Developer Data Modeling Mistakes: From Postgres to NoSQL
Sviluppi modellistici sulla propagazione degli incendi boschivi
1. Sviluppi modellistici
sulla propagazione degli incendi boschivi
Gianni PAGNINI
Borsista RAS
PO Sardegna FSE 2007-2013 sulla L.R. 7/2007
“Promozione della ricerca scientifica e dell’innovazione tecnologica in Sardegna”
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
2. Introduction
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
3. Turbulence Sources in Wildland Fire Front
Propagation (1)
Wildland fire propagates at the ground level and then it is
dependent on the dynamics of the Atmospheric Boundary
Layer, whose flow is turbulent in nature.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
4. Turbulence Sources in Wildland Fire Front
Propagation (2)
Moreover, in this atmospheric layer the turbulence is amplified
by the forcing due to the fire-atmosphere coupling
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
5. Turbulence Sources in Wildland Fire Front
Propagation (2)
... and by the appearing of the fire-induced flow.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
6. Importance of Turbulence Modelling in Wildland Fire
As a consequence of the turbulent transport of the hot air mass,
that can pre-heat and then ignite the area ahead the fire,
the fire front position becomes random.
Hence, it is of paramount importance for the prediction of the
fire motion to take into account turbulence.
Accounting for the effect of turbulence on the fire propagation
improves the usefulness of the operational models and thereby
increases the firefighting safety and in general the
efficiency of the fire suppression and control efforts.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
7. The Level-Set Method (1)
Let Γ(t) = (x(s, t), y(s, t)) be a parameterized evolving
interface.
Let ϕ(x, t) be a function such that the level-set ϕ = constant
corresponds to the evolving front Γ(t). Then the equation for
the evolution of ϕ corresponding to the motion of the interface
Γ(t) is given by
Dϕ
= 0. (1)
Dt
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
8. The Level-Set Method (2)
Dϕ ∂ϕ dx
= + · ϕ = 0. (2)
Dt ∂t dt
dx ϕ
= V(x, t) = V(x, t) n , n=− , (3)
dt || ϕ||
∂ϕ
= V(x, t) || ϕ|| . (4)
∂t
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
9. The Rate of Spread (1)
The rate of spread is the fire front velocity, firstly estabilished by
Rothermel (1972) as an operative approximation of a
theoretically based formula due to Frandsen (1971),
V(x, t) = V0 (1 + fW + fS ) , (5)
where V0 is the spread rate in the absence of wind,
fW is the wind factor and fS is the slope factor.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
10. The Rate of Spread (2)
IR ξ
V0 = , (6)
ρb ε Qign
IR : reaction intensity
ξ: propagation flux ratio, the proportion of IR transferred to
unburned fuels
ρb : oven dry bulk density
ε: effective heating number, the proportion of fuel that is heated
before ignition occurs
Qign : heat of pre-ignition.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
11. The Model: Deterministic Front
Let x(t, x0 ) be a deterministic trajectory with initial condition x0 ,
i.e., x(0, x0 ) = x0 , and driven solely by the rate of spread
V(x, t).
Moreover, let ϕ(x, t) be the function with values 1 and 0 such
that ϕ(x, t) = 1 markes the burned area Ω(t), i.e.,
Ω(t) = {x, t : ϕ(x, t) = 1}, and ϕ(x, t) = 0 markes the unburned
area, i.e., x ∈ Ω(t).
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
12. The Model: Random Front (1)
Let Xω (t, x0 ) = x(t, x0 ) + σ ω be the ω-realization of a random
trajectory driven by the noise σ, with average value
Xω (t, x0 ) = x(t, x0 ) and the same fixed initial condition
Xω (0, x0 ) = x0 in all realizations.
Hence, the ω-realization of the fire line contour follows to be
ϕω (x, t) = ϕ(x0 , 0) δ(x − Xω (t, x0 )) dx0 . (7)
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
13. The Model: Random Front (2)
Since the trajectory x(t, x0 ) is time-reversible, the Jacobian of
dx0
the transformation x(t, x0 ) = Ft (x0 ) is = 1, then formula
dx
(7) becomes
ϕω (x, t) = ϕ(x, t) δ(x − Xω (t, x)) dx . (8)
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
14. The Model: Randomized LS-Method
Finally, after averaging, the effective fire front contour emerges
to be determined as
ϕω (x, t) = ϕ(x, t) δ(x − Xω (t, x)) dx
= ϕ(x, t) δ(x − Xω (t, x)) dx
= ϕ(x, t) p(x; t|x) dx
= p(x; t|x) dx = ϕeff (x, t) , (9)
Ω(t)
where p(x; t|x) = p(x − x; t) is the probability density function
(PDF) of the turbulent dispersion of the hot flow particles with
average position x.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
15. Turbulent Premixed Combustion (1)
week ending
PRL 107, 044503 (2011) PHYSICAL REVIEW LETTERS 22 JULY 2011
Lagrangian Formulation of Turbulent Premixed Combustion
Gianni Pagnini and Ernesto Bonomi
CRS4, Polaris Building 1, 09010 Pula, Italy
(Received 4 November 2010; published 21 July 2011)
The Lagrangian point of view is adopted to study turbulent premixed combustion. The evolution of the
volume fraction of combustion products is established by the Reynolds transport theorem. It emerges that
the burned-mass fraction is led by the turbulent particle motion, by the flame front velocity, and by the mean
curvature of the flame front. A physical requirement connecting particle turbulent dispersion and flame front
velocity is obtained from equating the expansion rates of the flame front progression and of the unburned
particles spread. The resulting description compares favorably with experimental data. In the case of a zero-
curvature flame, with a non-Markovian parabolic model for turbulent dispersion, the formulation yields the
Zimont equation extended to all elapsed times and fully determined by turbulence characteristics. The exact
solution of the extended Zimont equation is calculated and analyzed to bring out different regimes.
DOI: 10.1103/PhysRevLett.107.044503 PACS numbers: 47.70.Pq, 05.20.Jj, 47.27.Ài
Turbulent premixed combustion is a challenging scien- In this Letter, the fresh mixture is intended to be a popu-
tific field involving nonequilibrium phenomena and play- lation of particles in turbulent motion that, in a statistical
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
ing the main role in important industrial issues such as sense, change from reactant to product when their average
16. Turbulent Premixed Combustion (2)
In premixed combustion all reactants are intimately mixed at the
molecular level before the combustion is started, while in
non-premixed combustion the fuel and the oxidant must be
mixed before than combustion can take place.
Premixed Combustion process can be described as the
following one-step irreversible chemical reaction
Fresh Gas → Burned Gas + Heat .
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
17. Turbulent Premixed Combustion (3)
Premixed combustion is describe by a single non-dimensional
variable named progress variable c(x, t), wich represents the
burned mass fraction and it is defined statistically defined in the
Lagrangian approach as
c(x, t) = p(x; t|x) dx . (10)
Ω(t)
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
18. Fire Front Propagation (1)
Reynolds transport theorem
d ∂Ψ
Ψ(x, t) dV = dV + ˆ
Ψ u · nS dS ,
dt V (t) V (t) ∂t S
Divergence theorem
ˆ
Ψ u · nS dS = · (u Ψ) dV .
S V
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
19. Fire Front Propagation (2)
Fire front evolution equation (1)
∂ϕeff ∂p
= dx + · (V p) dx . (11)
∂t Ω(t) ∂t Ω(t)
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
20. Fire Front Propagation (3)
Fire front evolution equation (2)
Mean front curvature: κ(x, t) = · n/2,
∂ϕeff ∂p
= dx + V· xp dx
∂t Ω(t) ∂t Ω(t)
∂V
+ p xκ ˆ
· n + 2 V(κ, t)κ(x) dx .
Ω(t) ∂κ
The fireline propagation is driven by:
i) turbulent dispersion (i.e., p(x − x; t)),
ii) rate of spread (i.e., V(x, t)),
iii) mean front curvature (i.e., κ(x, t)).
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
21. Non-turbulent Limit
If turbulence is negligible, hot air particles deterministically
move, i.e., p → δ(x − x),
and the fire front propagation equation becomes
∂ϕeff
= V(x, t) || ϕeff || , (12)
∂t
which is the Hamilton–Jacobi equation corresponding to the
Level-Set Method.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
22. The Model: Heating-before-burning Law (1)
The model is completed by a law for the ignition due to the
pre-heating by the hot air mass.
Let T (x, t) be the temperature field and Tign the ignition value.
The most simple law for the temperature growing, when
T (x, t) ≤ Tign , is
∂T (x, t) Tign − T (x, 0)
= , (13)
∂t τ
so that
t
T (x, t) = T (x, 0) + Tign − T (x, 0) . (14)
τ
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
23. The Model: Heating-before-burning Law (2)
In the proposed model, the heating in an unburned point x i.e.,
0 < ϕeff (x, t) ≤ 0.5, is due to the presence of hot air.
Formula (13) changes according to
∂T (x, t) Tign − T (x, 0)
= ϕeff (x, t) , T ≤ Tign . (15)
∂t τ
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
24. The Model: Heating-before-burning Law (3)
For a given characteristic ignition delay τ , the time of
heating-before-burning ∆t is such that it holds
∆t
τ= ϕeff (x, ξ) dξ . (16)
0
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
25. Turbulent Dispersion Model
The most simple model for turbulent dispersion of hot air mass
around the average fireline is
∂p 2
=D p, p(x − x; 0) = δ(x − x0 ) , (17)
∂t
and then
1 (x − x)2 + (y − y)2
p(x − x; t) = exp − , (18)
2πσ 2 2 σ 2 (t)
where σ 2 (t) is the particle displacement variance related to the
turbulent diffusion coefficient D, i.e.,
σ 2 (t) = (x − x)2 = (y − y )2 = 2 D t . (19)
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
26. Analytic Results (1)
Plane fire front
ˆ
When the normal to the front n is constant, then the curvature κ
is null. The fireline propagation is driven by
∂ϕeff 2
=D ϕeff + V(t) || ϕeff || . (20)
∂t
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
27. Analytic Results (2)
The exact solution of (20) is
1 x − LR (t) x − LL (t)
ϕeff (x, t) = Erfc √ − Erfc √ , (21)
2 2 Dt 2 Dt
where Erfc is the complementary Error function,
LR and LL are the right and left fronts, respectively, defined as
dLR dL
= − L = V(t) , Ω(t) = [LL (t); LR (t)] .
dt dt
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
28. Comparison
randomized level-ser
level-set
1
0.8
0.6
0.4
0.2
0
-10 -5 0 5 10
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
29. Analytic Results (2)
If V = constant then
the Level-Set fire front position is Lf = L0 + Vt and it holds
1 L0 + V t 1
ϕeff (LR , t) = 1 − Erfc √ < , 0 < t < ∞, (22)
2 Dt 2
hence the “cold” isoline ϕeff = 1/2 is slower than the Level-Set
contour.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
30. Analytic Results (3)
The “hot” isoline ϕeff = 1/2 can be faster than the Level-Set
contour. In fact, the elapsed time ∆t neccessery to meet
condition (16)
∆t
τ= ϕeff (x, t) dt , (23)
0
can be shorter than the elapsed time δt such that
δt
∂ϕeff 1
ϕeff (x, δt) = ϕeff (x, 0) + dt = . (24)
0 ∂t 2
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
31. Analytic Results (4)
In particular, with the initial condition
1 , x ∈ Ω(0)
ϕeff (x, 0) = , (25)
0 , x ∈ Ω(0)
the “hot” isoline ϕeff = 1/2 is faster than the Level-Set contour
when
∂ϕeff ϕ
< eff , (26)
∂t 2τ
so that
∆t ∆t
∂ϕeff ϕeff 1
dt < dt = . (27)
0 ∂t 0 2τ 2
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
32. Figures (1)
a) b)
Time [min] Time [min]
100 1e+03 100 800
800 600
600 400
80 400 80 200
200
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
c) d)
Time [min] Time [min]
100 600 100 500
400 400
200 300
80 80 200
100
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Evolution in time of the fire line contour, when τ = 10 [min], for
the level-set method a) and for the randomized level-set
method with increasing turbulence: b) D = 25 [ft]2 [min]−1 ,
c) D = 100 [ft]2 [min]−1 , d) D = 225 [ft]2 [min]−1 .
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
33. Figures (2)
a) b)
Time [min] Time [min]
100 1e+03 100 1e+03
800 800
600 600
80 400 80 400
200 200
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
c) d)
Time [min] Time [min]
100 1e+03 100 1e+03
800 800
600 600
80 400 80 400
200 200
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Evolution in time of the fire line contour, when τ = 50 [min], for
the level-set method a) and for the randomized level-set
method with increasing turbulence: b) D = 25 [ft]2 [min]−1 ,
c) D = 100 [ft]2 [min]−1 , d) D = 225 [ft]2 [min]−1 .
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
34. Figures (3)
a) b)
Time [min] Time [min]
100 1e+03 100 1e+03
800 800
600 600
80 400 80 400
200 200
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
c) d)
Time [min] Time [min]
100 1e+03 100 1e+03
800 800
600 600
80 400 80 400
200 200
60 60
40 40
20 20
0 0
0 20 40 60 80 100 0 20 40 60 80 100
Evolution in time of the fire line contour in the presence of a
firebreak, when τ = 100 [min], for the level-set method a) and
for the randomized level-set method with increasing turbulence:
b) D = 25 [ft]2 [min]−1 , c) D = 100 [ft]2 [min]−1 ,
d) D = 225 [ft]2 [min]−1 .
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
35. Conclusions (1)
A new formulation for modelling the wildland fire front
propagation is proposed.
It includes smallscale processes driven by the turbulence
generated by the Atmospheric Boundary Layer dynamics and
by the fire-induced flow.
It is based on the randomization of the level-set method for
tracking fire line contour by considering a distribution of the
contour according to the PDF of the turbulent displacement of
hot air particles.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
36. Conclusions (2)
This formulation is emerged to be suitable, more than the
ordinary level-set approach, to model the following two
dangerous situations:
i) the increasing of the rate of spread of the fire line as a
consequence of the pre-heating of zones ahead the fire front by
the hot air mass
ii) the overcoming of a breakfire by the fire because of the
diffusion of the hot air behind it, which is for the level-set
method a failed task because in the firebreak zone the rate of
spread is null since the absence of fuel.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
37. References
W.H. Frandsen, Fire spread through porous fuels from the
conservation of energy. Combust. Flame 16, 9–16 (1971).
R.C. Rothermel, A mathematical model for predicting fire
spread in wildland fires. USDA Forest Service, Research Paper
INT–115, (1972).
J.A. Sethian & P. Smereka, Level Set Methods for fluid
interfaces. Ann. Rev. Fluid Mech. 35, 341–372 (2003).
G. Pagnini & E. Bonomi, Lagrangian formulation of turbulent
premixed combustion. Phys. Rev. Lett. 107, 044503 (2011).
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012
38. Acknowledgements
Regione Autonoma della Sardegna
PO Sardegna FSE 2007-2013 sulla L.R. 7/2007
“Promozione della ricerca scientifica e
dell’innovazione tecnologica in Sardegna”.
Collana di Seminari per la Valorizzazione dei Risultati della Ricerca al CRS4, 22 marzo 2012