Beruflich Dokumente
Kultur Dokumente
CONSTITUTIVE EQUATIONS
Introducing the temperature field T(xi,t), the heat flux vector q(xi,t)→qn(xi,t) and the thermal
conductivity in the n direction λn, th following holds:
∂T
qn = −λn (4.1)
∂n
and in the (O,xi) space
∂T
q x = − λx
∂x
∂T
q y = −λ y (4.2)
∂y
∂T
q z = − λz
∂z
For an isotropic media
qi = −λT,i i = 1, 2,3.
(4.3)
THERMAL BALANCE
The balance of thermal energy in an infinitesimal volume 𝑑𝑉 = 𝑑𝑥𝑑𝑦𝑑𝑧 of the body in the x
direction for an infinitesimal time increment dt reads
dQin = qx dydzdt
⎛ ∂q ⎞
dQout = ⎜ qx + x dx ⎟ dydzdt
⎝ ∂x ⎠ (4.4)
∂q
dQin − dQout = x dVdt
∂x
Taking into account the heat fluxes in the other directions the thermal balance is
𝜕𝑞0 (4.5)
𝑑𝑄() − 𝑑𝑄+,- = − 𝑑𝑉𝑑𝑡 = 0
⟹
−𝑞(,( 𝑑𝑉𝑑𝑡 = 0
𝜕𝑥
Considering the case of heat generation within the body
dQgen = qgen dVdt
(4.6)
the equation becomes
𝜕𝑞0
𝑑𝑄() − 𝑑𝑄+,- + 𝑑𝑄78) = − + 𝑞78) 𝑑𝑉𝑑𝑡 = 0 ⟹ −𝑞(,( + 𝑞78) 𝑑𝑉𝑑𝑡 = 0 (4.7)
𝜕𝑥
FOURIER EQUATION
Introducing the body density ρ and its c specific heat for unit mass c, the Fourier Equation reads:
𝜕𝑇
−𝑞(,( + 𝑞78) = 𝑐𝜌
𝜕𝑡
𝑞<= = −𝑞( 𝑛(
on
𝑆B (4.8)
𝑇 = 𝑇
on
𝑆C
The previous equation can be recast (using the 4.3) in the form
𝜕𝑇
− −𝜆( 𝑇,( + 𝑞78) = 𝑐𝜌 (4.9)
,( 𝜕𝑡
and, in the homogeneous an isotropic case
𝜕𝑇 𝜕𝑇 (4.10)
𝜆𝑇,(( + 𝑞78) = 𝑐𝜌
or
− 𝑞(,( + 𝑞78) = 0,
with
𝑞78) = 𝑞78) − 𝑐𝜌
𝜕𝑡 𝜕𝑡
Finally
𝜕𝑇
𝜆∇K 𝑇 + 𝑞78) = 𝑐𝜌
(4.11)
𝜕𝑡
STATE VARIABLES
Displacement ui → T Temperature
Strains 𝜖 ij → T,i Temperature gradient
Stresses Ϭij → qi Heat Flux
LOADS
Internal Loads (per unit volume) Xi → qgen Generated Heat (per unit volume per unit time)
Surface Loads (per unit area) fi → 𝑞i Thermal flux on Sq
Prescribed Displacement 𝑢i → 𝑇 Prescribed Temperatures on ST
GOVERNING EQUATIONS
Equilibrium 𝜎(O,O + 𝑋( = 0
→
−𝑞(,( + 𝑞78) = 0
R
Compatibility 𝜖(O,O = 𝑢(,O + 𝑢O,(
→
grad
𝑇 = 𝑇,(
K
The latter equation can be obtained considering the thermal energy balance equation
− qi ,i + qgen = 0
(4.14)
and imposing the following expression to be zero
where 𝑇 is the virtual temperature field (null on ST). The following holds:
we have
− 𝑞( 𝑇,(
𝑑𝑉 + 𝑞( 𝑇 ,(
𝑑𝑉 = 𝑞78) 𝑇𝑑𝑉 (4.19)
Y Y Y
where
𝜕𝑇 𝜕𝑇 𝜕𝑇
𝑞C = 𝑞0
, 𝑞^
, 𝑞_
,
∇𝑇 C =
,
, (4.23)
𝜕𝑥 𝜕𝑦 𝜕𝑧
𝑇 𝑥, 𝑦, 𝑧 = 𝛼( 𝜑( 𝑥, 𝑦, 𝑧 (4.24)
(aR
where i is the number of grid points (the edges of our elements) defined at locations xi , yi , zi .
The functions are usually polynomials, but in some case (e.g. in acoustic analysis) can be useful
to use trigonometric functions. These functions are called interpolating functions or shape
functions. Unless in particular cases, the linear combination of these functions will not provide
an exact but only approximate solution. The set of interpolating functions can in theory be chosen
ad libitum, but in order for the convergence of the result to be monotonic with the condensation,
four fundamental requirements must be respected:
1. ability to have same variation of temperature of the element without the occurrence of
heat flux within (equivalent to rigid motions)
2. allowing states of constant gradients of temperature throughout the element (patch test)
3. compatibility along the contour lines or faces of contiguous elements, no breaks or
interpenetrations should occur
4. continuity or absence of singular points in the description of the temperature field.
One time the interpolating functions 𝜑( 𝑥, 𝑦, 𝑧 for each internal point of the element are
identified, it is possible to write
𝑇 𝑥, 𝑦, 𝑧 = 𝛼R 𝜑R 𝑥, 𝑦, 𝑧 + 𝛼K 𝜑K 𝑥, 𝑦, 𝑧 + ⋯ + 𝛼) 𝜑) 𝑥, 𝑦, 𝑧 (4.25)
or in matrix form
𝛼R
𝑇 𝑥, 𝑦, 𝑧 = 𝜑R 𝑥, 𝑦, 𝑧
𝜑K 𝑥, 𝑦, 𝑧
…
𝜑) 𝑥, 𝑦, 𝑧 𝛼K
…
𝛼) (4.26)
𝑇 𝑥, 𝑦, 𝑧 = 𝜑 𝑥, 𝑦, 𝑧
𝛼
We apply the previous interpolation on particular points, the nodes, considering all the degree-
of-freedom defined for each node
… …
ΘO = 𝛼R 𝜑R 𝑥O , 𝑦O , 𝑧O + 𝛼K 𝜑K 𝑥O , 𝑦O , 𝑧O + ⋯ + 𝛼) 𝜑) 𝑥O , 𝑦O , 𝑧O
… …
(4.27)
Θ = 𝐴
𝛼
𝛼 = 𝐴gR Θ
(4.28)
that replaced in the expression of the internal temperature field (4.27) bring to
𝑇 𝑥, 𝑦, 𝑧 = 𝜑 𝑥, 𝑦, 𝑧
𝐴gR Θ
(4.29)
𝑇 𝑥, 𝑦, 𝑧 = 𝑁 𝑥, 𝑦, 𝑧
Θ
(4.30)
Θ is the vector of the nodal temperatures of the element.
4.4 Finite element formulation
Introducing
𝜕 𝜕 𝜕
𝐷C =
,
, (4.31)
𝜕𝑥 𝜕𝑦 𝜕𝑧
we can write
∇𝑇 = 𝐷𝑇
⟹
∇𝑇 C = 𝑇𝐷C (4.32)
Considering a finite element approximation
𝑇 𝑥 = 𝑁 𝑥
Θ where 𝑁 = 𝑁R , 𝑁K , … , 𝑁)
,
ΘC = ΘR , ΘK , … , Θ)
(4.33)
where Θ is the nodal temperature vector and 𝑁 is the shape functions matrix:
C
∇𝑇 = 𝐷
𝑁
Θ
⟹
∇𝑇 C = 𝑁
Θ
𝐷C = ΘC 𝑁 C
𝐷C
(4.34)
The (4.22) become:
(4.35)
= ΘC
𝑁 C
𝑞78) 𝑑𝑉 + ΘC
𝑁 C
𝑞<=
𝑑𝑠
Y <=
Considering the expression 𝑞 = −Γ ∇𝑇 and the constitutive equations qi = −λiT,i we can write
and denoting
Kn = 𝐵C Γ
𝐵𝑑𝑉
and
𝑄 =
𝑁 C
𝑞78) 𝑑𝑉 +
𝑁 C
𝑞<=
𝑑𝑠 (4.39)
Y Y <=
we can write
K n
Θ =
𝑄
(4.40)
where K n is the thermal coefficient matrix.