Sensitivity Studies Wind tunnel simulations Alan Robins, Paul - PowerPoint PPT Presentation

Sensitivity Studies

Wind tunnel simulations Alan Robins, Paul Hayden, Hong Cheng, Matteo Carpentieri, Sando Baldi, Paolo Giambini, Tom Lawton, James Hamilton, Alex Nicolson, Nadia Bahar, James Fabian, Khaled Bashiti, et al Content: The EnFlo Wind tunnel Comparisons with field data Flow and dispersion characteristics Sensitivity studies Conclusions

The EnFlo wind tunnel Inlet and heater Chilled water 20 x 3.5 x 1.5 m section - 15 layers supply ~ 10oC working section 400 kW capacity; Twin ambient to ~ 80oC 0 - 4 m/s fans Mechanical simulation Source devices Heat Cooled rough Rough wall ~ 10oC exchanger wall Heater control Thermo- 18 channel FID Flow Speed couple air sampling system control control Gas system system system supplies Traverse and Computerised control, data turntable collection & data analysis control

Wind speed ratios Ratio of BT Tower and WCL roof top wind speeds essential data for relating wind tunnel and field data High wind speed limit is about 1:4

Comparison of concentrations The scatter is mainly due to the large repeat variability in the field data. C * = CUH 2 Q

Comparison of flow fields Component along Marylebone Rd, U t relative to U ref U t U ref Data from WCC House (at 18 m), U (BT) > 3 ms -1

Comparison of flow fields Component normal to Marylebone Rd, U n relative to U ref U n U ref Data from WCC House (at 18 m), U (BT) > 3 ms -1

Flow structure at an intersection Clarification of physical processes in support of field measurements Mean velocity vectors at height of 5 m - extract from full data set

Near-source concentrations in helical circulation Wind at 45 roof degrees to street level roof Δ x ~ W at level location of measurements Emission released upstream at street centre - dispersion prior to uniform mixing across the street Concentration contours in cross-section; H/W ≈ 1.0

Short Duration Emissions Characterisation of advection speed, U c , and along wind mixing. Time of flight, T 50 - rise - fall Rise time, Δ T 50 - fall time 7.5(R/H) U c ≈ U H /4 2.5(R/H) Δ T 50 ≈ T 50 /3

Concentration decay not fully mixed Concentration, C* = CUH 2 / Q R is source- receptor separation Separation, R / H The basic mean concentration decay function, CUH 2 / Q = 12 ( R / H ) -2

Fluctuation levels Concentration, C* Concentration fluctuation, c'/ C max Time, s instrument noise level Separation, R / H

Wind tunnel research EnFlo 20x3.5x1.5 boundary layer wind tunnel 0 to 3.5 m/s; inlet flow heating; surface heating and cooling Sensitivity studies examples

Basic sensitivity studies Dispersion repeat run to run variability better than ±10% - so take ± 20% as the target for singificant effects On this basis, results shown to be insensitive to:  modest variations in upwind boundary layer conditions including slightly stable approach flow ( L MO ~ H bl ~ 1 m)  addition of roof roughness  modification to zero-plane level at upwind edge of model  source design and emission rate (speed ratio)  Reynolds, U ( H ) H / ν , number from about 2,000 to 14,000

1. Model detail Comparison of results from low and higher resolution models; significant differences confined to short ranges.

Model detail

2. Street blockages - ineffective

Street blockage Concentrations along Marylebone Road Reference; Wind direction blockage at A, B or -90, degs C; blockage at A, B & C. Green, reference; Wind direction blue, blockage at -45, degs A; red, at D; yellow at A & D.

3. Street blockage - effective 7 9 blockage Wind direction source -45 degrees

Blockage Run 7, York St a canyon from west of Upper Montagu Street to east of Gloucester Place Wind direction -90, degs Run 9, Gloucester Place closed at York Street basically the same 2 CU ref H bl Q 9 not these 7 x ( mm )

4. Scale effects 1:200 scale model 1:500 scale model

Scale effects CU ( H ) H 2 / Q along Marylebone Road, source York Street 1:200 scale CU ( H ) H 2 / Q 1:500 scale Distance from intersection, m

5. Slightly unstable approach flow Unstable Neutral H bl / L MO ~ -2.7, u* / U ref = 0.10 u* / U ref = 0.06 w ' 2 2 U ref w ' 2 2 U ref

Slightly unstable approach flow H bl / L MO = -2.7 Marlylebone Rd CU ref H bl 2 / Q along Marylebone Road, source York Street 3.00E+01 2.50E+01 Unstable Neutral 2.00E+01 1.50E+01 neutral CUA/Q CU ref H bl 2 / Q unstable 1.00E+01 5.00E+00 0.00E+00 -600 -400 -200 0 200 400 600 800 1000 Distance from intersection, mm -5.00E+00 X mm

Sensitivity studies  C* = CU H H 2 / Q independent of Re ( U ) and Q over range examined  C* independent of model scale 1:200 to 1:500  C* repeatability (multi-user) better than ±10%  Effect of shallow boundary layer on C* ( x , y ,0) ~ -25% (not shown)  Effect of model detail on C* ≤ +30%  Effect of upwind detail on C* < ±25%  Effect of slightly unstable approach flow on C* ~ - 30%  … but C* ~ 12( R / H ) -2 robust