Você está na página 1de 124
SOLUTIONS MANUAL TO ACCOMPANY SPACEFLIGHT DYNAMICS SECOND EDITION WILLIAM E. WIESEL Air Force institute of Technology The McGraw-Hill Companies, Inc. New York St.Louis San Francisco Auckland Bogoté Caracas Uicbon London Madrid Mexico Cily Milan Montreal New Delhi San Juan, Singapore Sydney Tokyo Toronto CHAPTER 1 Problem 1-1. As shown in the sketch, the angle between the 4, and s, axes is wt, q. Either by referring to the standard form for a z axis rotation matrix (equation & (1.26)) ox by calculating the ca (1.23), the rotation matrix G which rotates from the s frame to the i frame is cos wt -sin ot 0 rt o= {sin ot cos ot 0 ° 0 1 The alleged inertial velocity, written in the unit vectors of the s ftaine, equation (1.17), is -R(6 + @) sine vos R(S + w) cos @ 0 Transforming this to the inertial frame unit vectors requires the matrix vector product cos wt -sin ot 0 -R(8 + w) sin 6 ri v8 2 fsin wt cos ot 0 R(d + w) Gos 0 o ° L 0 -R(Gtw) cos wt sin @ - R(8+e) sin at cos 6 -R(6tw) sin wt sin @ + R(6#e) cos wt cos @ ° Using the standard trigonometric relations for the sine 1 and cosine of a sum, this becomes -R(S+0) sin( @ + wt ) vig pst ye = R(Gtw) cos( 8 + wt ) 0 Comparison of this last result shows that it is identical to equation (1.18), and that therefore both vectors represent the same velocity, expressed in the unit vectors of differént frames. Problem 1-2. The acceleration A isthe inertial derivative of the velocity vector V 4 ecco: A= tv However, since the vehicle is flying in a trimmed condition, the velocity vector has a simple representation in the unit vectors of the body frame b: V= Vb. Now, the first derivative relation (1.37) applied to this problem becomes 4 b ied eed apt A= Seve tev g ata y The derivative of V in the body frame is just Pavyat = ¥b,, since unit vectors of the b frame are treated as constants when a b frame derivative is calculated. 2 Also, by the right hand rule, the angular velocity of the b frame with respect to the inertial frame 4 is et. pe, since y increases ‘upwards', and this is the direction of a positive b, angular velocity. The cross product ts then B4x v= > vB, x by => Vb). So, the inertial acceleration of the vehicle can be written as A = Ub. We will find this result quite useful in the chapters /on reentry and rocket performance. Problem 1-3. A rotation matrix can be thought of as a matrix assembled out of the unit vectors of one frame expressed in their components in the other frame. So, the matrix which performs the rotation from the s frame to the i frame is | The columns of a rotation matrix then behave as the basis vectors of a reference frame: the sum of the squares of any row or column is one ( eg, | s; | = 1), while if any column is multiplied, element by element, with another column, and then added, the result is zero (eg, seit Oo, aad ). In particular, then, the rate of change of R** can be found by calculating the inertial derivatives of the S$, unit vectors. This is most easily done by referring the inertial i frame derivatives back to the s frame itself. Since 3 the s frame unit vectors are constant in their own frame, Fas j/at = 0. Their inertial rate of change is then or on ee If we interpret the cross product of a vector with a matrix as the matrix whose columns are the cross products with the columns of the original matrix, the three derivatives of the s unit vectors can be written as Problem 1-4. The angular velocity can be found by considering each angular coordinate in turn, using the right hand rule to find the direction of a positive rate in that angle, and then adding together the resulting individual rates. Since the angular velocity B* is a vector, its individual component parts add 4 like vectors. As shown in the sketch, the rates due to the rotation of the earth, w, and the longitude rate < add, and are aligned with the earth's polar axis, while the latitude rate is along the s, unit vector. The angular velocity vector can then be written ga. 8 x Be Ss, + (ager k in a hybrid set of basis vectors. Resolving the unit vector k along s) and s,, we have the angular velocity expressed in the s frame as ast. og . : ar = Ss, + (gt Xr sind s, + (ws +X) cos és, Now, the position vector of the spacecraft can be written as ros (RH) 5, where R is the radius of the earth, The inertial velocity of the vehicle is given by the velocity rule (1.37) as 4a Sa gst eet ger + Be The s frame derivative of r is simply calculated: Sar/at = fis), while the necessary cross product is aa $2 83 Bar 2 & (wethsind — (wegth)cosd ° RH 0 CREH) Cok) cos 6 + sy (RH) S The total inertial velocity is then 4g a ert 78 CR +H) (a+) cops + + os, (RH) S Problem 1-5. In order to stationkeep at the focal point of the radio telescope as shown in the sketch, the receiver spacecraft must match the angular rates $ and 6 of the reflector, while maintaining a fixed distance R from the mirror. Writing the velocity law (1.37) i a : adeaain the spacecraft will remain ‘stationary! with respect to the reflector spacecraft if Sar/dt = 0, since the s frame is anchored in the reflector. The required inertial velocity is then given by just the cross product term. The position vector is r = R sj. The angular velocity, as shown in the sketch, consists of a contribution -@ s, due to the ‘elevation’ motion of the 7? $2 mirror, and @ k due to its ‘azimuth’ motion. The complete angular velocity vector is then the sum of these two terms, of course resolving the unit vector k along the s, and s, directions. This gives B= psines, - ds, + dcoses, The inertial velocity is then given by 1 2 tg ast 5 A ; oer = xr = | ésine -8 $cos0 R ° ° As a check, notice that the s; component of this result is zero, as it should be if R is to be constant. Problem 1-6. The position vector of the astronaut can be written as PES ca Rie cue yiE In order to decide the ‘safety’ issue of what side to put the ladder on, we will want equations of motion for the astronaut expressed in the unit vectors of the s frame. The inertial acceleration of the astronaut is 442 S42 s Oe Te ea ek BH Bey at at The angular velocity vector &* = ws, by the right hand rule. The term in d&/dt is zero, since the spin of the station is constant. ‘The derivatives of r in the rotating s frame are simply a * i get eevee S42 a - : Apr ks + Fs The first cross product needed for the acceleration expansion is & x r: s s. Bx re : -eys, + ow 0 0 ° : i * $2 x y o Because r and “dr/at look very similar in form, the Coriolis term can be immediately written by comparison with the above gi Sa . : 2 x Ger = -2obs, + 2oks, The final cross product needed is the centripetal acceleration term SS) 83 (tae ye 0 0 « ~wy ox 0 2, ~x 5) - oy sy We now have all of the pieces of the inertial acceleration expression, Since the astronaut is really floating freely in free space, there is no net force applied to him. Newton's second law, F = m a then reduces to a = 0. In the s frame unit vectors this becomes wx ae x g &, : + tox - wty = 9 These are the equations of motion written in their components along the s unit vectors. To decide which side of the spoke should have the ladder, rewrite the y equation of motion as As the astronaut 'falls' down the spoke, he will have a Positive % (down the spoke) and his y value will remain small. This means that his ¥ will be negative, and he will drift over to the left wall in Figure 1-19, or the trailing side of the spoke. The ladder should be installed on this side. Another way to look at this is from the inertial frame. As the astronaut ‘falls’ down the spoke, he is really moving in a straight line at constant velocity seen from the inertial frame. As he drops. down the spoke, he finds that the rotational component of his inertial velocity, which was appropriate to the sedate rotational speeds in the hub, is no longer adequate nearer the rim. ‘The spoke thus rotates faster than he does, and will get ahead of him as he goes further down the tube, It is thus the trailing wall that the astronaut will hit. Problem 1-7. The baseball is to be located by the three coordinates r, z, and ¢, which together form a cylindrical coordinate system. The s frame in Figure 1-20 has an angular velocity Se (o- hb) Sy since the s frame rotates not only with the colony ( }, but-also with the ball ( ¢ ). The angular velocity of the s frame. can change if with time, so there is an angular acceleration vector 10 sqgst s “a the problem is to write the inertial in the s unit vectors. Once again, acceleration of the baseball The acceleration can be written ig? 5, sqast Peace eer . at ast eM Be ny ‘a + 28x Seer The position vector r from the origin on the rotational axis of the colony is where, obviously, r # |r]. The s frame velocity and acceleration vectors are then Sq : : eros Fs, + ga, Sq? i Pf “arr Ey + Be, We are now in a position to begin calculating cross products. The first cross product of the centripetal acceleration term is -8le-bis, oo ’ a The angular acceleration and Coriolis virtually identical to the above: terms are eaaet i waa re bs, asi Sar +. . 2 x ae per ee ae time ee 3 The final cross product necessary is for the centripetal acceleration term s s 1 2 3 Bt (Bt ary = 0 wo-¢ 0 0 0 =r (w-$) es =r (w- $b)? 5, The baseball is in free fall in a force free environment, so F = 0, Newton's second law then becomes gE - (a-$)? 2 = 0 ste 0 s. ré - 2(0-$)% = 0 Problem 1-8. The acceleration necessary to stationkeep at the focal point of. the antenna is found by taking the derivative of the inertial velocity found in problem 5 12 v= Rocoses, + Ros, still treating R as a constant. As an inertial derivative can be calculated by ta Sa gst we can calculate the inertial acceleration a by taking the s frame derivative of v and performing one more cross product. The s frame derivative of the velocity is Sq A +3 “av 7 (Rdcos@ - RbSsine)s, + while the necessary cross product is S aa 83 Bx ye sine -6 é cos @ 0 Récos@ RS = sy - RS? ~ Rw cos’) + 5 -Rgdssine) 1 + s3(R @ sin 6 cos 0 ) Combining the two terms, the inertial acceleration of the receiver spacecraft, at constant R, is = s)(-R 8? - R$? coste ) + so RGcos@ - 2RGdsino) + s,(R8 + R# sine cos 6) 13 Problem 1-9. As in the previous problem, we could obtain the inertial acceleration by calculating another derivative of the inertial velocity vector, which was obtained for this system in problem 4. However, in this case we will use the acceleration expansion (1.40) and calculate the acceleration from the beginning. Recall from problem 4 that the angular velocity vector is Bt = Ss, + (wgth) sin bs, + (wth) cos 6 sy and that the position vector r is CRHEH) Ss) The inertial acceleration is given by the acceleration expansion as 1,2 s,2 s. Agr 7 eee eee at at 28 de 4 Be La Py The s frame derivatives of r are given by 5, a : wae HS es The derivative of the angular velocity vector, the angular acceleration vector, is s, ge OT = Ss, + CK sins + (wg) S cos 6) sy + (Xcos 6 - (wth) S sin 6) sy 14 We are now in a position to perform the cross Products needed for the acceleration formula. Without going into detail for each, these intermediate results are BA p= ( -(RHM) (agth) coed ) 5, + (REDS 53 ast. gat _ Sg x (oe xr) os (RHH) (tA) 6 sind S} + (“CRED (gt)? cos26 = (RHH) 8?) sy + ( (RHR) (wgth)? cos 6 sin 6) 8, ost Sq * * we 2 X ger = ~ 2H (wet) cos 6s) + 2H Ss, Sa asi = sk ae ® xP = -(RHH)( A cos 6 ~ (gtr) 6 sin 6) Sy + (Rt) 3 sy Adding together these intermediate terms, the final inertial acceleration vector is = sy( (RH) X coss + 2(RHH)(WgHK)S sind ~ Awd) cosé ) i = (RM) (weth)? cons - (RH) 3? ) + s3( (RH) 3 + (RH) (Ot) 7c086 sind + 283 ) If the vehicle is acted upon only by Newtonian inverse square law gravity, then the force on the spacecraft is a5 eee - SHE s, (RH)? The equations of motion then become m( (RH) X cosd + 2(RHH) (WgHIS siné - 2itlwethicosé ) = 0 6m (rea)? m( fi - (RH) (agHK)? cos?6 - (RH) m( (RH) & + (RAH) (wgth)?cos6 sind + 28S) = 0 These are hardly the best variables to express the equations of motion for the two body problem. They are, however, occasionally used for aircraft flight over a rotating spherical earth. Problem 1-10. To determine if the spherical pendulum conserves energy, it is necessary to have the inertial acceleration a, the inertial velocity v, and the total force F, Since the radius r has zero rate of change in the s frame, the inertial velocity is given by v= @ yr = posines, + Les. : 1 2 which can be extracted from equation (1.56). The force F is given by equation (1.51) Fo = -mgsin@s, + (mgcos@-T) 8, and the inertial acceleration by (1.59) 16 a = s(bdsine +2446 cose) + s,(L8 - 2 ¢ cos @ sine} + s3( - 16? - 1 3? cin’e ) With a11 of the pieces needed, begin by taking F = ma and dotting both sides with the inertial velocity v. The acceleration side becomes myeas mi? ogsin?otme a 2 1 fe {tee 2, 1 2 92 sine + 3m? é } 1 It should not take much work to recognize that mv + a is a perfect derivative, since it must always be the derivative of m v+v/2, the kinetic energy. On the other side, F + v is Fey = -mgbtsingéd = ~fe {m9 5 cos o} This is the negative of the potential energy. Since both sides are perfect derivatives, the total energy E is a constant, and is given by e- 2 {123% sino + 126) - matics = ain mab cos To see if angular momentum is constant, we need only calculate the torque M on the system ay Mo= rxF = mgbsines, since r= Ls. Since the torque M is not zero, the total angular momentum H changes with time, and cannot be constant. However, by inspecting Figure 1-9, notice that s; is a horizontal unit vector. Since M can never have a nonzero component in the vertical direction, the component of angular momentum H in the vertical direction is a constant of the motion. Problem 1-11, From problem 4 we have the inertial velocity of the spacecraft, as EE voe-s, (RHH) (ogth) cos 6 + sof + os, (RH) S while the inertial acceleration is a = 8, -(RHH) X cosd + 2(RHH) (wgth)S sind 1 - dfi(wgtXrcosd ) + spt Hi - (RH) (ogth)? cose -, (RIK) 8?) + s3( (RH) S + (RED (OgtHK)?c086 sind + 248 ) 3 Performing the tedious calculation of m a + ¥, we obtain the result marty = (RH)? (wth) K cos?6 = (RHH)?(w gry? 3 cos 6 sin 6 2. 6 + hi + cre? SS + (RFD (ogth)? Hi cos + (Ren) A 8? To check if this is a perfect time derivative of the 18 kinetic energy, it is simplest and easiest to proceed on faith and calculate the kinetic energy, and then compare its time derivative to the result above. The derivative of the kinetic energy a age . etc Geaprvey 4a at m 2; conan = {+ (RH)? (ath)? cos*6 + EL + BL creny? 3? } Comparison of the derivative above will show that it is identical to ma v, as it must be. The force side of F = m a, when dotted with v, becomes GMS Mitc ea (Le8y Zz eee eee which is the negative of the potential energy for Newtonian inverse square law gravity. 80, since both sides of ( F = ma )+v are perfect time derivatives, the total energy is conserved, and is given by B= —F mren)? (wth)? cos?s + bm i? + mr? 3 - Problem 1-12. If the angular momentum about point p is defined as 19 N Poe in ry iL where r, locates a particle with respect to point p, and F, {8 the velocity of particle i with respect to point p, then the inertial derivative of HP is we The last step occurs since a vector crossed with itself is zero. Now, substitute for r, the triangle relationship = Rk - Rk where R, is the inertial position vector of particle 1, and R locates point p. Actually, we are still interested in calculating moments about point p, so do not explicitly substitute for r,, but instead substitute for its second derivative N ae Pie 7 ow Ym {ry eR - my x} ces In the first term, we can use Newton's second law to replace the factor mR, = F,, where F, is the total force on particle 1. In the second tezm above, notice that R is independent of the summation, and can be extracted. We obtain 20 y SJ yt te mr isl isl The first summation is the total moment of the forces about point p, N Pos wl Pu «Fy asl while the second summation is related to the position o£ the center of mass, r,, with respect to point p N eee ew hy is. The inertial derivative of HP becomes which was to be shown: Now, if point p is itself an inertial origin, then it will not be accelerating with respect to the inertial origin shown in Figure 1-21. In this case, & = 0, identically, and the extra term vanishes. Alternately, if point p is the center of mass itself, then r., which locates the center of miss with respect to point p, will be zero. In either case, thé derivative of the angular momentum about point p reduces to The major use of this relation is in the case of rigid bodies, where it is often not convenient to use an inertial origin as the reference point p. In this 21 case, the center of mass of the satellite is usually much more useful as a reference point, and can be used as if it were an inertial origin! Problem 1-13. The simplification of the tethered satellite equations of motion for small ér hinges on the expansion of the gravity terms in (1.101) and (2.103). The first of these is Now, the term involving 6x? is much smaller than the term with one factor of ér, so we can drop the last term in the bracket above. Also, the second term is still far smaller than the first, so the quantity can be expanded using the binomial theorem, to give 1m, -3/2 3 2 Ite pe {2 -2Beee} Similarly, the other term which appears in the denominator expands to give The equation of motion of the center of mass (1,101) then becomes 22 after substituting for ry and r, in the numerator, and continuing to ignore terms of order 6r” and higher. After the obvious cancellations are carried out, the result is This is the simplified equation of motion for the center of mass of the tether system. Comparison with chapter 2 shows that this is the equation of motion for the two body problem. 80, the above equation states that the center of mass moves along a classical conic section orbit. The expansion of the relative equation of motion (1.103) is similar, involving the difference of the gravitational terms tg ee : cee aaa Seti 3 fa EF re Mee wre 3m ret re Bm Pet Or ry 5 5 urd Hx2 After the obvious simplifications, the relative equation of motion (1.103) becomes 23 KS ee m tm, - rit fe a Every term in this equation of motion is of order ér Beye ore a or, a c 6 = 6 Hef 5 5 } in contrast to the equation of motion for the center of mass, where the terms in Sr canceled, and the terms of ofder zero in 6r added together. 24 CHAPTER 2 Problem 2-1. ‘To calculate the speed of a satellite in a circular orbit, all we need to do is take the circumference of a circle of radius r c= anr and divide by the period of the orbit, given by Kepler’s third law, equation (2.37) pe 2 pen ve where a = r in a circular orbit. ‘The only reason this works for a circular orbit is that the radius is constant, so by Kepler’s law of areas, not just the areal rate da/dt, but the actual speed v, is constant. ‘$0, circular orbital speed is given by vos & oe le ¥ - Escape velocity from radius r is given by equation (2.17) as : p Comparing this to the expression for circular orbital speed, we see v= f2 ee e since //2 ».1,414, escape speed is about 40% larger than circular that 23 orbital speed at the same distance r. Of course, to achieve a circular orbit, a satellite must not only have v = v,, but it must also be moving parallel to the earth’s surface. on the other hand, if a spacecraft has v = v__, the direction of the velocity vector is not as important. It will escape so long as it avoids colliding with the earth In a typical low earth orbit, r = 6578 km, which corresponds to 200 km altitude, circular orbital speed is 7 5 4.8369 mi/sec = 25,539 ft/sec While escape velocity is = / 2 x 3.98601 x 10° = 11.008 km/sec peaein.) 14. ramen = 6.840 mi/sec 36,115 ft/sec These are functions of distance from the earth. The speed of the moon in its orbit, at r = 385,000 km, is vo( © = 385,000 km ) oc 1.017 kn/sec +632 mi/sec 3336 ft/sec This speed is considerably smaller than low earth orbital velocity The corresponding values for escape velocity from the moon’s orbit are Vi, = 14389 km/sec = .894 mi/sec 4720 ft/sec Problem 2-2. The energy conservation law for the two body problem 24 takes the form (2.15) and (2.36) 1 paced Grade 2a In a hyperbolic orbit, the radius r can go to infinity. As it does so, the potential energy term (-u/r) above approaches zero, and the speed of the satellite becomes asymptotically constant, say v, Evaluated at ‘r=’, the energy law becomes e tv 2-H = 2 Ye 2a Solving for the speed at infinity, we have = fu fre =a The total energy E is positive in a hyperbolic orbit, so the seminajor axis a is negative, and both forms above are well defined. Problem 2=3.. Since the problen is to calculate the position of Halley's comet at your current date, the actual solution depends on the date you are doing the problem. In this solution, we will assume a date of October 29, 1988. The semimajor axis of the orbit is 17.9564 A.U. = 2.68625x10° kn. The value of u for the sun is yw, = 1.32715 x 10!) km/sec”. ‘the period of the orbit is then given by Kepler’s third law (2.37) as 20 a”? = 2.40127 x 109 sec 27792.5 days 76 years 33.5 days at 86,400 sec/day (exact) and 365.25 days/year (approximate). The date of the next perihelion passage is then approximately March 13, 2062. To calculate the position of Halley’s comet on October 29, 1988, We need to find the time interval since perihelion passage, t - 7, 25 Qhis is, for the given date, 993 days, or 8.579 x 107 seconds. The, mean motion of Halley’s comet is mia = 2.6166 x 107? radians/sec The mean anomaly is then M = n( t - 7) = 0.22447 radians. It is necessary to iteratively solve Kepler’s equation to find the value of the eccentric anomaly E which duplicates this value of M. Begin with the approximate solution (2.65) E = M +e sin M = .4397805 radians The Newton - Raphson method to solving Kepler’s equation begins with the calculation of the error AM, (2.66) mM, = EH, > e sin E, - M = -0,196507 rad and the slope of Kepler’s equation (2.68) as SE 1-ecos EB = 0,124745 Then, the correction to the current value E, is found from Ay, - aM, / aM/az = 41.5757 rad k This correction is added to the first value of E to find the new value E,. Obviously, the correction was rather large, so the Newton -| Raphson method cannot be said to have converged. The table shows the next six iterations of the method. Convergence is slow at first, but| then becomes extremely rapid once the method has found the vicinity of! the true answer (about iteration 4). Once this happens, the convergence is quadratic, as can be seen by the approximate doubling of the number of leading zeros in the third and last columns of the table. 26 1 14397805 | ~0.196507 | 0.124745 1.575270 2 | 2.015054 0.917182 | 1.415733 | -0.647849 3 | 1.367204 0.195414 | 0.804423 | -0,242924 4 | 1.124279 0.027348 | 0.582295 | -0.0469671 5 | 1.077312 0.000954 |.0.541794 | -0.0017624 6 | 1.075550 9.0000013 | 0.540293 | -0.0000024 7 | 1.075547 0.0 he value of E must then ke converted into a true anomaly using (2.70) tan (v/2) =f tan ( B/2) qhis gives ‘v = 2.715791 radians, Inserting this into the conic section equation (2.71) a(_1- e”) Bee ee eer 9.70167 A.U. 1 +e cos v or, about the distance to the planet Saturn. The case of Halley’s comet is extreme, since theeccentricity of the elliptical orbit is quite high. The initial approxination (2.65) ie of relatively little value, and relatively many iterations of the Newton - Raphson method are needed. For a much smaller eccentricity, this is not true. Also, for a small eccentricity, the three anomalies M, E and v are much closer together than they are in the case of Halley’s comet. problem 2=4. As shown in the sketch, the angular momentum h can be written as H = xr vcos ¢$ Now, the tangential velocity component is v, = V 08 $, so this 27 component of the velocity vector is simply given by ‘The energy relation (2.16) gives the total speed as vis f 2B + au/r he radial velocity component r = v, = v sin ¢. This can be written, using cos ¢ = H/rv, as . > 2 Yy v fi -cos*@ = v fi - 4S r’y a z 2 S/o tei eee ap + HO. HE r Fs Still other forms of these (and most other) formulae for the two body problem are possible. To a large extent, the use of one form over another is up to the preference of the analyst. Problem 2=5. Examining Figure 2-8, the p component of the radius. vector can be expressed as -aep (the distance from the’ focus to the center in the -p direction), plus the base of the triangle, +a cosE p. The q component of r is the altitude of this triangle, a sink, reduced by the ratio b/a, since the ellipse is the auxiliary circle seen in projection. Equation (2.30) gives this ratio as b/a = ,/ 2, so 28 Fas (® Key the radius vector is r= a( cosE-e) p + a / sinE q To find an expression for the scalar radius, square and add the two components of r 2 2 x? = a*( cose - 2e cose + e” + sin"E - e’sin*E Replace cos"E + sin"E with 1, and in the last term replace sin? with 1 - cos"E, to find r? = a%( 1 - 2 cosk + oe” ~ e? + e%cos"E ) = a?( 1 - 2e cosk + e%cos’R ) 2 = (a0 1- © cose ) } or, r= a( 1-ecos E) Finally, to find an expression for the velocity, take a derivative of the expression for r v= -asinERp + a/i - cosE Eq Kepler’s equation ( equations (2.43) -(2.45)) is a He - = - i ™ SS ( t- 3, } E ~ e sink 29 Taking an implicit derivative of this expression, we find Hy = (1 -e cosk) & a Solving for & yields E 1 z / i __ _ @ cose Finally, this result is used to eliminate the offending & from the expression for the velocity, to find = —YH/a___f _ gs - e v qe sinEp + /1- 7 coskq Problem 2-6. The approximate solution procedure began by assuming e=0, soE® M, Substituting this into Kepler’s equation in the form E M + @ sinE led to equation (2.65) E+ Mtesinl To continue this process, substitute the above approximation for E into the right hand side of Kepler’s equation, E=M +e sing, to obtain Bow u+esin{w+esina } { function. This form is rather messy, and includes the unfamiliar sine of a sine If we apply the trigonometric identity for thesine of a sum, the above takes the form Ee Mm + of sintl cos( e sin} + cost sin( @ sink d} Now, in this form we can profitably apply the assumption that the 30 | orbital eccentricity e is small. If e << 1, then the cosine and sine above can be approximated as cos{ e sinM } R 1 sin( e sink } = e sinM ) These are just the usual small. angle approximations of sine and cosine, where the eccentricity e ensures that the ‘angle’ remains small. Inserting these into Kepler’s equation, we have Es M +e sinM + e” cost sint = ont esint + Lo? sin 2m This is the desired result, involving powers of the eccentricity and sines of multiples of the mean anomaly M. This process can be continued to any order of approximation. Problem 2-7. Any function of time can be expanded as a Taylor's series ae 1 ae act) = e(t,) + SE] (ety + pee] oe . t at? le 0 0 even when f(t) is a vector function of time. If we identify the function f£ = x, the position vector, then the first derivative df/at = v is the velocity vector, while the second derivative d7£/at? = a is the satellite’s acceleration. This, of course, can be obtained from the equations of motion for the two body problem, a = -ur/r?, Evaluating all of these quantities at time t,, and inserting into the Taylor’s expansion, we have . 2 ~t.)? Bye e lita) ear cess oC ee oda tate 0 x(t) = x9 31 If we have two position vectors r, at time t,, and r, at the slightly later time t,, we can interpret the right side of the above to be x,, and solve for the velocity vector rc, Me eer eee 0 x, Hw x °° Yo (tp - ty) The first term above is, of course, the straight line approximation, while the next term corrects for the acceleration of gravity. The expression on the right side can be evaluated in terms of known quantities t, t,, ry, ry. , With an approximate solution for the velocity vector v,, the orbital elements can be calculated. Of course, the Taylor’s series is an infinite series, and better results can be expected if we include more terms. The next term in the series requires the third derivative of r cee ee at This again can be evaluated in terms of the initial conditions r,, vy. However, if we include this term in the Taylor’s series, the result is no longer a simple linear function in the initial velocity vector v,, and solving for v, would require approximation techniques. Problem 2-8. Both radiation pressure and gravity obey inverse square 32 laws, with opposite signs. While gravity is attractive, radiation pressure is repulsive. If a solar sail spacecraft in orbit about the sun keeps its sail flat - on to the sun, the only forces are radial. (This is not true if the sail is angled to the radius, and in this latter case the solution is more complex.) Applying Newton’s second law, including gravity and radiation pressure = we 28a r bee fe oa 2 = 28a } xr - fp {w= 2B} ay This has exactly the same form as the equations of motion for the two body problem, with a slightly different ‘yu’. If we write the effective gravitational parameter as He = - 2S8A/om, then the equations of motion become which has exactly the form of the equations of motion for the two body problen. Since the equations of motion have exactly thesame form as the standard two body problem, the solution to the problem is the’ same as the usual two body problem, with only one small change. The effective fg value can take on either algebraic sign. The case where yu, > 0 (eg., M > 28A/on) corresponds exactly with the usual two body solution, except that the ‘sun’ is somewhat less massive. All of the usual types of conic sections are possible; including the circle, ellipse, parabola, and hyperbola. Tf u = 28A/om, then the effective u, = 0. The equations of motion reduce to This immediately integrates twice to give 33 x(t) = ry + vot a straight line. In this case the repulsive force of radiation pressure exactly balances the attractive force of gravity. Since both are inverse square laws, this remains true at all distances from the sun, and the only permissible conic section is the straight line at constant speed. A spacecraft with such a solar sail can ‘turn off’ the sun’s gravity by unfurling the sail. If the effective u, < 0, the pressure of sunlight on the sail overwhelms the gravitational force, and the effective acceleration of the vehicle is always outwards, away from the sun. In this case the vehicle always escapes, and the only permitted conic section is the hyperbola. In fact, it is the branch of the hyperbola normally forbidden in the usual two body problem which is the only permitted trajectory in this case. To see this, write the energy law for the two body problem Since 4, is negative, the total energy is always positive. solving for the speed v, < 0 2u, 1/2 {2 + 4s} is actually greatest at x = », since u, < 0. In this case, the spacecraft always escapes. If the sail is first unfurled when the vehicle has speed v, at distance r, from the sun, then 2 2 = vo and at r =o, the spacecraft will have speed 34 Problen 2=9. Start with the equations of motion for the N body problen (1.131) and, as suggested in the problem, interpret i = 1 as the earth, i = 2 as the satellite, and i = 3, 4, etc the rest of the solar system. When we make this change, a significant reduction happens in the order of the system. Since the satellite does not effect the motion of the other objects in the solar system, we can treat the vectors rj(t), i#2, as known functions of time. At worst, we will have to solve for the motion of the rest of the solar.system first, and then we can find the motion of the satellite. Now, since the vector r = r, - r, is the vector from the center of the earth to the satellite, we can find an equation of motion for r by simply subtracting the equations of motion for the satellite and the earth: Kem, (4, -3y) pin ry) See ype eee eee cece eee eee eee 2 bey? [x ae 2 5! ja a 1 j ‘The first summation has. a j=1 term, but no j=2 term; while for the second summation this is reversed. These two terms are the two body terms in the equations of motion: the acceleration of body 1 due to the gravitational attraction of body 2, and the acceleration of body 2 due to body 1. When these two terms are written separately 35 Gm, (4, ~ 2) I> d 1 -G(m +m, using the definition of r =r, - r,. But G(m, + m,) =H, so these are indeed just the terms needed to duplicate the two body equations of motion. The remaining sums now both begin with j = 3, and can be brought under one summation sign to give = - HE 2 N ee - Yen, - 353 aj where rj) = | x, - Yr, |, etc. this equation of motion for an earth satellite includes the effects of the rest of the solar system. By examining the extra terms, it is possible to see under what conditions these terms can be ignored. That is, under what conditions is the two body problem a valid approximation to the real equations of motion? First, if my is itself very small, then its term is negligible in the equation of motion, -A particle of dust orbiting the earth is not likely to produce significant terms in the equation of motion for another earth satellite. A second way in which the additional terms can be made negligible is to make the denominators r,, and x, very, very large. A star (other than the sun) would be a such a vast distance from the earth satellite that it would produce negligible effects on its notion, even though its mass mj would be very large. However, neither of these first two excuses are sufficient to eliminate the effects of the sun afd the moon-on an earth satellite. The moon, in particular, has a mass about 1/81 the mass of the earth, and is only 60 earth radii away. The sun is somewhat further out, but 36 is substantially more massive than the earth. But both of these objects produce negligible effects on the motion of an earth satellite if it is close enough to the earth so that rj, ~4r,;. This eliminates the extra term in the equations of motion by subtraction of two nearly equal terns, both in magnitude and direction. The two body problem is a valid approximation for the motion of a satellite close to the earth, but ceases to be valid when the satellite is a substantial fraction of the distance to the moon away from the earth. In this case the restricted problem of three bodies is a much more valid approximation to the equations of motion. 37 CHAPTER 3 Problem 3-1, The period of the orbit is given as T = 23" 56” 4.09% 86164.09 seconds. The radius of a circular orbit with this period can be found using Kepler’s third law (2.37) 2/3 aye | So = 42164.18 kn 2m : A Hohmann transfer from a low altitude 28° parking orbit to geosynchronous equatorial orbit is perhaps the most commonly performed transfer. Using a, = 6578 km as the radius of the parking orbit, the semimajor axis a of the Hohmann transfer ellipse is = 24371.09 km Then, using the notation of section 3.2 and Figure 3.1, the speed of the spacecraft in the parking orbit is given by the circular orbital speed at r = ay 7.7843 km/sec The required speed in ‘the transfer ellipse at perigee is given by equation (3.2) evaluated at r = ay a2 = au = y= { ard f } 10.2389. km/sec The difference between these two speeds gives the first Hohmann maneuver Avy, = vy, 7 y, 4 2.4546 km/sec ca The calculation of the second maneuver is somewhat similar, but we must perform the inclination change at apogee. Section 3.3 shows 38 that inclination maneuvers are much more expensive when performed at points where the satellite already possess a high velocity. The required final speed is just the circular speed at geosynchronous distance, r = a, = 3.0746 km/sec At apogee on the transfer ellipse, the spacecraft will have speed 1/2 = ee eee mee = v2 { 3 4 } 1.5973 km/sec As shown in the sketch, these two velocity vectors differ by the 28° difference in the planes of the orbits. The cheapest way to (i) circularize the orbit, and (ii) change the orbital plane is to do both as one maneuver by drawing the Av vector between the tips of the initial and final velocities. The required maneuver is then found from the cosine law for plane triangles as 2 ae 2 avy = { 2. + Vb - 2 Vo2¥, cos 28 = 1,8254 km/sec Any other strategy would be more expensive. For example, circulariaing the orbit first, and then performing the plane change is equivalent to following the dotted path in the sketch, and this will clearly be more costly. Problem 3-2, . The ascent trajectory flown by an Apollo Lunar Module 39 (LM) is the moral equivalent of a Hohmann transfer, where there is no lower circular orbit. Instead, the powered flight trajectory terminates directly on the ascent ellipse. The semimajor axis a of the ascent ellipse is a = (1740430) 3 17404250) | _ _1770 + 1990. i999 Km 2 where a= 1770 km and a= 1990 km. We will need the value of Hy for the moon. since 4 = GM, and the mass of the moon is 0.01213 the mass of the earth, u,, = 0.01213 n, = 4,.83503-x 10° km?/sec’ The burnout speed of the LM at perilune is the required velocity in the ellipse when r = a, 2u, Uy 1/2 yo { —- =} = 1.7004 km/sec The situation at rendezvous is the usual Hohmann transfer condition, since the LM must make an actual maneuver at this point. The speed of the Command Module (CM) in its circular orbit at r = aa is = 1.5587 km/sec 2 while the speed of the LM in the ascent ellipse at apolune will be 2H, My y1/2 = mae = x { = 7 } 1.5124 km/sec The required second maneuver is then Av, = 1.5587 - 1.5124 = 0.0462 km/sec = 46.2 m/sec. This is just slightly over 100 mph. In order to have the LM and CM come together directly after the second Hohmann burn, it is necessary to time the liftoff correctly. The analysis of section 3.4 does not apply, since the moon rotates , very slowly ( once per 29 days ), and Gorrections for rotation of the | moon would have been made by the CM. ‘The ascent trajectory is one half of an elliptical orbit, so the time which elapses from burnout to | rendezvous is given by Kepler’s third law divided by 2: 40 a} = 8021.5 sec of one orbit of the CM. If the CM is to be near the rendezvous point when the LM arrives there, it must be located 0.45911 x 360° = 165¢28 back from the rendezvous point at LM burnout. In other words, it must be 180° - 165928 = 14972 ahead of the LM burnout point at LM burnout. As a time difference, the CM must pass the burnout point 2, 14972 s n = 327.9 sec = 5" 28 3600 CM at before LM burnout. This is not the difference between the LM liftoff time and the passage of the CM over the landing site, since the actual dynamics of the powered flight trajectory must be known before this time can be calculated. Problem 3-3, The external tank is left in the tank disposal orbit, with an apogee of 100 km and a perigee of only 30 km, to ensure that the tank will burn up. As shown in the sketch, the semimajor axis of the tank disposal orbit is then a,4, = ( 2 x 6378.135 + 100 + 30 )/2 = 6443.135 km. (Of course, 6378.135 km is the radius of the earth. After the first maneuver, the shuttle orbiter will be in the ascent ellipsé, with perigee of 100 km and apogee of 250 km, The semimajor axis of the ascent ellipse is then a, = ( 2 x 6378.135 + 100 + 250 )/2 = 6553.135 km. Before the OMS1 burn (Orbital Maneuvering System 1 burn), the 41 OMS1 ascent 250 kan ellipse orbit OMs2 shuttle is in the tank disposal orbit, at radius r, = 6478.135 kn. Its speed before the maneuver is v, = 7.82278 km/sec a 1, before aeao Just after the OMS1 maneuver, the shuttle is located at the perigee of the ascent ellipse (which is of course the same spot), but now in the ascent ellipse with speed = 788887 km/sec 2 a, 1, after ‘a he difference between these two speeds is the OMS1 maneuver, so 0.06609 km.sec 4v, = Ya after ~ Y1,before = 66.09 m/sec This is an increase in speed. The OMS2 maneuver is the standard second maneuver in the Hohmann transfer. This is at radius r, = 6628.135 km. The velocity at apogee in the ascent ellipse is 2 Beeeee H V2 before * ( Fy aa = 7.71034 km/sec pe The speed after the OMS2 burn is just circular orbital speed at r = 42 So u 1/2 Y2,after ( Tr, } = 7.75485 km/sec So the maneuver is Ay, 0.04451 km/sec 2 7 Ya,atter ~ Y2,after = 44.51 m/sec Again, this maneuver is to speed up. Problem 3-4. ‘he acceleration of the vehicle is specified as A =-AWw, -90° < v < 90° A= AW, 90° < v <.270° where w is the orbit normal vector. From Figure 3-11, ' the radius vector r can be written as r= acosvp:'+asinvg Since this is a circular orbit, the unit vector p can be pointed along the line of nodes, while q points at the place on the orbit furthest above the equator. The vehicle acceleration program produces a specific torque ( torque per unit mass ) of Mo= rxA = tAacosyq #aAasinvp where the upper sign pertains to -90° < v < 90°, and the lower sign to the other half of the orbit. Now, (specific) torque M and (specific) angular momentum are related according to 43 The change in the orbital angular momentum can thus be found by integrating the torque over one orbit. T aq = fom at o However, our expression for the torque is in terms of the angle vy. To | change integration variables from time to vy, notice that in a circular orbit ) = n = /u/a®, where n is the orbital mean motion. ‘The integral above then becomes ae an wf ae Oy oO u 0 Inserting the expression for the torque, and separating the integrals over the two halves of the orbit, we find 90° aH = a? J (aa cos vq - Aa sin y p ) av # 50° 270 : a J (- Aa cos vq + Aa sin vy p-) av } ° 90' Performing the integrals leads to 90° 90° aH = Aa [® {+ einv | + cosy |p fs 90° 90° 270° 270° - sin | a - csv|p } 90° 90° = 4aa / q u Any change of the angular momentum vector along the q direction will change the orbital plane. As shown in the sketch, a change of H from its original position to the new H + AH will produce a change AL 44 HAH in the inclination. If Ai is small, then Ais sini = Lal [| We have the magnitude of AH from the calculation above, while the magnitude of H comes from the two body problem (2.33) as since the orbital eccentricity e = 0. This gives the change in the orbital inclination as anavay . 4a7a i Va Although we have not proven it, this is actually the optimal (minimum time) way to change the inclination of a low thrust spacecraft. Ais Problem 3-5. The circular orbital speeds are determined solely by the radii of the initial and final orbits a iE Yer / Similarly, the perigee and apogee speeds in the transfer ellipse, and v, respectively, are given by 1 ¥ 5 {& 7 ay? pa Yo,a a 45 3 Since these are fixed, the only question that remains is where is the best place to do the inclination change. The sketch shows the two velocity triangles, assuming that inclination change Ai, is for r, = a, and r, = a Ay, : Yen % ol Avs) LP va tA performed at the first maneuver, and Ain i- Ai, is performed at the second maneuver. Applying the cosine law of plane trigonometry to each triangle, the total velocity change is given by 2 2 Aaa hes av = Av, + Av, { Vert YB 7 2 Voy Vp cos Aiy } 2 : - yaya + { Yog +. Ya 7 2% 5 Vy cost 1 = Ad, ) } Since pr Yor? Var and Yon are all fixed, this is a function of the one variable which is at our disposal: Ai,. To minimize Av with respect to Ai,, take itsderivative and set it equal to zero Vg, Vp sin Ai o = dav 8 ca Yp a aay _ 7 _ \i72 { ort Yo 7 2 Voy Vp 808 ALy } if Von Yq Sin( i - Ai, ) 2 2 : aaa { %og + Va ~ 2 Yoq Yq cos( i - Al, ) } This is a very nonlinear function of Ai,, and its zeros, if any, give | the optimal inclination changes at the first and second maneuvers. Although solving the above equation for the optimal Ai, is not a 46 | trivial task, it is simple to show that Ai, = 0 (as considered in problen 3-1) is not the optimal solution. Evaluating the derivative | at Ai, = 0, we have ; ae = Vg Vg sin( 4) : i782 Se ate vag eva etiealvaaslv a oouCl4) this is only zero if i = 0, in which case there is no need to perform any inclination change at all. So Ai; = 0 is not optimal, although it is nearly so for the case considered in problem 3-1. Problem 3-6, During one orbit of the earth, taking time T given by Kepler’s third law, the earth rotates AA = w, T, where w, is the | rotation rate of the earth. Using the expression for Kepler’s third law, this becomes 21 w, a’ As shown in the sketch, divide the swath of the earth visible from the satellite in two, and notice that each side is a right triangle. Since, from either triangle, cos 0/2 = R,/a, the swath visible from the satellite has angular width The entire surface of the earth will be visible from any orbits a7 where © < Ad. Both of these-angles are functions of the orbital semimajor axis a, and they are shown in the plot. or satellites below about 116 km altitude, the swath visible from the satellite is smaller than the spacing between ground traces, and the spacecraft will no longer cover the entire earth within one day. Actually, the areas near the edge of the deg Seg gege cee Dae eee 0 200 400 600 800 1000 H, km 0 102030405060 earth (the ‘limb’ of the earth) are highly compressed and far away, both of which limit the, usefulness of images of the edge of the earth. Radio communication is not so badly affected long slant distances. Problem 3-7. Obviously the first condition is simply a) e-= 0. Then, using the expression for the nodal regression rate (3.53), the condition that the satellite be sun-synchronous is 2 ow 262 Rog i = -2.E_kadians = 32 cos i = 2H _tadlans 2a 365.25 days We have substituted that e = 0, and replaced the mean motion n with its equivalent in terns of a, n = ( u/a*)‘/?, Now, the ground track slips an amount in longitude of 48 3/2 AA = Va in the course of one orbit. After N orbits, we want the ground trace to close on itself. This means that the total slippage over N orbits, N AA, should be a whole number of revolutions, say 2 1 M, where M is an integer also. Then our third condition becomes 2nw, ad/? Naa = ——2—_w = ann vu Recalling the definition of the mean motion n, this states that Us)" =e = aes : : In words, this says that the orbital frequency (or equivalently period) should be a rational multiple (N/M) of the earth’s rotation rate. This supplies a value for the somimajor axis a. ‘Then, returning to the sun-synchronous condition, the only remaining unknown is cos i. It is possible to have a curcular orbit which is both sun-synchronous and whose ground trace repeats virtually exactly. Problem 3-8. The energy relationship (2.15) and (2.36) is During an impulsive maneuver, both v and a will change, but the position of the satellite is fixed, so r does not change. Calculating differentials of the energy law for small changes in a and v, we find = da Vo av = tae ‘o This can be solved to yield the velocity change as 49 av = —#—~ aa 2 v5 ag Now, the mean motion of an orbit is the average angular rate the satellite moves with respect to inertial space. If the original orbit was geosynchronous, then, the original mean motion n, = w,, the rotation rate of the earth. After the maneuver, the satellite has seminajor axis a, + da, and its mean motion is correspondingly changed. The average rate at which the satellite will drift in longitude is now A =n - w,, or | z a / 0 (a, + da) For small da, the denominator of the first term can be expanded as a binomial series to give oo The first term is the nominal mean motion in-geosynchronous orbit, no, and it cancels the term - w,. The remainder of the expression is os ag’? da Notice that this gives a negative (westwards) drift for a positive da. Since a positive da is generated by a positive av, a maneuver increasing the satellite’s speed ( positive dv, or a maneuver to the east ) will cause the satellite to drift towards the west. This is understandable if we remember that an increase in the satellite’s speed will increase its semimajor axis a, and this will decrease its average angular rate, allowing the earth to gain on the satellite. Alternately, slowing the satellite (maneuvering to the west) will drop the satellite into a slightly smaller orbit, where its period will be shorter, and the satellite will gain on the earth’s rotation, causing the drift to be eastward. 50 Problem 3-9, The alternative strategy given in the problem will require a first maneuver equal to the difference between circular orbital speed and escape speed from an orbit at radius a,. This is Zi z wy, yt - ({£ 1 aso ~ Yor Viney / a, laa fe = ( 2 fe } ay The second maneuver has zero cost, while the third maneuver from a tangent parabola into the, second circular orbit will involve a Av, of wy (JF -1)/E This is to be compared to the standard Hohmann transfer. For the case given, the lower orbit has a radiusof a, Gy 0869-47 km, while the moon is at a distance of a, = 382688 km. The Senimajor 194628 kn. The standard Hohmann axis of the transfer ellipse is a transfer maneuvers are given by (age) Ca)" = 10:9225 km/sec - 7.7894 km/sec = 3.13312 km/sec for the first maneuver, and avy = = 1.02057 km/sec ~ 0.18748 km/sec = 0.833096 km/sec for the second. ‘The total Av is thus 3.96621 km/sec for the Hohmann transfer. ‘The corresponding figure for the escape - and - return method is 51 wv = {/o - 2} { 7.7006 + 1.02007} = 3.64920 km/sec which is easily computed, since the circular velocities were already calculated as part of the Hohmann method. The escape and return method is cheaper, in this case, than the Hohmann transfer. ‘That this is not always the case can be seen from the accompanying figure, where the Av cost of the Hohmann transfer is compared to the escape - and - return method for 10 < a,/a, < 15. The crossover point is at about a, = 11.8 a,, so the Hohmann transfer is cheaper so long as the ratio of the radii of the circular orbits is not too large. + 0 oS ® Av Fa oa ° Oo - BEE Eee EERE e HEE Re eee eee Peo "10 12 14 16 18 20 a, /a, Some Additional Problems A) The shuttle deorbit maneuver is a Hohmann transfer from an orbital altitude of 250 km to an altitude of 40 km, where reentry begins. No second maneuver is performed: the atmosphere takes care of the rest. Calculate the Av for the deorbit maneuver. B) In space, a rocket doesn’t need a thrust to weight ratio greater 52 than one, so it is tempting to consider a rocket with a large fuel tank and a very small rocket engine. This saves the weight of a larger engine, but may require burn times so long that a standard Hohmann maneuver can not be done in a short orbital arc, and our impulsive burn assumption is violated. Consider doing Hohmann transfer maneuvers "a little bit at a time", over many orbits. Argue that repeated small maneuvers at perigee’ until the usual transfer ellipse is obtained, followed by repeated small maneuvers at apogee to circularize, is fully equivalent to the usual Hohmann transfer. It just takes longer. But consider alternating a small perigee burn with a small apogee burn, in effect building one large Hohmann transfer out of a series of small ones, Write a computer program to calculate the total Av of a series of small perigee-apogee~perigee~apogee....maneuvers. How does this method compare to.the usual Hohmann transfer? To the low thust spiral transfer? 53 CHAPTER 4 Problem 4-1 The three expressions (4.67) - (4.69) are linear in the angular rates $, 6 and J. ‘These three equations can be rewritten as the one matrix equation sin @ sin w cosy 0 a 4 sin @ cos -sin ~ -0 é] = cos 0 o aisle 6, These equations can be solved by Cramer’s rule of determinants. The first determinant we need to calculate is the determinant of the system matrix itself sine sin y cosy 0 D = | sin@ cosy -siny 0 cos @ 0 2 sin @ sin y cos wy 1 sin @ cos ¥ ~-sin w = -sin@sin*y - sine cos*y = - sine expanding by minors of the last column. Since the determinant is D = -sin 6, D » 0 when @ » 0 or 180°, and no solution for the three angular rates is possible under this condition The angular rates can now be found as the ratio of two determinants: the numerator determinant with the appropriate column replaced by the right hand side, divided by the denominator determinant D. The first angular rate is ee peda eee ances 1 [% cosy icles sala (cea cpateetn| tien el Be ate 1 54 sin + @, cos 2 sin @ which is undetermined at @ The angular rate 6 is sin@siny 0, 0 eae 1 6 = | sine cosy ow, cos @ 0, sino siny wu, ‘ a D | sin e cos y cos W ~ w, sin w This remains well determined even when @ » 0. Finally, the last angular rate is sin @ sin yp cosy wy b= —4—| sino cosy -siny o, oe cos 6 oO Os _ sin y cos @ cos cos @ sin 6 oy + sin 6 Lyn 3 which is not finite when @ = 0 or 180°. Notice that the two angles which have difficulty, # and y, share a common dividing line: which becomes undefined when @ = 0. Problem 4-2. The statement that when expanded in the principal axis body frame becomes Euler's 55 equations. If instead we use the fact that H 1 opt and expand the derivative in the inertial frame, the result is i i 4 gee ae(ana pi a_ gbi Peddie { ae 7 } ae eo The inertial derivative of the angular velocity vector needs no explanation. However, the derivative of the moment of inertia matrix is most easily found by using the derivative expansion rule b, Gets Ger + BP I The derivative of I in the body frame b is, of course, zero. While the cross product of a vector with a matrix is not defined, the alternate interpretation of the cross product as a matrix multiplication (see Appendix A) leads to the correct result 0 ®, is wy r= ~ I fa ae ~w, 0, 0 Finally, inserting this back into the torque equation gives x {ac x pais rape While this result is correct, it is rarely if ever used. The value of the moment of inertia tensor I in the inertial frame must be used in the above equations, and this forces us to integrate the equation of motion ¢ = 8°! x 1 as well as the moment equation. ‘This increases the order of the differential equations by 3, and is usually unwarranted. Problem 4-3. Write the position vector of a mass element dm with respect to an inertial origin as R,, +r. Then a time derivative 56 with respect to the inertial frame gives i i i a ( } ‘a a v= 4 (aver Sao, t cher ae ¢ Reon at Room ae In the body frame the vector r locates the mass element dm. Since the body is rigid, the body frame derivative of r, Par/dt = 0. ‘then, using the velocity law bi v= Mogg + BPE The kinetic energy can be written as = z bi bi t= EE (cont Bx }e( Yoon + 8 xx} an Expanding the dot products, we obtain three integrals 1 sbi t= > J von * Yoon ® + J Yoon * (BPA x = | an + J (ah ce )o[ ahh ae | am In the first integral, Yoon is independent of the integral and can be removed, The remaining integral is fdm = M, the mass of the body. The first term then becomes M Yoon ° Yoon/2 the kinetic energy of the center of mass as if the body were a mass particle. In the second integral, both the center of mass velocity and the angular velocity can be removed, leaving 87 J veo * [Be] an = vegg Bx fe an But, as we have seen several times, the integral f r dm = 0 when r is measured from the center of mass. So the second term vanishes. The third term is just the integral in equation (4.29), and this integral gives the usual rigid body kinetic energy tern (4.32). So HE 1 sbi obi pict ver Ruia(e-laea)-. Veeder ieadeaeeeset Problem 4-4. ‘The pivot point on the side of the shuttle payload bay is a fixed point in inertial space. From an origin placed at this; point, the center of mass of the satellite is located at r = -r by, and the satellite revolves about the pivot point with angular velocity! @=wb,. The usual velocity expansion then becomes vie Bx r = ob, x(-rb,) = orb The acceleration about the center of mass can then be calculated by applying the velocity expansion rule to the velocity vector above, using the b frame: b, = Ja pb a= ev ¢ BM x v ‘ 2 = orb, + w rb, 58 } However, the satellite is not free to accelerate in the by direction, so we will concentrate on the vertical (b,) motion of the satellite. The horizontal motion is forbidden by another constraint force at the pivot point. Now, the center of mass of the satellite moves as if all of the external forces acted through that point, and all of the mass of the satellite were concentrated there (equation (4.5)). Thus, since the center of mass obeys F = m a, and the two external forces are Fer the spring force, and ¥,, the pivot force, we have However, the pivot reaction force Fp is not really known. In order to set up the rotational dynamics about the center of mass, we need to calculate the torque about the center of mass. The spring force acts through the center of mass, and will exert no torque. the torque from the pivot force is.M= r F, bj. Then, since M = dH/dt = I} by, the rotational dynamics about the center of mass takes the form Té = rR, ‘This latter relation furnishes, the pivot force directly as Fy = I ® / © . But the acceleration of the center of mass in the b, direction is related to ® by & = a/r, so the pivot force is Fo=ra/ r®, Inserting this into the equation of motion for the center of mass gives a form free of the pivot force: I 2 (»+S)a-r z Ss In a sense, the retarding effect of the rotational dynamics, responsible for the pivot force For is to add an additional equivalent mass I / r* to the actual mass m of the satellite. There is also an equivalent rotational form of this equation of motion. If we use a = r @ to eliminate. a in the translational equation of motion, we have 59 {= + r)o- re, after multiplying by r. Notice that r F, is the torque of the spring, about the pivot point, ‘This suggests an alternate way to derive the! rotational form above. It is permissible to write rotational equations of motion about the pivot point, since this is a fixed point. Using the parallel axis theorem, the equivalent monent of inertia about the pivot point is 1’| = I+ mr, Since the pivot point is now the origin, the pivot reaction force now exerts no torque on the satellite, and the moment of the spring force is r F, b,. ‘The statement that M = I @ becones {= + re = rf, as before. Problen 4-5, If the thrusters are located two meters off the rotation axis, and produce 20 N sec per pulse each, then the total angular impulse M At is | Mat | = 2 thrusters x 2 meters x 20 N sec / pulse = 80 kg m/sec per pulse. This is also the change in the satellite angular monentun| per pulse, since M At = ff At = AH. To achieve the desired final spin state, the angular momentum must be changed by oH = Tu, - Tuy, 3400 kg m* x 0.95 rad@/sec = 3230 kg m* /sec ‘The total number of pulses is then the required change in the angular! momentum, divided by the change per pulse: 7 60 2 wo = 3220 Kg m/sec _ 49, 575 80 kg m/sec It is possible, of course, to fudge the numbers in the problem so that this result would come out as an integer. ‘hat is never true, however, in the real world. Problem 4-6, To begin use the Skylab as the reference frame. ‘The velocity of the center of mass with respect to the b frame shown in Figure 4~13 is since the Skylab is at rest in its own reference frame. since the approach speed is 10 m/sec, this gives v__ - 2.307 b, m/sec. To calculate the total angular momentum about the center of mass, it is necessary to calculate the location of the centers of mass of the Apollo and Skylab with respect to the center of mass of the assembly. ‘The sketch shows the location of these three points at the moment of docking, as well as the velocity components with respect to the center of mass. For example, we calculated above. the velocity of the center of mass with respect to Skylab. This implies that the Smeters 3,847 moters 1,153 m —_—siiineters_ ge hs ‘Apollo speed of Skylab with respect to the system center of mass is - Vioqs and the velocity of the Apollo with respect to the center of mass is 10 - 2.307 = 7.693 m/sec b,. The lever arms for the cross products r 61 x v are the horizontal displacements shown in the sketch. The total angular momentum is 2 w= Day xy isi 15000 x 3.846 x 7.693 b, + 50000 x 1.157 x 2.307, b, = 8.772 x 10° kg n@/sec At the moment of docking, two large contact forces arise within the system. While the crew will find them quite noticeable, they occur as an equal and opposite pair within the system, and cannot change the total angilar momentum. So H is constant both before and after the docking. To calculate the moment of inertia of the docked configuration it is necessary to use the parallel axis theorem to translate the) moment of inertia matrices of both the Skylab and the Apollo to their joint center of mass. Once there, they can be added together. The! parallel axis theorem says that the new moment of inertia matrix I’ is related to I by Ay*+a2? -ayax -AzAx If = I #mq4 -AxAy Ax?+hz2 — ~AzAy wAxdz © -Ayhz — Ax?+ay* For the Skylab, Ax = 1.154 m, Ay = .9230 m, and Az.= 0. ‘The new moment of inertia tensor is +8519 -1.065 0 IL, = Ig, # 50000 {-1.065 1.331 0 ° 0 2.183 132.89 -53.24 0 = | -53.24 116.55 0 x 10° kg nm? o 0 159.18 Similarly, for the Apollo the displacements are Ax = -3.846 m, Ay =| 3.077 m, and Az = 0 + So its new mor 62 ment of inertia matrix is 142.00 -177.53° 0 Ty = I, + {-177.53 221.87 0 x 107 kg m? o 0 363.94 162.00 -177.53 0 = 4-177.53 233.87 0 x 10 kg nm? ° 0 383.94 The total moment of inertia matrix for the docked assembly is then Teor = Tat Tg1 294.59 -230.77 0 = 1230.77 360.42 ° x 10° kg nm? ° 0 543.42 Since H is along the b, axis, and by inspection of the total moment of inertia matrix, the b, axis is still a principal axis, the docked assembly will initially rotate about this axis. Extracting the b, diagonal moment of inertia, the rotation rate will be Seca o,-1H#L - 8:69 x10" kg m’/see . 1.962 raa/sec 133 5.4312 x 10° kg m’ However, this may not be a stable situation. Although by is a principal axis, it could be the intermediate axis (unstable) or the minor axis (unstable with energy’ dissipation, see Chapter 5). Calculating the other two principal moments of inertia involves solving the eigenvalue problem 2.9459 x 20° eet. -2.3077 x 10° -2.3077 x 10° 3.5042 x 10° - a - 6.4502 x 107 A + 4.9975 x 109? = 0 63 Using the quadratic formula, the other two principal moments are 3 2 = 1109.9 10° kg m* , 7 2 180.10 x 10° kg m’ and it is now obvious that b, must be the new intermediate axis of inertia. As we will see in Chapter 5, rotation about this axis is violently unstable. Problem 4-7. Before we can apply the Euler equations to this problem, we must calculate the torque on the rocket due to the thrust vector. For small gimbal angles @, and @,, the thrust vector is approximately T = -T6,b, + TO, b, + TH, while the vector from the center of mass to the engine is r = -@ bj. bs : ie be "a The torque is then Moar x T= mb, + 10, b, = THK, b + TeKw. ba after the feedback relationships are used. It is now possible to write Euler’s equations as 64 Mz = TKe, = Ad, + (A-C) wo, My = 0 = Ca, + (A-A) wo, The third equation immediately reduces to @,= 0, which has the consequence that w, = w,,, a constant. If this constant is zero, the other two Euler equations become in Ad, =TEKa, Ab, =TEKo, These are first order constant coefficient equations, and furthermore are decoupled. If we solve the first equation by separation of variables, we have = [meat = meK( t - €) 10 This can be solved for the behavior of w, as ®, = 1, exp( THK t - t,) The exponential function will decay to zero, and the system will be stable, only if TK <0. Other than the absurd possibilities of T <0 or &< 0, we need the feedback gain constant K < 0. fxoblem 4-8. In the case where the constant angular velocity about the b, axis, w,,, is not zero, the situation is somewhat more complex. Tf we recall the two Euler equations from problem 6, and divide by A, ve have 65 These are still linear constant coefficient equations, but they are now coupled. Define @ = (C - A) ©,,/A, a constant. The two equations above can then be put into matrix / vector form 10 oy -TK/A =. o, ‘ * - ~TEK/A faa o4 oy a PEK / ” Substituting the standard form of assumed solution ©, fa] gett 2 leads to the characteristic equation A ~ TEK/A a a -a A = TEK/A Expanding the determinant gives the quadratic equation 2 2TeK TeK ee me 4 (BH). gk = 0 Applying the quadratic formula gives the two possible values of A as 2TeK/A ante 2 : After the obvious cancellation: inside the radical, and dividing throughout by 2, the system roots A become 66 The final radical is, of course, imaginary, and simply gives oscillatory behavior to the solution. This will damp out or grow exponentially according to the sign of the first term, which is the real part of A. Again, if K < 0, the solution will decay to zero, and is therefore stable. Some Additional Problems A) A satellite booster is accelerating in a vacuum with acceleration A,+ The sketch shows half of an aerodynamic fairing, or shroud of mass m separating from the booster. The shroud first pivots about the hinge, and then at a given angle ¢, it separates, It is quite desirable that the shroud not collide with the booster afterwards. ¢ Ay i) During the hinge phase, argue that the "weight" of the shroud is mA,, and use Buler’s equations to show that Tyo = mALr sing where I,, is the moment of inertial about the hinge. 4i) After. separation show that the shroud tumbles about its center of mass at a constant rate dg/dt, and that the center of mass follows a parabola with respect to the vehicle axis. 67 CHapTeR 5 Problen 5-1, Since attitude control thrusters are pulsed devices, with a mininum pulse, the minimum change that the thrusters can make in the satellite angular momentum is one pulse. Integrating M over the very short (typically millisecond) pulse duration 2 2 a = fuat = Sax frat = SR xy : jet = 2pRb, from one pulse, where b, is the unit vector out of the paper. The quantity p = f F dt is termed the impulse of the thruster. Since one pulse is the minimum amount possible, the thruster system cannot change the satellite’s angular momentum by any less than this amount. Once the angular momentum is reduced to less than this value, the thrusters cannot reduce it any further. The satellite will possess, on the average, an angular momentum of the order of Hoe | AH | = 2pR to within a factor of 2 or so. If the satellite’s moment of inertia about the by axis is I, then since H = Iw, the average angular velocity of the satellite is « 2pR le] T = t in either direction. The thrusters fire every time the satellite reaches the edge of the angular window, of width @ (this is termed the deadband of the system). The satellite will take about to drift across the deadband. This is, then the average interval 68 between thruster firings. The larger the impulse from the thruster, the larger the fuel usage, assuming the propulsion technology is held constant. so, the fuel used per pulse is also proportional to p. The number of pulses per second is simply 1/At. The average rate at which fuel is consumed is then proportional to 2 “fine “pulse “Gecone” © Pht = AER This has several implications for the goal of minimizing fuel usage. A small deadband @ causes frequent thruster firings, and a large fuel usage. A large radius R gives a thruster more torque, and produces a larger change in the satellite’s angular momentum. It is the minimum impulse bit p, however, which has the greatest impact on fuel expenditures. Since p appears in the above as p”, halving p reduces the fuel usage by a factor of 4. Small p thrusters thus greatly extend the lifetime of a thruster controlled spacecraft. Problem 5-2. Euler’s equations (4.60) for a general rigid body with no applied torque are M, = 0 = + (C-B) ww, M, = 0 = BO, + (A-C) wy wy My = 0 = Ca, + (B-A) ww If both w, and w, are very small (the body is nearly in a state of pure rotation about the b, axis), then the last term in the b, equation, involving the square of these small quantities, is negligible. The b, Euler equation then becomes 69 This immediately integrates to the statement that w, = 0,,, a constant. The remaining two equations, when the constant value of w,, is substituted, are now constant coefficient linear equations. They can be put into the matrix / vector form ao yay] | 0 (C-B)H5 ]f : o cf] a, (B-A)w2, ° ®, Inserting the usual solution for linear equations, this leads to the characteristic equation AA (c-B)a, | 2 9 (B-A)o,, a or, expanding the determinant aca? - (B-a)(c-B) w2, = 0 This gives the system frequencies as If b, is the intermediate axis, then, say, A>B>C. This means that both B-A and C-B are negative, and their product is positive. The radical is then real, and one of the A values is. both real and positive. Rotation about the intermediate axis is unstable. (On the other hand, nothing in this calculation is specialized to b, being the intermediate axis. If it is the major or minor inertia axis, then one of B-A or C-B is negative, while the other is positive, Both A values are then purely imaginary, and the system is linearly stable.) 70 With the specified control torques, the coupled equations become Ao a, 0c |] 4, + i (M20 oy a = wy (BAe 3 oy where the gain constant K must still be determined. The system characteristic equation becomes DA (C-B)W-K ag (B-A)wy.-K cA This expands to yield aca? = { ( (e-Br%29°* J [ (B-A)W26-K J} = 0 The system roots become we [ EPI ag) (BAI HV099) a Ac or System stability changes when either of the individual quantities within the radical changes sign. In order to keep the system stable, then, K/o29, must be between C-B and B-A. Problem 5-3. The angular momentum is a vector, and must be added for both rigid bodies, the satellite and the rotor, to obtain the total angular momentum of the system. The inertial angular momentum of the satellite is (W,, W,, ,)", and its moments of inertia are A, B, and C in the principal axis b frame. If these moments of inertia include the rotor, then we only need concern ourselves with the spin of the rotor itself, Os for the second body. The total angular momentum vector is then 72 A 0 07]f wy - - Uf - H = |o Bol|jw} # J- - - | - oo cllw. -- 1 = Aw bj + Busby, + (To + Cus) by As we saw in the case of both the. axisymmetric and general rigid bodies, H, although it must be constant in the inertial frame, need not appear to be constant seen from the body frame. In the body frame, only the magnitude of H will be constant. This magnitude is 2 + (Tw. +e 0, ) Dividing by H? and rearranging, this becomes 2 2 2 : ey : w2 : ( % + T,.0,/C ) (Hyay? (HB)? (aye)? This is the equation for a triaxial ellipsoid, whose center is now shifted to 8 = ob, + Ob, ~ (1,0, /C)b,. The total kinetic energy T is a scalar, and can be directly calculated for each body, and then added. The result is 72 : A 0 O}fw T = > (Hy, Oy, 0) 0 BO |} a, oo cl}l a, ul + Gereuren Opi er nannane| yan To TELL ey, eH fad ead sede nee} Rearranging, we can define the kinetic energy within the satellite, Ty, as sae 2 2 2 Tw = {aed + pod + cu? } Dividing both sides by T,, the result is 2 2 2 0! 0! 0; tt aac eect pe eee 20, /A 20,/B 2t/e This is the equation of an ellipsoid with its center at the origin of the coordinate frame. The tip of the angular velocity vector of the satellite, = w,b, + Wb, + W,b, must lie on the surface of both ellipsoids at once. so, just as in the case of the triaxial rigid body, the motion of @ can be visualized by constructing these intersection curves. in analogy with the single triaxial rigid body, they can be termed polhodes. Problem 5-4. It is sometimes necessary to spin a satellite, for example to enable instruments to scan the sky, and yet the satellite must still be maneuvered to point in different directions during the course of the mission. If the satellite had a large net angular monentun, reorientation maneuvers (as in section 5.5) can become quite 73 expensive. On the other hand, if the angular momentum of the spinning spacecraft is subtracted out by a counterrotating momentum wheel, the entire satellite has a zero net angular momentum. Using the results of the last problem, this will occur when + 80, where 8, is the (presumed small) additional angular rate away from the required nominal value producing zero net angular momentum, The magnitude of the total angular momentum is then, using the results of the last problem The term from the rotor has been canceled by the appropriate choice of angular rate for the body of the satellite. In other words, the satellite has total angular monentum + ¢ 8u5b, H = Awb, + B wb, The angular rates ©,, w, and 8w, will probably be small, and only present during spacecraft reorientation maneuvers. Since torque equals rate of change of angular momentum Ms it applying the derivative rule to the expression for H above will yield Euler‘s equations in the variables w,, #,, and 6v,. The satellite can be controlled just as if it had zero net spin rate. However, its natural motion is somewhat changed. The angular momentum ellipsoid is given by 2 2 2 1s fr + einai a neeo ae (aay (HB)? (ayoy* 74 which is once again centered at the origin. However, the kinetic energy ellipsoid can be put into the form 2 2 2 o o ( 60, - Tw./e ) Safe ana ar/B 27/7 ¢ which is displaced from the origin. again, it is possible to define pohodes for this case, as well, but they will not be the ‘same as for a single torque free body. Problem 5-5. The usual stable region for a gravity gradient satellite is when C > B> A. Since the moment of inertia matrix for the shuttle is 1.29 0 «0 I= 0 9.68 0 x 10 0 0 10.2 © kg m, the major axis is b,, the intermediate axis is b,, and the minor axis is b,. ‘The stable attitude involves aligning the orbit normal with the normal vector to the wings (b,), pointing one wing into the velocity vector (b,), and putting the nose (b,) either up or down. In a 90 minute period orbit the spacecraft’s mean motion is most easily calculated as an a = 25% rad/min = 0.06981, rad/min The pitch frequency n, is directly given by equation (5.70) n= P This corresponds to-a pitch oscillation frequency of 7, = 2n/n; = 57.01 minutes. The roll / yaw motion of the vehicle is coupled, and the remaining two frequencies must be calculated from equation (5:74). 1/2 } = 0.1102 rad/min The dimensionless quantities ky, k, are = 0.3255, k= 0.9102 ir The characteristic equation (5.74) then becomes a4 + 4.0265 22 + 1.2849 = 0 The roots of this equation are easily calculated from the quadratic, formula to be 2 a? =~ 2.01322 1.6998 = { 197398 -3.7067 Since both values of a* are negative, the four a values are purely imaginary, and represent sinusoidal oscillations. ‘The two roll / yaw frequencies are then 1/2 } = £0.5654 i rad/min aya 7 { -o-sn97 1/2 } = £1,925 i rad/min ee { -3.7067 These modes have periods of about 11.1 and 3.26 minutes; respectively. Problem 5-6 The velocity calculation for the mass is still valid so| long as m << M, the mass of the satellite. We can thus pick up the derivation at equations (5.43) and (5.44) for the energy and angular) momentum, respectively, remembering that there is now only one mass n. The angular momentum is 2 u = [e+m?] = corm[o (nse +P tyr] while the total energy is 76 | = bee? + Bu [otet + @curtsni?] Divide both of these equations by m R*, and define the constant K as we obtain the result a2 K( a -0) = 3 (otrbsR) R for angular momentum, and e2 ») = + cortsry? R aieuea K (08 - for the energy equation. Just as for the double mass case, if we divide the energy equation by the angular momentum equation, and use the identity for the sum and difference of squares, the result is wy, = bse after a simple cancellation. Again, the cord unwraps at a constant rate. Using this result to eliminate ! from the energy equation, and solving for the cord length £, we have the result ‘| w, -w, \]/? = r|x}—2?-£ £ Wy F Op which is identical to the double mass result, with the slightly different definition of K. If the final spin rate is to be we = 0, 77 then the cord length needed to effect complete despin is te =R SK = f+ c/n Finally, equation (5.40) gives the velocity of the mass at release as (since w = 0) v be Wy I This is the velocity of the mass m relative to the satellite of mass M. The velocity of the satellite itself, relative to the center of mass of the system, will be = m - wew ¥ This is usually quite small, since m << M, and its effects on the orbit of the satellite can be ignored. Problem 5-2 The angular velocity of the earth is 20 -5 —pectoo rad/sec = 7.272 x 10°? rad/sec a Since the entire 23° tip of the earth’s polar axis is to be removed,| the angle @ for the coning maneuver is @ = 11.5°. ‘The time to complete this reorientation is given by equation (5.27) as x (Age) sees = 138.81 seconds = 2.31 minutes This time is ridiculously low (although correct!) because the earth is so close to being a perfect sphere. A sphere is, of course, completely indifferent to the location of the spin axis within the body. Needless to say, the two impulses necessary to perform this, 78 maneuver are far beyond human engineering capabilities. Although the problem is somewhat fanciful, it does have implications for satellite design: don’t make nearly spherical satellites! Problem 5-8 ‘The moment of inertia matrix of the satellite proper and the antenna masses can be calculated by use of the parallel axis theorem, equation (4.51). Since the masses move out along the b, axis, the position of the masses is given by Ax = ¢, Ay = Az = 0. The total moment of inertia matrix is then given by o 0 0 = 2 Jege = To + amy oo & or oo @ Or, adding in the moment of inertia matrix of the satellite, we have 500 0 o fi 2 2 Leot 7 © = 600 + 2me’ ae kg mw’ ° ° 300 + 2mé since the masses m are one kg each. The initial angular momentum is H, = 60 kg m*/sec”. since this is conserved, and assuming it works as planned, the final angular velocity is w, = 0,00132 radians per second. There are two major flaws with this design. When ¢ is small, the satellite is spinning about its minor inertia axis. This is acceptable so long as energy dissipation within the body of the satellite is negligible. This will not be true when the flexible wire antennas begin to deploy, and energy dissipation in the wires will most likely force the system to becone unstable. Should the satellite somehow survive this difficulty during deployment, another, nore violent disaster awaits. when the b, and b, monents of inertia become equal 500 = 300 + 267 + £'= 10 meters, 79 | the b, axis becomes the intermediate inertia axis, and the aesombly becomes violently unstable. The motion at this point becone: exceedingly difficult to model, since the wires are not rigid. ‘Th most likely outcome is for the satellite to end up inside a rat’s ial of tangled antenna wire, still conserving its original angular momentum. To make it work correctly, the satellite should be designeq as a major axis spinner from the outset. Some Additional Problems A) In the recovery of the Westar and Palapa satellites (which were stable major axis spinners about 2 meters long), astronauts were sent out wearing a Manned Maneuvering Unit (MMU) and a large stinger attached to their front. In theory, they could capture the ey by inserting the stinger into the (empty) rocket nozzle while th satellite was spinning. Then, pulling a lever, the stinger would expand to secure the astronaut to the rotating satellite. Then activating the "attitude hold" mode of the MMU would despin the satellite. Lene The first time it was tried, it worked as expected. The second tine, the astronaut/satellite combination tumbled dramatically, with thd first motion of the astronaut being upwards, out of the page in the sketch above. What went wrong the second time? (Hint: the second astronaut was a considerably larger and more massive individual than the first...) 80 Chapter 6 Rroblen 6-1. ‘he unit accelerates upwards with acceleration a = a bj. In order to accelerate the mass m, there must be an upwards force ma b, on the mass, and an equal and opposite force therefore appears on the inner gimbal. Since the mass is located at r = - d b,, the torque on the inner gimbal is Mos rxF = -db)x(-mab,) = -madb, The innér gimbal can move by either a rotation about the vertical outer gimbal axis, with an angular velocity of %, = $ b, or a Fotation about the inner gimbal axis with an angular velocity 3,, = 6 b,- Since angular velocities add, and the rotor angular velocity is assumed to be constant, the rate of change of H is i b. 5 ‘a = a He ou Sen + Be = (ob, +b, ) x Hb, = Héb, - Héb. However, the only torque component on the inner gimbal is in the b, 81 direction, so the gimbal will not accelerate in the @ direction, so @ = 0. Equating the inertial rate of change of the angular momentum H t the torque M, we have Hé = -maad Separating variables, this leads to $ This integrates to give 6 - % = { ve - vitg) J So, this instrument integrates the vehicle (and gravity) accelerations in the b, directions with time, and the ¢ displacement indicates thé total velocity change in the b, direction. Problen 6=2. ‘The inertial acceleration of the center of mass is a but the accelerometers are located at position vector R from th center of mass, and so read an acceleration A. Now, using equation (1.40), the acceleration read by the instrument package’ is b, api, Par agi . ee { As ao- -P xR = 2 - Bt ( BER) However, the instrument package is strapped down in the body frame, si Paryat = 0. he inertial acceleration is then given by age a + -# xR + By (dR) This expression is most easily evaluated in the unit vectors of 82 the body frame. The position vector R is directly available in this frame, and since the rate gyros are also bolted to this frame, they give the components of 3! directly in the body frame. | Strapdown accelerometers will furnish A in this frane, also. The last remaining piece needed is the angular acceleration dBP4yat, which. could be obtained by numerical differentiation of the angular velocity readings. Problen 6-3. The vertical error in an inertial navigation unit propagates according to equation (6.67) : eee 62 - ht 62 = z 0 ro Ignoring coupling to the 6x coordinate and the z accelerometer error da,, this becomes a 26M vo az- | —— + —2 bz = 0 *o Fo Now, this equation was originally derived assuming that the vehicle will be on a great circle course at constant speed. In a boost to orbit, this will not be true. We can obtain approximate results by using the average values of ry = 6390 km and v, = 4 km/sec. “The o coefficient in the equation above is then equal to +3.447 x 1076 sec". Since the 42 equation is now a constant coefficient differential equation, we can assume a trial solution of the form 6z = a e%*, y Numerically the values of « are + 1.856 x 10° sec™?, Now, both values of @ are possible, and with linear equations the general solution is a sum of both possibilities. The general solution is then Substitution shows that 26M a = #}/——— + 83 oz Where the two constants A, and A, must be evaluated when iti conditions are obtained. Notice that since one value of a ii positive, this is an unstable system. However, we are not give initial conditions. Instead, we are interested in setting limits on guidance system accuracy. After sufficient time has elapsed, only the positive a term above will remain. Since we wish 8z s 100 m at v burnout, t 480 sec, we have 100 m 100 n oar = w= 41 meters Since t = 0 at launch, 62 = A, = 41 meters at liftoff. If the erro: is greater than this, the 100 meter criterion at burnout cannot be met. 84 CHAPTER 7 Problem 7-1. The exhaust velocity is V, = g I, = 4439 m/sec = 4.439 km/sec. Writing the rocket equation (7.9) as we know everything except the payload mass m,. Put g@ = VyVe 0.9664. ‘Taking the exponential of both sides, the result is a linear equation for the payload mass, which can be solved to yield 8 my( 1.- e) + ce ae Se = 9525 kg cel when the calculation is carried out. For a round trip, the tug will carry payload on the outbound leg, but will return without payload. In order to return, we need propellant m,, for the return leg sufficient for a velocity change of 4.29 km/sec. The rocket equation becomes y, avg tn ( 78 *ee-) for the return leg. This gives a propellant mass of my, ng(o® = 1) = 2117 kg to return the tug to low earth orbit. For the outward leg, then, this‘ propellant must be considered to be payload. The available fuel for the outward trip is m,, =m, - mp, = 13883 kg. Recalling our earlier result See npn Te HL pats ag Fai However, this value of m, includes the fuel for the return trip. The actual payload delivered to geosynchronous orbit is then 7225 kg - 2117 kg = 5108 kg This is about 2/3 of the mass delivered to geosynchronous orbit if the tug is expended. The economics of this situation must consider 85 the cost of the tug itself, and the cost of its fuel delivered to low earth orbit, ‘The extra half of a shuttle payload required to deliver the fuel for a second trip must be considered against the cost of the tug itself delivered to the parking orbit. Problem 7-2. Recall the definitions of burnout speed and the initial and final masses from section 7.6: N y+ DVy an Also, note that specific impulse of stage k is I, = Vey / I Taking implicit partial derivatives with respect to 14,4 and m,, we have "ok a, = 0 = a an ( 7 ) 5px M5 * aces J} "tj "ej since the specific impulse of stage k is only present in one tern, while the payload mass is present in all terms of the sum. mq variation of V, is zero since we cannot permit this to vary. Solving for the desired result, we have 86 An ( my / Mey ) am, “Tak ¥ Y Y%ey C27 M37 17 Mey) which was to be shown. Using the numbers in parenthesis in Table 7~3 for individual components of the shuttle, the denominator above becomes 1 sete 207» 9.8 ( rresz00 - TolUz | + 4959.8 ( gla - aagsoo" 2.622 x 107 “8 | aa6i00 143300 : meters/sec/kg. he tradeoff ratios then become on, ar, = 54-61 kg/sec ‘sp,1 am, iH = 663.6 kg/sec ‘sp,2 The question of which to try to improve first, the solid rocket boosters or the main engines, depends on economics and how feasible the improvement is. The solids certainly are the least efficient part of the system, but the main engines offer the greatest payback per second of I5,- Problem 7-3. Using the numbers in Table 7-2, the tradeoff ratios for the Von Braun three stage shuttle are: k om, /OM gy om, /2M yy. 8m 4/28 gp a 0.0288 0.0096 367.98 2 0.324 0.0556 399.31 3 1.0 0.2600 203.66 If the specific impulse of the Von Braun upper two stages increase to 295 sec, we have AI,,, = +9 sec for both stages two and three. A seven percent penalty in structural mass for these stages is 87 Amz, = +4888 kg =, «Ams, = 1536 kg The change in the payload will then be ony on, om * “a ope * Taps Sapa amy an, + cma se + aa Mes After inserting the numbers, this gives am, © + 2307 kg Since this is a positive number, it is definitely worth doing. In fact, it is almost a 7% increase in the payload for this vehicle. Problem 7-4, Starting at the top of the vehicle, and using Table 7-5, the masses we need are Men = My tmz, = 291 kg Mop = Meg + Mpg = 706 kg Mey = Moo + My = 819 kg Mr 7 Mey tM, = 1986 kg Since we wish to consider two burn strategies, we will work out the burnout speeds separately for each stage. The rocket law gives 1 vy = Y, a ( 7 } = 2447.9 m/sec 1 en Rey A | va = Yep an ( a } = 2449.3 m/sec 'p2 88 The fact that this is an optimized rocket is shown by the fact that each stage shoulders an equal portion of the burnout speed. In the burn - burn - coast strategy, the total burnout velocity is vj, = 4897.2 m/sec. Working out the conservation of energy per unit mass “Feo = 2-199 x 107 m/sec? = g Hay gives a value of Fuax = 1223 km, indicating that we have exceeded the bounds of the constant gravity assumption: this number is too large a fraction of one earth radius! Using inverse square law gravity, energy is given by 4+ - + = -50.50 km/sec” starting from the surface of the earth, r, = 6378 km. Since at apogee v = 0, the energy must equal -u/r,,,, giving a value of r.., = 7893 km, or Hy, = 1515 km, ‘This value is larger than that using the constant g assumption, since gravity drops off with altitude in the real earth. For the burn - coast - burn - coast method successive applications of energy conservation give, for the first burn Ave = -59.50 xn@/see? =~ HL ‘oO ty 6699.1 and, for the second burn bv? - = 56.800 ton?/see! ‘ oe Sree *2 > Fy = 7055 km This translates to a maximum altitude of H,., = 677 km. This is much less than the burn - burn - coast method. Since the vehicle trades kinetic energy for altitude, doubling the burnout velocity results in much more altitude than achieving the same burnout speed twice. 89 Problem 7-5, We are asking what payload a certain configuration car place in a low earth orbit. Using the rocket equation, the (required burnout velocity is found from 2 2 | V, = 31,000 ft/sec = J vy in (+2 } ek Fx 81900 + 35000 + 703000 + my, 4 x 584600 + 35000 + 703000 + m, = 32.2 x 290 In ana 35000 ¥ my 35000 + 703000 + m, + 32.2 x 455 In It is not necessary to be careful in the argument of the logarithms| and use pounds. The arguments are dimensionless, and m, will come out in kg. Performing the operations above not requiring m,, we have 3076400 + m, 31,000 = 9338 In | —————— 1065600 + m, 738000 + m, + 14651 In ft/sec : 35000 + m, (ii) Now, the payload mass is certainly much less than 3 x 10 kg. It is also probably negligible compared to 1 x 10° kg, and tan should probably be neglected compared to 700,000 kg as well, for thi sake of consistency. It is certainly not small compared to the dry tank mass, 35000 kg, since this is the nominal payload of the usual shuttle! Ignoring m, in these places, we have 31,000 ft/sec ~ 9900 ft/sec 738000 + 14651 In 35000 + m, Solving for mn, yields the value of 90 ® 139,800 kg = 307,000 1b or almost 5 normal shuttle payloads. This vehicle has a slightly larger payload than the Apollo Saturn 5 moon rocket. (The exact solution is m, = 167092 kg.) Problen 7-6. Equation (7.49) gives the optimal payload ratios as oe Cle OLAV ) Solving for the value of A, we find -e ae 2 {——*_ - ek ( (1 - ey) mm Although we would normally solve equation (7.50) for the one value of A needed to optimize the entire rocket design, in this problem we are looking at two designs after the fact, to see how close to optimal they are. Using the data from Table 7~2 for the Von Braun rocket, we have k TM ey Vex R/sec Ae 1 +14 +128 2254 9.055 x 10* 2 1144 +0909 2803 76.044 x 10% 3 +272 +240 2803 -7.709 x 10 Similarly, using the data (for equivalent parallel stages) from Table 7-3 for the Space Shuttle, the corresponding results are x ™% ei Voxr m/sec ry 2.332 +144 3057 4.927 x 107 2 +161 062 4459 -3.163 x 10 The orbital maneuvering system is too small to be considered here. 91 The numbers for the shuttle are quite close, and almost certainl: represent an optimized vehicle. The numbers for the Von Braun sere are somewhat more disparate, and may not be optimized. Actually since the optimization must be done early in the design process, ant changes inevitably must be added when it is too late to reoptimize the vehicle, perfectly equal values of 4 cannot be expected: Problem 7-7. A cargo third stage for the Von Braun rocket mus! deliver the same total AV as the shuttle third stage for this same design. This will avoid disturbing the design of the remainin( stages, and keeps development costs to a minimum. This means that, using the same propulsion system, the initial and final masses fo) this stage must be the sane. From Table 7-2, these are m, = 129700 k and m, = 60225 kg. ‘Their difference must be the mass of propellant so this is the same as before. Recalling the definition of © equation (7.13) ™s + n+ Wy, ss" P Using the © = 0.0909 value from the second stage, and noting that m, My - Mp = 69475 kg, we can find the structural mass m, em, nz = oP = 6947 kg Knowing the propellant and structural masses for this stage, the remainder is payload: Mm, = My- mm = 53278 kg Compared to the original payload of 35,400 kg, we have almost doubled the lifting capability of the vehicle. Some Additional Problems A) why do virtually all aircraft take off horizontally, and virtuall) all rockets take off vertically?? What is there about aircraft that 92 puts a vertical takeoff vehicle (e.g., the Harrier) at a disadvantage, and what is there about rockets that makes horizontal takeoff difficult, if not virtually impossible?? B) Combining sections 1.7 and 7.11, write equations of motion for an aerospace plane, including lift, drag, and engine thrust (only within the atmosphere). C) Simulate aerospace plane ascents on the computer, using .the results of problem (B). The aerospace plane gains considerably in performance in not having to carry its own oxygen, but how much must it give back in drag losses? D) A single stage to orbit vehicle will use hydrogen / oxygen propulsion, and is to be designed with a nominal 40,000 1b payload (18150 kg), and allowance for a 25% growth in its dry weight during development. (This is the current x-33 specification.) If the best structural factor achievable is © = 0.08, how big must the vehicle be to include the growth allowance? How much payload could it launch if no weight growth occurs? 93 CHAPTER 8 Problem 8-1. The velocity equation of motion for inverted flight (8.47) is av S eee ee L v Separating the variables in this expression, we have eH D Rane erae ae: Once variables are separated, it is possible to simply integrate the resulting expression: Yaw vf [e-[ ee where V, and y, are the speed and flight path angle at the rollover point, respectively. Performing the integrals results in £p Inv - Inv, = a (* - x) Finally, taking the exponential of both sides and rearranging, we have ¢, > v= Vv. exp ze (r --7) f { cr, 7 This is the general form of the solution, with initial conditions Vat ¥ = 4. However, these conditions are not arbitrary. ‘The form of the right side up solution is ( equation (8.55)) c, D exp] --g ( r- 7, } { c, i Evaluating this at the rollover point, y = ¥ gives another expression for V.. Substituting this into the upside down solution gives v= 94 which is the speed solution in terms of the entry speed and rollover flight path angles. The speed of the reentry vehicle at its maximum altitude, where ¥ = 0, is and entry which is the desired result. Problem 8-2 In order to enter even a low altitude orbit, the altitude must be high enough that air density is essentially negligible. Recalling the expression for the rollover angle te equation (8.71) 95 - + Ef eme(-HT 7) ~ exp(-Hy/H,) } | both exponential terms are very small. The first, at the peak altitude H, since we wish to establish a minimum orbit. The second since we can define the entry point to be above the sensible atmosphere. The rollover angle then becomes 1 cos yy = | 1 + cos 7; This is just the expression for the critical rollover angle, equation (8.72). For small angles, we can replace the cosine functions with the first two terms of their Taylor’s series. This produces « aed feaaed eo g(tta- 3) After sone rearrangement, this becomes le 2 2 zi} Sede eae or, since 7, and 7, have opposite signs (¥, is traditionally negative, since we are entering the atmosphere) This result can be inserted into the result of the last problem, and the speed at maximum altitude equated to circular orbital speed Sere c R, +H This gives the result 96 since y; is intrinsically negative, this can be written as ¢, m(3) 22 Ble] ial L wx, finally solving for the entry angle 1 ae ( 42 } 1+ fz a This result gives the required entry angle to fly a rollover majectory to direct entry into a low circular orbit. such trajectories are under study to return a reusable chemical tug to its wiginal orbit, without having to carry fuel for the final Hohmann circularization burn all the way to geosynchronous orbit and back. ‘or nearly parabolic entry speed, Vo/¥5 x1 / fe and the required entry angle is y; = -4° for c,/c) = 0.5, so the small angle assumption s justified. Notice that in this application you don’t want a good glider ‘large C,/Cy). This will force steep entry angles and enormous pullup ‘orces (see problem 3). The basic objective is to loose energy, and a poor glider is better at this. Also, it is desirable to reach maximun ltitude at a speed slightly larger than circular speed. The vehicle can then follow a Hohmann transfer to a more reasonable orbital 1titude, and then circularize the orbit. «zoblem 8-3, The wings of a skip reentry vehicle must be designed to withstand an acceleration of ees a nm ~ D a, using the standard expression for the lift force (8.48). Inserting he expression for the exponential atmosphere and our solution for the 97 speed, equation (8.557), this becomes Peceta 2C, 1 L mmm 2 . SD = aT Po & vi exp { a (x x)} Using the solution for the altitude (8.57) in the exponential atmosphere term, it becomes K, i = ony {= an { tote YP 95 tg Finally, placing this result back into the acceleration expression, and remembering the definition of K,, equation (8.53) we have a, A very efficient glider will have a small drag to lift ratio Cp/Cy. The exponential term above can then be ignored. In this case wt = +f (cose +B) a, hice HS Either by inspection, or by calculating the derivative da/dy and setting it equal to zero, this expression has its maximum value at 7 = 0. Using (8.58) for B, and ignoring the exponential term B® = cos 4 the maximum acceleration is vi a, * (2 - ces x |} eee fe TES te 4 For the given conditions, V, = 20,000 ft/sec and y, = -20°, the acceleration due to lift is = 1048 ft/sec* = 32.5 ‘gees’ a, eesnax Tt is this fact that made the Sanger bomber impractical, since the pullup from any. moderately steep entry angle requires enormous 98 accelerations: more than either man or machine could stand. of course, the maxinum acceleration rapidly becomes more reasonable as 7; ecomes smaller. The shallow skip flown in an Apollo reentry involves far smaller acceleration. Problem 8-4. Making the shallow entry angle (small ¥) and small iltitude assumptions, we obtain the modified eqns (8.28): cl A au, aves De 2 ae VN, ae aa PV" 2 cA v 2 Pee eEE eee Ry The next step is to change the independent variable from time t to speed V by multiplying by the inverse of the second equation. this gives the two equations an Hea Te ~av— “GAapv dy 20m ( v eee eg bad c Apv Wow, using the definition of y, and the differential relations dH = “Hy dy / y and av = - V dx, and using ¥? = v* g R,, these become z He - - fate fo ‘inally, taking another x derivative of the first equation and substituting the second, then using V = e* gives the modified varoshevskii equation 2, wy = 3 [e%- a} - Hee eee ax" 7 Fores The drag only Yaroshevskii equation showed that planar ballistic 99 reentry at a shallow angle was independent of most of the physical parameters that supposedly determined the problem. ‘The addition of lift to the vehicle introduces only one additional parameter, the lift to drag ration C,/C). Problem 8-5. This is a computer exercise, and a complete answer cannot be given here. Several modern computer mathematics and analysis packages make this a fairly simple task. Problen 8-6, The heating rate is proportional to p V°. Fron Yaroshevskii’s substitution for y we get oy amy p= Cy A SRS while the dimensional velocity is v = /gR,V = /g R, e %. Then, substituting these relations and the solution (8.41) for y(x) gives Cae = ae lal (CH 2 | 12) v2 rtex + 3px?) } exp( -3 x ) This is somewhat too complicated for the author to wish to take an analytic derivative with respect to x. ‘he portion of the equation above which depends only on x is plotted below. From this, and examining the computer output, the maximum heating rate occurs at x ~ 0.36 and y # 1.11, somewhat earlier than the maximum deceleration point, But like the maximum deceleration point, the maximum heating rate occurs at, a definite fixed point on the entry trajectory. The value of the maximum heating rate does depend on the drag factor Cp. unlike the maximum deceleration. The larger the drag factor the smaller the maximum heating rate, Blunt is better! (For another reason too: blunt bodies cause the heat to appear mainly in the shock, which is further away from the body when it is blunt.) 100 0.2 0.3 0.4 0.0 O21 x Some Additional Problems A) The plot below gives lift c, and drag C, coefficients for a typical reentry vehicle as a function of angle of attack, for very high speeds. eT eee a Tee ee eee 0.0 0.2 0.4 0.6 0.8 1.0 1.2 1.4 1.6 AoA, radians -2 0 2.4 6 8 10 Consider a reentry trajectory consisting of a skip reentry until horizontal flight is reached, then a constant altitude phase. Initial entry conditions are V = 7.8 km/sec, 7 = -5°, the mass is m = 50,000 101 kg, and A= 45 n®, the initial altitude is H, = 50 km. Fly the skip portion at the maximum lift point, not the maximum lift-to-drag point. At what speed and altitude is the pullup complete? At the bottom of the pullup, the angle of attack is increased to lower the lift force to what is needed to balance gravity minus) "centrifugal force". (The angle of attack is increased since only one side of the vehicle is covered with heavy thermal -protection. Increasing the angle of attack also increases the drag, too, and it is our goal to slow down.) As the vehicle decelerates, the angle of| attack is gradually decreased to increase lift, supporting the! increasingly "heavy" vehicle. When will this no longer be possible? At what speed will the altitude begin to drop again? 102 CHAPTER 10 Problem 10-1. Jacobi’s integral is given by equation (10.34): c= ty) sp As shown in the sketch, if the spacecraft starts in the vicinity of the earth ( r, = 1/60 distance units ), then two terms above are quite small. ‘The gravitational potential of the moon ( -1/r, ) is of order uy while the centrifugal potential term (x*ty”)/2 is of order yu’. This latter term is of order u” since the spacecraft is near the earth, which itself is a distance y from the inertial origin. Ignoring these terms, Jacobi’s integral becomes 1 “1.59411 = ® Solving for the. speed, we get v = 10.740 du/tu, where the distance unit is 384,400 km, and 2n time units is 27.322 days. ‘This makes 1 du/tu = 1,023 km/sec, so the speed above is very close to the two body = 11 km/sec: escape velocity v.., Problem 10-2. This is very similar to the previous problem, but the text does not contain the numerical value of C(L,). In order to calculate this value, recali that the triangular points are located at r, = rz, = 1 distance unit. As seen in the sketch, the x and y 103 coordinates of the equilibrium are then x = -0.5 + m = -0.48787 and y, = 1 sin 60° = 0.86602 . Inserting these values in the expression for 2 Jacobi’s integral (with v* = 0), the required value of C(L,) 5 -1.49400. > tp-O Then proceeding as in the last problem, the potential of the moon, and the centripetal term are negligible when the spacecraft departs from the vicinity of the earth. we have io = ee pee 1.49400 c 7 Vv yeo- This gives a value of v = 10.749 du/tu, or only slightly more expensive than reaching L,. Problem 10-3. Equation (10,18) must be satisfied at the L, point. Since x - 4 < 0 and x + 1-4 > 0 at L,, between the earth and the moon, this equation becomes x + —2=H# 7 - —_# i so (Cx-H) (x+i-H) Substituting x = -0.83702, we have -0.83702 + 1.37003 - 0.533052 = ~0.00004 80 L, is at approximately this position. Since L,.is near x, = -0.83702, y, = 0, we can evaluate the coefficients in the linearized equations (10.22) and (10.23). They 104 become 8k - 28y - 11,2941 6x = 0 sy + 28k + 4.147 by Define the state vector x = ( éx, dy, 8%, 8y )". ‘Then the above equations can be brought into first order form k= BX where the system matrix B is ° 0 LO, ee ° o Ona 11.2941 ° o 2 ° 74.147 -2 0 0 Inserting an assumed solution of the forn xX = A e*t to the eigenvector / eigenvalue problem , We are led a ° lic ° a Ogee ae aoe 11.2941 Oona ° “4.147 -20 =a To have non - trivial eigenvectors A, the determinant of - the coefficient matrix must be zero. This determinant yields the characteristic equation a4 = 3.2471 a2 - 46.8366 = 0 This is a quadratic equation in a2. the eigenvectors A are found by substituting the four A values back into the eigenvalue problem, and solving for the three unknown components. The system eigenvalues and eigenvectors are 105 A, = +2.9319 Ay = -2.9319 0.29325 0.34210 0.13494 0.15742 -0.85979 1.00299 0.39564 0.46153 Ag g = # 2.334281 0.25608 + oi 7 i 0 =F 0.918424 34 OF 0.597775 -2.14382 + oi The eigenvectors have not been normalized. Since we have one positive veal eigenvalue, the equilibrium point is unstable. The purely imaginary pair of frequencies generate undamped oscillatory behavior, which is about the most stable behavior possible in a conservative systen. Problem 10-4, ‘the inertial velocity of the spacecraft is dr Far si ae" get BM where Sarsat = ( %, } )7 and the angular velocity is 3! = 1%. the cross product is then 3°* x r = -yi + xj, and the inertial velocity components, resolved along the rotating axes, are given by x key, € =P ex Solving these for the rotating frame velocity components, we obtain x= X%+y, y= 1 - x, and substituting into Jacobi’s integral, equation (10.34), we have e- (ky) 106 = (+P) eee - after expanding and simplifying. When the third object is far from the second primary, the last term above is small, and can be neglected. The first term above is the inertial kinetic energy per unit mass, while the remaining potential term is the potential energy of the larger primary. Together, they are the usual two body total energy E 4 (#+ #) - settee tee tela: 1 2a using equation (2.36) for the two body energy in terms of the semimajor axis. The quantity u.. is the gravitational constant ¢ times the mass of the larger primary. Since G = 1, and we are ignoring quantities of order u ( of. the restricted problem ), u.,, * 1. The remaining terms are the cross product of the inertial velocity with the inertial position vector ky - fe = ver [s-H =- /ula (1 - e%) using the expression (2,33) for the angular momentum in terms of the two body orbital elements. The Jacobi integral then becomes cw p-H® +e - Saa-e’) which is Tisserand’s criterion. Before numerical integration of the orbital equations of motion was feasible, this was used to check the identity of comets suspected of having made close approaches to Jupiter, with resulting large changes in their orbits. Problem’ 10 The z equation of motion is given by equation (10.14) 107 wz a | 2 I Expanding the denominators about an equilibrium point within the 2 = 0 plane will produce results of the form { terms linear in 6x, 6y and z } | However, when linearizing about an equilibrium point, terms that are | quadratic in small quantities are discarded. ‘The numerator of the z | equation of motion already has one (small) z, ‘so the extra terms in, the radius expansion contribute nothing. 1 The linearized z equation is then - z | | 3 3 Fle | 4 This has the form of a harmonic oscillator, since the bracketed term above is positive. The vertical motion about any of the five equilibria is stable, and uncoupled from the x and y motion. Problem 10-6, In problem 10-3 we reduced the problem of relative motion about the L, point to a constant coefficient linear system, and saw that the L, equilibrium point is unstable. In order to stationkeep a spacecraft near this point, a control systen must be | devised. Remember that xX = ( 8x, sy, 8%, Sy )T, and that the uncontrolled system obeys X= BX. Augnent this system with the usual | control term = BK +#cu where u is the (scalar) control amplitude, and c is the eee distribution matrix. It is only possible to control a spacecraft by thrusting, which means that ¢ must have zeros in its first two | elements. Assume that the thrusters are aligned with the earth - moon line, so that ¢ = ( 0, 0, 1, 037, 108 | Now, the eigenvectors of the uncontrolled system can be assembled column by column into a matrix A. If we define new variables ¥ = A‘X, the variables Y are termed modal variables for the system. Since the first eigenvalue in problem 10-3 was the unstable real value, it is the first component of Y¥ that we wish to suppress. If we make the control proportional to this component of Y, the stationkeeping system will, with the proper gain, act to suppress any tendency to leave L, via the outgoing eigenvector. The remaining modes are either oscillatory or decay, and can be considered already stable. So, chose control of the form =k { at \x where the bracketed quantity is the first row of the inverse of the A matrix, and k is an undetermined gain constant. ‘the row of a? we need is { at } = ( -2.0325, 0.3434, -0.5277, -0.2428 } 1 The closed loop system now becomes k= px + ke { res } x 1 ° 0 aH. ee 0 ° o aly 11.2941 ° o 2 ° “4.147 -2 0 0 0 o ° o 0 ° ° + | <2,0325, 0.3434, -0.5277, -0.2428 | * ° 0 ° o 109 The stability of this system can be studied by calculating the eigenvalues as a function of the gain k. ‘The attached plot details the behavior of the unstable A value, showing that a minimum gain of about +5.6 is needed to make this root stable. Feeding back one mode, as we have done, ensures that the other three A values do not change. Of course, this much calculation is best left to a computer. 2 3 i 0 -2-1 k 110 CHAPTER 11 Problem 11-1. Consider a particle of dust on the moon‘s surface. If the moon is much closer to the earth, the difference in the gravitational attraction of the earth on the dust an the earth on the moon (the ‘tidal’ force) becomes large. This is exactly the situation Earth “ty ¢ discussed in section 11.2 . Using the formula for the radius of the activity sphere ee { nat ie R 3H we can interpret r as the distance between the dust particle and the center of the moon: the moon’s radius. ‘The quantity ( m/3M }'? = 0.1602. So, interpreting r as the radius of the moon, r = 1738 km, and solving for R, the distance between the earth and the moon, we have = —FrH = R SST 10848 km This is termed the Roche Limit for the moon. If the moon ever approached the earth closer than this, it would be torn apart by the tidal component of the earth’s gravity. (Adding the radii, the surfaces of the earth and moon would be only about 2700 km apart!) The rings of Jupiter, Saturn, and Uranus all lie inside the Roche Limit for their respective planets, and may represent disrupted moons, or moons that were never permitted to form. aa. Problem 11-2. Following the argument in section 11.2, the inertial acceleration of the test mass-is now as Bx (Bx (RH ri) = -SH* perp while the gravitational acceleration is PSs eeepseeec case eee 2 ej g (R + x)? oe Since, as shown in Figure 11-8, the test mass is now outside the orbit of the planet.. Equating the inertial acceleration to the acceleration produced by gravity, we have GM (pyr) cM R? (R +r)? Expanding the denominator of the sun’s gravity term, we find for small r (Re ry? RF? - aR, t cee Inserting this back into the acceleration balance equation, and canceling the term -GM/R’, the result is ~ GME 2 4 2 Gir R R Solving for the ratio r/R, we have 112 This result is identical to the result in section 11.2 for the inner (L,) point, Since they are equal, this is some justification for claiming that the activity ‘sphere’ really is a sphere. Problem 11-3.’ For the direct method shown in Figure 11-9, one single maneuver injects the spacecraft into the desired hyperbolic orbit. Energy conservation in the hyperbolic orbit requires that the speed at perigee, v, be related to the speed at infinity, v,, by Ay? - wb ew bye 2 wo. Ly? aaauD) ag 2 Ye oe ~ 2 Ye Solving for the perigee speed in the hyperbola, we have Vetee ie The speed before the maneuver is, of course, circular orbital velocity at a radius a, ‘The Olberth method requires two maneuvers. The first of these is identical to the final maneuver in a Hohmann transfer. The speed before the first maneuver is circular speed, v_. The transfer ellipse has a senimajor axis of a= ( a, + rp)/2, so the energy of the transfer ellipse is 113 Equating this to the ellipse’s energy at apogee, v2/2 ~ u/a,, where v, ay Pa 2 q = os + AYaiee is the speed in the ellipse at apogee, the speed after the first Olberth maneuver is Zo Mote = ‘a a FE P The first Olberth maneuver is then OY aa SIY acuity al = He a a5 ai The speed before the second maneuver is found similarly. If v, is how the speed in the ellipse at perigee, rn v= a 2 *p ‘0 Replacing the radius a, with the radius r,, in the hyperbolic orbit energy equation above, the speed after the second maneuver, vj, is Yh ec a Finally, this second maneuver is 114 The first Olberth maneuver slows the spacecraft down, while the second speeds it up. Problem 11-4. Circular orbital velocity at the distance to the moon, a, = 384,000 km, is v= 1.018 km/sec. The required speed at perigee in the escape hyperbola is v, = 3.3100 km/sec, so the velocity change required by the direct maneuver is BV sect 7 Vp 7 Voy = 24292 km/sec For the Olberth maneuvers, the quantity a, + x, = 390,578 km. The apogee speed in the transfer ellipse is v, = 0.1868 km/sec, so the first Olberth maneuver requires a velocity change of = 0.8311 km/sec This maneuver is made to slow the spacecraft down. ‘The second Olberth maneuver begins with a speed of v, = 10.8987 km/sec at the perigee of the transfer ellipse. The hyperbola with this perigee and the correct v, has a perigee speed of v, = 11.4049 km/sec, so the second Olberth maneuver requires a velocity change of av, = Vy ~ Yp = 05062 km/sec Adding the two Olberth maneuvers, the total Av required is Av, = 1.337 km/sec This is substantially less than the Av, = 2.292 km/sec requi~ed by the direct strategy. It is a general property of the two body 115 problem that large velocity changes can be achieved by (relatively) small maneuvers made in the close proximity of a planet. \ Problem 11-5. The parking orbit radius r, is likely to be well known, while errors in the rocket burn will produce errors in the burnout speed v,. These will, in turn, produce errors in the spacecraft’s speed at the activity sphere crossing, v,. So, taking differentials of the energy equation, and holding r, constant, we have Vy WV, = Vo av, This easily yields Since v, is substantially larger than v,, the error in the crossing speed is larger than the error in the burnout velocity. This gets still worse. The aphelion distance is R, = 2 a, + Ry, so the error in aphelion distance is dR, = 2 da,, since the radius of the earth’s orbit is known, The differential. of the semimajor axis of the transfer ellipse is a a ata Now, V, = V, + v, by velocity addition at the activity sphere, so av, = dv,. Replacing da, and av, with the equivalent. expressions in terns of dR, and dv,, respectively, we find sg ¥ M4 3 (vi72 - 47% | aR, = a 2 Inserting the error expression for the departure hyperbola links this all the way back to burnout in the parking orbit. This is not a complete analysis, since the injection velocity can also have the wrong direction, as well. 116 | Problem 11-6. The required departure speed at infinity for a Mars mission is v, = 2.98 km/sec, as given in problem 11-4, The necessary | departure speed v, from a low altitude parking orbit (say rg ~ 6500 km) is given by equation 11.19 7 uy y1/2 Yo = {% + =} = 11.46 km/sec So, the error growth in the departure hyperbola is av, = av, = 3.84 av, However, things get much worse during the heliocentric portion of the flight. Using #, = 1.33 x 10" km°/sec” and R, = 1.5 x 10° km for the earth’s distance from the sun, the earth’s orbital velocity is about 1/2 Ve = {Ho 7B fo = 29477 kn/see and the heliocentric injection speed for the spacecraft is V, = V, + v, = 92.75 km/sec. fvaluating the error growth in the heliocentric phase, we have H, V. ar, 2 1, w, 2 (72 - wR) 32 4.358 x 10°" av, = 3.54 x 107 av, 1.227 x 10° Combining these two error propagation equations, the aphelion error at Mars arrival is related to the burnout velocity error by a 8 AR, = 1.36 x 10° dv, if dR, is in kilometers and dv, is in km/sec. If the error at Mars arrival is 100 km, we need to achieve the correct burnout speed to within 117 av, = 200 ____ = 7.3 x 1077 km/sec i 1.36 x 10 This is an error of only 0.073 cm/sec, far more than can be expected of booster hardware. In order to actually fly an interplanetary trajectory, provision must be made for -midcourse correction maneuvers. Since these are usually performed by very small rocket engines attached to (relatively) massive spacecraft, they can be done to high precision. Of course, the actual trajectory must first be determined to high accuracy before the error can be eliminated. Problem 11-7. ‘The injection from an incoming hyperbola to a (nearly) parabolic orbit about the planet requires dropping from periapse speed in an hyperbola (see problem 11-3, direct method) to (just below) escape speed, v= /Zi 7%, 80, the required maneuver is z 2n i HL won fet / Now, this is a function of v,, the speed at which we enter the activity sphere of the planet, and r,, the periapse radius in the t¥ approach hyperbola. The first quantity is not under our control, 118 since it will be mandated by the heliocentric orbit on which the spacecraft approaches the planet. However, as shown in the sketch, small changes in this heliocentric orbit can produce major changes in x. It is of interest to ask if there is an optimun r, Pp the maneuver Av with respect to r,, p bY minimizing ‘Taking the derivative of the maneuver with respect to Xpr we have Hee 2 ee eee y, 172 P 2 { vis ee } For a minimum maneuver, we want this derivative to be zero. However, it is never zero. If x, is very large (undesirable, since the spacecraft is supposed to approach the planet), the derivative does approach zero. In another fashion, it ‘approaches’ wero as x,» 0, as well. In this case, the expression above reduces to d Av/ dr, =» - ©, Reexamining the original maneuver expression, however, shows that as r, 0, the term in v? becomes negligible, and the required maneuver approaches 2 2i ee eraeeeeee z= 7 P PB Since x, cannot really go to zero, the best we can do is approach as close to the planet as is safe. This was actually done in the case of Mariner 9, the first spacecraft to orbit another planet. The orbit of the probe about Mars was highly elliptical, with a low perimartem distance. 119

Você também pode gostar