FIELD EXPERIMENT MUST Short Term Scientific Mission, COST 732 Efthimiou George 1, Silvia Trini Castelli 2, Tamir Reisin 3 31 March - 5 April 2008, Torino, Italy 1 Department of Engineering and Management of Energy Resources, University of West Macedonia, Kozani, Greece 2 Institute of Atmospheric Sciences and Climate National Research Council, Torino, Italy 3 SOREQ NRC, Yavne, Israel Thessaloniki May 2008

Purpose of this STSM  Mock Urban Setting Test (MUST) is one of the most successful field experiment containing a rich and comprehensive dataset. Largely used by the scientific community, it includes detailed information about tracer concentrations and turbulence.  COST 732 WGs used mainly Wind Tunnel data (Bezpalcova, 2005).  The purpose of this STSM was to process the field campaign’s data in order to prepare a specific data set to further validate CFD and non-CFD codes for the field experiment conditions. Thessaloniki May 2008

Description of the work carried out during the visit  General description of the MUST field experiment (buildings, equipment).  Examination of existing meteorological and concentration data sets.  Development of software to handle data.  Processing of Velocity and Concentration Time Series – Statistics. Thessaloniki May 2008

General description of the MUST field experiment (buildings, equipment)  The geometry and the coordinates of the Wind Tunnel experiment is supposed to be the same as used in the Field experiment [0 degree case]. Accordingly: The Shipping Containers. The VIP van for the collection of wind and concentration data. The 32-m tower near the centre of the container array. The 6-m towers in each of the four quadrants. The measurements of concentrations in the four sampling lines. Thessaloniki May 2008

Meteorological Measurements & Tracer Detection  Concentration Measurements Ultraviolet Ion Collectors (UVIC). Digital Photoionization Detection (digiPID).  Meteorological Measurements Dugway Proving Ground (DPG) data. Defense Science and Technology Laboratory (DSTL) data. Thessaloniki May 2008

Ultraviolet Ion Collectors (UVIC)  These files include time series of concentration in ppm with time interval 0.01s.  There are 24 UVICS mounted on the four 6-m towers A, B, C, D.  On each of these 6-m towers, 6 UVICs were deployed at the following levels: 1, 2, 3, 4, 5 and 5.9 m. Thessaloniki May 2008 Yee, E. and Biltoft, 2004

Ultraviolet Ion Collectors (UVIC)  There are 2 other files: Tip.dat (2 m above the ground on the 32 m tower) Wake.dat (2 m above the ground, 1m behind the center of the building H4) Thessaloniki May 2008 Yee, E. and Biltoft, 2004

Digital Photoionization Detection (digiPID)  These files include time series of concentration in ppm with time interval 0.02s.  There are 48 ascii files which correspond to horizontal and vertical profiles of concentration.  Horizontal profiles of concentration fluctuations were measured using 40 dPIDs which were arrayed along the four horizontal sampling lines that were parallel to and centred in the street canyons.  The concentration detectors along the four horizontal sampling lines were placed at a height of 1.6 m. Thessaloniki May 2008

Digital Photoionization Detection (digiPID)  Vertical profiles of concentration statistics were characterized by 8 dPIDs deployed on the 32-m lattice tower near the centre of the obstacle array at heights of 1m, 2m, 4m, 6m, 8m, 10m, 12m, and 16m. Thessaloniki May 2008 Yee, E. and Biltoft, 2004

DPG wind data  These files include time series of velocities and temperature with time interval 0.1s.  Measurements of the vertical profiles of the mean horizontal wind velocity and temperature in the upwind flow obtained from a 16-m telescoping pneumatic mast.  Similar 2-D sonic anemometer/thermometers were mounted at the 4-, 8- and 16-m levels of a 16-m pneumatic mast downwind of the back of the obstacle array.  Vertical profiles of mean wind speed and temperature were obtained from the 32-m lattice tower located near the center of the obstacle array. Thessaloniki May 2008

DPG wind data  Also there are 4 more positions with measurements inside the domain.  V2In front of the building G5 1.15m  V4Between the buildings G6 and G7 1.15m  V3Between the buildings G6 and H6 1.15m  V12.5m Northwest of the building Η6 1.15m Thessaloniki May 2008 Yee, E. and Biltoft, 2004

DSTL wind data  These files include time series of velocities and temperature with time interval 0.05s.  There are 8 ascii files which correspond to velocities and temperatures at heights 2 and 6 m.  These data belongs to the four towers A, B, C, D. Thessaloniki May 2008

Examination of existing meteorological and concentration data sets  There are two main sets of data acquired during the trials, namely: The dispersion data which were obtained using 74 high- speed photoionization detectors (48 DPIDs and 26 UVICs). The meteorological dataset (i.e. the wind velocity and sonic temperature), which was obtained using 22 sonic anemometers (14 DPG and 8 DSTL). Thessaloniki May 2008

Examination of existing meteorological and concentration data sets  We selected a first sub-set of data, collected during two days (25 and 26 September 2001) and corresponding to a neutrally stratified atmospheric surface layer (ASL) according to Monin Obukhov Length.  We chose the experiment that corresponds to the release starting at 18:30 and ending at 18:45 on 25 of September Thessaloniki May 2008

Development of software to handle data Thessaloniki May 2008  A tailored Fortran code was written as a flexible tool that allows reading the time series of velocities, concentration and temperature, thus calculating mean values and variances for any averaging time, chosen by the user.  The output files, as time series and averaged fields can be used by the COST WGs, CFD and non-CFD, for numerical model simulation.

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008  The concentration time series were acquired over sampling times of 15 minutes for most of the continuous release experiments.  The MUST dataset authors made the following processing of the data:  Because background meteorological conditions may change over the 15-minute sampling time duration, it was necessary to apply conditional sampling to the concentration time series.

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008  For this reason they extracted 3- to 5- minute period from each record of 15-minute duration with a minimal variation of mean wind direction.  Finally they used this 3- to 5- minute period as the standard sampling period for computation of the plume concentration statistics.

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008  According to the above, two periods from trial 25 September 2001 were chosen for analysis.  These two periods ( seconds and seconds) were the same both for velocities and concentrations and primarily based on the stationarity (i.e., speed and direction) of the wind over the period.

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008 Mean value o

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008  We performed the same analysis on the original data as carried out by the MUST data referees, checked and compared our results with their averaged data.  For velocities and temperatures we chose also to analyze a 30 minutes period and we calculated the statistics producing a time series of data averaged over one minute. For concentrations we performed an analogous analysis but for a period of 17 minutes.

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008 Velocities – Temperatures seconds (18:30:40 – 18:44:00), values averaged over 800 s seconds (18:34:00 – 18:37:20), values averaged over 200 s 30 minutes period (18:15:00 – 18:45:00), time series of data averaged over 1 minute

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008 Velocities – Temperatures For each data record from each sonic anemometer, we computed the following quantities:  Mean velocity in each direction:,, and (m s -1 ). Note that W is not available for the two-axis sonic anemometers mounted on the pneumatic masts just upstream and downstream of the MUST array.  Mean direction:  Velocity standard deviations of the velocity fluctuations in the x, y, z directions:,, and (m s -1 ).

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008 Velocities – Temperatures  Turbulence kinetic energy:  Mean temperature: (K)  Covariances: and.  Temperature flux: in ms -1 K.  Friction velocity:

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008 Velocities – Temperatures  Local free convection velocity scale: where g=9.8 m s -2 and z is the height (m) of the anemometer above the ground surface.  Monin-Obukhov length: where κ = 0.4 von Karman’s constant.  Sensible heat flux: (W m -2 ) where ρ=1.2 kg m -3 is density of air, and C pa =1005 J kg -1 K -1 is specific heat capacity of dry air at constant pressure.

Processing of Velocity and Concentration Text Files Thessaloniki May 2008 Meteorological variables (200s)

Processing of Velocity and Concentration Text Files Thessaloniki May 2008 Meteorological variables (15 min)

Meteorological plots Thessaloniki May 2008  Velocity U, V, W  Direction  Temperature  Turbulence Kinetic Energy

Inflow wind 0 Thessaloniki May 2008

Inflow wind 0 Thessaloniki May 2008 ? 0

Inflow wind 0 Thessaloniki May 2008

Inflow wind 0 Thessaloniki May 2008

Inflow wind 0 Thessaloniki May 2008

Inflow wind 0 Thessaloniki May 2008

Inflow wind Thessaloniki May 2008

Inflow wind 0 Thessaloniki May 2008

Inflow wind 0 Thessaloniki May 2008

Inflow wind 0 Thessaloniki May 2008

Calculation of turbulence kinetic energy in upwind mast (consistency check) Thessaloniki May 2008  Because the upwind mast consists only from 2-D sonic anemometer-thermometer we did not have the 3rd component of velocity (w) and we calculate Turbulent Kinetic Energy in four ways:  We calculate first the time series of the variances of the velocities fluctuations,. Then we calculate the time series of turbulent kinetic energy according to the relation: and finally:

Calculation of turbulence kinetic energy in upwind mast Thessaloniki May 2008  Like in the first way but this time we account also for the prime of velocity w΄w΄ according to the relation (Yee and Biltoft, 2004). The time series of turbulent kinetic energy becomes:

Calculation of turbulence kinetic energy in upwind mast Thessaloniki May 2008  In the following process we calculate the mean values of the standard deviations of wind velocity fluctuations and denoted as varu, varv where

Calculation of turbulence kinetic energy in upwind mast Thessaloniki May 2008  Like in the third way but at this time we account also the mean value of the prime w΄w΄ according to the relation (Yee and Biltoft, 2004) and the mean value of turbulent kinetic energy becomes: South Tower Numerical Simulation 4m m m

Calculation of turbulence kinetic energy - mistakes Thessaloniki May 2008  From the MUST data we noticed that turbulent kinetic energy in the statistics file is erroneously calculated as follows: South Tower MUST data 4m m m

Calculation of turbulence kinetic energy - mistakes Thessaloniki May 2008 The part of the script DPGSONIC.MAT which refers to TKE: Ubar = mean U ;/* mean x component*/ \ Vbar = mean V ;/* mean y component*/ \ Wbar = mean W ;/* mean z component*/ \ Tbar = mean T ;/* mean temperature*/ \ Abar = RADTOD * {atan Ubar Vbar} ;/* mean bearing (deg)*/ \ Sbar = sqrt{{Ubar*Ubar}+{Vbar*Vbar}} ; /* mean wind speed*/ \ /* compute deviations from mean*/ \ dU = U - Ubar ; dV = V - Vbar ; dW = W - Wbar ; dT = T - Tbar ; dA = A - Abar ; /* compute variances*/ \ U2 = mean{dU * dU} ; V2 = mean{dV * dV} ; W2 = mean{dW * dW} ; T2 = mean{dT * dT} ; A2 = mean{dA * dA} ;

Calculation of turbulence kinetic energy - mistakes Thessaloniki May 2008 /* compute standard deviations*/ \ U1 = sqrt{ U2 } ; V1 = sqrt{ V2 } ; W1 = sqrt{ W2 } ; T1 = sqrt{ T2 } ; A1 = sqrt{ A2 } ; TKE = {0.5}*{sqrt{ U2+{V2+W2} }} ;/* turbulent kinetic energy*/ \

Calculation of mean direction - mistakes Thessaloniki May 2008  In the file explaining how to calculate the direction they suggest first to calculate the instantaneous direction and then to average these time series to obtain the mean value.  The procedure described does not output the values as in the file of statistics M  We apply the correct way to average the wind direction: for every averaging period, we calculated the mean values of wind components and after calculate the corresponding wind direction on the averaged and

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008 Concentrations seconds (18:30:40 – 18:44:00), values averaged over 800 s seconds (18:34:00 – 18:37:20), values averaged over 200 s 17 minutes period (18:29:00 – 18:46:00), time series of data averaged over 1 minute

Processing of Velocity and Concentration Time series - Statistics Thessaloniki May 2008 Concentrations After the conditional sampling of concentration, we computed the following concentration statistics:  Mean concentration: (ppm).  Concentration standard deviation of the concentration fluctuation:  Concentration fluctuation intensity:

Processing of Velocity and Concentration Text Files Thessaloniki May 2008 Concentrations (200s)

Processing of Velocity and Concentration Text Files Thessaloniki May 2008 Concentrations (17 min)

Concentration plots Thessaloniki May 2008  Mean concentration

Thessaloniki May 2008 Inflow wind Position of the source

Thessaloniki May 2008 Inflow wind Position of the source

Thessaloniki May 2008 Inflow wind Position of the source

Inflow wind Thessaloniki May 2008

Inflow wind Thessaloniki May 2008

Further discussion Thessaloniki May 2008  Except from the known measurements points in COST there are others for which we have the data but we do not know their exact positions inside the domain. Milliez and Carissimo, 2008