32 Scientific Technical Review, 2010,Vol.60,No.2,pp.

32-43

UDK: 66.045.1:512.2:517.9
COSATI: 20-13, 12-01

Comparative Analysis of Different Methods in Mathematical

Modelling of the Recuperative Heat Exchangers

Dalibor Stević 1)

The heat exchangers are frequently used as constructive elements in various plants and their dynamics is very
important. Heat exchangers are used in military aircraft of all sizes, flying different missions, in many different
aircraft systems Their operation is usually controlled by manipulating inlet fluid temperatures or mass flow rates. On
the basis of the accepted and critically clarified assumptions, a linearized mathematical model of the cross-flow heat
exchanger has been derived, taking into account the wall dynamics. The model is based on the fundamental law of
energy conservation, covers all heat accumulation storages in the process, and leads to the set of partial differential
equations (PDE), the solution of which is not possible in a closed form. In order to overcome the solution difficulties,
this paper analyses different methods for modeling the heat exchanger: the approach based on the Laplace transform,
the approximation of partial differential equations based on the finite difference method, the method of physical
discretization and the transport approach. Specifying the input temperatures and output variables, under constant
initial conditions, the step transient responses have been simulated and presented in a graphic form in order to
compare the results for the four characteristic methods considered in this paper and to analyze their practical
significance.

Key words: heat exchanger, mathematical model, comparative analysis, Laplace transform, finite difference method,
discretization method.

Introduction lose their veracity as well.

In this paper the four characteristic methods for
T HE heat exchangers are designed to achieve certain
requirements in the steady state which implies that the
transient response of the heat exchanger must be known to
modelling heat exchangers are considered:
1. Analytical approach using the Laplace transform.
define a correct control strategy. 2. Approximation of PDE based on finite differences.
The heat exchangers are frequently used in military and 3. The method of physical discretization.
civil aircraft as oil coolers and fuel heaters with heat 4. Transport approach.
exchangers placed inside fuel tanks. In large transport The comparative analysis of these methods and their
aircraft the cabin pressurization systems cool the hot engine practical significance is carried out on the model of a cross
bleed air to a temperature suitable for use within the cabin flow heat exchanger.
with a heat exchanger. The air conditioning system is also Finally, the step responses of the derived mathematical
based on a heat exchanger, known as an air-cycle machine, models are presented in a graphic form in order to compare
in order to provide air at the desired temperature to the these results with different methods.
cabin and flight deck.
The mathematical model of the heat exchanger must be
known in order to determine the transient response. Analytical approach based on the laplace
Many researchers have been working on this problem transform
within the last decade such as Čermak et al. (1968), In this paper, the recuperative cross-flow heat exchanger
Roetzel, Huan (1992), Romie (1984), Spiga, Spiga (1987), is observed, shown in Fig.1, as a process with distributed
Tan, Spinner (1991). parameters. Among many kinds of water-to-air heat
Solving of this problem in the mentioned papers is based exchangers, the cross-flow geometry is very common. The
on two approaches: geometry of cross-flow heat exchangers can be
1. numerical solving of PDE which pulls drawbacks such complicated, but in this paper we observe the case with the
as convergence problems, stability, stiffness, etc. simplest geometry that can be easily computed. This heat
2. Laplace transform which is complicated in this case and exchanger consists of a single tube with the fluid (water or
demands a numerical inversion to the original time do- other fluid) flow inside and the cross flow of hot air
main. outside.
These problems and the similar ones have motivated the The mathematical model of the cross flow heat
research to invent new approaches for modeling these exchanger, shown in Fig.1, is carried out on the basis of the
processes in order to obtain wider practical use and not to following assumptions:

1)
University of Belgrade, Faculty of Mechanical Engineering, Department of Control Engineering, Kraljice Marije 16, 11120 Belgrade, SERBIA
 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS 33

- Fluid in the tube is incompresible and viscous. where ρ f is the fluid (water) density, c f is the specific
- Temperature field fluid in the tube is one-dimensional.
- Temperature field of the tube wall is one-dimensional. heat of the fluid in the tube, ri is the inner radius, θ f is the
- Fluid enthalpy can be expressed by means of a corre- fluid temperature, α wf is the heat transfer coefficient
sponding temperature. between the wall and the fluid, and G f is the mass-flow
- Inlet air temperature is a function of time.
- Specific heat of fluid, wall and air has constant values. rate of the fluid.
- There is convective heat transfer between the cross flow air - Finally, in the wall of the tube:
and the tube wall, conduction along the tube wall, and con-
∂θ w ( t , ξ ) ∂ 2θ w ( t , ξ )
vection between the tube wall and the fluid in the tube wall. ρ w cwπ ( r02 −ri2 ) = λwπ ( r02 −ri2 ) +
- Newton’s law of heat convection determines exactly ∂t ∂ξ 2 (4)
enough of the amount of heat exchange in the steady state +2π r0α aw (θ a ( t ) − θ w ( t , ξ ) )−2π riα wf (θ w ( t , ξ )−θ f ( t , ξ ) )
and the transient state.
- Heat transfer coefficients on both sides of the tubewall where ρ w is the wall density, cw is the specific heat of the
have constant values. wall, λw is the wall termal conductivity.
- The fluid flow in the tube is one-phased.
The boundary and initial conditions are:

θ f ( t , 0 ) = θ inf ( t ) , θ f ( t , L ) = θ out
f (t ) , (5)

θ f ( 0, ξ ) = 0 , θ w ( 0, ξ ) = 0
Introducing relative deviations in this form:
θ f ( t , ξ ) − θ fN (ξ )
Δθ f ( t , ξ ) =
θ fN (ξ N )

θ w ( t , ξ ) − θ wN (ξ )
Δθ w ( t , ξ ) = (6)
θ wN (ξ N )

in θ ain ( t ) − θ aN
in
Δθ a ( t ) =
θ aN
in

where θ fN , θ wN , θ aN
in
are nominal values.
Figure 1. Symbolic-functional scheme and the diagram of the process of From eq. (2), (3), (4), (5) the mathematical model of the
the cross-flow heat exchanger cross-flow heat exchanger is obtained in the following
On the basis of these assumptions, the mathematical form:
model of the cross- flow heat exchanger after some simple
∂Δθ f ( t , ξ ) ∂Δθ f ( t , ξ )
mathematical transformations can be written in the = a1 Δθ w ( t , ξ )−a2 Δθ f ( t , ξ ) − a3 (7)
following form. ∂t ∂ξ
- On the outside of the tube:
∂Δθ w ( t , ξ ) ∂ 2 Δθ w ( t , ξ )
Ga ca (θ ain ( t ) − θ aout ( t ) ) = 2α awπ ro L (θ a ( t ) − θ w ( t , ξ ) ) (1) = b1 − b2 Δθ w ( t , ξ ) +
∂t ∂ξ 2 (8)
in
where L is the length of the tube, Ga is the mass-flow rate + b3 Δθ f ( t , ξ ) + b4 Δθ a (t )
of air, ca is the specific heat of air, θ ain and θ aout are
where:
incoming and outgoing air temperatures, α aw is the heat
2α wf θ wN
transfer coefficient beween the hot air and the tube wall, r0 a1 = ,
ρ f c f ri θ fN
is the outer radius of the tube, θ a is the air temperature
sorrounding the tube, and θ w is the tube wall temperature. 2α wf
For convenience, the air temperature can be assumed to be a2 = ,
ρ f c f ri
approximately:
Gf
θ ain ( t ) + θ aout ( t ) a3 = ,
θa (t ) = (2) ρ f ri2π
2
This can be substituted in eq. (1).
- In the water: λw
b1 = , (9)
ρ w cw
∂θ f ( t , ξ )
ρ f c f ri2π = α wf 2π ri (θ w ( t , ξ ) − θ f ( t , ξ ) ) −
∂t (3) 2Ga ca ( r0α aw + riα wf ) + 2α awα wf r0 riπ L
∂θ f ( t , ξ ) b2 = ,
−G f c f
∂ξ ρ w cw ( r02 − ri2 ) ( Ga ca + α waπ r0 L )
34 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS

2riα wf θ fN a3b1σ 3 + ( b1 s + a2 b1 ) σ 2 − ( a3 s + a3b2 ) σ −

b3 = , (18)
ρ w cw ( r0 − ri ) wN
2 2 θ
− s 2 − ( a2 + b2 ) s + a1b3 − a2b2 = 0

2r0α awGa ca θ aN
in In order to obtain the step response for the mathematical
b4 = . model of the cross-flow heat exchanger it is necessary to
ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 L ) θ wN take another inverse Laplace transform with respect to time,
Taking the Laplace transform of (6), (7) first with L−s 1→t . Considering the complexity of the right side of Eqs.
respect to t, and then with respect to the spatial coordinate (14) and (15) this can be done using the numerical methods
ξ , Lξ →σ , with initial conditions (5) we find: for the inverse Laplace transform, Abate, Choudhury
(1999), Abate, Valko (2004).
Applying the Gaver-Stehfest algorithm for the inverse
( s + a2 + a3σ ) Δθ f ( s, σ ) = a1 Δθ w ( s, σ ) + a3 Δθ f ( s, 0 ) (10) Laplace transform of (14),(15) we find:

( s − b1σ 2 + b2 ) Δθ w ( s, σ ) = Δθ f ( t , L ) ln ( 2 ) ⎛ k ln ( 2 ) ⎞
2M

in (11) Δθ f ( t , 0 )
= f (t , M ) =
t ∑ς F ⎜⎝k
t ⎟⎠
(19)
= b3 Δθ f ( s, σ ) − b1σ Δθ w ( s, 0 ) + b4 Δθ a ( s ) k =1

Sollution of the system (10), (11) for θ f ( s, σ ) has the Δθ f ( t , L ) ln ( 2 )

2M
k ln ( 2 ) ⎞
following form: Δθ
in
a (t )
= g (t , M ) =
t ∑ς G ⎛⎜⎝
k =1
k
t ⎟⎠
(20)

a3b1σ − M ( s )2
θ f ( s, σ ) = Δθ f ( s , 0 ) − where
a3b1σ 3 +P ( s ) σ 2 − Q ( s ) σ − R ( s )
k ∧M
j M +1 ⎛ M ⎞ ⎛ 2 j ⎞ ⎛ j
−
a1b4
a3b1σ + P ( s ) σ − Q ( s ) σ − R ( s )
3 2
in
Δθ a ( s ) + (12) ς k = ( −1)
M +k
∑
j = ( k +1) / 2]
[
M ! ⎜⎝ j ⎟⎠ ⎜⎝ j ⎟⎠ ⎜⎝ k −
⎞
j ⎟⎠
(21)

a1b1σ
+ Δθ w ( s, 0 )
a3b1σ + P ( s ) σ 2 − Q ( s ) σ − R ( s )
3 with [( k + 1) / 2] being the greatest integer lower than or

where equal to ( k + 1) / 2 , k ∧ M ≡ min {k , M } and the positive

integer M.
M ( s ) = a3 s + a3b2 , P ( s ) = b1 s + a2b1 ,

Differential-discrete mathematical model of the

Q ( s ) = a3 s + a3b2 , (13) cross-flow heat exchanger
The system of PDE is very complex to be solved
R ( s ) = s 2 + ( a2 + b2 ) s + a2 b2 − a1b3 analytically as it was considered in the analytical aproach
for modeling the cross-flow heat exchanger. A numerical
Taking the inverse Laplace transform of (12) with method based on the finite differences is developed to
respect to the spatial coordinate, Lσ−1→ξ , and switching approximate infinite-dimensional equations by finite-
dimensional ones. The spatial coordinate of the cross-flow
ξ = L , appropriate transfer functions are obtained in the heat exchanger is discretized by means of the finite
following form: difference method. In this manner the system of PDE is
transformed to the system of ODE. Discretization of the
Δθ f ( s, L )
W1 ( s, L ) = = F (σ 1 , s , L ) + spatial variable is carried out by means of substituting
Δθ f ( s , 0 ) , (14) partial derivation with respect to the spatial coordinate in
+ F (σ 2 , s , L ) + F (σ 3 , s , L ) certain numbers of points. In the physical sense, the
observed heat exchanger is divided into equal p-cells with
the spatial coordinate discretization.
Δθ f ( s , L ) The mathematical model of the cross-flow heat
W2 ( s, L ) = = G (σ 1 , s , L ) +
in
Δθ a ( s ) (15) exchanger is derived in the previous section, described with
Eqs. (1), (3), and (4). The finite difference method has a
+ G (σ 2 , s , L ) + G (σ 3 , s , L ) different form for the first, k-th (k = 2,3,..., p-1) and the last
cell. This method has a lot of different forms. Regarding a
where
one-dimensional problem, the finite difference method has
a3b1σ n2 − U ( s ) the following form:
F (σ n , s , L ) = e−σ n L , (16)
3a3b1σ n2 + V ( s ) σ n − U ( s ) Δξ = ξ k +1 − ξ k = l , θ f , k ( t ) = θ f ( t , k ) , k=1,2,..., p

−b4 a1 ∂θ f ( t , ξ )
G (σ n , s , L ) = e −σ n L
3a3b1σ n2 + V ( s ) σ n − U ( s ) ≅ 1 (θ f , k +1 ( t ) − θ f , k −1 ( t ) ) , (22)
∂ξ k
2l
and
∂ 2θ f ( t , ξ )
U ( s ) = a3 s + a3b2 , V ( s ) = 2 ( b1s + a2 b1 ) (17) k ≅ 12 (θ f , k +1 ( t ) − 2θ f , k ( t ) + θ f , k −1 ( t ) ) .
∂ξ 2 l
σ n , n = 1, 2,3 are the roots of the equation
 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS 35

Substituting (22) into (3), (4) the differential-discrete θ f , p +1 ( t ) = 0 , θ f ,0 ( t ) = θ inf ( t ) .

model of the cross-flow heat exchanger is obtained in the
following form: Introduce relative deviations and define the state
- for the first cell: variables in this form:
dθ f ,1 ( t ) 2α wf 2α wf θ f , k ( t ) − θ fN , k
= θ (t ) − θ (t ) − Δθ f , k ( t ) = = xk ( t ) ,
dt ρ f c f ru w,1 ρ f c f ru f,1 θ fN , k
(23)
G f ⎛ θ f ,2 ( t ) − θ inf ( t ) ⎞
− ⎜ ⎟ θ w, k ( t ) − θ wN , k
ρ f ru2π ⎝ 2l ⎠ Δθ w, k ( t ) = = xk* ( t ) ,
θ wN , k
dθ w,1 ( t ) λ ⎛ θ w,2 ( t ) − 2θ w,1 ( t ) ⎞
= w ⎜
ρ w cw ⎝ ⎟+ in θ inf ( t ) − θ infN
dt l2 ⎠ Δθ f ( t ) = = u1 ( t ) , (30)
2rsα aw Ga ca θ infN
+ θ ain ( t ) +
ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 L )
2riα wf in θ ain ( t ) − θ aN
in

+ θ f ,1 ( t ) − (24) Δθ a ( t ) = = u2 ( t ) ,
ρ w cw ( r02 − ri2 ) θ aN
in

2Ga ca ( r0α aw + riα wf ) + 2α awα wf r0 riπ L

− θ w,1 ( t ) ⎡ xk ( t ) ⎤ ⎡u (t ) ⎤
ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 L ) xk ( t ) = ⎢ * ⎥ , u ( t ) = ⎢ 1 ⎥ .
⎣ xk ( t ) ⎦ ⎣u2 ( t ) ⎦

k-th cell: The mathematical model of the cross-flow heat

exchanger is obtained in the state space:
dθ f , k ( t ) 2α wf 2α wf - for the k-th cell:
= θ (t ) − θ (t ) −
dt ρ f c f ri w, k ρ f c f ri f, k
(25) x k ( t ) = Ak1 x k −1 ( t ) + Ak2 x k ( t ) + Ak3 x k +1 ( t ) + Bk u ( t ) (31)
G f ⎛ θ f , k +1 ( t ) − θ f , k −1 ( t ) ⎞
− ⎜ ⎟
ρ f ri2π ⎝ 2l ⎠ where
a
Ak1 = ⎛⎜ k 3
0 ⎞ 2 ⎛ −ak1 ak 4 ⎞ ,
⎟ , Ak = ⎜ b ⎟
⎝ 0 bk 3 ⎠ ⎝ k 4 −bk1 ⎠
dθ w, k ( t ) λ ⎛ θ w, k +1 ( t ) − 2θ w, k ( t ) + θ w, k −1 ( t ) ⎞
= w ⎜ ⎟+
dt ρ w cw ⎝ l2 ⎠ −a 0 ⎞ ⎛0 0 ⎞
2r0α awGa ca Ak3 = ⎛⎜ k 2 ⎟ , Bk = ⎜ 0 b ⎟ ,
+ θa (t ) +
in
⎝ 0 bk2 ⎠ ⎝ k5 ⎠
ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 L )
2riα wf (27) K = 2,3,…, p-1 (32)
+ θ f ,k ( t ) −
ρ w cw ( r02 − ri2 ) and
2Ga ca ( r0α aw + riα wf ) + 2α awα wf r0 riπ L
− θ w, k ( t ) 2α wf Gf θ fN ,k +1
ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 L ) ak1 =
ρ f c f ri
, ak 2 = ,
ρ f ri π 2l θ fN , k
2

- last cell: Gf θ fN , k −1 2α wf θ wN ,k
ak 3 = , ak 4 = ,
dθ f , p ( t ) 2α wf 2α wf θ
ρ f ri π 2l fN , k
2 ρ f c f ri θ fN , k
= θ (t ) − θ (t ) +
dt ρ f c f ri w, p ρ f c f ri f, p
G f θ f , p −1 ( t ) ⎛ 2λ 2Ga ca ( r0α aw + riα wf ) + 2α awα wf r0 riπ L ⎞
+ bk1 = ⎜ w
+ ⎟
ρ f ri2π 2l ⎜ ρ w cw l
⎝
2
ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 L ) ⎟⎠

dθ w, p ( t ) λ 2θ w, p ( t ) − θ w, p −1 ( t ) λw θ wN , k +1 θ
dt
=− w
ρ w cw l2
+ bk 2 = , bk 3 = λw 2 wN , k −1 ,
ρ w cwl θ wN , k2
ρ w cwl θ wN , k
2r0α aw Ga ca
+ θ ain ( t ) +
ρ w cw ( r0 − ri2 ) ( Ga ca + α awπ r0 L )
2
2riα wf θ fN ,k
2riα wf bk 4 = , k=1,2,..., p (33)
+
ρ w cw ( r02 − ri2 )
θ f , p (t ) − ρ w cw r0
2
( − ri ) θ wN , k
2

2Ga ca ( r0α aw + riα wf ) + 2α awα wf r0 riπ L

− θ w, p ( t ) 2r0α awGa ca θ aN
in
ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 L ) bk 5 = .
ρ w cw r0
2
( − ri ) ( Ga ca + α awπ r0 L ) θ wN , k
2

where The mathematical model of the first and the last cell in the
state space will be slightly modified, because these two cells,
θ f ( t , L ) = θ f , p ( t ) , θ w,0 ( t ) = 0 , (29) through control boundary, have contact with enviroment.
36 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS

The mathematical model of the first cell has the the first an the last cell which through control boundary have
following form: contact with environment. The flowing index indicates the
position of the cell in the heat exchanger.
x1 ( t ) = A12 x1 ( t ) + A13 x 2 ( t ) + B1u ( t ) (34) In the physical sense, the observed cross-flow heat
exchanger is divided into equal p cells with the spatial
where coordinate discretization. Assuming that the number of
−a a −a 0 ⎞ cells is big enough, each of these cells can be considered as
A12 = ⎛⎜ 11 14 ⎞⎟ , A13 = ⎛⎜ 12 ⎟, (35) a process with certain parameters. The fundamental law of
⎝ b14 −b11 ⎠ ⎝ 0 b12 ⎠ energy conservation is derived for every cell, so the
mathematical model of the cross-flow heat exchanger can
a 0 ⎞
B1 = ⎛⎜ 15 ⎟, be reperesented as a system of ODE.
⎝ 0 b15 ⎠ Considering the k-th cell, the ballance equation can be
written in the following form:
while this model in the state space for the last cell can be
written as follows: dθ f , k ( t ) Gf
= (θ f ,k −1 ( t ) − θ f ,k ( t ) ) +
x p (t ) = A1p x p −1 (t ) + Ap2 x p (t ) + Bpu (t ) (36) dt ρ f ri2π l
(39)
2α wf
where + (θ ( t ) − θ f ,k ( t ) )
ρ f c f ri w, k
⎛a 0 ⎞ ⎛ −a p1 a p 4 ⎞
A1p = ⎜ p 3 2
⎟ , Ap = ⎜ b ⎟, (37)
⎝ 0 b p3 ⎠ ⎝ p 4 −bp1 ⎠ dθ w, k ( t ) λ
dt
= w
ρ w cwl
(θ w,k −1 ( t ) − θ w,k ( t ) ) +
⎛0 0 ⎞ 2r0α aw
Bp = ⎜ ⎟. + (θ a ( t ) − θ w,k ( t ) ) − (40)
⎝ 0 bp5 ⎠ ρ w cw ( r02 − ri2 )
The state equation and the output of the cross-flow heat 2riα wf
− (θ w,k ( t ) − θ f ,k ( t ) )
exchanger are given with: ρ w cw ( r02 − ru2 )
x ( t ) = Ax ( t ) + Bu ( t ) (38)

xi ( t ) = cT x ( t ) , (θain ( t ) − θaout ( t ) ) = 2αGawa cπar0l (θa ( t ) − θ w,k ( t ) ) (41)

⎛ A12 A13 0 0 … 0 0 0 0 ⎞ where

⎜ 1 2 3 ⎟
⎜ A2 A21 A22 03 … 0 0 0 0 ⎟ θ ain ( t ) + θ aout ( t )
⎜ 0 A3 A3 A3
A=⎜
… 0 0 0 0 ⎟ θa (t ) = (42)
⎟ 2
⎜ ⎟
⎜ 0 0 0 0 … 0 A1p −1 Ap2 −1 A3p −1 ⎟ Using Eqs. (39) - (42) the mathematical model of the
⎜ 0 0 0 0 … 0 0 A1p 2 ⎟
Ap ⎠ cross flow heat exchanger can be obtained in the following
⎝
form:
⎡0⎤ - k-th cell:
⎢0⎥
⎡ B1 ⎤ ⎢…⎥ dθ f , k ( t ) ⎛ 2α wf Gf ⎞
⎢ B2 ⎥ ⎢ ⎥
⎡ x1 ( t ) ⎤
= −⎜ + ⎟θ f ,k ( t ) +
⎢0⎥ dt ⎝ ρ f c f ri ρ f ri π l ⎠
2
⎢ B ⎥ ⎢ x (t ) ⎥ (43)
⎢0⎥ 2α wf
B=⎢ 3 ⎥, c = ⎢ ⎥ , x (t ) = ⎢ 2 ⎥ , +
Gf
θ f , k −1 ( t ) + θ (t )
⎢ ⎥ … ⎢ ⎥ ρ f c f ri w, k
⎢ ⎥ ρ f ri π l
2
⎢ B p −1 ⎥
⎢ ⎥ ⎢⎣ x p ( t ) ⎥⎦
⎢ Bp ⎥
⎣ ⎦ ⎢…⎥
⎢1⎥ dθ w , k ( t ) 4α aw2 2
r0 π l
= θ w, k ( t ) −
⎢⎣ 0 ⎥⎦ dt ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 l )
λ 2r0α aw + 2riα wf
xi ( t ) = Δθ f , p ( t ) . − w θ w, k ( t ) − θ w, k ( t ) +
ρ w cw l ρ w cw ( r02 − ri2 )
(44)
2riα wf λw
+ θ f ,k ( t ) + θ (t ) +
The method of physical discretization ρ w cw ( r02 − ri2 ) ρ w cwl w, k −1
To avoid partial differential equations, the method of 2r0α aw ( Ga ca − α awπ r0 l )
+ θ ain ( t )
physical discretization suggested by Grujić (1978) is used. ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 l )
By this method the observed heat exchanger is divided into
the same p cells with the spatial coordinate discretization. - for the first cell:
These cells are assumed to have homogenous fields of
specific physical values. The qualitative description of the
dθ f ,1 ( t ) ⎛ 2α wf Gf ⎞
process with distributed parameters can be obtained taking = −⎜ + ⎟ θ f ,1 ( t ) +
dt ⎝ ρ f c f ri ρ f ri π l ⎠
2
into consideration a big enough number of cells. (45)
The advantage of such an approach is avoiding complex Gf 2α wf
PDE to be set and solved. Ballance differential equations are + θ inf ( t ) + θ (t )
ρ f ri π l
2 ρ f c f ri w,1
set for an arbitrary cell, taking especially into consideration
 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS 37

dθ w,1 ( t ) 4α aw
2 2
r0 π l 2α wf θ wN , k
= θ w,1 ( t ) − ak 3 = ,
dt ρ w cw ( r0 − ri ) ( Ga ca + α awπ r0 l )
2 2 ρ f c f ri θ fN , k
λ 2r0α aw + 2riα wf
− w θ w,1 ( t ) − θ w,1 ( t ) +
ρ w cwl ρ w cw ( r02 − ri2 ) 4α aw r0 π l
2 2
(46) bk1 = −
+
2riα wf
θ f ,1 ( t ) +
ρ (
w cw r0
2
− ri ) ( Ga ca + α awπ r0 l )
2
, (52)
ρ w cw r0( − ri )
2 2
−
λw
−
2r0α aw + 2riα wf
2r0α aw ( Ga ca − α awπ r0 l ) ρ w cw l ρ w cw ( r02 − ri2 )
+ θ (t )
in
ρ w cw r0(
2
− ri 2
) ( Ga ca + α awπ r0l ) a

λw θ wN , k −1 2riα wf θ fN ,k
- for the last cell: bk 2 = , bk 3 =
ρ w cw l θ wN ,k ρ w cw ( r0 − ri ) θ wN ,k
2 2

dθ f , p ( t ) ⎛ 2α wf Gf ⎞
= −⎜
ρ
+ ⎟θ f , p ( t )
dt ⎝ f f u ρ f ri π l ⎠
c r 2
(46) 2r0α aw ( Ga ca − α awπ r0 l ) θ gN
u

Gf 2α wf bk 4 = .
+ θ f , p −1 ( t ) + θ (t ) ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0l ) θ zN , k
ρ f ri π l
2 ρ f c f ri w, p
The mathematical model for the k-th cell of the cross-
dθ w , p ( t ) 4α πl
2 2 flow heat exchanger is obtained in the state space:
aw ro
= θ w, p ( t ) −
dt ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0 l ) dx k ( t )
= Ak1 x k −1 ( t ) + Ak2 x k ( t ) + Bk u ( t ) (53)
λ 2r0α aw + 2riα wf dt
− w θ w, p ( t ) − θ w, p ( t ) +
ρ w cw l ρ w cw ( r02 − ri2 ) (48) where
2riα wf λw
+ θ f , p (t ) + θ (t ) + a 0 ⎞ ⎛ −ak1 ak 3 ⎞ ,
ρ w cw ( r02 − ri2 ) ρ w cwl w, p −1 Ak1 = ⎛⎜ k 2 2
⎟ , Ak = ⎜ b ⎟
⎝ 0 bk ⎠
2 ⎝ k 3 bk1 ⎠
2r0α aw ( Ga ca − α awπ r0 l )
+ θ ain ( t )
ρ w cw ( r02 − ri2 ) ( Ga ca + α awπ r0l ) 0 0 ⎞
Bk = ⎛⎜ ⎟, (54)
Introducing relative deviations and defining the state ⎝ 0 bk 4 ⎠
variables in this form:
x k ( t ) = ⎡⎣ xk ( t ) xk* ( t ) ⎤⎦ , u ( t ) = [u1 ( t ) u2 ( t )]
T T
θ f , k ( t ) − θ fN , k
Δθ f , k ( t ) = = xk ( t ) ,
θ fN , k For the first cell of the cross-flow heat exchanger the
state space equation has the following form:
θ w, k ( t ) − θ wN , k
Δθ w, k ( t ) = = xk* ( t ) , (49) x1 ( t ) = A12 x1 ( t ) + B1u ( t ) (55)
θ wN , k
where
in θ (t ) − θ
in in
−a a a 0 ⎞
Δθ f ( t ) = = u1 ( t ) , A12 = ⎛⎜ 11 13 ⎞⎟ , B1 = ⎛⎜ 12
f fN
⎟ (56)
θ infN ⎝ b13 b11 ⎠ ⎝ 0 b14 ⎠
For the last cell the state space equation can be written as
in θ in ( t ) − θ in follows:
Δθ ( t ) = a in aN = u2 ( t ) .
a
θ aN
x p ( t ) = A1p x p −1 ( t ) + Ap2 x p ( t ) + B p u ( t ) (57)
The mathematical model for the k-th cell is obtained in
the following form:
⎛a 0 ⎞ ⎛ −a a ⎞
A1p = ⎜ p 2 ⎟ , Ap2 = ⎜ p1 p 3 ⎟ , (58)
dxk ( t ) ⎝ 0 bp 2 ⎠ ⎝ bp 3 bp1 ⎠
= − ak1 xk ( t ) + ak 2 xk −1 ( t ) + ak 3 xk* ( t ) (50)
dt
⎛0 0 ⎞
Bp = ⎜ ⎟.
dxk* ( t ) ⎝ 0 bp 4 ⎠
= bk1 xk* ( t ) + bk 2 xk* −1 ( t ) + (51)
dt
+bk 3 xk ( t ) + bk 4 u2 ( t ) The state equation and the output of the cross-flow heat
exchanger are given with:
where k = 1, 2,… , p and x ( t ) = Ax ( t ) + Bu ( t ) , (59)
2α wf Gf G f θ fN , k −1
ak1 = + , ak 2 = , xi = cT x ( t )
ρ f c f ri ρ f ri π l
2
ρ f ru2π l θ fN ,k
38 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS

⎛ A12 0 0 0 … 0 0 ⎞ Qwf ( t ) = α wf Awf (θ w ( t ) − θ f ( t ) ) .

⎜ A1 A2 0 0 … 0 0 ⎟
⎜ 2 2 ⎟ Assuming that fluid temperature in the cell can be
A = ⎜ 0 A31 A32 0 … 0 0 ⎟ expressed in the folowing form:
⎜… … … … … 0 0 ⎟
⎜ 2⎟ θ inf ( t ) + θ out
f (t )
1
⎝ 0 0 0 0 … Ap Ap ⎠ θ fsr ( t ) = . (62)
2
⎡0⎤ The heat content in the fluid for the observed cell can be
⎢0⎥ expressed in the following form:
⎢…⎥
⎡ B1 ⎤ ⎢ ⎥
⎢ B2 ⎥ ⎢0⎥ ⎡ x1 ( t ) ⎤ θ inf ( t ) + θ out
f (t )
⎢ x (t ) ⎥ Q ( t ) = m f c f θ fsr ( t ) = m f c f , (63)
⎢0⎥ 2
B = ⎢ B3 ⎥ , c = ⎢ ⎥ , x (t ) = ⎢ 2 ⎥ ,
⎢ ⎥ … ⎢ ⎥
⎢ ⎥ ⎢ ⎥ The total amount of heat accumulated in the observed
⎢ ⎥ ⎢
⎣ x ( t ) ⎦⎥ cell for the time interval ( ( t + Δt ) − t ) = Δt can be written
⎣⎢ B p ⎦⎥
p
⎢…⎥
⎢1⎥ as follows:
⎢⎣ 0 ⎥⎦
Δ Q ( Δt ) = Q ( t + Δ t ) − Q ( t ) =
(64)
xi ( t ) = θ f , p ( t ) = x p ( t ) . = Q in ( Δt ) − Q out ( Δt ) + Q wf ( Δt )

where

The transport approach θ inf ( t ) + θ inf ( t + Δt )

Qin ( Δt ) = c f G f Δt ,
The transport approach represents the special metohod 2
for mathematical modelling of heat exchangers. This
f ( t ) + θ f ( t + Δt )
θ out out
approach is based on dividing the observed heat exchanger
Q out ( Δt ) = c f G f Δt , (65)
into the same p cells. 2
The fundamental law of energy conservation is derived
by means of the finite differences for every cell with α wf Awf Δt ⎛ θ in ( t ) + θ out
f (t ) ⎞
respect to the finite time interval Δt . In this manner the Q wf ( Δt ) = ⎜ θw (t ) − f ⎟+
2 ⎝ 2 ⎠
system of partial differential equations is transformed to the
system of algebric equations which are easy to solve by α wf Awf Δt ⎛ θ inf ( t + Δt ) + θ out
f ( t + Δt ) ⎞
+
2 ⎜ θ w ( t + Δt ) − 2 ⎟
numeric, itterative methods, Ilić (2002). ⎝ ⎠
Let us observe the k-th cell, k=1,2,..., p, of the cross-flow
heat exchanger. Considering the heat ballance within the The ballance equation for the observed cell is obtained in
observed cell, the heat input and output by fluid flow the following form:
through the boundary of the cell can be identified as well as - for the fluid in the tube:
the heat exchange between the fluid in the cell and the wall.
m f c f θ fsr , k ( t + Δt ) − m f c f θ fsr , k ( t ) =
Consider the changes of the specific physical values in a
c f G f Δt in
particular interval defined with t and t + Δt . The amount of
2
(θ f ,k ( t ) − θ inf ,k ( t + Δt ) ) −
heat exchange into the k-th cell for the time interval Δt can c f G f Δt out (66)
be written in the following form: −
2
(θ f ,k ( t ) − θ out
f , k ( t + Δt ) ) +

Q ( t ) + Q ( t + Δt ) α wf Awf Δt ⎛ θ in ( t ) + θ outf ,k ( t ) ⎞
Q ( Δt ) = Δt , + ⎜ θ w, k ( t ) − f , k ⎟+
2 ⎝ 2 ⎠
2
α wf Awf Δt ⎛ θ inf ,k ( t + Δt ) + θ out
f , k ( t + Δt ) ⎞
+ ⎜ θ w, k ( t + Δt ) − ⎟
Qin ( t ) + Qin ( t + Δt ) 2 ⎝ 2 ⎠
Q in ( Δt ) = Δt , (60)
2
- for the wall:
Qout ( t ) + Qout ( t + Δt ) mw cw (θ w, k ( t + Δt ) − θ w, k ( t ) ) =
α aw Aaw Δt
(θ a ( t ) − θ w,k ( t ) ) +
Q out ( Δt ) = Δt , 2
2 α aw Aaw Δt
+
2
(θ a ( t + Δt ) − θ w,k ( t + Δt ) ) −
(67)
Q ( t ) + Qwf ( t + Δt ) α wf Awf Δt ⎛ θ inf , k ( t ) + θ outf ,k ( t ) ⎞
Q wf ( Δt ) = wf Δt . −
2 ⎜ θ w, k ( t ) − 2 ⎟−
2 ⎝ ⎠
α wf Awf Δt ⎛ θ inf , k ( t + Δt ) + θ out
f , k ( t + Δt ) ⎞
where Q in ( Δt ) is the heat change at the cell inlet for the −
2 ⎜ θ w , k ( t + Δt ) − 2 ⎟
⎝ ⎠
time interval Δt , Q out ( Δt ) is the heat change at the cell
- for the air:
outlet and Q wf ( Δt ) the heat exchange between the wall
Ga ca (θ ain ( t ) − θ aout ( t ) ) = α aw Aaw (θ a ( t ) − θ w, k ( t ) ) (68)
and the fluid in the cell.
The heat flow can be presented as: The mathematical model for the observed cell after a
simple mathematical transformation can be obtained in the
Qin ( t ) = c f G f θ in
f ( t ) , Qout ( t ) = c f G f θ out
f ( t ) , (61) following form:
 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS 39

The observed heat exchanger is divided into five cells. If the

f , k ( t + Δt ) = a1θ f , k ( t ) + a2θ f , k ( t ) +
θ out out in
(69) results are unsatifactory the number of cells must be increased.
+ a3θ w, k ( t ) + a4θ f , k ( t + Δt ) + a3θ w, k ( t + Δt )
in
Table 1.

θ w, k ( t + Δt ) = b1θ w, k ( t ) + b2θ ain ( t ) + Parameter Dimension Value

1. cw kJ/kgK 0.53
+b2θ ain ( t + Δt ) + b3θ inf , k ( t ) + b3θ inf , k ( t + Δt ) + (70)
2. ρw Kg/m 3
7.85 ⋅ 103
+b3θ outf , k ( t ) + b3θ f , k ( t + Δt )
out
3. λw W/mK 0.04 ⋅103
where 4. r0 m 0.016
2m f c f − 2c f G f Δt − α wf Awf Δt 5. ri m 0.012
a1 = ,
2m f c f + 2c f G f Δt + α wf Awf Δt 6. wf m/s 0.5
7. Gf kg/s 0.12
2m f c f + 2c f G f Δt − α wf Awf Δt
a2 = , 8. Ga kg/s 1.8
2m f c f + 2c f G f Δt + α wf Awf Δt
9. ρf Kg/m 3
952.38

2α wf Awf Δt 10. cf kJ/kgK 4.233

a3 = , (71) α aw
2m f c f + 2c f G f Δt + α wf Awf Δt
2
11. W/(m K) 220
12. ρa Kg/m3 0.748
13 α wf W/(m2K) 3000
b1=
( 2mw cw −α aw Aaw Δt−α wf Awf Δt )( 2Ga ca +α aw Aaw )+α aw
2 2
Aaw Δt
14. ca kJ/kgK 1.097
( w w aw aw
2 m c +α A Δt+α A
wf wf Δt ) a a aw aw aw aw Δt
( 2G c +α A )−α 2
A 2
15. L m 10
16. θ infN C 110
2c f G f Δt − 2m f c f − α wf Awf Δt θ aN
in
a4 = , 17. C 220
2m f c f + 2c f G f Δt + α wf Awf Δt
Results of the simulation for the analytical approach based
2α aw Aaw ΔtGa ca on Laplace transformation
b2 =
( 2mwcw +α aw Aaw Δt+α wf Awf Δt )( 2Ga ca +α aw Aaw )−α aw
2 2
Aaw Δt Fig.2 and 3 show the step response of the outlet
temperature of the vfluid in the tube for the step change of
α wf Awf Δt ( 2Ga ca + α aw Aaw ) the inlet fluid temperature and the inlet air temperature.
b3 = Since the temperature function obtained using the approach
2 ( 2mw cw +α aw Aaw Δt+α wf Awf Δt )( 2Ga ca +α aw Aaw )−2α aw
2 2
Aaw Δt
explained in Section 2 is complex valued, the amplitude of
this function is used for the simulation.
Eliminating θ w, k ( t + Δt ) from Eqs. (69) and (70) the
outlet temperature from the k-th cell can be expressed in the
following form:

f , k ( t + Δt ) = c1θ f , k ( t ) + c2θ f , k ( t ) + c3θ w, k ( t ) +

θ out out in
(72)
+c4θ ain ( t ) + c5θ inf ,k ( t + Δt ) + c4θ ain ( t + Δt )
where
a1 + a3b3 a + a3b3 a +a b
c1 = , c2 = 2 , c3 = 3 3 1 , (73)
1 − a3b3 1 − a3b3 1 − a3b3

a3b2 a + a3b3
c4 = , c = 4 .
1 − a3b3 5 1 − a3b3
Figure 2. Step response of the outlet temperature of the fluid in the tube
for the step change of the inlet fluid temperature
Eq. (72) is used for the simulation of the process in the
cross-flow heat exchanger.
Assuming, in the first step, θ inf , k ( t + Δt ) = θ inf , k ( t ) and
θ ain ( t + Δt ) = θ ain ( t ) where Δt is small enough, the outlet
temperature is determinated for the k-th cell, k=1,2,..., p.
Repeating this procedure for different values of Δt the
fluid temperature of cross-flow heat exchanger at the cells
outlet is obtained as a function of time.

Simulation and comparative analysis

This section presents the analysis of the mathematical model
of the cross-flow heat exchanger with respect to different
methods considered in the previous sections. The characteristic Figure 3. Step response of the outlet temperature of the fluid in the tube
of a real cross-flow heat exchanger are given in Table 1. for the step change of the inlet air temperature
40 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS

Results of the simulation for the differential discrete Results of the simulation for the mathematical model based
mathematical model based on the finite difference method on the physical discretization
Figures 4, 5, 6 and 7 show the step response in the fluid Fig. 8, 9, 10 and 11 show the step response in the fluid
outlet temperature and the temperature profile in the outlet outlet temperature and the temperature profile in the outlet
of each cell for the step change in the fluid inlet of each cell for the step change in the fluid inlet
temperature and the step change in the air inlet temperature. temperature and the step change in the air inlet temperature.

Figure 4. Step response of the outlet temperature of the fluid in the tube Figure 8. Step response of the outlet temperature of the fluid in the tube
for the step change of the inlet fluid temperature for the step change of the inlet fluid temperature

Figure 9. Temperature profile in the outlet of each cell for the step change
Figure 5. Temperature profile in the outlet of each cell for the step change in the fluid inlet temperature
in the fluid inlet temperature

Figure 10. Step response of the outlet temperature of the fluid in the tube
Figure 6. Step response of the outlet temperature of the fluid in the tube for the step change of the inlet air temperature
for the step change of the inlet air temperature

Figure 7. Temperature profile in the outlet of each cell for the step change Figure 11. Temperature profile in the outlet of each cell for the step
in the air inlet temperature change in the air inlet temperature
 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS 41

Results of the simulation for the mathematical model based

on the transport approach
Fig. 12 and 13 show the step response in the fluid outlet tempe-
rature and the temperature profile in the outlet of each cell for the
step change (from 220ºC to 257ºC) in the air inlet temperature.

Figure 15. Temperature profile in the outlet of each cell for the step
change in the fluid inlet temperature

Finally, it should be pointed out that the temperature

changes obtained by the aplication of the transport
approach are in total coordinates while the step responses
obtained by the application of previous methods are
Figure 12. Step response of the outlet temperature of the fluid in the tube
represented by the relative deviations.
for the step change of the inlet air temperature

Conclusion
In this paper some characteristic methods in mathematical
modelling of the heat exchangers are considered.
Analyzing the analytical approach using the Laplace
transform it can be concluded that the system of PDE is very
complex to be solved analytically and this model has only
academic significance while its practical usage is limited.
In order to overcome the solution difficulties, the
procedure of differential discrete modelling is applied,
leading to the set of ordinary differential equations of a
rather high order. This procedure is based on the
discretization of spatial coordinates using the finite
difference method.
One of possible methods to avoid partial differential
equations is the method of physical discretization which
implies dividing the observed heat exchanger into an
appropriate number of cells. If the number of cells is big
enough it can be assumed that the process within one cell is
Figure 13. Temperature profile in the outlet of each cell for the step a process with certain parameters. This procedure results in
change in the air inlet temperature ordinary differential equations of a high order that are still
simple enough to be solved.
The transport approach is based both on the spatial and
Fig. 14 and 15 show the step response in the fluid outlet tempe- time discretization which transforms the system of partial
rature and the temperature profile in the outlet of each cell for the differential equations to the system of the algebric
step change (from 110ºC to 127ºC) in the fluid inlet temperature. equations that can be solved using numerical, iterative
procedures. This procedure should be carried out taking
into account the conditions for the convergence of
numerical procedures.
The graphical results of simulations of the different
mathematical models of cross-flow heat exchangers are
presented in this section.
Comparing the step responses obtained by the
application of different methods for mathematical
modelling of the heat exchanger it can be concluded that
results agree very well with each other.

References
[1] DEBELJKOVIĆ,LJ.D, SIMEUNOVIĆ,G.V., MULIĆ,V.S.:
Mathematical Model of Objects and Plants in Automatic Control
Systems, Planeta print, Belgrade, 2006.
[2] GRUJIĆ,Lj.T., DEBELJKOVIĆ,LJ.D.: Differential Discrete
Figure 14. Step response of the outlet temperature of the fluid in the tube Modelling and Controllability and Observability Analysis of Heat
for the step change of the inlet fluid temperature Exchangers, Technika, (1978), No.7-8.
42 STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS

[3] ALOTAIBI,S., SEN,M., GOODWINE,B., YANG,K.T.: [14] DAVIS,P.J., RABINOWITZ,P.: Methods of Numerical Integration,
Controllability of Cross-Flow Heat Exchangers, International Journal 1984, Academic Press.
of Heat and Mass Transfer, (2004), Vol.47, pp.913-924. [15] DEBELJKOVIĆ,LJ.D., GAZIKALOVIĆ,J.V.: ,Mathematical
[4] SPIGA,G., SPIGA,M.: Two - Dimensional Transient Solution for modelling and simu-lation of dynamical behavior of steam
Cross-Flow Heat Exchangers with Neither Gas Mixe, Journal of Heat generating-power plant, Preprints IFAC/IMACS International
Transfer, May, (1987), Vol.109, pp.281-286. Symposium on Simulation of Control Systems, Vienna, Austria,
[5] ROMIE,F.E.: Transient Response of the Counter-Flow Heat September 22 - 26, (1986), pp.369 - 373.
Exchanger, J. Heat Transfer 106 (3), (1984), pp.620-626. [16] EDELMAN,A.: Eigenvalues and condition numbers of random
[6] ROETZEL,W., XUAN,Y.: Transient/Response of Parallel and Counter- matrices, SIAM J., Matrix anal. appl.,1988.
Flow Heat Exchangers, J. Heat Transfer, 114 (2), 510-512, (1992). [17] KLAMKA,J.: Controllability of Dynamical Systems, Kluwer
[7] ROETZEL,W., XUAN,Y.: Transient/Response of Parallel and Academic Publishers, Netherlands, 1991.
Counter-Flow Heat Exchangers, J. Heat Transfer, 114 (2), 1992, [18] PALM,W.J.: Modeling, analysis and control of dynamical systems, J.
pp.510-512. Wiley, New York, 1983.
[8] BEJAN,A.: Entropy Generation Minimization, CRC Press, 1996. [19] RAKE et.al., Development, test and applications of improved models
[9] KLAMKA,J.: Controllability of Dynamical Systems, Kluwer for the dynamics of large crossflow heat exchangers-case study,
Academic Publishers, Netherlands, 1991. Identification and System Parameter Estimation, Washington D. C.,
USA, (1982), pp.205-210.
[10] ABATE,J., VALKO,P.P.: Multi-precision Laplace Inversion, Int. J.
Numer. Meth.,Engng. 60,979-993, 2004. [20] SADIBAŠIĆ,A.: Regulacija temperature kod razmenjivača toplote
kao procesa sa raspodeljenim parametrima, Magistarski rad,
[11] ILIĆ,K.: Dynamical behaviour of the heat exchangers, M.Sc. Thesis, Tehnološko-metalurški fakultet, Beograd, 1981.
Faculty of Mechanical Engineering, Sarajevo,1986.
[12] ANTSAKLIS,P.J., MICHEL,A.N.: Linear Systems, McGraw-Hill, 1997.
[13] BAI,Z., DEMMEL,J., MCKENNEY,A.: On computing Condition
Numbers for the Nonsymetric Eigenproblem, ACM transactions on Received: 01.03.2010.
Mathematical Software, June (1993), Vol.19, No.2, pp.202-223.

Komparativna analiza različitih metoda matematičkog modeliranja

rekuperativnih razmenjivača toplote
Rekuperativni razmenjivači toplote se veoma često koriste kao konstruktivni elementi različitih postrojenja tako da je
poznavanje njihove dinamike veoma važno. Razmenjivači toplote se takođe koriste i u vojnim letilicama svih veličina,
različitih namena, kao delovi različitih sistema ovih letilica. Njihovo funkcionisanje je najčešće kontrolisano
regulacijom ulaznih temperatura radnih fluida ili veličine njegovog masenog protoka. Na bazi realno usvojenih
pretpostavki izveden je linearizovani matematički model razmenjivača toplote sa unakrsnim strujanjem radnih
fluida, uzimajući u obzir i dinamiku zida razmenjivača. Model je zasnovan na fundamentalnom zakonu konzervacije
energije, uzimajući u obzir sve akumulatore toplote ovog procesa i predstavljen je pomoću sistema parcijalnih
diferencijalnih jednačina (PDE), čije rešenje nije, u opštem slučaju, moguće u zatvorenom obliku. Kao jedan od
mogućih načina da se izbegne rešavanje sistema parcijalnih diferencijalnih jednačina, u ovom radu se razmatraju
različite metode matematičkog modeliranja razmenjivača toplote: prilaz zasnovan na primeni Laplasove
transformacije, aproksimacija parcijalnih diferencijalnih jednačina pomoću metode konačnih razlika, metoda fizičke
diskretizacije i transportni prilaz. Za konkretno usvojene vrednosti parametara razmenjivača toplote izvršena je
simulacija njegovog rada i dati su grafički prikazi odskočnih odziva za sve razmatrane metode i analiziran je njihov
praktični značaj.

Ključne reči: razmenjivač toplote, matematički model, uporedna analiza, Laplasova transformacija, metoda konačnih
razlika, metoda diskretizacije.

Сравннительный анализ различных методов математического

моделирования рекуперативных теплообменников

Рекуперативные теплообменники очень часто использованы в роли конструктивных элементов различных

оборудований, из-за чего очень важным является познание их динамики. Теплообменники тоже
употребляются во военных летательных аппаратах всех размеров, с различными назначениями и в роли
составных частей различных систем этих летательных аппаратов. Их функционирование в большинстве
случаев контролировано регулированием входящих температур рабочих жидкостей или значением их
протока массы. На основании действительно принятых предположений выведена линейная математическая
модель теплообменника с перекрёстным потоком рабочих жидкостей, учитывая и динамику стены
теплообменника. Модель обоснована на фундаментальном законе консервации энергии, учитывая все
аккумуляторы теплоты этого процесса и представлена путём систем частичных дифференциальных
уравнений (ЧДУ), чьё решение в общем случае возможно только в закрытой форме. В роли одного из
возможных способов во избежание решений систем частичных дифференциальных уравнений, в настоящей
работе рассматриваются различные методы математического моделирования теплообменников: подход
обоснован на трансформации Лапласа, приближение (апроксимация) частичных дифференциальных
STEVIĆ,D.: COMPARATIVE ANALYSIS OF DIFFERENT METHODS IN MATHEMATICAL MODELLING OF THE RECUPERATIVE HEAT EXCHANGERS 43

уравнений при помощи метода конечных разниц, метода физической дискредитации и транспортный подход.
Для конкретно приемлемых значений параметров теплообменника проведена симуляция его работы и даны
графики отскакивающих отзывов для всех рассматриванных методов и анализировано их практическое
значение.

Kly~evwe slova: tеплообменник, математическая модель, сравннительный анализ, трансформация Лапласа,

метод конечных элементов, метод дискретизации.

Analyse comparative des différentes méthodes de modélisation

mathématique des échangeurs récupérables de chaleur

Les échangeurs récupérables de chaleur sont très souvent utilisés comme les éléments constructifs pour les différentes
installations et la connaissance de leur dynamique est très importante. On utilise les échangeurs de chaleur dans les
avions militaires de toutes tailles, de diverses missions, comme les parties de différents systèmes de ces avions. Leur
fonctionnement est contrôlé le plus souvent par le réglage des températures d’entrée des fluides moteurs ou par la
grandeur de leur écoulement de masse. A la base des hypothèses réellement adoptées on a dérivé le modèle
mathématique linéarisé de l’échangeur de chaleur à l’écoulement croisé de fluides moteurs considérant la dynamique
du paroi de l’échangeur. Le modèle est conçu sur la loi fondamentale de la conservation d’énergie tenant compte de
tous les accumulateurs de chaleur de ce processus et il est représenté par le système des équations partielles
différentielles (EPD) dont la solution , généralement, n’est pas possible en forme fermée. Comme un des moyens
d’éviter la solution du système des équations partielles différentielles, dans ce travail on considère les différentes
méthodes de la modélisation mathématique pour les échangeurs de chaleur : approche basée sur l’application de la
transformation de Laplace, approximation des équations partielles différentielles par la méthode des différences
finies, méthode de discrétisation physique et approche de transport. Pour les valeurs adoptées des paramètres des
échangeurs de chaleur, on a effectué la simulation de leur fonctionnement et on a donné les formes graphiques des
réponses pour toutes les méthodes étudiées en analysant leur importance pratique.

Mots clés: échangeur de chaleur, modèle mathématique, analyse comparative, transformation de Laplace, méthode
des différences finies, méthode de discrétisation.