Analytical Mechanics This page intentionally left blank
Download 10.87 Mb. Pdf ko'rish
|
v p L 0 Fig. 7.6 Poinsot motion. To complete the proof we only need to show that the plane π is fixed. We compute the distance h from O: h = |A − O| cos α = |A − O| ω · L 0 ωL 0 . Recalling now (7.9), which gives the construction of the ellipsoid of inertia, we find that |A − O| = λ/ √ I, where I is the moment of inertia with respect to the axis of instantaneous rotation. Since ω · L
0 = Iω
2 = 2T = 2T 0 , it follows that h = λ L 0 ω · L
0 (Iω
2 ) 1/2 = λ L 0 (2T
0 ) 1/2 . (7.55)
Hence the plane π has prescribed orientation and distance from O, and is therefore fixed. Since the contact point between the ellipsoid and the plane lies on the axis of instantaneous rotation, the ellipsoid does not slide. C orollary 7.2 The motion is determined by the rotation of a curve moving with the ellipsoid of inertia (polhode) on a fixed plane curve (herpolhode). Remark 7.4 The polhode is the intersection of the body Poinsot cone with the ellipsoid of inertia and the herpolhode is the intersection of the fixed Poinsot cone with the fixed plane. The equations of these curves can be obtained by remarking that a polhode is the locus of the points of the ellipsoid of inertia with the property that the plane tangent to the ellipsoid at these points has a fixed distance h from O. In the principal frame of inertia the equations of this locus are 3 i =1 J i x 2 i = λ 2 , (7.56) 3 i =1 J i ( J i − J 0 )x 2 i = 0, (7.57) 7.8 The mechanics of rigid bodies: dynamics 253 where we define J 0 = (λ/h) 2 . (7.58) Equation (7.57) is the equation of the body Poinsot cone. For a generic ellipsoid ( J 1
J 2
J 3
1 direction, and the minimal axis is in the x 3 direction. Let h max , h
med , h
min be the lengths of the three semi-axes. Then the constant h, determined by the initial conditions, varies in the interval [h min , h
max ] and correspondingly J 0
1 , J 3 ]. In the extreme cases h = h min ( J 0 = J 3 ), h = h
max ( J 0 = J 1 ), equation (7.57) implies that the polhodes degenerate to the vertices of the minimal or maximal axes, respectively. The other degenerate case is h = h med (
0 = J 2 ), because in this case we do not have the x 2 term in equation (7.57), which then represents the pair of planes J 1 ( J 1 − J 2 )x 1 ± J
3 ( J 3 − J
2 )x 3 = 0, (7.59)
symmetrically intersecting on the axis x 2 . These planes produce four arcs of an ellipse on the ellipsoid of inertia, called limiting polhodes. These arcs are separated by the vertices on the intermediate axis, which are degenerate polhodes (Fig. 7.7). In the generic case, the polhodes can be classified into two classes: (a) h
min < h < h med
( J 2 < J 0 < J 3 ), the body Poinsot cone has as axis the minimal axis of the ellipsoid of inertia; Polhodes around the maximal axis Polhode degenerating Limiting polhodes Polhodes around the minimal axis
2
1
3 Fig. 7.7 Classification of polhodes. 254 The mechanics of rigid bodies: dynamics 7.9 (b) h
med < h < h max
( J 1 < J 0 < J 2 ), the body Poinsot cone has as axis the maximal axis of the ellipsoid of inertia. The limiting polhodes are the separatrices of the two families. As we have already noted in Section 7.7, the phase space for Euler equations can be identified in the case of precessions by inertia with the space (ω 1 , ω 2 , ω
3 ). From this point of view, the study of polhodes is particularly interesting in view of the following fact. P roposition 7.10 The polhodes represent the trajectories in the phase space of equations (7.54). Proof
We verify that (7.56), (7.57) are still valid after the substitution x → ω.
Let us write the equations requiring that T and the absolute value of L(O) are constant: 3 i
J i ω 2 i = 2T 0 , (7.60) 3 i =1 J 2 i ω 2 i = L 2 0 . (7.61)
By eliminating h between (7.55) and (7.58) we find J 0 = L 2 0 2T 0 . (7.62)
It follows, again using (7.60), that we can write equation (7.61) in the form 3 i =1 J i ( J i − J 0 ) ω 2 i = 0, (7.63) and hence equations (7.56), (7.57) and (7.60), (7.63) coincide except for an inessential homothety factor. As mentioned already, these considerations are also valid in the case of a generic ellipsoid. When the ellipsoid of inertia is a surface of revolution, we have J 1 = J 2 or J 2 = J 3 , and it is easy to see that the polhodes and herpolhodes are circles (Fig. 7.8). The limiting polhodes do not make sense any more: every point of the circle obtained by intersecting the ellipsoid with the plane through O orthogonal to the rotation axis is a degenerate polhode. The case that the ellipsoid becomes a sphere ( J 1
J 2 = J 3 ) is trivial: all points of the sphere are degenerate polhodes. 7.9
Permanent rotations T heorem 7.2 If ω(0) = ω 0 has the direction of a principal axis of inertia, the corresponding precession by inertia is reduced to the uniform rotation ω = ω
0 .
7.9 The mechanics of rigid bodies: dynamics 255 Polhodes
Median cross-section Fig. 7.8 Polhodes for the ellipsoid of revolution. Conversely, if a precession by inertia is a rotation, the latter must be uniform and must be around a principal axis of inertia. Proof
The first claim is trivial, as it is immediately verified that ω = ω
0 is the only solution of equations (7.54). Suppose now that the motion is a rotation and let us examine first the case when ω, which has constant direction by our hypotheses, has at least one zero component. Suppose, e.g. that ω 1 = 0. Then equations (7.54) imply that ω 2 and ω
3 are constant, and the rotation is uniform. In addition we find that ( J
−J 3 )ω 2 ω 3 = 0, and hence either J 2 = J 3 , which implies that every diametral axis, including the axis of rotation, is a principal axis of inertia, or else one of the components ω 2 , ω 3 is also zero. This implies that ω has the direction of a principal axis of inertia. Consider finally the case that ω 1 ω 2 ω 3 = / 0. Since our hypotheses imply that ˙ ω = 0 or else that ˙ω ω, we can always write ˙ ω i ω i = f (t), i = 1, 2, 3, (7.64)
and we make f (t) appear on the left-hand side of equations (7.54) by rewriting them in the form J i
2 i f (t) = ( J j − J k ) ω
1 ω 2 ω 3 , where {i, j, k} are the three even-order permutations. Summing term by term, we obtain 2T f (t) = 0, which yields that f (t) = 0, and hence ˙ ω = 0, and the rotation is uniform. 256 The mechanics of rigid bodies: dynamics 7.10 In conclusion, the right-hand side of all equations (7.54) vanish. This is compatible only with J 1 = J 2 = J 3 , which is the case that every line through O is a principal axis of inertia. The rotations considered in the previous theorem are called permanent rotations. Such rotations are associated with degenerate polhodes. In the case of a generic ellipsoid of inertia, there exists an important qualitative difference between the permanent rotations around the extreme axes of the ellipsoid and those around the intermediate axis. T heorem 7.3 The rotations around the extreme axes of the ellipsoid of inertia are stable with respect to perturbations of ω 0 ; those around the intermediate axis are unstable. Proof We use the geometrical analysis of the polhodes of the previous section, and the fact that these are the trajectories, in phase space, of equations (7.54) (Proposition 7.10). This shows that, for a fixed neighbourhood of the degenerate polhodes lying on the extreme axes, we can consider perturbations of ω 0 of such small amplitude that the corresponding polhodes remain inside the chosen neighbourhood. This is not possible for the degenerate polhodes lying on the intermediate axis, as every neighbourhood of such curves is crossed by polhodes which rotate around the maximal axis, as well as by polhodes which rotate around the minimal axis. In the case of ellipsoids of revolution we can easily prove that any rotation around the rotation axis of the ellipsoid is stable, while any rotation around the diametral axis is unstable. Remark 7.5 Stable phenomena are not the only observable phenomena. Try to make a cyl- inder, of height much larger than the radius, rotate around a diametral axis; in spite of any error in initial conditions, the rotation will appear stable. This is not in contradiction with what has just been proved: if the radius of the polhode is much larger than the radius of the herpolhode the contact point must rotate many times along the latter to make any tangible progress along the polhode, in agreement with the instability of this phenomenon, although such instability can only be observed over long time intervals. 7.10 Integration of Euler equations We consider again the first integrals (7.60) and (7.63). Eliminating once ω 3 and once ω 1 , we find the following equations: J 1 ( J 1 − J 3 ) ω
2 1 = 2T 0 ( J 0 − J
3 ) − J 2 ( J 2 − J
3 ) ω
2 2 , (7.65) J 3 ( J 3 − J 1 ) ω 2 3 = 2T 0 ( J 0 − J
1 ) − J 2 ( J 2 − J
1 ) ω
2 2 . (7.66) 7.10 The mechanics of rigid bodies: dynamics 257 In the generic case that J 1
J 2
J 3 we deduce that ω 2 1 = A 2 1 (ν 2 1 − ω 2 2 ), ω 2 3 = A
2 3 (ν 2 3 − ω 2 2 ), (7.67) with
A 2 1 = J 2 J 1 J 3 − J
2 J 3 − J 1 , A 2 3 = J 2 J 3 J 2 − J
1 J 3 − J 1 , (7.68) ν 1 = 2T 0 J 2 J 3 − J
0 J 3 − J 2 1/2 , ν 3 = 2T 0 J 2 J 0 − J
1 J 2 − J 1 1/2 . (7.69)
The dimensionless coefficients A 1 , A 3 contain information only on the geometric structure of the system, while the frequencies ν 1 , ν 3 are also determined by the initial conditions. We assume that J 0 = / J 1 , J 3 (we have already considered the case that these quantities are equal). Note also that k = ν
3 ν 2 1 = J 0 − J
1 J 2 − J 1 · J 3 − J 2 J 3 − J 0 (7.70) is greater than one if J 0 > J 2 and less than one if J 0 < J 2 . We exclude temporarily the case that J 0
J 2 . Using equations (7.67) we obtain from the second of equations (7.54) a differential equation for ω 2 :
ω 2 = ± A(ν 2 1 − ω 2 2 ) 1/2
(ν 2 3 − ω 2 2 ) 1/2
, (7.71)
with A =
( J 2 − J 1 ) ( J 3 − J 2 ) J 1 J 3 1/2 . (7.72) Note that the initial condition for ω 2 must be such that |ω 2 (0) | ≤ min(ν 1 , ν 3 ) and the same inequality is satisfied for |ω 2 (t) |. In addition, the constant solution |ω 2 | = min(ν 1 , ν 3 ) must be discarded; indeed if, for example, ν 1
3 , it follows that ω 1 = 0 (cf. (7.67)) and ω 2 ω 3 = / 0, contradicting the first equation of (7.54). 4 We can now compute the integral of equation (7.71) corresponding to the initial data ω 2 (0) = 0, and distinguish between two cases: (a) J 0 ∈ (J 1 , J 2 ), that is k < 1: t = ±τ 1 F ω 2 ν 3 , k ; (7.73)
4 Besides the constant solutions ω 2
± min(ν 1 , ν 3 ), equation (7.71) also admits non-trivial solutions which periodically take these values.
258 The mechanics of rigid bodies: dynamics 7.10 (b)
J 0 ∈ (J 2 , J 3 ), that is k > 1: t = ±τ
F ω 2 ν 1 , k −1 . (7.74) Here F is the elliptic integral of the first kind (cf. Appendix 2): F (z, k) = z 0
− η 2 ) (1 − kη 2 )] −1/2 dη,
|z| ≤ 1 (7.75)
and τ 1 = 1 ν 1 A , τ 3 = 1 ν 3 A . (7.76) The sign in (7.73), (7.74) must be chosen according to the initial conditions, and must be inverted every time that ω 2 reaches the extreme values (respectively, ± ν 3 and ± ν 1 ). The solution is periodic of period 4τ 1 K(k) (see Appendix 2) along the polhodes in the family described by J 0 ∈ (J 1 , J 2 ), and 4τ 3 K(k
−1 ) along those in the family J 0
2 , J 3 ). Finally, we examine the case that J 0 = J 2 (motion along the limiting polhodes). In this case the frequencies ν 1 and ν 3 coincide (k = 1): ν 1
3 ≡ ν = (2T 0 /
2 ) 1/2 . (7.77)
Since lim k →1 − K(k) =
∞ we expect that the motion is no longer periodic. Equation (7.71) can be simplified to ˙ ω
= ±A(ν
2 − ω
2 2 ), (7.78) where A is still given by (7.72). We choose the initial data ω 2 (0)
∈ (−ν, ν), since we are not interested in the extreme values, which correspond to permanent rotations. By separating variables we easily find ν + ω
2 ν − ω 2 = ν + ω 2 (0)
ν − ω
2 (0)
e ±2t/τ
, (7.79)
with τ =
1 νA = 2T 0 J 1 J 2 J 3 ( J 3 − J 2 ) (
J 2 − J 1 ) 1/2 . (7.80)
It follows that ω 2 (t) tends monotonically to ± ν, depending on the sign of ˙
ω 2 , which is determined by the initial conditions and by the second of equations (7.54). 7.11 The mechanics of rigid bodies: dynamics 259 7.11
Gyroscopic precessions In the previous section we integrated equations (7.54) in the generic case that J 1
J 2
J 3
the gyroscopic axis. Suppose that J 1 = J 2 = J and consider initially the simple case of precessions by inertia. Setting η =
J 3 J − 1, (7.81)
and excluding the trivial case η = 0, equations (7.54) become ˙ ω 1 = − ηω 2 ω 3 , ˙ ω 2 = ηω
3 ω 1 , ˙ ω 3 = 0,
(7.82) and hence the gyroscopic component of the rotational velocity, ω 3 , is constant, and the system (7.82) is linear ; ω 1 and ω 2 oscillate harmonically with frequency ν = |ηω
(0) 3 |/2π (7.83) (we refer to the generic case that ω(0) does not have the direction of the gyroscopic axis). We also find the first integral ω 2 1 + ω
2 2 = [ω 1 (0)]
2 + [ω
2 (0)]
2 . (7.84) It follows that the trajectory in the phase plane (ω 1 , ω 2 ) is a circle centred at the origin. The vector ω e with components (ω 1 , ω 2 , 0) is called the equatorial component of ω; it rotates uniformly around the gyroscopic axis with frequency ν. It is interesting to study the perturbations introduced by the presence of a moment normal to the gyroscopic axis. To illustrate the main qualitative properties of the motion, we consider a simple example for which the equations of motion are easily integrable. Consider the case of a moving torque, which we suppose has the same direction as x
2 : M(O) = (0, M, 0). (7.85) Again, ω
3 is constant and we must integrate the system ˙ ω
= − ηω
(0) 3 ω 2 , ω 1 (0) = ω
(0) 1 , (7.86) ˙ ω 2 = ηω
(0) 3 ω 1 + M/
J , ω 2 (0) = ω (0) 2 , (7.87) which is equivalent to ¨ ω
+ (ηω (0)
3 ) 2 ω 1 = −ηω (0)
3 M/ J , (7.88) ¨ ω 2 + (ηω
(0) 3 ) 2 ω 2 = 0, (7.89)
260 The mechanics of rigid bodies: dynamics 7.12 with the additional conditions ˙ ω 1 (0) = −ηω (0)
3 ω (0) 2 , ˙
ω 2 (0) = ηω (0) 3 ω (0) 1 + M/ J . Instead of (7.84) we find the first integral (ω 1
1 ) 2 + ω 2 2 = constant, (7.90)
with ω 1 = −M/[(J
3 − J ) ω
(0) 3 ], (7.91) and we can immediately deduce the integrals for (7.88), (7.89): ω 1
1 = C cos(2πνt + α), (7.92) ω
= C sin(2πνt + α), (7.93)
where C =
{[ω (0)
1 − ω
1 ] 2 + [ω (0)
2 ] 2 } 1/2
, (7.94)
tan α = ω (0) 2 ω (0) 1 − ω
1 . (7.95) We summarise the properties of the perturbed motion. (a) Amplitude and phase perturbations are measured by ω 1 . If initially |ω e |/|ω (0) 3 | 1 (7.96)
and if |ω 1 |/|ω (0)
3 | 1 (7.97) then, because of (7.90), equation (7.96) is satisfied at every time. If e 3 is the
unit vector of the gyroscopic axis, its variation is described by de 3 dt = ω × e 3 = ω e × e
3 , i.e. |de 3 /dt | = |ω e |, which implies that the motion of the gyroscopic axis is much slower than that around the same axis, and the effect of the torque M is smaller for larger ω (0) 3
(b) We note also that the vector ω e varies, with respect to the moving reference frame, with frequency ν, proportional to ω (0) 3
(c) Over a period of ω e the average of ω 2 is zero, while the average of ω 1 is ω
1 . It follows that by taking the average over a period of ω e de 3 dt = ω e × e 3 = M ( J 3 − J ) ω (0)
3 , (7.98) and hence the mean displacement of the gyroscopic axis is in the direction of the torque (tendency to parallelism). 7.12 The mechanics of rigid bodies: dynamics 261
3
2
1
w Download 10.87 Mb. Do'stlaringiz bilan baham: |
ma'muriyatiga murojaat qiling