MISKAM simulation of the Oklahoma City case

Márton Balczó COST 732 MC /WG meetingMadrid, December 4, 2008

Budapest University of Technology and Economics (BME)

Department of Fluid Mechanics

Part I. - Model setup

Motivation

• MISKAM – a model in operation (100 users in Europe)

• not as sophisticated like e.g. Fluent – simple grid generation, fewer user settings

faster way to the results | no tuning.

• I am a user, not a model developer

• Model was already used in high-rise building project

The model

• RANS with k- turbulence closure, modified as suggested by Kato & Launder

(1993) and López (2002)

• Dispersion : Reynolds averaged advective diffusion equation (passive scalar)

• Cartesian non-equidistant grid, buildings blocked out of the grid

• Versions used in the OKC test: 5.02a (official); 6 beta3 (not yet released)

Domain

• Same as in the FLUENT grid of Jörg Franke

• 5H - inflow, top and lateral 10H – outflow (H max = 152m Chase Tower F9 )

• 2460 x 1200 x 912 m (full scale simulation)• Orientation: 180.7° || long axis

Grid

• MISKAM: Building accuracy depends on grid resolution

• Investigated area and close investigation area determined

• Building input : Rhino 3D model Background bitmap building polygons

prjection to grid in WinMISKAM

• 460 x 480m, 4x4m

• Outside: growth rate 1.25

• 284 x 156m, 2x2m • At building walls 1m

Grid - vertical resolution

• 1.5m up to the height of 75m, spreading to 5m at 125m and from then growth rate of 1.2

• Refinement to 0.5m at ground level and at the top of lower Park Ave buildings

(48-52m) to resolve the top separation bubble

246 x 216 x 108 = 5.74 mio.cells (OKC_3.inp)

Garages

• Openings only on one side of buildings (|| x or || y) possible• Garages outside of the investigated area or with too small openings not modelled.

No. of open floorsWT MISKAM

E08 Metro PG 7 4 25.5 YE11 Plaza PG 4 2 13.5 X / YE34 unknown 7 0 -F3 Broadway/Kerr PG 6 3 16.5-19.5 YF7 Santa Fé Plaza PG 5 2 22.5 X

F8B First National PG 7 5 25.5 YF10 Sheraton Hotel/Century Center PG 1 0 13.5 -

Building code

Building nameBuilding

height [m]Flow

direction

• Less floors due to rough grid, opening area ~ same• Floors fully open, wall structural elements

like in the wind tunnel were not modelled.

Source

• Canopy could not be modelled• Simple volume sources without momentum• Size of source region similar to the size and height of the canopy

Source geometry in wind tunnel (from the drawing of Bernd Leitl)

Sources in MISKAM

Boundary conditions

0

10

20

30

40

50

60

70

80

90

0.5 0.6 0.7 0.8 0.9 1 1.1

U/Uref [-]

Z [

m]

Wind tunnel, uw measured

Wind tunnel, uv measured

MISKAM z0 = 0.24m

MISKAM z0 = 0.48m

0

50

100

150

200

250

300

0 0.005 0.01 0.015 0.02 0.025 0.03

k/Uref2 [-]

Z [

m]

MISKAM z0 = 0.24m

MISKAM z0 = 0.48m

Wind tunnelinterpolation polynom

•Surfaces : z0 at the ground 0.12m, building walls 0.01m (for the wall functions)

•Inlet velocity: 8 m/s at 81m height

•Profile: logarithmic mean flow, k and generated by the model from equilibrium

•Lateral and outflow: no-flux, top: constant values from the top of the inlet profile

Numerical schemes

MISKAM version 5.02a 6 beta3

Convection terms    

Momentum equation

UpstreamPredictor-corrector method suggested by

MacCormack (1969)

Turbulence equations*

UpstreamMPDATA algorithm (Smolarkiewicz and

Grabowski, 1989)

Transport equation MPDATA algorithm

Diffusion terms    

  ADI (Alternating Direction Implicit)-method of Douglas and Rachford (1956)

Run ID Grid Description

OKC_01 OKC_3 MISKAM 6 beta3 with garages, z0 = 0.48

OKC_02 OKC_3 MISKAM 6 beta3 without garages, z0 = 0.48

OKC_03 OKC_3 MISKAM 6 beta3 with garages, z0 = 0.24

OKC_04 OKC_2* MISKAM 5.02a with garages, z0 = 0.48

Simulation runs

To be continued…

To be continued…