Despite its importance for natural cooling, natural ventilation is rarely considered in building thermal optimization. Reduced-order ventilation models, such as EnergyPlus’s Airflow Network, enable rapid assessment of thermal performance, making them valuable tools for designing naturally ventilated buildings. However, their adoption has been limited because creating Airflow Network models requires manually replicating building layouts. Moreover, compared to other factors affecting thermal performance, building layouts have received little attention in optimization studies.As a solution, we introduce AURA, a method for quickly replicating existing building layouts and generating Airflow Network-enabled EnergyPlus models. Through a case study of three apartment units, we demonstrate that building layouts can have an order of magnitude larger impact in mediating airflow and cooling than construction materials or window area. This work enables more rigorous analysis of naturally cooled buildings, providing researchers and practitioners with a tool to investigate the role of modifications in sustainable building design.

This paper validates Large Eddy Simulation (LES) for predicting wind-induced pressures on low-rise buildings in urban areas. Validation data for pressure coefficients on the building was obtained from experiments on a 1:100 scale model of Stanford’s Y2E2 building in the NHERI Wall of Wind (WoW) facility at Florida International University, and simulations were conducted using the CharLES code. The study first ensured an accurate representation of WoW surface layer velocity statistics in the LES. Next, simulations of the surface layer wind flow interacting with the buildings showed a close agreement between LES and wind tunnel data for wind pressure coefficient statistics (mean, RMS, peak, skewness, kurtosis) on the building surface. The LES can accurately identify areas where surrounding buildings create more negative peak pressure coefficients than would occur on the isolated building. The changes in the peak pressure coefficients were found to be induced by changes in the mean flow velocity magnitude and direction, including new regions of flow separation, acceleration, and vortex formation. In conclusion, LES is a valuable tool for analyzing wind pressures on realistic low-rise buildings in complex urban environments, offering reliable estimates for local peak pressure coefficients and insight into the flow physics causing these peaks.

This study revisits aerodynamic admittance functions for roof pressures on buildings subjected to turbulent wind conditions. A new model for pressure–velocity admittance is proposed, expressed as a product of space- and frequency-dependent factors. Unlike classical approaches, the proposed formulation does not rely on the mean pressure coefficient and is therefore better suited to handle separated flow regions. The spatial dependency is linked to the local standard deviation of roof pressure, while the frequency-dependent factor varies slowly across the roof surface. Both factors can be readily identified from numerical or experimental data, as demonstrated in this paper, using state-of-the-art simulations of flow over a generic building. Two simulations with differing turbulence intensities in the incoming flow are analyzed. The proposed model accurately captures the pressure fluctuations on the roof and on the rear façade of the building. The observed slow spatial variation of the frequency-dependent factor suggests that this model provides a robust framework for physics-based classification and offers practical benefits for wind engineering design and codification.

The urban canopy affects wind in complex ways, making it challenging to predict wind-driven natural ventilation and cooling in buildings. Using large eddy simulations of coupled outdoor and indoor airflow, we study how the surrounding urban canopy and wind angle influence ventilation rates through four ventilation configurations: cross, corner, dual-room and single-sided. Flow visualisations demonstrate how both large-scale flow patterns and local interference effects can influence ventilation rates by 50 %–85 %. In general, lower density canopies give higher ventilation rates, and wind angles that align with a direct path between two openings also lead to higher ventilation rates. However, interference effects from surrounding buildings can significantly change the local wind speed and direction, thus also changing ventilation rates. The magnitude of these interference effects depends on both the wind angle and surrounding building geometries. The effect of wind angle is less pronounced in a higher density canopy, where the urban canopy geometry more strongly guides the flow. The results demonstrate that the canopy’s effect on ventilation rates is much more complex than those suggested by existing natural ventilation parametrisations.

The study of contrails has gained widespread attention because their effective radiative forcing (ERF) is comparable to that of carbon dioxide from aviation, yet current ERF estimates remain highly uncertain. To enhance predictions of contrail formation for both conventional and alternative fuels, we have developed a high-fidelity numerical framework for the simulation of the jet and vortex interaction phases of contrails. Our approach combines three-dimensional large-eddy simulations of an Eulerian–Lagrangian two-phase flow in the compressible flow solver charLES with detailed mixing, turbulence, and microphysics models. We conduct temporally resolved simulations of early single contrail formation and compare the effects of atmospheric, aircraft, engine, and modeling parameters on the number of nucleated ice crystals and estimated net radiative forcing. Adding bypass flow shifts the plume away from the ideal core-atmosphere mixing line and reduces the number of initially nucleated ice crystals, while a more complete microphysics treatment alters the nucleation rate but has a limited lasting effect for sufficiently low atmospheric temperatures. Our simulations show the strongest sensitivity to aircraft size, followed by the soot number emission index and atmospheric temperature. Atmospheric aerosols also produce a nonlinear low-soot regime, indicating that ambient aerosol can contribute appreciably to ice crystal formation.

This review first examines how urban wind flow impacts the sustainability and resilience of cities and identifies the three main challenges in predictive modeling of urban flows: the complexity of the flow physics, the variability and uncertainty in the flow conditions, and the diversity and multiscale nature of urban geometries. To review the complexity of the flow physics, the typical flow patterns observed in canonical urban flows are summarized, and related modeling challenges and opportunities in both wind tunnel experiments and simulations are highlighted. Next, opportunities to predict realistic urban flows by addressing the other challenges are explored through the lens of a modeling framework with uncertainty quantification. The important role of field measurements, supporting the more accurate characterization of uncertainties in the flow conditions, as well as enabling validation with real-world data, is emphasized. The review concludes with two specific examples that demonstrate how integrated use of field measurements and computational models can improve the understanding and modeling of real urban flows to ultimately support sustainable development goals for urban areas.

Wind-resistant design requires evaluating pressure coefficients (Cp) on building facades and roofs. These coefficients are either obtained from building standards or codes, which have limited accuracy, or from wind tunnel testing, which has limitations related to spatial resolution, geometrical scaling, as well as cost and scheduling constraints. Computational fluid dynamics (CFD), particularly large-eddy simulation (LES), offers a promising alternative to overcome these challenges, but the simulations are computationally expensive. A significant computational speedup is essential for the routine use of LES in wind loading calculations across the full wind rose.Multi-fidelity (MF) modeling can provide an effective approach to reducing computational costs by combining data from many evaluations of a cheap low-fidelity (LF) model with few evaluations of an expensive high-fidelity (HF) model to provide predictions with HF accuracy across the full parameter space. In this work, we investigate the use of an MF framework to provide wind loading predictions for a high-rise building across the full wind rose. We combine wind loading predictions for 10 wind directions from low-resolution LESs with predictions for 5 wind directions from high-resolution LESs to provide predictions for mean, root-mean-square (rms), and peak Cp across the full wind rose with accuracy similar to that of high-resolution LESs. We fit the discrepancy between LF and HF predictions using LF models with different accuracy and two reduced-order surrogate models, namely kriging (KRG) and co-kriging (CoKRG), which are then used to correct the low-resolution LES results. The results demonstrate that both surrogate models provide accurate predictions of the HF model response at wind directions where HF data is unavailable. The MF framework predicts mean, rms, and peak max Cp within reasonable bounds of HF data for more than 93% of the building surface and peak min Cp for more than 82% of the surface. The approach is also robust to dataset selection, performing well with equispaced datasets without requiring optimization of the HF evaluation points. While some discrepancies between MF and HF predictions remain, the framework demonstrates potential to be used as an early-stage design tool for wind-resilient building design.

Rapid urbanization demands accurate and efficient monitoring of turbulent wind patterns to support air quality, climate resilience and infrastructure design. Traditional sparse reconstruction and sensor placement strategies face major accuracy degradations under practical constraints. Here, we introduce Diff-SPORT, a diffusion-based framework for high-fidelity flow reconstruction and optimal sensor placement in urban environments. Diff-SPORT combines a generative diffusion model with a maximum a posteriori (MAP) inference scheme and a Shapley-value attribution framework to propose a scalable and interpretable solution. Compared to traditional numerical methods, Diff-SPORT achieves significant speedups while maintaining both statistical and instantaneous flow fidelity. Our approach offers a modular, zero-shot alternative to retraining-intensive strategies, supporting fast and reliable urban flow monitoring under extreme sparsity. Diff-SPORT paves the way for integrating generative modeling and explainability in sustainable urban intelligence.

Wind farm layout optimization (WFLO) studies often aim to maximize the annual energy production (AEP) of a wind farm by choosing an arrangement of turbines that minimizes wake interactions. One way to reduce the cost of WFLO studies is by using more computationally efficient AEP models. The cost of standard AEP modeling approaches, based on the numerical integration of low-fidelity engineering wake models, scales poorly with the number of simulated discrete wind conditions. A second way to reduce cost when using a gradient-based algorithm is to supply exact gradient information instead of finite-difference estimates. However, analytical functions for the derivatives of AEP with respect to turbine positions are not always available in the conventional modeling approach. FLOWERS is a computationally inexpensive, analytical model for wind farm AEP that is specifically developed for WFLO applications. In this paper, we analyze the performance of the FLOWERS AEP model with analytic gradients in a layout optimization study compared with a reference optimization framework across three wind farm case studies. We find that the FLOWERS-based approach reduces computation time by a factor of 50–4000 and improves optimal AEP by about 0.3% with less than half of the variability in AEP across instances with randomized initial conditions. We also find the optimal layouts to be insensitive to model parameter tuning, making FLOWERS-based layout optimization a streamlined, user-friendly approach.

Large eddy simulations (LES) can aid the prediction of wind loading on buildings, provided that representative inflow turbulence properties are prescribed. This study conducts LES to assess the sensitivity of the mean, root mean square (rms) fluctuation, and peak pressure coefficients ( C p ) on building surfaces to the uncertainties in the incoming flow turbulence. Compared to wind tunnel measurements, the simulated mean C p is well predicted, and the variation in inflow turbulence has a negligible effect. The rms C p increases with increasing turbulence intensities and increasing turbulence length scales. Increasing inlet values of turbulence intensities and turbulence length scales reduces the root mean square errors (RMSE) of rms C p from 0.049 to 0.026 and from 0.047 to 0.024, respectively, on the side surfaces with flow separation. The minimum C p responds similarly, where the RMSE is reduced from 0.382 to 0.280 and from 0.385 to 0.286. The maximum C p on the windward surface achieves the lowest RMSE of 0.089 at nominal inlet values. The agreement between LES and experiment improves significantly after incorporating uncertainties in the input turbulence properties by repeating simulations with smaller and larger values from the estimated turbulence inputs. Wind tunnel experiments often do not measure the complete turbulence properties of the incoming flow, thereby obscuring the validation process of simulation results. The findings recommend wind tunnel experiments to measure and report the complete turbulence properties of the incoming flow for accurate prediction of wind loading.

Computational simulation is a critical tool for assessing the impacts of natural hazards and informing risk mitigation and resilience strategies. The NHERI SimCenter has developed an open-source, modular framework that integrates performance-based engineering methodologies with regional-scale assessments to enable multi-hazard, multi-scale simulations. This paper presents the conceptual foundation and current capabilities of the SimCenter platform, covering hazard characterization, structural response analysis, damage and loss estimation, and recovery modeling. By leveraging high-performance computing, standardized data schemas, and open-source tools, the platform facilitates transparent, reproducible research while bridging local and regional analyses. Key contributions include improved inventory generation, damage simulation, and recovery analysis, with applications extending across multiple hazard domains. The paper also discusses challenges in implementing high-resolution, high-fidelity simulations, advancing multi-hazard assessments, and enhancing accessibility for a broad user base. Looking ahead, expanding hazard models, refining regional-to-local modeling techniques, and fostering community collaboration will be essential for advancing computational simulation in natural hazards engineering. Through continued development, the SimCenter aims to provide researchers and practitioners with scalable, adaptable tools to enhance disaster risk assessment and resilience planning.

Recent advancements in computational wind engineering have demonstrated the potential of Large-Eddy Simulations (LESs) as a design tool for wind loading predictions. However, the computational cost associated with finely resolved LESs capable of capturing peak pressures can become a limiting factor when the analysis requires evaluating the wind loads for all wind directions. The same limitation arises in other LES applications that require predictions across a parameter space. This study proposes a Multi-Fidelity (MF) framework for LES that leverages Neural Networks (NN). A MF data set selection strategy and a MF loss function are proposed to obtain NNs that provide optimal MF model performance. The framework is applied to predict wind loading on a high-rise building across the entire wind rose. Data from 10 low-resolution LESs is combined with data from 5 high-resolution LESs, to provide MF predictions for the point-wise mean, rms, and peak pressure coefficients across the entire wind rose. The MF framework consistently reduces the error associated with the low-resolution LESs for each wind direction, achieving accuracy close to the high-resolution LESs at half of the computational cost. The MF model error averaged across all building facades is consistently lower than the error of the low-resolution model for every wind direction and Quantity of Interest (QoI). The framework can be readily applied to other LES applications.

Contrails have recently gained widespread attention, as their estimated warming potential is in the same order as aviation's CO2 and is largely more uncertain. Our research is motivated by concerns about future hydrogen-fueled aircraft, as they will emit considerably more water vapor, albeit no soot. In this study, we compare the ice crystal number and net radiative forcing of contrails forming behind kerosene- and hydrogen-fueled aircraft with large-eddy simulations (LES) of the jet-vortex interaction phases of contrails. We simulate the jet-vortex interaction using prescribed axial and vortical velocity fields and employ Lagrangian particle tracking and microphysical models for ice crystal formation and growth. We find that hydrogen contrails have a fraction to half the number of ice crystals of kerosene contrails, except when the engine operates in the soot-poor regime and flies at lower altitudes. Higher atmospheric aerosol concentrations may also increase the net radiative forcing of contrails from hydrogen-fueled aircraft with respect to an equivalent kerosene-fueled aircraft.

The accuracy of large-eddy simulations (LESs) for predicting wind-induced pressure loads remains an important topic of inquiry. This paper aims to advance this topic by validating an LES workflow for predicting wind pressures on a realistic low-rise building model exposed to a suburban neutral surface layer. We compare two wind tunnel data sets and LES predictions, obtained using a two-step workflow. First, we ensure that an accurate representation of the surface layer wind flow is obtained at the building location. Next, we assess the resulting wind loads on the building model. Using this workflow, we demonstrate consistent agreement between LES predictions and wind tunnel tests, where the discrepancies between the LES and wind tunnel results mimic the discrepancies between the two wind tunnel tests. This finding underscores that the pressure signals in certain locations are sensitive to inevitable, small differences in the approach flow. LES-based flow visualization uncovered that the most negative pressure peaks, which occur on the building roof, arise from hairpin-like vortices that are lifted from the separation region near the upstream roof edge. The results shed light on the complex dynamics of wind-induced pressure loads and contribute to quantifying the reliability of LES for wind load estimation.

Accurate prediction of wind flow fields in urban canopies is crucial for ensuring pedestrian comfort, safety, and sustainable urban design. Traditional methods using wind tunnels and Computational Fluid Dynamics, such as Large-Eddy Simulations (LES), are limited by high computational cost, and time requirements. This study presents a deep neural network (DNN) approach for fast and accurate predictions of urban wind flow statistics, reducing computation time from O(10) hours on 32 CPUs for one LES evaluation to O(1) seconds on a single GPU using the DNN model. We employ a U-Net architecture trained on high-fidelity LES data including 252 synthetic urban configurations at seven wind directions (0 degrees to 90 degrees in 15 degrees increments). The inflow profiles of the LES represent an atmospheric surface layer over a suburban terrain. The model predicts two key quantities of interest: mean velocity magnitude and streamwise turbulence intensity, at multiple heights within the urban canopy, unlike other surrogates that focus on single-plane predictions. The U-net uses 2D building representations at 3 different heights augmented with signed distance functions and their gradients as inputs, forming a 256 x 256 x 9 tensor. In addition, a Spatial Attention Module and a global binary skip connection are used for feature transfer through skip connections. The custom loss function combines the root-mean-square error of predictions, their gradient magnitudes, and L2 regularization. An ablation study shows that these model design choices improve the model prediction. Model evaluation on 50 test cases demonstrates high accuracy with an overall mean relative error of 9.3% for velocity magnitude and 5.2% for standard deviation. This research shows the potential of deep learning approaches to provide fast, accurate urban wind assessments essential for creating comfortable and safe urban environments. Code is available at https://github.com/tvarg/Urban-FlowUnet.git.

Accurate prediction of wind loading on buildings is crucial for structural safety and sustainable design, yet conventional approaches such as wind tunnel testing and large-eddy simulation (LES) are prohibitively expensive for large-scale exploration. Each LES case typically requires at least 24 hours of computation, making comprehensive parametric studies infeasible. We introduce WindMiL, a new machine learning framework that combines systematic dataset generation with symmetry-aware graph neural networks (GNNs). First, we introduce a large-scale dataset of wind loads on low-rise buildings by applying signed distance function interpolation to roof geometries and simulating 462 cases with LES across varying shapes and wind directions. Second, we develop a reflection-equivariant GNN that guarantees physically consistent predictions under mirrored geometries. Across interpolation and extrapolation evaluations, WindMiL achieves high accuracy for both the mean and the standard deviation of surface pressure coefficients (e.g., RMSE $\leq 0.02$ for mean $C_p$) and remains accurate under reflected-test evaluation, maintaining hit rates above $96\%$ where the non-equivariant baseline model drops by more than $10\%$. By pairing a systematic dataset with an equivariant surrogate, WindMiL enables efficient, scalable, and accurate predictions of wind loads on buildings.

Despite its relevance to natural cooling, natural ventilation is rarely considered when optimizing building thermal performance. Reduced-order ventilation models, such as EnergyPlus's airflow network (AFN) model, are ideal tools for design and analysis of naturally ventilated buildings because they enable rapid assessment of ventilation performance. However, such analysis requires replicating the building layouts of interest, a time-consuming task. We present a semi-automated methodology for replicating existing building layouts, an intermediate step for the rapid generation of AFN-enabled EnergyPlus models. This method converts user-input SVG files of floor plans to data containing the dimensions and connectivity of rooms using Python. We demonstrate the merits of our proposed methodology on a case study apartment unit, successfully replicating the plan layout while ensuring that the data will be suitable for EnergyPlus. Overall, our work aims to decrease the manual effort needed to evaluate and optimize the thermal performance of buildings, contributing to the growing movement to create more energy-efficient buildings.

Annual energy production (AEP) is commonly used in objective functions for wind farm layout optimization. AEP is proportional to wind farm power production integrated over an annual distribution of free-stream wind conditions. Physics-based estimates of wind farm power production typically rely on low-fidelity engineering wake models that approximate the steady-state wind farm flow field. AEP estimates are then obtained by performing independent simulations for discrete wind conditions and using rectangular quadrature to account for each condition's expected frequency of occurrence. Depending on the number of simulated discrete wind conditions, this numerical integral could be hampered by poor accuracy or high computational costs. The FLOWERS AEP model instead poses an analytical integral of the engineering wake model over the variable wind conditions, yielding a closed-form, analytical function for wind farm AEP. This paper derives the analytical functions for FLOWERS AEP and its derivatives with respect to turbine position, which are useful for gradient-based wind farm layout optimization, in nondimensional form. We then analyze the benefits of the FLOWERS AEP model over conventional reference models, focusing on its low cost, adequate wake loss predictions, and smooth design space. Although the FLOWERS approach is found to predict the exact value of AEP with some error relative to the reference model (within 14% on average), it dramatically reduces computation time by an order of magnitude, produces a qualitatively similar design space at relatively low resolution, and yields comparable optimal layouts. This significant speed improvement is critical in layout optimization applications, where determining an optimal layout in an efficient manner is more important than precise AEP prediction.

Contrails have recently gained widespread attention due to their large and uncertain estimates of effective radiative forcing, i.e., warming effect on the planet, comparable to those of carbon dioxide. To study this aircraft-induced cloud formation in the context of current conventional fuels and future alternative fuels, we have developed a numerical framework for simulating the jet and early vortex interaction phases of contrail formation. Our approach consists of high-fidelity, 3D large-eddy simulations (LES) of an Eulerian-Lagrangian two-phase flow using the compressible flow solver charLES. We perform temporal simulations of the early contrail formation phases for a single linear contrail and compare the sensitivity of the results to modeling choices and atmospheric, aircraft, and engine parameters. Specifically, we discuss how these choices and parameters affect the number of nucleated ice crystals and estimated net radiative forcing (based on an optical depth parameterization). Our simulations show the most significant sensitivity to the aircraft size, followed by the soot number emission index and fuel consumption. Adding atmospheric aerosol as a precursor for future studies of sustainable fuels evidences a non-linear relation previously highlighted in the literature between number of emitted soot and nucleated ice crystals.

The majority of wind damage is to building envelope components. Large eddy simulations (LESs), which can predict flow fields at high resolution, have significant potential for analyzing component loads. This study aims to (1) demonstrate that LESs can elucidate flow phenomena behind peak pressure loads, and (2) quantitatively compare LESs to full-scale measurements performed on the roof of the 184 m tall Space Needle. The simulations revealed unsteady flow features responsible for pressure signals observed in separation regions and shear layers. Furthermore, predictions for fluctuation pressure coefficients were quantitatively representative of field data. The main discrepancy observed was a more pronounced variability of measured, compared to simulated, shear layer peak pressures. Detailed measurements of the incoming wind field would be needed to further investigate this discrepancy. The results demonstrate the value of joint field measurements and LESs for studying wind effects.