Beruflich Dokumente
Kultur Dokumente
Modeling and
Parameter
Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Mathematical Modeling and Parameter Bargo
Department of Mathematics
University of the Philippines Diliman
Ma. Cristina R.
I Prepare the setup for the heat conduction experiment, Bargo
data gathering
I Create a model for heat conduction on a metal rod,
under the following assumptions:
I constant heat flux at one end of the rod (due to the
heat source)
I heat loss along the sides of the rod
I heat loss at the other end of the rod
I Find the parameters (thermal conductivity, heat transfer
coefficient, specific heat capacity) of the metal rod by
minimizing the difference between the actual
temperature and computed temperature values
Mathematical
Objectives Modeling and
Parameter
Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
I Prepare the setup for the heat conduction experiment, Bargo
data gathering
I Create a model for heat conduction on a metal rod,
under the following assumptions:
I constant heat flux at one end of the rod (due to the
heat source)
I heat loss along the sides of the rod
I heat loss at the other end of the rod
I Find the parameters (thermal conductivity, heat transfer
coefficient, specific heat capacity) of the metal rod by
minimizing the difference between the actual
temperature and computed temperature values
Mathematical
Objectives Modeling and
Parameter
Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
I Prepare the setup for the heat conduction experiment, Bargo
data gathering
I Create a model for heat conduction on a metal rod,
under the following assumptions:
I constant heat flux at one end of the rod (due to the
heat source)
I heat loss along the sides of the rod
I heat loss at the other end of the rod
I Find the parameters (thermal conductivity, heat transfer
coefficient, specific heat capacity) of the metal rod by
minimizing the difference between the actual
temperature and computed temperature values
Mathematical
Experimental Setup Modeling and
Parameter
Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
Ma. Cristina R.
Bargo
Ma. Cristina R.
Bargo
Ma. Cristina R.
Bargo
Ma. Cristina R.
Bargo
Ma. Cristina R.
Bargo
I Selection
I compute the fitness of each individual
I form the “roulette”
I create the “mating pool”
I Recombination
I randomly select two individuals from the mating pool
I “mating” (to produce population of “offsprings”)
I probability of mating is determined by a fixed probability
Mathematical
Genetic Algorithm Modeling and
Parameter
Processes Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Selection
I compute the fitness of each individual
I form the “roulette”
I create the “mating pool”
I Recombination
I randomly select two individuals from the mating pool
I “mating” (to produce population of “offsprings”)
I probability of mating is determined by a fixed probability
Mathematical
Genetic Algorithm Modeling and
Parameter
Processes Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Mutation
I introduce small changes in the parameters of an
individual (“mutate”)
I probability of mutation is determined by a fixed
probability
I the resulting (mutated) pool will make up the initial
population for the next generation
I Optional process: Elitism
I the fitness of the next generation doesn’t necessarily
improve (because of recombination and mutation)
I if the best “parent” has disappeared, introduce it again
by removing 1 arbitrary offspring
Mathematical
Genetic Algorithm Modeling and
Parameter
Processes Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Mutation
I introduce small changes in the parameters of an
individual (“mutate”)
I probability of mutation is determined by a fixed
probability
I the resulting (mutated) pool will make up the initial
population for the next generation
I Optional process: Elitism
I the fitness of the next generation doesn’t necessarily
improve (because of recombination and mutation)
I if the best “parent” has disappeared, introduce it again
by removing 1 arbitrary offspring
Mathematical
Steady-State Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I temperature u of the rod does not change over time:
∂u
∂t = 0, hence u is a function of the position x
(measured from the end with the heat source)
I given parameters:
I r (cross-sectional radius of the rod)
I ` (length of the rod)
I ua (ambient temperature)
I parameters to be estimated:
I k (thermal conductivity of the rod)
I h (heat transfer coefficient)
I Q (heat flux provided by the soldering iron)
Mathematical
Steady-State Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I temperature u of the rod does not change over time:
∂u
∂t = 0, hence u is a function of the position x
(measured from the end with the heat source)
I given parameters:
I r (cross-sectional radius of the rod)
I ` (length of the rod)
I ua (ambient temperature)
I parameters to be estimated:
I k (thermal conductivity of the rod)
I h (heat transfer coefficient)
I Q (heat flux provided by the soldering iron)
Mathematical
Steady-State Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I temperature u of the rod does not change over time:
∂u
∂t = 0, hence u is a function of the position x
(measured from the end with the heat source)
I given parameters:
I r (cross-sectional radius of the rod)
I ` (length of the rod)
I ua (ambient temperature)
I parameters to be estimated:
I k (thermal conductivity of the rod)
I h (heat transfer coefficient)
I Q (heat flux provided by the soldering iron)
Mathematical
Steady-State Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I the model is given by
2h
00
ku (x) − r (u(x) − ua ) = 0, x ∈ (0, `)
u 0 (0) = − Qk , x =0 (1)
0 h
u (`) = − k (u(`) − ua ) , x =`
I general solution
q
to (1): q
2h 2h
− x x
u(x) = c1 e rk+ c2 e rk + ua , where c1 and c2 are
constants (determined by the boundary conditions of the
problem)
Mathematical
Steady-State Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I the model is given by
2h
00
ku (x) − r (u(x) − ua ) = 0, x ∈ (0, `)
u 0 (0) = − Qk , x =0 (1)
0 h
u (`) = − k (u(`) − ua ) , x =`
I general solution
q
to (1): q
2h 2h
− x x
u(x) = c1 e rk+ c2 e rk + ua , where c1 and c2 are
constants (determined by the boundary conditions of the
problem)
Mathematical
Steady-State Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
q
Q 2h h
k −rk k Q
c1 = q q q q + qk
2h 2L 2h 2h h 2h
rk e rk
rk + k − 2h h
rk + k rk
and
q
Q 2h h
k − rk k
c2 = q q q q
2h 2L 2h 2h h 2h h
rk e rk
rk + k − rk + k
Mathematical
Steady-State Case Modeling and
Parameter
Parameter Estimation Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
Q T
I vector of unknown parameters: q = k , kh
I steady-state data points: û (x1 ) , û (x2 ) , . . . , û (xN )
I u (xi ; q): solution of (1), using the parameter q and
evaluated at xi
I find:
N
1 X
min J(q) = min |u (xi ; q) − û (xi )|2
q∈R2 q∈R 2 N
i=1
Mathematical
Steady-State Case Modeling and
Parameter
Parameter Estimation Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
Q T
I vector of unknown parameters: q = k , kh
I steady-state data points: û (x1 ) , û (x2 ) , . . . , û (xN )
I u (xi ; q): solution of (1), using the parameter q and
evaluated at xi
I find:
N
1 X
min J(q) = min |u (xi ; q) − û (xi )|2
q∈R2 q∈R 2 N
i=1
Mathematical
Time-Dependent Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
I temperature u is a function of the position x and time t Bargo
∂u (x, t) ∂ 2 u (x, t) 2h
ρc p = k − (u (x, t) − ua )
∂t ∂x 2 r
∂u (0, t) Q
=−
∂x k
∂u (`, t) = − h (u (`, t) − u )
a
∂x k
(2)
Mathematical
Time-Dependent Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
I temperature u is a function of the position x and time t Bargo
∂u (x, t) ∂ 2 u (x, t) 2h
ρc p = k − (u (x, t) − ua )
∂t ∂x 2 r
∂u (0, t) Q
=−
∂x k
∂u (`, t) = − h (u (`, t) − u )
a
∂x k
(2)
Mathematical
Time-Dependent Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
I temperature u is a function of the position x and time t Bargo
∂u (x, t) ∂ 2 u (x, t) 2h
ρc p = k − (u (x, t) − ua )
∂t ∂x 2 r
∂u (0, t) Q
=−
∂x k
∂u (`, t) = − h (u (`, t) − u )
a
∂x k
(2)
Mathematical
Time-Dependent Case Modeling and
Parameter
The Model Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
I temperature u is a function of the position x and time t Bargo
∂u (x, t) ∂ 2 u (x, t) 2h
ρc p = k − (u (x, t) − ua )
∂t ∂x 2 r
∂u (0, t) Q
=−
∂x k
∂u (`, t) = − h (u (`, t) − u )
a
∂x k
(2)
Mathematical
Time-Dependent Case Modeling and
Parameter
Solution to the Model Estimation on a
Heat Conducting
Rod Experiment
Z ` Z `
∂u (x, t) k
φ (x) dx = − u 0 (x, t) φ0 (x) dx
0 ∂t ρcp 0
Z `
2h h
− u (x, t) φ (x) dx − u (`, t) φ (`)
r ρcp 0 ρcp
2hua `
Z
hua Q
+ φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Solution to the Model Estimation on a
Heat Conducting
Rod Experiment
Z ` Z `
∂u (x, t) k
φ (x) dx = − u 0 (x, t) φ0 (x) dx
0 ∂t ρcp 0
Z `
2h h
− u (x, t) φ (x) dx − u (`, t) φ (`)
r ρcp 0 ρcp
2hua `
Z
hua Q
+ φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Solution to the Model Estimation on a
Heat Conducting
Rod Experiment
Z ` Z `
∂u (x, t) k
φ (x) dx = − u 0 (x, t) φ0 (x) dx
0 ∂t ρcp 0
Z `
2h h
− u (x, t) φ (x) dx − u (`, t) φ (`)
r ρcp 0 ρcp
2hua `
Z
hua Q
+ φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Weak Formulation Estimation on a
Heat Conducting
I Let V = H 1 (0, `) and H = L2 (0, `) with the following Rod Experiment
2hua `
Z
hua Q
F (φ) = φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Weak Formulation Estimation on a
Heat Conducting
I Let V = H 1 (0, `) and H = L2 (0, `) with the following Rod Experiment
2hua `
Z
hua Q
F (φ) = φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Weak Formulation Estimation on a
Heat Conducting
I Let V = H 1 (0, `) and H = L2 (0, `) with the following Rod Experiment
2hua `
Z
hua Q
F (φ) = φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Weak Formulation Estimation on a
Heat Conducting
I Let V = H 1 (0, `) and H = L2 (0, `) with the following Rod Experiment
2hua `
Z
hua Q
F (φ) = φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Weak Formulation Estimation on a
Heat Conducting
I Let V = H 1 (0, `) and H = L2 (0, `) with the following Rod Experiment
2hua `
Z
hua Q
F (φ) = φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Weak Formulation Estimation on a
Heat Conducting
I Let V = H 1 (0, `) and H = L2 (0, `) with the following Rod Experiment
2hua `
Z
hua Q
F (φ) = φ (x) dx + φ (`) + φ (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Weak Formulation Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Show:
I solution to (3) exists
I equivalence of the solution to (2) and (3)
Mathematical
Time-Dependent Case Modeling and
Parameter
Weak Formulation Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Show:
I solution to (3) exists
I equivalence of the solution to (2) and (3)
Mathematical
Time-Dependent Case Modeling and
Parameter
Galerkin Method Estimation on a
Heat Conducting
Rod Experiment
I Project the solution of (3) in a finite dimensional space
V n = span {φ1 , . . . , φn } Ma. Cristina R.
Bargo
I Finite-dimensional problem: Find u n ∈ C (0, ∞; V n )
such that
n
∂u , φ = −σ (u n , φi ) + F (φi ) , i = 1, . . . , n
i
∂t H
n
u (0) = u0
(4)
X n
I Write u n (x, t) = αi (t) φi (x) and substitute to (4):
i=1
Ma. Cristina R.
T Bargo
α (t) = [α1 (t) , α2 (t) , . . . , αn (t)]
Z `
k
[A]ij = − φ0 (x) φ0i (x) dx
ρcp 0 j
Z `
2h h
− φj (x) φi (x) dx − φj (`) φi (`)
r ρcp 0 ρcp
Z `
[M]ij = φj (x) φi (x) dx
0
`
2hua
Z
hua Q
[F (t)]i = φi (x) dx + φi (`) + φi (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Galerkin Method Estimation on a
Heat Conducting
The variables in (5) are defined as follows: Rod Experiment
Ma. Cristina R.
T Bargo
α (t) = [α1 (t) , α2 (t) , . . . , αn (t)]
Z `
k
[A]ij = − φ0 (x) φ0i (x) dx
ρcp 0 j
Z `
2h h
− φj (x) φi (x) dx − φj (`) φi (`)
r ρcp 0 ρcp
Z `
[M]ij = φj (x) φi (x) dx
0
`
2hua
Z
hua Q
[F (t)]i = φi (x) dx + φi (`) + φi (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Galerkin Method Estimation on a
Heat Conducting
The variables in (5) are defined as follows: Rod Experiment
Ma. Cristina R.
T Bargo
α (t) = [α1 (t) , α2 (t) , . . . , αn (t)]
Z `
k
[A]ij = − φ0 (x) φ0i (x) dx
ρcp 0 j
Z `
2h h
− φj (x) φi (x) dx − φj (`) φi (`)
r ρcp 0 ρcp
Z `
[M]ij = φj (x) φi (x) dx
0
`
2hua
Z
hua Q
[F (t)]i = φi (x) dx + φi (`) + φi (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Galerkin Method Estimation on a
Heat Conducting
The variables in (5) are defined as follows: Rod Experiment
Ma. Cristina R.
T Bargo
α (t) = [α1 (t) , α2 (t) , . . . , αn (t)]
Z `
k
[A]ij = − φ0 (x) φ0i (x) dx
ρcp 0 j
Z `
2h h
− φj (x) φi (x) dx − φj (`) φi (`)
r ρcp 0 ρcp
Z `
[M]ij = φj (x) φi (x) dx
0
`
2hua
Z
hua Q
[F (t)]i = φi (x) dx + φi (`) + φi (0)
r ρcp 0 ρcp ρcp
Mathematical
Time-Dependent Case Modeling and
Parameter
Parameter Estimation Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
Q cp T
I vector of unknown parameters: q = k , kh , k
I set of data points:
{û (xi , tj ) | i = 1, . . . N, j = 1, . . . , Nt}
I u n (xi , tj ; q) is the numerical solution of (2) using the
parameter q and evaluated at xi at time tj
I find:
Nt N
1 XX n
min J(q) = min |u (xi , tj ; q) − û (xi , tj )|2
q∈R3 q∈R3 N · Nt
j=1 i=1
Mathematical
Time-Dependent Case Modeling and
Parameter
Parameter Estimation Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
Q cp T
I vector of unknown parameters: q = k , kh , k
I set of data points:
{û (xi , tj ) | i = 1, . . . N, j = 1, . . . , Nt}
I u n (xi , tj ; q) is the numerical solution of (2) using the
parameter q and evaluated at xi at time tj
I find:
Nt N
1 XX n
min J(q) = min |u (xi , tj ; q) − û (xi , tj )|2
q∈R3 q∈R3 N · Nt
j=1 i=1
Mathematical
Results Modeling and
Parameter
Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Steady-state case
I Copper:
Q
I
k
= 69.859011
h
I
k
= 0.0354198
I Aluminum:
Q
I
k
= 41.409499
h
I
k
= 0.0402905
I Time-dependent case: to follow!
Mathematical
Results Modeling and
Parameter
Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Steady-state case
I Copper:
Q
I
k
= 69.859011
h
I
k
= 0.0354198
I Aluminum:
Q
I
k
= 41.409499
h
I
k
= 0.0402905
I Time-dependent case: to follow!
Mathematical
Results Modeling and
Parameter
Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Steady-state case
I Copper:
Q
I
k
= 69.859011
h
I
k
= 0.0354198
I Aluminum:
Q
I
k
= 41.409499
h
I
k
= 0.0402905
I Time-dependent case: to follow!
Mathematical
Results Modeling and
Parameter
Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
I Steady-state case
I Copper:
Q
I
k
= 69.859011
h
I
k
= 0.0354198
I Aluminum:
Q
I
k
= 41.409499
h
I
k
= 0.0402905
I Time-dependent case: to follow!
Mathematical
Results Modeling and
Parameter
Copper Rod: actual temperature and computed temperature Estimation on a
Heat Conducting
Rod Experiment
Ma. Cristina R.
Bargo
320
data
318 computed
316
314
temperature (K)
312
310
308
306
304
302
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0
position (m)
Ma. Cristina R.
Bargo
313
data
312
computed
311
310
309
temperature (K)
308
307
306
305
304
303
302
0.0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.0 1.1
position (m)
Ma. Cristina R.
Bargo
THE END