Solution of Plate Bending Equation Uniform Load Simply Supported Free to pull in via sinusoidal loading y x p( x, y) := po⋅sin π⋅ ⋅sin π⋅ a b loading pxy w=0 for mx = my = 0 2 mx = my = 0 => d 2 dx w= x=0 2 d dy 2 1 w=0 y=0 x=0 x=b y=0 y=a x=b y=a w( x, y) := C⋅sin π⋅ all boundary conditions satisfied if take y a ⋅sin π⋅ x b substitute in plate equation: 4 d 4 w(x,y) + 2⋅ dx 4 d 4 2 2 d d 2 dx dy w( x, y) + 2⋅ dx 2 d 2 d 2 dx dy 2 2 w(x,y) + 4 d dy 4 d w( x, y) + dy 4 4 w(x,y) = y x po⋅sin π⋅ ⋅sin π⋅ a b D w( x, y) → 4⋅ C⋅sin( π⋅y ) ⋅sin( π⋅x) ⋅π 4 after collecting terms: 4 d 4 w( x, y) + 2⋅ dx = 2 d 2 d 2 dx dy 2 4 d w( x, y) + dy 4 w( x, y) = C⋅sin π⋅ a π2 π2 C⋅sin π⋅ ⋅sin π⋅ ⋅ + a b b2 a2 y is a solution if y ⋅sin π⋅ x π4 ⋅ b b4 + 2⋅ π 4 2 2 + a ⋅b π a 4 4 2 x C⋅π ⋅sin π⋅ 4 y a ⋅sin π⋅ x b ⋅ 2 2+ 2 = a b 1 1 y x po⋅sin π⋅ ⋅sin π⋅ a b D is a solution if 2 po 1 1 C⋅π ⋅ + = 2 2 D a b 4 w( x, y) := po 4 ⋅ 1 2 ⋅sin π⋅ 1 + 1 2 2 a b d2 mx := −D⋅ dx2 w(x,y) + D⋅π D or .... y a ν ⋅sin π⋅ x b d2 w(x,y) dyy 2 2 C= po D⋅π 4 1 ⋅ 1 + 1 2 2 a b 2 is the displacement for a sinusoidal loading in x and y moments and stresses are determined from: d2 2 d w(x,y) + ν w(x,y) 2 2 dxx dy my := −D⋅ and σ x := mx I ⋅zzmax I := t 3 12 this result can first be generalized to: w( x, y) := po D⋅π 4 ⋅ 1 n2 m2 + b2 a2 2 ⋅sin mπ⋅ y a mx t ⋅ I 2 σ x := for a loading of a higher order sinusoidal loading ⋅sin nπ⋅ x y x p( x, y) := po⋅sin mπ⋅ ⋅sin nnπ⋅ a b b now consider a uniform load: po represent po in a double fourier series: 6 σ x := mx⋅ 2 t ∞ p = f(x, y) = ∞ ∑ ∑ amn = 16⋅po amn⋅ sin odd coefficients even = 0 2 π ⋅m⋅n a m⋅π⋅y ⋅sin n⋅π⋅x a b for example on a square plate with infinity = 20 i.e. 20 terms in the series: eachpxy sinusoidal element (m,n) is: amn⋅sin m= 1 n= 1 coefficients amn can be determined and are: N ≡ 20 M ≡ 20 ⋅ sin b note here that we are using m, and n odd 3 the displacement for each loading element 1 w( x, y) := D⋅π is ... amn 4 ⋅ n2 m2 + b2 a2 2 m⋅π⋅y ⋅sin n⋅π⋅x a b ⋅sin from above this result can first be generalized to: w( x, y) := po D⋅π 4 ⋅sin mπ⋅ 1 ⋅ n2 m2 + b2 a2 2 y a for a loading of a higher order sinusoidal loading ⋅sin nπ⋅ x y x p( x, y) := po⋅sin mπ⋅ ⋅sin nnπ⋅ a b b so by superposition of lots of the components of the fourier expansion of po is ... w( x, y) := 1 4 ∞ ⋅ ∞ ∑ ∑ π ⋅D m = 1 n = 1 substituting for amn ∞ w( x, y) := ∞ ∑ ∑ m= 1 n= 1 amn m2 n2 + a2 b2 2 m⋅π⋅y ⋅sin n⋅π⋅x a b ⋅sin 16⋅po m2 n2 D + π ⋅D⋅m⋅n⋅ a2 b2 2 ⋅sin m⋅π⋅ 6 y a ⋅sin n⋅π⋅ amn := 16⋅po 2 n π ⋅m⋅n x b m, n, odd => substitute 2*m-1 and 2*n-1 for m, n ∞ w( x, y) := ∞ ∑ ∑ 16⋅po 2 4 ⋅sin(2⋅m − 1)⋅π⋅ y x ⋅sin(2⋅n − 1)⋅π⋅ a b ∑ ∑ m= 1 n= 1 (2⋅ m − 1) 2 (2⋅ n − 1) 2 π ⋅D⋅(2 D ⋅ m − 1)⋅(2⋅ n − 1)⋅ + 2 2 a b 6 2 a b if we want to look at maximum deflection: x = b/2 y = a/2 expand here from previous lecture: d2 w(x,y) + 2 dx mx := −D⋅ ν d2 w(x,y) dyy 2 d2 2 d w(x,y) + ν w(x,y) 2 2 dxx dy my := −D⋅ and we can solve just as above for the single half waves: plot as a function of a/b i.e. a with b = 1 a := 1 .. 10 M := 10 b := 1 ν := 0.3 N := 10 M mx_po_b_sq( a) := N ∑ ∑ m= 1 n= 1 2 (2⋅ m − 1) 2 (m ⋅ (2⋅ n − 1) + ν⋅ ⋅( −1) 2 2 (2⋅ m − 1) 2 a 4 2 + (2⋅ n − 1) π ⋅(2⋅ m − 1)⋅(2⋅ n − 1)⋅ b 2 a b 16 0.1 mx_po_b_sq(10) = 0.125 mx_po_b_sq( a) 0.05 0 0 5 what length (ratio) is needed to declare a plate long??? 10 a see how many terms we need to obtain convergence: M mx_po_b_sq(a , M , N) := N ∑ ∑ m= 1 n= 1 2 (2⋅ m − 1) 2 ⋅ (2⋅ n − 1) + ν⋅ ⋅( 2 2 2 a (2⋅ m − 1) 4 2 + (2⋅ n − 1) π ⋅(2⋅ m − 1)⋅(2⋅ n − 1)⋅ b 2 a b 16 5 M := 1 .. 10 0.2 mx_po_b_sq(10, M , M) 0.15 0.1 0 5 10 M the corresponding y direction moment is: ν is associated with the m term vs. the n term see above my) M := 10 N := 10 b := 1 M my_po_b_sq( a) := a := 1 .. 10 N ∑ ∑ m= 1 n= 1 ν := 0.3 max at a/2 b/2 16 ⋅ ν⋅ (2⋅ n − 1) (2⋅ m − 1) 2 4 2 + (2⋅ n − 1) π ⋅(2⋅ m − 1)⋅(2⋅ n − 1)⋅ 2 a b 2 2 + (2⋅ m − 1) a b 2 2 ⋅(− 0.05 my_po_b_sq(10) = 0.037 0.045 my_po_b_sq( a) 0.04 0.035 mx_po_b_sq(10, 10, 10)⋅ν = 0.037 0 5 10 a restating original form of mx/po*b^2 M mx_po_b_sq( a) := N ∑ ∑ m= 1 n= 1 2 (2⋅ m − 1) 2 (m ⋅ (2⋅ n − 1) + ν⋅ ⋅( −1) 2 2 (2⋅ m − 1) 2 a 4 2 + (2⋅ n − 1) π ⋅(2⋅ m − 1)⋅(2⋅ n − 1)⋅ b 2 16 6 a b part of figure 9.5 in Hughes mx_po_b_sq( a) 0.1 my_po_b_sq( a) 2 0.05 0 0 5 10 a stress is related to the moment as before σ x := mx I ⋅zzmax I := t 3 σ x := 12 mx t ⋅ I 2 6 σ x := mx⋅ 2 t at maximum mid point: considering: b σ x = k⋅ po⋅ t therefore: σx 2 b 2 po⋅ t =k k= similarly kx( a) := 6⋅mx_po_b_sq( a) 1 6 mx⋅ 2 t b 2 p ⋅ ot = 6⋅ mx_po_b_sq ky ( a) := 6⋅my_po_b_sq( a) kx(10) = 0.748 kx( a) ky( a) ky (10) = 0.221 0.5 0 0 5 10 kx( 10)⋅ν = 0.224 a the clamped situation is considerably more complicated the results are developed in Timoshenko results are shown in the plot: 0.1 0.5 M x_clamped 0.08 kx_clamped 7 b M y_clamped ky_clamped 0.06 0.04 1 1.5 2 2.5 a_over_b 0.343 0.4 0.3 1 1.5 2 a_over_b N.B. this NOT the same My and ky as for the simply supported case. In the clamped case, there is an axial stress in the y (long direction) even for a long plate. The y axis stress is the maximum at the midpoint of the short side (x = b/2 at the edge y = 0 and y = a) the x axis stress is maximum at the midpoint of the long side (y = a/2 at the edge x = 0 and x = b) 8 2.5 the bottom line plot for simply supported and clamped/clamped plates under uniform pressure is figure 9.6 Hughes 0.8 k in relation stress = k*p*(b/t)^2 0.7 0.6 0.5 0.4 0.3 0.2 1 1.5 2 2.5 3 ratio of a/b 3.5 4 4.5 x simply supported - > 0.75, max in center y simply supported -> 0.225 max in center x clamped -> 0.5 max at side mid long length y clamped -> 0.34 max at side mid short length Mx is max (k = 0.5) at x = a/2, i.e. at ends of short side, middle of long side My is max (k = 0.34) at y = b/2, i.e. at ends of long side middle of short side (think of situation in square => both are max (k = ~0.3) and equal on sides in middle) 9 5