Automated terrain generation for precise atmospheric boundary layer simulation in the wind tunnel

Abstract

This study presents a two-stage framework to characterize boundary layer wind tunnel (BLWT) approach flows naturally developed over grid roughness for partial atmospheric boundary layer (ABL) simulation. The first stage applies curve fitting techniques to a comprehensive series of high-resolution spatially-averaged velocity profile measurements to estimate aerodynamic roughness parameters (ARPs) for a wide range of homogeneous (i.e., equal height) roughness element configurations. For this study, an automated (i.e., computer-controlled) 62 ​× ​18 roughness element array called the Terraformer was used to generate 33 unique roughness element fields. The mean flow structure was captured downwind to the Terraformer, where key ARPs—i.e., the urban canopy attenuation coefficient, zero-plane displacement height, shear (friction) velocity, roughness length, and Coles’ wake strength coefficient—were estimated. In contrast to previous ABL modeling methods that primarily focused on curve fitting of the inertial sublayer (ISL), the proposed approach applies the urban canopy exponential profile within the roughness sublayer (RSL), the log law in the ISL, and the law of the wake in the outer wake layer to model full-depth (i.e., floor to freestream) rough-wall turbulent boundary layers. Further, the method explicitly captures potential variability of Reynolds shear stress in the ISL and the wake strength in the outer layer to generalize characterization of naturally-developed BLs produced by traditional tunnel designs. The second stage applies a morphometric model for each ARP—calibrated with estimates from Stage 1—to predict flow characteristics for a wide range of roughness element configurations, with the goal of producing a deterministic solution for selecting an element configuration to satisfy user-specified aerodynamic objectives for the approach flow. The calibrated models effectively interpolate between estimates, e.g., ARPs estimated for open and suburban terrains can be applied in the second stage model calibration to predict ARPs for a “rough-open” condition without further experimentation. The findings of this study demonstrate that coupling the proposed framework with a mechanized roughness element grid can significantly reduce the trial-and-error required to commission a BLWT, while improving the quality of flow characterization.