22-Mec-B9 Advanced Engineering Structures · December 2018
Question 3 of 8: Unsymmetrical bending of a thin-walled channel
Nivaar worked solution (AI-drafted; not reviewed by a licensed engineer)
Notes on this paper
Paper format. National Exams, December 2018 — 16-Mec-B9 Advanced Engineering Structures. Three hours, open book, any non-communicating calculator permitted. Eight questions are printed and the rubric states that any five constitute a complete exam paper, with all problems of equal total value (20 marks each). All eight are worked here. Marks for the individual parts are those printed inside each problem.
Reference texts. T. H. G. Megson, Aircraft Structures for Engineering Students, 6th ed. (thin-walled open and closed sections, shear flow, shear centre, multi-cell torsion, crack growth, laminated plates); A. P. Boresi and R. J. Schmidt, Advanced Mechanics of Materials, 6th ed. (yield criteria, unsymmetrical bending, torsion of non-circular sections); N. E. Dowling, Mechanical Behavior of Materials, 4th ed. (Paris law, linear elastic fracture mechanics); R. C. Hibbeler, Mechanics of Materials, 10th ed. (statically indeterminate axial members, thermal stress).
Check: the direction of the 500 N load in Question 3. On the isometric view the 500 N arrow lies antiparallel to the direction in which both flanges project (both make the same 28° angle with the horizontal on the drawing, in opposite senses). The solution therefore takes the 500 N as acting away from the flange tips, so that its bending contribution puts the lower-flange tip A into tension and adds to the contribution of the 900 N load. Had the arrow been read the other way, the stress at A would be 122.5 MPa compressive instead of 155.3 MPa tensile; every section property and both moment components are unchanged, so only the sign of the 500 N load’s contribution (138.9 MPa of the 155.3) moves.
Check: two rubric details not printed on the paper. (i) Question 8 gives no mark split for its parts (a) and (b); an even 10 + 10 is assumed, consistent with every other two-part question on this sitting. (ii) In Question 4(b) “120 mm to the left of the shear center” is taken as 120 mm further forward, i.e. towards the semicircular nose, which is what “left” means on the printed figure. Moving the load the other way would simply reverse the sign of the superposed torsional flow.
Question 3: Unsymmetrical bending of a thin-walled channel (20 marks)
Set up axes in the plane of the section with $z$ horizontal, positive towards the flange tips, and $y$ vertical, positive upward, with the origin at the foot of the web. In that frame the wall mid-lines are: the web from $(0,0)$ to $(0,120)$; the upper flange from $(0,120)$ to $(50,120)$; the lower flange from $(0,0)$ to $(100,0)$. Point A is the lower-flange tip at $(100,0)$.
Find. The direct bending stress at point A on the section 1500 mm from the loaded end.
Question 3 — the channel cross-section on its mid-line idealisation, with the centroid C, the reference axes, the computed neutral axis and point A at the tip of the wider lower flange. The two tip loads are shown to the right in the same $z$–$y$ frame.
Approach. Because the section has no axis of symmetry, $I_{yz} \neq 0$ and the stress must come from the general unsymmetrical bending formula; the loads act at the shear centre, so there is no torsion to superpose.
Locate the centroid. Treating each wall as a line of area $t\,L$ concentrated on its mid-line, the three wall areas are $3.0\times120 = 360$, $3.0\times50 = 150$ and $1.5\times100 = 150$ mm$^2$, totalling $A = 660$ mm$^2$. Then$$\bar z = \frac{\sum A_i z_i}{A} = \frac{0+150(25)+150(50)}{660} = 17.045\ \text{mm},\qquad\bar y = \frac{360(60)+150(120)+0}{660} = 60.0\ \text{mm}$$The two flanges have equal areas — $50\times3.0 = 100\times1.5$ — which is why the centroid falls at exactly mid-web depth. That is worth noticing, because it removes a whole class of arithmetic slips from the rest of the calculation.
Compute the second moments about the centroidal axes. For a straight thin wall running between centroid-relative end points $(a_1,b_1)$ and $(a_2,b_2)$ the exact three-point formulae are$$I = \frac{tL}{3}\big(b_1^2+b_1b_2+b_2^2\big),\qquad I_{yz} = \frac{tL}{6}\big(2a_1b_1+a_1b_2+a_2b_1+2a_2b_2\big)$$Summing over the three walls gives$$I_{zz} = 1.5120\times10^{6}\ \text{mm}^4,\qquad I_{yy} = 4.3324\times10^{5}\ \text{mm}^4,\qquad I_{yz} = -2.2500\times10^{5}\ \text{mm}^4$$The product of inertia comes almost entirely from the lower flange, which sits far from both centroidal axes; the web contributes nothing because it straddles the horizontal axis symmetrically.
Find the two bending moments at the section. Take the free-end portion of length $a = 1500$ mm as a free body. With $F_y = +900$ N and $F_z = -500$ N applied at its tip, moment equilibrium about the centroid of the cut gives$$M_z \equiv \int \sigma y\,\mathrm{d}A = -a F_y = -1.350\times10^{6}\ \text{N}\cdot\text{mm},\qquad M_y \equiv \int \sigma z\,\mathrm{d}A = -a F_z = +7.50\times10^{5}\ \text{N}\cdot\text{mm}$$The negative $M_z$ is the familiar cantilever result — an upward tip load puts the top fibres into compression — while the offset of the shear centre from the centroid contributes only torque, not bending moment, and so does not enter here.
Write the stress as a plane and invert. With no axial force the stress must be linear in position, $\sigma = \mathcal{A}y + \mathcal{B}z$. Substituting that into the definitions of $M_z$ and $M_y$ gives two simultaneous equations whose solution is$$\mathcal{A} = \frac{M_zI_{yy}-M_yI_{yz}}{I_{zz}I_{yy}-I_{yz}^2},\qquad\mathcal{B} = \frac{M_yI_{zz}-M_zI_{yz}}{I_{zz}I_{yy}-I_{yz}^2}$$With $I_{zz}I_{yy}-I_{yz}^2 = 6.0426\times10^{11}$ mm$^8$ this evaluates to $\mathcal{A} = -0.68845$ MPa/mm and $\mathcal{B} = +1.37360$ MPa/mm.
Evaluate the stress at point A. Point A lies at $(z,y) = (100,0)$, so relative to the centroid $z_A = 82.955$ mm and $y_A = -60$ mm:$$\sigma_A = \mathcal{A}y_A + \mathcal{B}z_A = (-0.68845)(-60) + (1.37360)(82.955)$$The two terms are $+41.31$ MPa and $+113.95$ MPa. These are the $\mathcal{A}y$ and $\mathcal{B}z$ terms, not the separate load contributions: because $I_{yz}\neq0$, each coefficient mixes both moments. Their sum gives$$\boxed{\sigma_A = +155.3\ \text{MPa (tensile)}}$$
Audit the result. Two checks cost nothing. First, integrating the assumed linear field over the three walls must return zero net axial force and reproduce both applied moments — it does, to four significant figures. Second, the neutral axis is the line $\mathcal{A}y+\mathcal{B}z = 0$ through the centroid, of slope $\mathrm{d}y/\mathrm{d}z = -\mathcal{B}/\mathcal{A} = 1.995$, i.e. inclined at $63.4^{\circ}$ to the horizontal. Point A is the section point furthest from that line on the tension side, which is precisely why the examiner asks for it.
The split between the two load contributions is the engineering lesson here. The 500 N horizontal load is only 56 per cent of the 900 N vertical one, yet it supplies 89 per cent of the stress at A. Evaluated one load at a time, the 900 N load alone gives $+16.37$ MPa at A and the 500 N load alone gives $+138.88$ MPa. This is because it bends the channel about its weak axis — $I_{yy}$ is only 29 per cent of $I_{zz}$. Open thin-walled sections are extremely sensitive to loads applied out of the plane of the web, and a designer who sizes such a member for the vertical load alone will underestimate the peak stress by a factor of more than nine (155.3 against 16.4 MPa). The steep neutral axis is the geometric statement of the same fact: the section bends about an axis nowhere near the one the loads suggest.