CFD and ventilation research

Improving CFD simulations by local machine-learned correction

Appropriate CFD Turbulence Model for Improving Indoor Air Quality of Ventilated Spaces