URBAN CFD EXTREME WIND AIR QUALITY SIMULATIONS FOR

URBAN CFD EXTREME WIND AIR QUALITY SIMULATIONS FOR INDIANAPOLIS Indiana GIS Conference, Remote sensing Workshop March 12, 2007 Dr. Erdal Yilmaz, CFD Laboratory, Mechanical Engineering Department , IUPUI Email: eyilmaz@iupui. edu Web: http: //engr. iupui. edu/cfdlab

Urban CFD ¨ ¨ CFD (Computational Fluid Dynamics): numerical solution of the gas/liquid flow equations Solving flow equations for urban atmospheric conditions ¤ ¤ ¨ ¨ Low speed (incompressible) compared to aerospace and most mechanical engineering problems Turbulent flow by nature Comparatively big geographic area (> 1 mile x 1 mile) Need bigger grid size (>>1 millions elements) NOT a tornado simulation study Can be coupled to atmospheric prediction models

Where to use Urban CFD ¨ Air quality simulation ¤ ¤ ¨ Dispersion of NOX emission from vehicles Gaseous pollutant (SO 2, CO 2, NOX etc. ) release from plants Security ¤ ¤ ¨ Hazardous plume simulation, Homeland Security concerns On demand support for emergency response teams City planning ¤ ¤ ¨ Wind pattern analysis for existing building clusters Simulations for city expansion Exterior architectural and structural design ¤ ¨ (HVAC systems installations etc… ) Assist dispute resolution for wind damage ¤ Insurance: CFD tools to identify effect of building topology

CFD in IUPUI ¨ ¨ More than 20 years of experience Several simulation software, Fluent, STAR-CD etc and in-house tools Parallel computing with more than 2048 processors (IU BIGRed parallel cluster) Projects from Rolls-Royce, Cummins, Carrier, Eli-Lilly, NASA, Indiana State and more.

CFD Steps ¨ Generate CAD model ¤ ¤ ¤ ¨ Architectural resources: All buildings have external geometry in the form of a CAD file LIDAR data: Point clouds but needs to be converted to surface. Pictometry and others Generate CFD Grid/Mesh ¤ ¤ ¤ Surface/volume CAD model is needed. Sufficient grid density and boundary layer grids are needed. Surfaces are triangulated, volume is defined by tetrahedral or cubic elements Flow boundary conditions are defined Grid size may vary 1 to 100 millions elements

CFD Steps (cont) ¨ Flow solution ¤ ¤ Flow model feature: steady/unsteady, turbulence model, flow solver parameters ¤ ¨ Boundary conditions are defined: wind inflow, outflow, wall (ground/buildings) condition, gas (pollutant) flow are and mass fractions, surface temperatures, surface roughness etc. Solution may take 10 -100 CPU hours. Post-Processing ¤ Usually most fun part of whole process!!! ¤ Data is huge needs powerful computer (memory, speed, graphics) ¤ Solution layers/planes at different altitude or sections, extraction of flow properties, and 3 D virtual reality display. ¤ Data extraction can be generated by batch runs ¤ Can be integrated into GIS mapping services

CAD and CFD models Li. DAR point clouds CAD geometry Auto. Cad drawing (reduction is needed) CFD grid, (GAMBIT) CFD flow solution (FLUENT)

Extreme Wind Condition Weather data from the National Oceanic and Atmospheric Administration (NOAA) shows that, on April 15, 2006, Indianapolis had winds as high as 85 mph which created damage to Regions Bank building in the downtown.

Extreme Wind Solutions

Extreme Wind Solutions (cont)

Regions Bank Tower: CFD Solutions

Regions Bank Tower: CFD Solutions (cont) Straight wind velocity at the top of the west face of the Regions Bank tower reaches to 100 mph, +15 miles more than actual wind speed, due to partial blockage of the other buildings.

Wind Force and Pressure What does 800 pascal mean? ~10 people standing on a glass panel Note that at the NW corner loads are at the same strength from both sides

Air Quality Simulations Motivations Building canyons or clusters affect dispersion of the gaseous pollutants in urban areas. There are medical studies causing child asthma hospitalization at low concentrations of SO 2 (100 -250 ppb, Ref. 1, 2) References: [1] Toxicological Profile For Sulfur Dioxide, US Department of Health and Human Services, Public Health Service, Agency for Toxic Substances and Disease Registry, ATSDR, December 1998 [2] “Effect of short-term exposure to gaseous pollution on asthma hospitalization in children: a bi-directional case-crossover analysis, ” M Lin, Y Chen, R T Burnett, P J Villeneuve and D Krewski , J. Epidemiol. Community Health 2003; 57; 50 -55

Pollutant Sources: Geographic Locations NO 2 CO SO 2 Based on EPA (Environmental Protection Agency) records for 1999.

Ambient Air Quality Standard The Clean Air Act, 1990, requires Environmental Protection Agency (EPA) to set National Ambient Air Quality Standards for pollutants considered harmful to public health and environment. Pollutant Standards Averaging Times Carbon Monoxide 9 ppm , 10 mg/m 3 35 ppm, 40 mg/m 3 8 -hour 1 -hour Nitrogen Dioxide 0. 053 ppm, 0. 1 mg/m 3 Annual Sulfur Dioxide 0. 03 ppm, 0. 14 ppm 0. 50 ppm, 1. 3 mg/m 3 (2 nd) Annual 24 -hour 3 -hour Quantities are not to be exceeded once a year, except annual averaging times.

Dispersion of SO 2 A local emission source iso-surfaces at EPA standard values iso-surface of the concentration > 0. 5 ppm Wind speed= 19 mph Wind Direction: SW (225 deg) iso-surface of the concentration > 0. 14 ppm Note: All dispersion simulations have been done with low resolution CFD grid due to memory limitation of the computer. Current grid size is 1. 2 million cells. However, parallel simulation is planned with finer grid scale, hence better solution quality.

Effect of Building Topology Downwind around building surfaces causes pollutants diffuse into street level. Colored contours are in ppm Upwind Downwind

CFD Solutions for GIS Mapping Services ¨ CFD can provide city level fine details of the wind flow patterns and dispersions of the gaseous pollutant ¨ CFD solution database can provide instant access combined with GIS mapping services. ¨ Firefighters, homeland security response teams, city planners, architects, and environmental study/monitoring groups can benefit from CFD integration

Geo-referenced CFD Extreme wind CFD solution is imported into ARCGIS software (velocity vectors)

Geo-referenced CFD (cont) Extreme wind CFD solution is imported into ARCGIS software (velocity contours)

Geo-referenced CFD (cont) Dispersion of SO 2 mapped on Aerial Image iso-surface of the concentration > 0. 5 ppm iso-surface of the concentration > 0. 14 ppm

CFD Database for GIS Map Services Following database is proposed from CFD results: ¨ Wind direction: 5 degrees interval, total of 72 individual cases, ¨ Wind speed: 5 -85 mph, total of 20 individual cases, ¨ CFD solution time: 24 hours/case/processor for sufficient mesh density Total # of CFD case runs = 72 x 20 = 1440 cases Total CPU hours = 1440 x 20 = 34560 hours/proc. , or = 45 days on 32 CPU parallel cluster. ¨ 2 D Images from the solution: 20 slices along z-direction ¨ 3 D images from the solution: 4 different angles ¨ Properties to display: 5 (pressure, velocity vectors, temperature etc… ) ¨ Number of image extraction for each case = (20+4)* 5 = 120 Total # of images =1440 x 120 = 172, 800 images ¨ User input parameters to request the data: 1) wind direction, 2) wind speed, and 3) 3 D section parameters.

Conclusions ¨ Building topology changes the wind patterns hence causing more complicated wind flow structure in the downtown area. ¨ Regions Bank tower was exposed to higher wind speed due to upstream building blockage hence causing stronger wind forces on the window panels. It was also exposed to highly recirculating unsteady wind structure. ¨ Integration of the CFD solutions into GIS mapping services can provide street level wind patterns, pressures, temperatures, etc. for a wide range of GIS users. ¨ Building topology affects street level dispersion of the pollutants. Some buildings have upwind on the front face while some have downwind. ¨ Comparison with actual ambient monitoring will be necessary to validate the model with finer grid resolutions. In addition, other source of SO 2 in the area should be included into the model for more accurate representation of the results. ¨ This is an ongoing research. Therefore, no conclusion regarding air pollution from any emission sources has been drawn yet.

Future Work q Comprehensive CFD modeling with fine grid scale q CFD solutions as a GIS layer in GIS mapping services

Acknowledgments ¨ MURI (Multidisciplinary Undergraduate Research Institute) in IUPUI for supporting this research, ¨ IMAGIS (Indianapolis Mapping and Geographic Infrastructure System) for providing Li. DAR data and Autocad Model of the Downtown Indianapolis, ¨ Environmental Affairs, CTE Perry K. Steam Plant, for providing flue gases data.