The “teapot in a city”: A paradigm shift in urban climate modeling

Abstract

Urban areas are a high-stake target of climate change mitigation and adaptation measures. To understand, predict, and improve the energy performance of cities, the scientific community develops numerical models that describe how they interact with the atmosphere through heat and moisture exchanges at all scales. In this review, we present recent advances that are at the origin of last decade’s revolution in computer graphics, and recent breakthroughs in statistical physics that extend well-established path-integral formulations to nonlinear coupled models. We argue that this rare conjunction of scientific advances in mathematics, physics, computer, and engineering sciences opens promising avenues for urban climate modeling and illustrate this with coupled heat transfer simulations in complex urban geometries under complex atmospheric conditions. We highlight the potential of these approaches beyond urban climate modeling for the necessary appropriation of the issues at the heart of the energy transition by societies.

Path integrals formulations leading to propagator evaluation for coupled linear physics in large geometric models

“…A main strength of the Stardis code used in the following practical illustration is its ability to deal with huge amounts of physical and geometrical data. In [4], the city is simulated with interacting buildings, each of them described in full details, room per room, and the main message is that the computation time is fully insensitive to the level of refinement of this description 6 . The same observation can be made concerning the computation of propagators: the computation times required for the estimation of the propagators, and for their use in external codes with new sets of sources, are also both insensitive to the refinement level.…”

“…Presently, the solver is used along the Stardis command line tools 3 which are intented to be reference implementation of a complete workflow using stardis-solver. In addition, the stardis-solver is implemented in SYRTHES 4 , the general thermal software developed by Electricté de France R&D. Stardis was recently used to explore ways toward the definition of new climate services for analysts and designers anticipating climate change in urban area. In this context, Stardis strength was its ability to deal simultaneously with all the spatial and temporal scales involved in the modeling of energy exchanges, from the milimeter scale of windows and heat seals to the kilometers extensions of cities, and from the minute scale of wind and solar fluctuations to typically fifty-year lifetimes of the ground installations to be planed.…”

“…In this context, Stardis strength was its ability to deal simultaneously with all the spatial and temporal scales involved in the modeling of energy exchanges, from the milimeter scale of windows and heat seals to the kilometers extensions of cities, and from the minute scale of wind and solar fluctuations to typically fifty-year lifetimes of the ground installations to be planed. Radiative transfer inside and outside buildings could be modeled together with full three-dimension heat diffusion inside the solid structures, as well as convection inside each fluid cell, providing the first reported attempt to model energy transfers in a complete city, without any compromise on its geometrical description [4]. The corresponding approach was name "teapot in a city" by reference to the "teapot in a stadium" paradigm in computer graphics where it was indeed shown that ray tracing Monte Carlo algorithms could deal with quasi-infinite scale ratios on an industrial basis when producing cinema and video-game images.…”

“…The corresponding approach was name "teapot in a city" by reference to the "teapot in a stadium" paradigm in computer graphics where it was indeed shown that ray tracing Monte Carlo algorithms could deal with quasi-infinite scale ratios on an industrial basis when producing cinema and video-game images. Retaining Stardis for illustration of the present theoretical discussion of propagator estimation in complex systems, we want to highlight an essential idea to be added to the perspective statements of [4]: using Feynman-Kac strategy to extend the computer graphics strategies to other physics than light transport allows one to model the heat transfer in geometrically complex systems, but also the systematic computation of all the propagators associated to each energy source. We will concentrate on standard heat transfer physics, but the statement is valid for all the physical contexts where same path-integral Monte Carlo strategies were already reported, either using similitudes with slightly modified versions of Stardis (e.g.…”

Toward machine-assisted tuning avoiding the underestimation of uncertainty in climate change projections

“…Providing climate change projections with an estimation of this modeling uncertainty is an essential role of the Coupled Model Intercomparison Project (CMIP) exercises, which are conducted about every 7 years, in advance of Intergovernmental Panel on Climate Change (IPCC) reports. The multi-model CMIP ensemble is used as an entry for so-called impact studies, often using physical and statistical downscaling approaches, and recent advances in Monte Carlo methods pave the way to a systematic use of the CMIP simulations ensembles in such studies ( 8 ).…”