4. Calculation of the warping constant#
The warping constant is used by the warping beam model (modeling POU_D_TG and POU_D_TGM), which is important to take into account for beams with thin open sections (cf. [R3.08.04]).
This coefficient (noted \({I}_{w}\) in [R3.08.04], in \({m}^{6}\)) is involved in the expression of the virtual work of internal forces on torsional terms:
\({W}_{\text{int}}={\int }_{0}^{2}({\theta }_{x,x}^{\text{*}}\mu \text{.}C\text{.}{\theta }_{x,x}+{\theta }_{x,\text{xx}}^{\text{*}}\text{.}E\text{.}{I}_{\omega }\text{.}{\theta }_{x,\text{xx}})\text{dx}\)
Taking the approach of [§3.1], and by placing ourselves in a coordinate system linked to the center of torsion \(C\), the kinematics of the torsion of any section is:
\(u(M)\text{=}\left[\begin{array}{c}u\\ v\\ w\end{array}\right]\text{=}\left[\begin{array}{c}\frac{\partial {\theta }_{x}}{\partial x}x\\ 0\\ 0\end{array}\right]\wedge \left[\begin{array}{c}x\\ y\\ z\end{array}\right]\text{+}\left[\begin{array}{c}\frac{\partial {\theta }_{x}}{\partial x}\xi (y,z)\\ 0\\ 0\end{array}\right]\) \(\text{=}(\begin{array}{c}\frac{\partial {\theta }_{x}}{\partial x}\text{}\xi (y,z)\\ -\frac{\partial {\theta }_{x}}{\partial x}xz\\ \frac{\partial {\theta }_{x}}{\partial x}xy\end{array})\)
where \(\xi (y\text{,}z)\) is the warping function (which is only cancelled in the case of a circular section).
The expression for the stress field is (in elasticity):
\(\begin{array}{}{\sigma }_{\text{xx}}=E{\varepsilon }_{\text{xx}}=E\xi (y,z)\frac{{\partial }^{2}{\theta }_{x}}{\partial {x}^{2}}\\ {\sigma }_{\text{xy}}=2\mu {\varepsilon }_{\text{xy}}=\mu \frac{\partial {\theta }_{x}}{\partial x}(\frac{\partial \xi (y,z)}{\partial y}-z)\\ {\sigma }_{\text{xz}}=2\mu {\varepsilon }_{\text{xz}}=\mu \frac{\partial {\theta }_{x}}{\partial x}(\frac{\partial \xi (y,z)}{\partial z}-y)\end{array}\)
Unlike [§4], second-order terms in \(\frac{{\partial }^{2}{\theta }_{x}}{\partial {x}^{2}}\) are no longer overlooked.
The first equilibrium relationship \({(\text{div}\sigma )}_{x}={\sigma }_{\text{xx},x}+{\sigma }_{\text{xy},y}+{\sigma }_{\text{xz},z}=0\) then involves the following condition on the warping function: \(\Delta \xi =0\)
On the other hand, without external loading on the outline of the section, we must have \(\sigma \otimes n\), which can be written: \(\frac{\partial \xi }{\partial y}{n}_{y}+\frac{\partial \xi }{\partial z}{n}_{z}=z\mathrm{.}{n}_{y}-y\mathrm{.}{n}_{z}\), where \({n}_{y}\) and \({n}_{z}\) are the two components of the normal, or even in vector form: \(\text{grad}\xi \text{.}n=\frac{\partial \xi }{\partial n}=(n\wedge \text{CM}).x\)
This determines the warping function to the nearest constant. To overcome this indeterminacy, for example, we write the expression for normal force (for a section where twisting produces warping):
\(N=\underset{S}{\int }{\sigma }_{\text{xx}}\text{ds}=\underset{S}{\int }E\xi \frac{{\partial }^{2}{\theta }_{x}}{\partial {x}^{2}}\text{ds}=0\)
so the additional condition on the warping function is \(\underset{S}{\int }\xi \text{ds}=0\). In practice, in MACR_CARA_POUTRE, we place ourselves first and foremost in a coordinate system linked to the center of torsion \(c\).
We then calculate \(\xi\) which must check:
\(\begin{array}{}\Delta x\text{=0}\\ \text{grad}x\text{.}n=\frac{\partial \xi }{\partial n}=(n\wedge \text{CM})\text{.}x\\ \underset{S}{\int }x\text{ds}=0\end{array}\)
Warping inertia \({I}_{\omega }\) is then obtained by: \({I}_{\omega }=\underset{S}{\int }{\xi }^{2}\text{ds}\)
MACR_CARA_POUTRE uses the following basic commands:
Translation of the coordinates of the nodes into the coordinate system linked to the center of torsion (calculated previously in table TCARS):
CREA_MAILLAGE (MAILLAGE = my,
REPERE = _F (TABLE = TCARS, NOM_ORIG = “TORSION”))
Assignment of a model (thermal plane), of a material field:
AFFE_MODELE (MAILLAGE = my,
AFFE = _F (TOUT = “OUI”, PHENOMENE = “THERMIQUE”,
MODELISATION =” PLAN “))
AFFE_MATERIAU (MAILLAGE = my,
AFFE = _F (TOUT = “OUI” MATER: mat))
Boundary conditions on the outer contour \(\mathrm{G0}\): \(\frac{\partial \xi }{\partial y}{n}_{y}+\frac{\partial \xi }{\partial z}{n}_{z}=z\mathrm{.}{n}_{y}-y\mathrm{.}{n}_{z}\)
F1= DEFI_FONCTION (NOM_PARA =, VALE = (0., 0., 10., -10.))
F2= DEFI_FONCTION (NOM_PARA =, VALE = (0., 0., 10., 10.))
CH1 = AFFE_CHAR_THER_F (MODELE = mod,
FLUX_REP = _F (GROUP_MA = G0, FLUX_X = F1, FLUX_Y = F2))
Condition on the solution field: \(\underset{S}{\int }\xi \text{ds}=0\): creation of a unit source term throughout the mesh, and of the associated second member vector. LIAISON_CHAMNO then makes it possible to impose the desired condition.
CHS = AFFE_CHAR_THER (MODELE =
SOURCE = _F (TOUT = “OUI” SOUR = 1.))
VS = CALC_VECT_ELEM (OPTION = “CHAR_THER” CHARGE = CHS…)
MS = CALC_MATR_ELEM (MODELE =… OPTION = “RIGI_THER”)
NUM = NUME_DDL (MATR_RIGI = MS)
GO = ASSE_VECTEUR (VECT_ELEM = VS NUME_DDL = NUM)
CH2 = AFFE_CHAR_THER (
LIAISON_CHAMNO = _F (CHAM_NO = VA COEF_IMPO = 0.))
Calculation of the warping function \(x\):
THER_LINEAIRE (MODELE =…
EXCIT = (
_F (CHARGE: CH1),
_F (CHARGE: CH2),
)
)
Calculation of the warpage constant \({I}_{\omega }=\underset{S}{\int }{\xi }^{2}\text{ds}\) and enrichment of the table:
TCARS = POST_ELEM (MODELE =…
CARA_POUTRE = _F (CARA_GEOM = TCARS,
LAPL_PHI = KSI OPTION: “CARA_GAUCHI”));