SPRAY-WEB: Lagrangian Community model a fini di ricerca - Enrico Ferrero - Univ.PO

1. enrico.ferrero@uniupo.it
SPRAY-WEB 1.0
Un community model Lagrangiano per la ricerca
Enrico Ferrero
Università del Piemonte Orientale

2. enrico.ferrero@uniupo.it
Convenzione per la REALIZZAZIONE DELLA VERSIONE PUBBLICA DEL CODICE DI
DISPERSIONE LAGRANGIANO SPRAY (SPRAY-WEB)
Tra
•Ricerca sul Sistema Energe3co - RSE SpA
•CNR – ISAC, Consiglio Nazionale delle Ricerche, Is3tuto di Scienze dell’Atmosfera e del
Clima U.O.S. di Torino
•Università del Piemonte Orientale, Dipar3mento di Scienze e Innovazione Tecnologica
•ARIANET Srl,

3. enrico.ferrero@uniupo.it
Un po’ di storia
• Nell’arco degli ul3mi 30 anni, ques3 en3 hanno collaborato nello
sviluppo e nella messa a punto di un modello Lagrangiano “a par3celle”
denominato “SPRAY”.
• aPraverso una sessione di lavoro di due seQmane presso i laboratori di EDF – Direc3on des
Etudes et Reserch a Chatou (Parigi - Francia) a cui parteciparono Jacques Moussafir (EDF),
Domenico Anfossi (CNR-ICGF) di Torino, Giuseppe Brusasca, (ENEL-DSR) di Milano e Paolo
ZanneQ (EnviroComp);
• tale versione fu completata e resa opera3va
presso il Servizio Ambiente dell’ENEL-DSR di
Milano dal Gianni Tinarelli in collaborazione con
Giuseppe Brusasca e Domenico Anfossi.
•Tale codice “nasce” ufficialmente nel 1987

4. enrico.ferrero@uniupo.it
• Ques3 tre ricercatori cos3tuiscono il nucleo base di sviluppo e validazione
del codice nei primi anni
• successivamente nel 1990, si aggiunge al gruppo di sviluppo Enrico Ferrero
dell’Università del Piemonte Orientale e nel 1994 Silvia Trini Castelli del CNR-
ISAC (già ICGF) di Torino.
• Nel 1996-97 Stefano Alessandrini svolge la sua tesi di laurea sul codice, e
collaborerà allo sviluppo del codice, tramite finanziamenti del Fondo di
Ricerca per il Sistema Elettrico, come dipendente RSE, presso cui è assunto
nel 2001.
La storia continua………
• Jacques Moussafir seguirà sempre gli sviluppi del codice, in par3colare quelli
lega3 all’applicazione alla micorscala, e sarà lui a dare il nome SPRAY al codice di
calcolo quando uscirà da EDF per fondare la società ARIA Technologies SA.

5. enrico.ferrero@uniupo.it
Physica A 388 (2009) 1375–1387
Contents lists available at ScienceDirect
Physica A
journal homepage: www.elsevier.com/locate/physa
A hybrid Lagrangian–Eulerian particle model for reacting pollutant
dispersion in non-homogeneous non-isotropic turbulence
Stefano Alessandrinia
, Enrico Ferrerob,⇤
a
CESI RICERCA via Rubattino 54, 20134 Milano, Italy
b
Dipartimento di Scienze e Tecnologie Avanzate, Università del Piemonte Orientale, via Bellini 25/g, 15100, Alessandria, Italy
a r t i c l e i n f o
Article history:
Received 23 July 2008
Received in revised form 16 October 2008
Available online 16 December 2008
Keywords:
Dispersion
Turbulence
Stochastic model
a b s t r a c t
Lagrangian stochastic models are recognized as being powerful tools for pollutant
dispersion at different scales in complex terrain and at different stability conditions. One
of the still unresolved problems is the difficulty of including chemical reactions when, for
example, NO2 or O3 concentrations have to be predicted in the presence of NOx emissions.
In this work, a Lagrangian stochastic (single particle) model is modified in order to account
for simple chemical reactions and tested against measured data in a wind tunnel. It is
well-known that, in the single particle models the trajectories are considered independent
and hence the concentration correlations and fluctuations cannot be calculated. However,
these models can be simply modified to account for the segregation throughout a proper
parameterisation derived from measurements. Further, in order to avoid the use of the large
amount of computational resources, which would be necessary due to the release of an high
number of particles filling the whole domain, needed to reproduce the ozone background
concentration, we mark the particles with a deficit of ozone instead of its concentration. A
numerical experiment is carried out and the results of the comparisons between calculated
and measured concentrations of different species are presented and discussed.
© 2008 Elsevier B.V. All rights reserved.
1. Introduction
Lagrangian stochastic models are nowadays recognised as being the most powerful tool for predicting atmospheric
pollution at high resolution in complex terrain and different meteorological conditions. As a matter of fact, they
have demonstrated their ability in simulating concentration fields taking advantage of the sophisticated meteorological
measurements and/or of meteorological model forecasts [1].
The behaviour of a single passive tracer particle in a turbulent flow is well known both from theoretical and modelling
points of view. Since the early work of Taylor [2] a theory has been developed and the models have been applied to
different kinds of turbulent flows including non-homogeneous and non-isotropic turbulence. The most exhaustive work is
by Thomson [3], who proposed a complete theory for the one-particle dispersion in 3D turbulence based on the concept of
a Markovian stochastic process. In this frame, the dynamics of a passive tracer particle is described by a couple of stochastic
differential equations: the Langevin equation for the turbulent velocity and the Fokker–Planck equation for the Eulerian
probability density function (PDF) of the stochastic process. Based on this fundamental work many numerical models have
been developed aiming to simulate the dispersion of pollutants in the atmospheric boundary layer in different stability
conditions (see for instance Refs. [4–8]). These models are able to account for higher order turbulence moments of the
atmospheric PDF and complex turbulence flow dynamics [9,1], and have demonstrated that they accurately reproduce the
mean concentration field of the dispersed tracer.
⇤ Corresponding author.
E-mail address: enrico.ferrero@unipmn.it (E. Ferrero).
0378-4371/$ – see front matter © 2008 Elsevier B.V. All rights reserved.
doi:10.1016/j.physa.2008.12.015
Le principali pubblicazioni…..
A new Lagrangian method for modelling the buoyant plume rise
Stefano Alessandrini a,*, Enrico Ferrero b,c,1
, Domenico Anfossi c,2
a
RSE, Ricerca Sistema Energetico, via Rubattino 54, 20134 Milano, Italy
b
Università del Piemonte Orientale, Dipartimento di Scienze e Innovazione Tecnologica, viale Teresa Michel 11, 15121 Alessandria, Italy
c
ISAC-CNR, Istituto di Scienze dell’Atmosfera e del Clima, Corso Fiume 4, 10133 Torino, Italy
h i g h l i g h t s
The paper deals with the problem of the plume rise in the dispersion model.
A new technique based on temperature difference is used for plume rise computation.
Comparison of the model with experiments shows a satisfactory agreement.
a r t i c l e i n f o
Article history:
Received 5 October 2012
Received in revised form
12 April 2013
Accepted 26 April 2013
Keywords:
Plume rise
Pollution dispersion
Lagrangian model
a b s t r a c t
A new method for the buoyant plume rise computation is proposed. Following Alessandrini and Ferrero
(Phys A 388:1375e1387, 2009) a scalar transported by the particles and representing the temperature
difference between the plume and the environment air is introduced. As a consequence, no more par-
ticles than those inside the plume have to be released to simulate the entrainment of the background air
temperature. A second scalar, the vertical plume velocity, is assigned to each particle. In this way the
entrainment is properly simulated and the plume rise is calculated from the local property of the flow.
The model has been tested against data from two laboratory experiments in neutral and stable stratified
flows. The comparison shows a good agreement.
Then, we tested our new model against literature analytical formulae in a simple uniform neutral
atmosphere, considering either the case of a single plume or the one of two plumes from adjacent stacks
combining during the rising stage. Finally, a comparison of the model against an atmospheric tracer
experiment (Bull Run), characterized by vertically non-homogeneous fields (wind velocity, temperature,
velocity standard deviations and time scales), was performed. All the tests confirmed the satisfactory
performance of the model.
Ó 2013 Elsevier Ltd. All rights reserved.
1. Introduction
The computation of plume rise is one of the basic aspects for a
correct estimation of the transport and dispersion of airborne
pollutants and for the evaluation of ground level concentration.
A buoyant plume rises under the action of its initial momentum
and buoyancy. It experiences a shear force at its perimeter, where
momentum is transferred from the plume to the surrounding air,
and ambient air is entrained into the plume. This phenomenon,
entrainment, is responsible of the plume diameter increase, of the
decrease of its mean velocity and of the average temperature dif-
ference between air and plume as well. In the first stage the plume
also spreads under the action of the buoyancy-generated turbu-
lence but progressively the effect of ambient turbulence becomes
predominant. In a calm atmosphere, plumes rise almost vertically,
whereas in windy situations they bend over. In this case, the ve-
locity of any plume parcel is the vector composition of horizontal
wind velocity and vertical plume velocity in the first stage and then
approaches the horizontal wind velocity.
In the Eulerian dispersion models, the calculation of plume rise
is based on the fluid dynamic equations, namely on the mass,
momentum and energy conservation equations. A complete,
Contents lists available at SciVerse ScienceDirect
Atmospheric Environment
journal homepage: www.elsevier.com/locate/atmosenv
Atmospheric Environment 77 (2013) 239e249
Atmospheric Environment 40 (2006) 7234–7245
Tracer dispersion simulation in low wind speed conditions with
a new 2D Langevin equation system
D. Anfossia,Ã, S. Alessandrinib
, S. Trini Castellia
, E. Ferreroc
,
D. Oettld
, G. Degraziae
a
C.N.R., Istituto di Scienze dell’Atmosfera e del Clima, Torino, Italy
b
CESI S.p.A., Milano, Italy
c
Dipartimento di Scienze e Tecnologie Avanzate, Universita` del Piemonte Orientale, Alessandria, Italy
d
Institute for Internal Combustion Engines and Thermodynamics, Graz University of Technology, Graz, Austria
e
Universidade Federal de Santa Maria, Santa Maria, RS, Brazil
Received 17 January 2006; received in revised form 10 April 2006; accepted 23 May 2006
Abstract
The simulation of atmospheric dispersion in low wind speed conditions (LW) is still recognised as a challenge for
modellers. Recently, a new system of two coupled Langevin equations that explicitly accounts for meandering has been
proposed. It is based on the study of turbulence and dispersion properties in LW. The new system was implemented in the
Lagrangian stochastic particle models LAMBDA and GRAL. In this paper we present simulations with this new approach
applying it to the tracer experiments carried out in LW by Idaho National Engineering Laboratory (INEL, USA) in 1974
and by the Graz University of Technology and CNR-Torino near Graz in 2003. To assess the improvement obtained with
the present model with respect to previous models not taking into account the meandering effect, the simulations for the
INEL experiments were also performed with the old version of LAMBDA. The results of the comparisons clearly indicate
that the new approach improves the simulation results.
r 2006 Elsevier Ltd. All rights reserved.
Keywords: Low wind speeds; Meandering; Dispersion; Tracer experiments; Langevin equation system
1. Introduction
It is well known that atmospheric dispersion in
low wind speed (LW hereafter) conditions is mainly
governed by meandering (low frequency horizontal
wind oscillations). In such conditions, the airborne
pollutants are generally dispersed over rather wide
angular sectors. This fact together with the problem
of defining a precise mean wind direction in such
conditions makes the simulation of dispersion
rather difficult. In particular, it appears that
standard Gaussian plume models are inadequate
(Sagendorf and Dickson, 1974; Wilson et al., 1976;
Brusasca et al., 1992; Oettl et al., 2001) and more
appropriate types of models, taking explicitly into
account the meandering phenomenon, must be
used. With reference to this, we briefly recall that
ARTICLE IN PRESS
www.elsevier.com/locate/atmosenv
Atmospheric Environment Vol. 27A, No. 9, pp. 1443 1451, 1993. 00046981/93 $6.00+0.00
Printed in Great Britain. q') 1993 Pergamon Press Ltd
A SIMPLE WAY OF COMPUTING BUOYANT PLUME RISE
IN LAGRANGIAN STOCHASTIC DISPERSION MODELS
D. ANFOSSI
Istituto di Cosmogeofisica del CNR. C.so Fiume 4, 10133, Torino, Italy
E. FERRERO
Istituto di Fisica Generale, Via Giuria 1, 10125 Torino, Italy
and
G. BRUSASCA,A. MARZORATI,and G. TINARELLI
ENEL/CRTN, Servizio Ambiente, Via Rubattino 54, 20134 Milano, Italy
(First received 16 June 1992 and in finaljbrm 25 November 1992)
Abstract--A simple and easy to use method to include Eulerian plume rise in Lagrangian particle models is
presented. This approach takes into account the vertical variation of wind and stability. Its ability to
realistically simulate plume rise and spread, both through numerical experiments under typical atmo-
spheric conditions and by comparison with actual plumes detected by a Differential Lidar in a case study, is
shown. Despite its simplicity, our method proved to describe the main plume rise characteristics in a
satisfactory way and to yield a fair agreement among observed and predicted plume centreline heights and
standard deviations.
Key word index: Plume rise, Lagrangian particle models, plume spread, elevated releases, field experiment.
F
Fi
F'
9
n(t)
He
Hs
r
s
t
ra
rf
rLi
U
U I
v
y;
w
W~
w b
Wo
Z
At
AZ
Oo/Oz
P
NOTATION
buoyancy flux (m4 s- 3)
buoyancy flux of each particle (m4 s- 3)
buoyancy flux fluctuation (m4 s- a)
acceleration of gravity (m s- 2)
mean plume centreline height at a particular
cross-section (m)
plume rise as a function of t (m)
final plume rise (m)
stack height (m)
stack radius (m)
stability parameter (s-2)
time (s)
ambient temperature (K)
plume exit temperature (K)
Lagrangian time scales (s)
longitudinal component of wind speed (m s- 1)
mean horizontal wind speed at a particular cross-
section (m s- 1)
wind fluctuation (m s- 1)
crosswind component of wind speed (m s- 1)
horizontal downwind distance of a cross-section
(m)
vertical component of wind speed (m s- 1)
fluctuation of the vertical component of wind
speed (m s- 1)
buoyancy velocity (m s 1)
plume exit velocity (m s-1)
vertical coordinate (m)
time step (s)
plume rise increment (m)
ambient potential temperature gradient (K m- 1)
random forcing (m s- 1)
mean range (m)
o-u
ov
ow
{7Z
longitudinal component of wind speed standard
deviation (m s- 1)
crosswind component of wind speed standard
deviation (m s i)
vertical component of wind speed standard devi-
ation (m s 1)
vertical standard deviation (m)
horizontal standard deviation (m).
I. INTRODUCTION
A correct estimation of buoyant plume rise is one of
the basic requirements for the determination of
ground level concentrations of airbone pollutant
emitted by industrial stacks. In fact maximum ground
level concentration is roughly inversely proportional
to the square of the plume final height H e.
In Eulerian models, H e is generally computed by
means of simple analytical expressions like the well-
known Briggs' formulae (1975) and inserted in the
model as an input parameter. In contrast, the inclu-
sion of plume rise in Lagrangian particle models is not
as straightforward as in the Eulerian ones, so that a
complete and rigorous method is not yet available.
This is mainly due to the intrinsic impossibility of
Lagrangian particle models to simulate correctly the
entrainment phenomenon. To do this, these models
should take into account all the fluid simultaneously.
In fact the buoyancy and, as a consequence, the
1443

6. enrico.ferrero@uniupo.it
principali caratteristiche dell’attuale versione
di SPRAY-WEB
Reazioni chimiche
Diverse versioni di plume rise dinamico
Deposizione secca e umida
Equazione di Langevin
per la temperatura
223 ONOONO
k
+→+ 322 ONOhONO
J
+→++ ν
Author
Figure 3. Comparison of the measured (Weil et al. 2002) and simulated (both with Anfossi et al.,
1993 and Bisignano and Devenish, 2015 plume rise methods) vertical profiles of dimensionless
crosswind-integrated concentration as a function of dimensionless downwind distance for F∗=0.1.
Buoyant plume rise in Lagrangian particle models 1449
IIII.
911.
8111.
711.
61111.
• '~llll.
411.
N
310.
21111.
100.
I.
II.
' I ~ I I ' I ' I ' I ' I ' I ' I
- - - ~- -.:-_--::.-':;: .... i.--; .;:::~: ~:----.;':~-:.--..::
• -~ =:t-~..=- -. .; .:. , -.. ~.:.~. ...~_,_:~:..:,::_;,.7..7L..._2.....,~.
_.. -~-:*: '-;-:.'.'--:~---. ::- L- ..... ..:. -.;:-;i;--?:%:~:. !S-:: :---~:.-:.--:- L
OOWNVlND DISTANCE (ml
Fig. 4. As in Fig. 2 but for a non-homogeneous multi-layered atmosphere. A deep fog fills the
first layer, from 0 to 250 m, where almost calm conditions prevail (u= 1ms-t); an inversion
layer with a large temperature increase (8°) lies between 250 and 400 m; the superior layer is
isothermal.
LIOAR
1000
. •
soo I- I~ :,
21s~[
0
123:
500 l 1000x (m)
200~
1115~i
Fig. 5. Plain view of relative Lidar--stack positions,
cross-sections and plume simulation.
Table 1. Sostanj Power Plant specifications on 2 April 1991
Stack Stack Exit Exit
Stack height radius velocity temperature
operating H+ r wo Tf
(m) (m) (m s - ~) (K)
5 230 3.1 9.2 450
1,2,3 100 3.25 6.2 450
in the first cross-section (185 °) only the plume emitted
by stack 5 was tracked; in the second cross-section
(200°) both plumes (not yet combined) were detected
and in the third one (215°) the merged plume was
measured.
Wind (u, v, w) information was continuously collec-
ted by the ENEL Doppler Sodar (Elisei et al,, 1986),
averaging every 30 min, from a minimum height of
50 m to a maximum height of about 1000 m (space
resolution of about 50 m). However, as the Doppler
Sodar (located near the DIAL) was more than 1000 m
away from the stacks and the terrain was rather
complex, the (u, v, w) data obtained by the mass
consistent model MINERVE (Geai, 1987), initialized
by the Sodar profile and by six low level wind records

7. enrico.ferrero@uniupo.it
Ai potenziali u-lizzatori che scaricheranno il codice dalla rete verrà richiesto di
aderire alle policy di uso indicate, a7raverso la presa in visione e firma di un
documento con il quale viene preso ufficialmente l’impegno di rispe7arle.
Art. 1 Finalità
versione pubblica denominata SPRAY-WEB del modello SPRAY per fini di ricerca.
Le versioni di SPRAY-WEB saranno distribuite con obbligo di u3lizzo per aQvità
di ricerca senza scopi commerciali.

8. enrico.ferrero@uniupo.it
Le par3 concordano sul faPo che SPRAY-WEB debba diventare un
community model, che incen3vi il libero scambio di informazioni e di
innovazione riguardo gli sviluppi teorici e favorendo le potenziali
applicazioni pra3che con modelli di questo 3po.
Un comunity model Lagrangiano per la ricerca

9. enrico.ferrero@uniupo.it
L'aQvità è finalizzata alla realizzazione di una
piaDaforma informaGca online su cui risiederà il
codice SPRAY-WEB, compreso di sorgen- e script
necessari alla compilazione, nonché di case studies
per testare il codice.
Art. 2 Descrizione dell’aQvità
La versione del codice che verrà messa a disposizione sul sito sarà
concordata tra gli en3 aderen3 alla convenzione e le decisioni in
proposito dovranno essere prese all'unanimità.
Si intende creare, aPraverso questo
strumento un gruppo di sviluppatori e
uGlizzatori del codice

10. enrico.ferrero@uniupo.it
Verrà realizzato un blog per la discussione tra i
gestori del sito (vedi En3 partecipan3) e gli
u3lizzatori, ed analogamente per ciò che riguarda
aggiornamen3, nuove rou-nes o versione del codice
stesso.
Non si prevede di fornire agli u3lizzatori strumen3
per il pre e post processing, così come i da3
meteorologici di input.
Sarà messo a disposizione un esempio di codice di
interfaccia tra un modello meteorologico e SPRAY-
WEB.
Come funziona?

11. enrico.ferrero@uniupo.it
u3lizzo di Git (Torvalds et al.) come sistema di controllo
distribuito per la ges3one dello sviluppo di SPRAY-WEB;
u3lizzo della piaPaforma web GitHub o GitLab, dedicata
allo sviluppo e alla distribuzione di strumen3 FOSS (Free/
Libre Open-Source Sorware).
๏versione “single processor” per sistema opera3vo
“Linux”,
๏opportuno “test case”
๏“quick manual” (in italiano ed inglese),
๏ nonché l’aggiornamento del Manuale
In dettaglio…..

12. enrico.ferrero@uniupo.it
Attuali collaborazioni