16-Civ-B9 The Finite Element Method · December 2016
Question 2 of 3: Beam ABCD propped by a two-bar truss — nodal displacements, shear and moment diagrams
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Notes on this paper
98-Civ-B9 — Applications of the Finite Element Method, National
Examinations, December 2016. Three hours, open book, any non-communicating
calculator permitted. Five printed pages: a front page of instructions, three
equal-value problems, and Appendix A carrying the strain-displacement matrix and the
stiffness matrix of the constant-strain triangle. The front page instructs the candidate
to attempt only two of the three; because this set is a study resource, all three
are solved in full below.
Reference texts for this subject.
Logan, D. L., A First Course in the Finite Element Method, 6th ed., Cengage
— bar and beam elements (Ch. 3–4), the constant-strain triangle and thermal
loading (Ch. 6 and §6.5), isoparametric quadrilaterals (Ch. 10).
Cook, R. D., Malkus, D. S., Plesha, M. E. and Witt, R. J., Concepts and
Applications of Finite Element Analysis, 4th ed., Wiley — continuity requirements,
the Jacobian, integration order, spurious zero-energy modes and element quality.
Bathe, K.-J., Finite Element Procedures, 2nd ed. — the variational basis,
convergence, and the meaning of the weak form.
Timoshenko, S. P. and Goodier, J. N., Theory of Elasticity, 3rd ed., McGraw-Hill
— the compatibility equation and the Airy stress function used in Problem 1.
Hibbeler, R. C., Structural Analysis, 10th ed., Pearson — shear and moment
diagram conventions and fixed-end actions used to check Problem 2.
CSA A23.3, Design of Concrete Structures, and the ISIS Canada design manuals
— the Canadian code framework behind the reinforcement-modelling question.
Problem 2: Beam ABCD propped by a two-bar truss — nodal displacements, shear
and moment diagrams (one of three equal-value problems)
Given. A three-span beam of constant $EI$ built in at both ends, propped
at its two interior nodes by a pair of equally inclined bars meeting at a roller-supported
apex, and carrying a point load $P$ at each propped node.
Given data (Figure 2 and the printed instruction)
Quantity
Symbol
Value
Outer spans AB and CD
$L_{1}$
$L$
Central span BC
$L_{2}$
$2L$
Height of apex O above the beam
$h$
$L$ (from the $45^\circ$ bars)
Length of each bar OB, OC
$L_{b}$
$\sqrt{2}\,L$
Beam flexural rigidity
$EI$
constant
Bar axial rigidity
$EA$
$EI/6L^{2}$
Point load at B and at C
$P$
downward
Supports
—
A, D encastré; O on a horizontal roller
Find. (a) $v$ and $\theta$ at A, B, C, D and O; (b) the shear force
diagram; (c) the bending moment diagram.
[Figure not reproduced: Figure 2.1 — Figure 2 redrawn with the discretisation used below: three beam elements AB, BC, CD carrying $(v,\theta)$ at each node, and two axial-only bars OB, OC. Sign convention: $v$ positive up, $\theta$ positive counter-clockwise, sagging moment positive. See the official exam paper.]
Approach. Assemble the beam with the printed $4\times 4$ matrix, replace
each inclined bar by the equivalent vertical spring it presents to the beam, exploit the
symmetry of the structure and of the load to cut the system to two unknowns, then recover the
element end actions to draw the two diagrams.
Part (a) — identify the degrees of freedom the given element library
actually has. The beam element printed on the paper carries only $(v,\theta)$ at
each node: it has no axial term. The bars carry only $(u,v)$. A and D are built in, so
$v_{A}=\theta_{A}=v_{D}=\theta_{D}=0$; the roller at O gives $v_{O}=0$. Because the beam is
continuous into two immovable walls and has no axial flexibility in this model, it is axially
rigid, so $u_{B}=u_{C}=0$ and the horizontal pull of each bar is carried straight into the
walls. That leaves five free degrees of freedom:
$$\{d\}=\left\{v_{B},\ \theta_{B},\ v_{C},\ \theta_{C},\ u_{O}\right\}^{T}$$
Replace each bar by an equivalent vertical spring. Bar OB runs from
$B$ to $O$, a horizontal distance $L$ and a vertical distance $L$, so its length is
$\sqrt{2}L$ and its direction cosines are $(c,s)=(1/\sqrt{2},\,1/\sqrt{2})$. With
$u_{B}=u_{O}=v_{O}=0$ the elongation of the bar is $e=-v_{B}/\sqrt{2}$, so its strain energy
is $\tfrac{1}{2}\left(EA/\sqrt{2}L\right)\left(v_{B}^{2}/2\right)$ and the bar behaves
exactly as a vertical spring of stiffness
$$k_{v}=\frac{EA}{2\sqrt{2}\,L}=\frac{1}{2\sqrt{2}L}\cdot\frac{EI}{6L^{2}}
=\boxed{\frac{\sqrt{2}}{24}\,\frac{EI}{L^{3}}=0.058926\,\frac{EI}{L^{3}}}$$
Use symmetry to halve the problem. The geometry, the supports and the two
loads are all symmetric about the mid-point of BC, so $v_{C}=v_{B}$, $\theta_{C}=-\theta_{B}$
and $u_{O}=0$. Only $v_{B}$ and $\theta_{B}$ remain.
Assemble the two remaining equations. Span AB, with A fixed, contributes
its lower-right $2\times 2$ block. Span BC contributes nothing to the vertical equation
— under equal end deflections and equal-and-opposite end rotations the shear it
transmits is zero — and contributes $(4-2)EI/L_{2}=EI/L$ to the rotation equation.
Adding the spring to the vertical equation:
$$\begin{aligned}
\left(\frac{12EI}{L^{3}}+k_{v}\right)v_{B}-\frac{6EI}{L^{2}}\theta_{B}&=-P\\
-\frac{6EI}{L^{2}}v_{B}+\left(\frac{4EI}{L}+\frac{EI}{L}\right)\theta_{B}&=0
\end{aligned}$$
Eliminate the rotation. The second equation gives
$\theta_{B}=6v_{B}/5L$ directly, and substituting it into the first collapses the
bending contribution from $12EI/L^{3}$ to $\left(12-\tfrac{36}{5}\right)EI/L^{3}
=\tfrac{24}{5}EI/L^{3}$:
$$\left(\frac{24}{5}+\frac{\sqrt{2}}{24}\right)\frac{EI}{L^{3}}\,v_{B}=-P
\qquad\Longrightarrow\qquad 4.858926\,\frac{EI}{L^{3}}\,v_{B}=-P$$
Solve for the nodal displacements and rotations. Writing the result as an
exact fraction and then as a decimal:
$$\boxed{v_{B}=v_{C}=-\frac{120}{576+5\sqrt{2}}\,\frac{PL^{3}}{EI}
=-0.205807\,\frac{PL^{3}}{EI}}$$
$$\boxed{\theta_{B}=-0.246968\,\frac{PL^{2}}{EI},\qquad
\theta_{C}=+0.246968\,\frac{PL^{2}}{EI}}$$
and $v_{A}=\theta_{A}=v_{D}=\theta_{D}=0$ at the built-in ends, with $u_{O}=v_{O}=0$ at the
apex — the roller would allow O to slide, but symmetry gives it no reason to.
Recover the bar forces and the reaction at the apex. The elongation of
OB is $e=-v_{B}/\sqrt{2}=0.145527\,PL^{3}/EI$, so with $EA=EI/6L^{2}$
$$N_{OB}=N_{OC}=\frac{EA}{\sqrt{2}L}\,e=\boxed{+0.017151\,P\ \text{(tension)}}$$
Each bar therefore pulls its node upward with $N\sin 45^\circ=0.012127P$ and pulls the wall
horizontally with the same amount, while the roller at O must resist
$2\times 0.012127P=0.024255P$ acting downward.
Part (b) — element end actions for span AB. Post-multiplying the
printed matrix by $\{0,\,0,\,v_{B},\,\theta_{B}\}^{T}$ gives the four end actions of AB
directly:
$$\begin{aligned}
V_{A}&=\frac{EI}{L^{3}}\left(-12v_{B}+6L\theta_{B}\right)=\left(2.469682-1.481809\right)P
=0.987873P\\
M_{A}&=\frac{EI}{L^{3}}\left(-6Lv_{B}+2L^{2}\theta_{B}\right)=0.740905\,PL\\
M_{B}&=\frac{EI}{L^{3}}\left(-6Lv_{B}+4L^{2}\theta_{B}\right)=0.246968\,PL
\end{aligned}$$
so the vertical reactions are $\boxed{R_{A}=R_{D}=0.987873P}$ and the fixing moments
$0.740905PL$. Global equilibrium checks:
$2(0.987873)+0.024255=2.000P$, exactly the applied load.
Build the shear diagram. Span AB carries the constant shear
$+0.987873P$. At B the applied $P$ acts down and the bar pulls $0.012127P$ up, a net
downward step of $0.987873P$, which takes the shear to zero; span BC therefore carries no
shear at all, which is what symmetry demands. The same step at C carries the shear to
$-0.987873P$ in span CD.
$$\boxed{V_{AB}=+0.9879P,\qquad V_{BC}=0,\qquad V_{CD}=-0.9879P}$$
Part (c) — build the bending moment diagram. With zero shear in BC
the moment there must be constant, so the diagram is completely described by its two end
values and the plateau between the propped nodes:
$$\boxed{M_{A}=M_{D}=-0.740905\,PL\ \text{(hogging)},\qquad
M_{B}=M_{C}=+0.246968\,PL\ \text{(sagging)}}$$
The moment varies linearly across each outer span, so it passes through zero where
$-0.740905+0.987873\,x/L=0$, that is at $x=0.750L$ from each built-in end.
Sanity-check against the beam without the truss. Deleting the two bars
leaves a fixed-ended beam of span $4L$ with loads $P$ at $L$ and at $3L$, for which the
classical result is $M_{A}=\sum Pab^{2}/L_{\text{tot}}^{2}=
\left(9+3\right)PL/16=0.75PL$, $M_{B}=0.25PL$ and $v_{B}=5PL^{3}/24EI
=0.208333\,PL^{3}/EI$. Our answers sit just inside those values, and the relief is only
$1.21\%$ — exactly what should be expected, because $EA=EI/6L^{2}$ makes
$k_{v}=0.0589EI/L^{3}$ barely one per cent of the $4.8EI/L^{3}$ the beam already offers at
B.
Figure 2.2 — Shear force and bending moment diagrams. Sagging is plotted upward.
The shear steps down by $0.9879P$ at each propped node and is identically zero in span BC,
so the moment there is the constant plateau $+0.2470PL$.
Check: the printed beam element has no axial
stiffness, so the model as supplied cannot resolve $u_{B}$ and $u_{C}$ on its own —
left free, the two bars plus the apex form a mechanism. The solution above adopts the only
consistent reading, that the beam is axially rigid between the two built-in walls, which is
the standard assumption behind a bending-only beam element. Giving the beam a finite $EA$
instead would let B and C draw together very slightly and would reduce the bar forces, but
the paper supplies no beam area, so no such data exist.