Development of CFD-based multi-fidelity surrogate models for indoor environmental applications DOI Creative Commons
Nina Morozova

Опубликована: Июнь 20, 2022

This thesis presents a methodology for CFD-based multi-fidelity surrogate models indoor environmental applications. The main idea of this work is to develop model that has accuracy comparable CFD simulations but at considerably lower computational cost. It can perform real-time or faster than environments using ordinary office computers. be divided into three parts. In the first part, rigorous analysis feasibility affordable high-fidelity environment design and control carried out. chapter, we analyze two representative test cases, which imitate common airflow configurations, on wide range different turbulence discretization methods meet requirements cost, run-time, accuracy. We apply knowledge growth in power advances numerical algorithms order possibility performing accurate yet no-model LES with staggered discretizations studied show best performance. conclude computers are too slow used as primary tool control. Taking account laws computer prediction, estimate these applications next decades. second part dedicated developing data-driven predict comfort-related flow parameters ventilated room. chapter uses previously tested cavity heated floor case. developed predicts set parameters, such average Nusselt number hot wall, jet separation point, kinetic energy, enstrophy, temperature, were also comprehensively previous thesis. based gradient boosting regression, chosen due its performance among four machine learning methods. inputs temperature velocity values locations, act sensor readings. locations sensors determined by minimizing prediction error. does not require repetition applied since structure input data imitates Furthermore, low cost execution good makes it an effective alternative where rapid predictions complex configurations required, predictive third extension part. implement approach reduce training dataset generation. Gaussian process regression (GPR), capable handling data. variable fidelity constructed coarse- fine-grid model. approaches: GPR trained both high- low-fidelity without distinction, linear correction, co-cringing. approaches compared GPRs only All successfully generation while maintaining required level co-cringing demonstrates trade-off between Esta tesis presenta una metodología para modelos sustitutos de fidelidad múltiple basados en aplicaciones ambiente interior. La principal este trabajo es desarrollar un modelo que tenga precisión las simulaciones pero costo computacional considerablemente inferior. metodologia permite realizar tiempo real o más rápido utilizando ordinadores oficina ordinarios. Este se puede dividir tres partes principales. En la primera parte, lleva cabo análisis viabilidad asequibles alta el diseño y ambientes interiores. capítulo, analizamos dos casos, imitan configuraciones comunes flujo aire interior, amplia gama diferentes turbulencia métodos discretización. Aplicamos conocimiento sobre crecimiento potencia analizar posibilidad precisas Los sin con discretizaciones escalonadas muestran mejor rendimiento. Concluimos son demasiado lentas ser utilizadas como herramienta Teniendo cuenta leyes predicción del computacional, estimamos estas durante próximas décadas. segunda parte esta está dedicada al desarrollo sustituto basado datos predecir los parámetros habitación ventilada. El desarrollado predice conjunto flujo, número promedio pared caliente, punto separación chorro, energía cinética promedia, entrofia promedia temperatura promedia. basa regresión aumento gradiente, elegida debido su rendimiento preciso entre cuatro aprendizaje automático probados. Las entradas valores velocidad ubicaciones, actúan lecturas sensor. ubicaciones estos sensores determinaron minimizando error predicción. no requiere aplicación repetición ya estructura entrada imita Además, bajo ejecución buena lo convierten alternativa eficaz requieren predicciones rápidas complejas, predictivo modelo. tercera extensión parte. implementamos enfoque reducir generación entrenamiento. procesos gaussianos capaz manejar múltiple. construye CFD. Probamos enfoques múltiple: entrenado baja distinción, corrección lineal co-krigeaje. comparan solo fidelidad. Todos probados reducen éxito conjuntos comparación mientras mantienen nivel requerido precisión. co-krigeaje demuestra compensación

A review on indoor airborne transmission of COVID-19– modelling and mitigation approaches DOI Creative Commons
Saeed Rayegan, Chang Shu, Justin Berquist

и другие.

Journal of Building Engineering, Год журнала: 2022, Номер 64, С. 105599 - 105599

Опубликована: Ноя. 26, 2022

In the past few years, significant efforts have been made to investigate transmission of COVID-19. This paper provides a review COVID-19 airborne modeling and mitigation strategies. The simulation models here are classified into infectious risk numerical approaches for spatiotemporal transmissions. Mathematical descriptions assumptions on which these based discussed. Input data used in previous studies assess dispersion extracted reported. Moreover, measurements performed study within indoor environments introduced support validations anticipated future studies. Transmission strategies recommended recent include modifying occupancy ventilation operations, using filters air purifiers, installing ultraviolet (UV) disinfection systems, personal protection compliance, such as wearing masks social distancing. application various building types, educational, office, public, residential, hospital, is reviewed. Recommendations works also discussed current apparent knowledge gaps covering both approaches. Our findings show that different measures were environments; however, there no conclusive work reporting their combined effects level may be achieved. further should conducted understand better balance between mitigating viral transmissions buildings energy consumption.

Язык: Английский

Процитировано

53

Evaluating SARS‐CoV‐2 airborne quanta transmission and exposure risk in a mechanically ventilated multizone office building DOI
Shujie Yan, Liangzhu Wang,

Michael J. Birnkrant

и другие.

Building and Environment, Год журнала: 2022, Номер 219, С. 109184 - 109184

Опубликована: Май 13, 2022

Язык: Английский

Процитировано

39

A review of Modelica language in building and energy: Development, applications, and future prospect DOI

Kaiying Qiu,

Junlu Yang,

Zhi Gao

и другие.

Energy and Buildings, Год журнала: 2024, Номер 308, С. 113998 - 113998

Опубликована: Фев. 20, 2024

Язык: Английский

Процитировано

10

Towards the use of data-driven methods for indoor airflow field reconstruction: A systematic review DOI

A. Olivas,

Jurng‐Jae Yee

Journal of Building Engineering, Год журнала: 2025, Номер unknown, С. 112581 - 112581

Опубликована: Апрель 1, 2025

Язык: Английский

Процитировано

0

Design of convertible patient care unit for both non-pandemic and pandemic times: prototype, building spatial layout, and ventilation design DOI

Yufan Chang,

Zhengtao Ai, Pawel Wargocki

и другие.

Building and Environment, Год журнала: 2024, Номер 258, С. 111597 - 111597

Опубликована: Апрель 30, 2024

Язык: Английский

Процитировано

4

A CFD-based multi-fidelity surrogate model for predicting indoor airflow parameters using sensor readings DOI
Nina Morozova, F. Xavier Trias, Vladimir Vanovskiy

и другие.

Building and Environment, Год журнала: 2025, Номер unknown, С. 112533 - 112533

Опубликована: Янв. 1, 2025

Язык: Английский

Процитировано

0

A Markov chain-based approach for assessing respiratory infection risk in a multi-zone office building DOI

Qi Zhen,

Anxiao Zhang,

Regina Bokel

и другие.

Journal of Building Engineering, Год журнала: 2024, Номер 90, С. 109328 - 109328

Опубликована: Апрель 21, 2024

Язык: Английский

Процитировано

3

A multi-zone spatial flow impact factor model for evaluating and layout optimization of infection risk in a Fangcang shelter hospital DOI
Luping Ma, Xiaohong Zheng, Yong Guo

и другие.

Building and Environment, Год журнала: 2022, Номер 214, С. 108931 - 108931

Опубликована: Фев. 23, 2022

Язык: Английский

Процитировано

15

Demonstrating the use of absolute pressure sensors for monitoring stack-driven pressure differences in high-rise buildings DOI

Jiajun Jing,

Sungmin Yoon, Jaewan Joe

и другие.

Building and Environment, Год журнала: 2024, Номер 270, С. 112500 - 112500

Опубликована: Дек. 26, 2024

Язык: Английский

Процитировано

2

A virtual sensor network for pressure distribution inside a multi-zone building based on spatial adjacency relationships and multivariate adaptive regression spline DOI

Jiajun Jing,

Dongseok Lee,

Jaewan Joe

и другие.

Journal of Building Engineering, Год журнала: 2023, Номер 80, С. 108059 - 108059

Опубликована: Ноя. 1, 2023

Язык: Английский

Процитировано

6