Artificial Neural Networks for Modeling of Chemical Source Terms in CFD Simulations of Turbulent Reactive Flows

Porumbel, Ionut; Petcu, Andreea Cristina; Florean, Florin; Hritcu, Constantin Eusebiu

doi:10.4028/www.scientific.net/amm.555.395

Cited by 4 publications

(4 citation statements)

References 23 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…computationally demanding physical processes (notably convection) (Krasnopolsky et al, 2005(Krasnopolsky et al, , 2010Krasnopolsky, 2007;Jiang et al, 2018;Gentine et al, 2018;Brenowitz and Bretherton, 2018). Machine learning has also been used to replace the chemical integrator for other chemical systems such as those found in combustion and been shown to be faster than solving the ODEs (Blasco et al, 1998;Porumbel et al, 2014). Recently, Kelp et al (2018) found order-of-magnitude speed-ups for an atmospheric chemistry box model using a neural network emulator, although their solution suffers from rapid error propagation when applied over multiple time steps.…”

Section: Introductionmentioning

confidence: 99%

Application of random forest regression to the calculation of gas-phase chemistry within the GEOS-Chem chemistry model v10

Keller

Evans

2019

Geosci. Model Dev.

View full text Add to dashboard Cite

Abstract. Atmospheric chemistry models are a central tool to study the impact of chemical constituents on the environment, vegetation and human health. These models are numerically intense, and previous attempts to reduce the numerical cost of chemistry solvers have not delivered transformative change. We show here the potential of a machine learning (in this case random forest regression) replacement for the gas-phase chemistry in atmospheric chemistry transport models. Our training data consist of 1 month (July 2013) of output of chemical conditions together with the model physical state, produced from the GEOS-Chem chemistry model v10. From this data set we train random forest regression models to predict the concentration of each transported species after the integrator, based on the physical and chemical conditions before the integrator. The choice of prediction type has a strong impact on the skill of the regression model. We find best results from predicting the change in concentration for long-lived species and the absolute concentration for short-lived species. We also find improvements from a simple implementation of chemical families (NOx = NO + NO2). We then implement the trained random forest predictors back into GEOS-Chem to replace the numerical integrator. The machine-learning-driven GEOS-Chem model compares well to the standard simulation. For ozone (O3), errors from using the random forests (compared to the reference simulation) grow slowly and after 5 days the normalized mean bias (NMB), root mean square error (RMSE) and R2 are 4.2 %, 35 % and 0.9, respectively; after 30 days the errors increase to 13 %, 67 % and 0.75, respectively. The biases become largest in remote areas such as the tropical Pacific where errors in the chemistry can accumulate with little balancing influence from emissions or deposition. Over polluted regions the model error is less than 10 % and has significant fidelity in following the time series of the full model. Modelled NOx shows similar features, with the most significant errors occurring in remote locations far from recent emissions. For other species such as inorganic bromine species and short-lived nitrogen species, errors become large, with NMB, RMSE and R2 reaching >2100 % >400 % and <0.1, respectively. This proof-of-concept implementation takes 1.8 times more time than the direct integration of the differential equations, but optimization and software engineering should allow substantial increases in speed. We discuss potential improvements in the implementation, some of its advantages from both a software and hardware perspective, its limitations, and its applicability to operational air quality activities.

show abstract

Section: Introductionmentioning

confidence: 99%

Application of random forest regression to the calculation of gas-phase chemistry within the GEOS-Chem chemistry model v10

Keller

Evans

2019

Geosci. Model Dev.

View full text Add to dashboard Cite

show abstract

“…Machine learning is becoming increasingly popular within the natural sciences (Mjolsness and DeCoste, 2001) and specifically within the Earth system sciences to emulate computationally demanding physical processes (notably convection) (Krasnopolsky et al, 2005(Krasnopolsky et al, , 2010Krasnopolsky, 2007;Jiang et al, 2018;Gentine et al, 2018;Brenowitz and Bretherton, 2018). Machine learning has also been used to replace the chemical integrator for other chemical systems such as those found in combustion and been shown to be faster than solving the ODEs (Blasco et al, 1998;Porumbel et al, 2014). Recently, Kelp et al (2018) found order-of-magnitude speedups for an atmospheric chemistry model using a neural network emulator, albeit their solution suffers from quick error propagation when applied over multiple time steps.…”

Section: Introductionmentioning

confidence: 99%

Application of random forest regression to the calculation of gas-phase chemistry within the GEOS-Chem chemistry model v10

Keller¹,

Evans²

2018

Preprint

View full text Add to dashboard Cite

Abstract. Atmospheric chemistry models are a central tool to study the impact of chemical constituents on the environment, vegetation and human health. These models are numerically intense, and previous attempts to reduce the numerical cost of chemistry solvers have not delivered transformative change. We show here the potential of a machine learning (in this case random forest regression) replacement for the gas-phase chemistry in atmospheric chemistry models. Our training data consists of one month (July 2013) of output of chemical conditions together with the model physical state, produced from the GEOS-Chem chemistry model (v10). From this data set we train random forest regression models to predict the concentration of each transported species after the integrator, based on the physical and chemical conditions before the integrator. The choice of prediction type has a strong impact on the skill of the regression model. We find best results from predicting the change in concentration for long-lived species and the absolute concentration for short-lived species. We also find improvements from a simple implementation of chemical families (NOx = NO + NO2). We then implement the trained random forest predictors back into GEOS-Chem to replace the numerical integrator. The machine learning driven GEOS-Chem model compares well to the standard simulation. For O3, error from using the random forests grow slowly and after 5 days the normalised mean bias (NMB), root mean square error (RMSE) and R2 are 4.2 %, 35 %, 0.9 respectively; after 30 days the errors increase to 13 %, 67 %, 0.75. The biases become largest in remote areas such as the tropical Pacific where errors in the chemistry can accumulate with little balancing influence from emissions or deposition. Over polluted regions the model error is less than 10 % and has significant fidelity in following the time series of the full model. Modelled NOx shows similar features, with the most significant errors occurring in remote locations far from recent emissions. For other species such as inorganic bromine species and short lived nitrogen species errors become large, with NMB, RMSE and R2 reaching >2100 % >400 %, <0.1 respectively. This proof-of-concept implementation is 85 % slower than the direct integration of the differential equations but optimisation and software engineering would allow substantial increases in speed. We discuss potential improvements in the implementation, some of its advantages from both a software and hardware perspective, its limitations and its applicability to operational air quality activities.

show abstract

“…Recent attempts have also used machine learning to replace the use of traditional integrators [16]. For example, using a neural network emulator for an atmospheric chemistry box model, an order-of-magnitude speed-up was found, but once applied over several time intervals, the new implementation suffered from fast error reproduction [15].…”

Section: Background and Rationalementioning

confidence: 99%

Dynamic Complex Network Analysis of PM2.5 Concentrations in the UK, Using Hierarchical Directed Graphs (V1.0.0)

Broomandi¹,

Geng²,

Guo³

et al. 2021

Preprint

View full text Add to dashboard Cite

The risk of a broad range of respiratory and heart diseases can be increased by widespread exposure to fine atmospheric particles on account of their capability to have a deep penetration into the blood streams and lung. Globally, studies conducted epidemiologically in Europe and elsewhere provided the evidence base indicating the major role of PM2.5 leading to more than four million deaths annually. Conventional approaches to simulate atmospheric transportation of particles having high dimensionality from both transport and chemical reaction process make exhaustive causal inference difficult. Alternative model reduction methods were adopted, specifically a data-driven directed graph representation, to deduce causal directionality and spatial embeddedness. An undirected correlation and a directed Granger causality network were established through utilizing PM2.5 concentrations in 14 United Kingdom cities for one year. To demonstrate both reduced-order cases, the United Kingdom was split up into two southern and northern connected city communities, with notable spatial embedding in summer and spring. It continued to reach stability to disturbances through the network trophic coherence parameter and by which winter was construed as the most considerable vulnerability. Thanks to our novel graph reduced modeling, we could represent high-dimensional knowledge in a causal inference and stability framework

show abstract

Artificial Neural Networks for Modeling of Chemical Source Terms in CFD Simulations of Turbulent Reactive Flows

Cited by 4 publications

References 23 publications

Application of random forest regression to the calculation of gas-phase chemistry within the GEOS-Chem chemistry model v10

Application of random forest regression to the calculation of gas-phase chemistry within the GEOS-Chem chemistry model v10

Application of random forest regression to the calculation of gas-phase chemistry within the GEOS-Chem chemistry model v10

Dynamic Complex Network Analysis of PM2.5 Concentrations in the UK, Using Hierarchical Directed Graphs (V1.0.0)

Contact Info

Product

Resources

About