Environmental Engineering Reference
In-Depth Information
1.6 Implementation of Efficient Two-Way Mesoscale-
Microscale Coupling Using Interpolating Metamodels
G. Tsegas, Ph. Barmpas, I. Douros, and N. Moussiopoulos
Laboratory of Heat Transfer and Environmental Engineering, Aristotle University, Greece
Abstract
Coupling between prognostic mesoscale and microscale flow models
has been suggested as a way to improve the accuracy of calculated flows around
artificial structures and over densely-built areas. Nevertheless, implementations of
full two-way mesoscale-microscale coupling are burdened by large scale mismatches
as well as the formidable computational cost of on-line microscale calculations. In
the present work, an efficient implementation of an effective two-way coupling
between the mesoscale model MEMO and the microscale model MIMO, based on
interpolating metamodels, is introduced. The performance of two different meta-
model formulations is assessed using vertical flow profiles obtained from a two-way
coupled mesoscale-microscale calculation in the urban area of Athens, Greece.
Keywords
Metamodelling, urbanisation, mesoscale, microscale, two-way coupling
1. Introduction
The main obstacles for an efficient implementation of two-way coupling between
mesoscale and CFD flow models are the large gap of spatial and temporal scales
as well as the prohibitive chasm between the execution speeds of the two classes
of models. Metamodelling techniques (Piñeros Garcet et al., 2006) offer a potential
solution by approximating numerically expensive physical modelling with multi-
dimensional interpolators. As part of a mesoscale-microscale coupling scheme, a
metamodel's function is to approximate the area-averaged outflow response of
each microscale domain for various inflow conditions without performing a full
microscale calculation at each timestep. In the present work, two metamodel-based
schemes for implementing two-way coupling of the mesoscale model MEMO
(Moussiopoulos et al., 1993) and the microscale model MIMO (Ehrhard et al.,
2000) are developed and evaluated using a mesoscale test case.
