OPTIMUM TRANSFERS FROM CIRCULAR ORBITS by GIM ER JEW B.Sc. Queen Mary College, London University, 1974 S. M. Massachusetts Institute of Technology, 1975 SUBMITTED IN PARTIAL FULFILLMENT OF THE REQUIREMENTS FOR THE DEGREE OF DOCTOR OF PHILOSOPHY at the MASSACHUSETTS INSTITUTE OF TECHNOLOGY June 1978 Massachusetts Institute of Technology Signature of Author Signature Redacted Department Certified by%. 1978 (Aeronautics and Astronautics Signature Redacted Thesis Supervisor Certified by Signature Redacted Thesis Supervisor Certified by Signature Redacted Thesis Supervisor Signature Redacted Accepted by Chairman, Yepartmental Graduate Committee Archives MfTTs JUL 6 TITUTE 1978 LIBRARIES Optimum Transfers From Circular Orbits by Der Gim Jew Submitted to the Department of Aeronautics and Astronautics on in partial fulfillment of the requirements February 17, 1978 for the Degree of Doctor of philosophy ABSTRACT a The problem circular orbit orientation. the minimum impulse transfer is to determine to an elliptic The optimum number shape and of arbitrary size, orbit from 4 3, location, magnitude and direct- ion of the impulses are to be determined. The basic H.W. approach combines technique a numerical developed by in his Ph.D. thesis at Stanford University with a number of Small analytic approximations. estimate of An analytic optimal the trajectory provides approximation and Smallos formulation a first allows rapid convergence to an exact numerical solution. The analytic ing of this approximations problem guidance technique. circular orbit to which could an improve the theoretical provide the basis Knowledge of elliptic orbit the optimum could for understanda subsequent transfers improve the envelope and fuel economy of the space tug. Thesis Supervisor: Wallace E. Vander Velde Title: Professor of Aeronautics and Astronautics Thesis Supervisor: Richard H. Battin Title: Lecturer of Aeronautics and Astronautics Thesis Supervisor: Walter M. Hollister Title: Associate Professor of Aeronautics and Astronautics from a operaional iii In Memory Of My Teacher Therodore Neil Edelbaum iv Acknowledgments I T.N. wish to express deepest my gratitude and respect to Edelbaum, who suggested and supported the work of this thesis. and It is a distinct pleasure to thank Professors W.M. Hollister for many helpful criticisms. L.L. Sackett R.H. Battin for his valuable for his contributions. teaching and support W.E. Vander Velde I am grateful to I am also indebted to through out my graduate studies. This research under contract was supported I.R.F. 18705. by the C.S. Draper Laboratory v List Of Symbols semi-major axis d. J defined by eq. 49 e eccentricity e unit vector h angular momentum per unit mass * a h reference angular momentum per unit mass H variational Hamiltonian H DH/Du u i inclination p adjoint vector P parameter (semi-latus rectum) Q defined by eqs.52c and 60 R radius vector t tan(Au/2) u central angle v true anomaly V velocity vector AV characteristic velocity x state vector x. state variable defined by eq. 49 O. coefficients defined by eq. 54 1 thrust unit vector X. J adjoint variable defined by eq. 49 X primer vector (Lawden) A. functions of state and adjoint variables defined in eq. 50 vi Y thrust angle w.r.t. eL axis (Fig. 1) thrust angle w.r.t. e1 axis (Fig. 1) w argument of pericenter argument of the ascending node T nondimensional velocity y gravitational constant subsfripts: i initial orbit s final orbit vii Table of Contents Abstract Acknowledgments List Of Symbols Page Chapter 1. Introduction 1 2. General Analysis 4 3. Small Inclination Approximations 8 4. Small Eccentricity Approximations 17 5. Large Inclination And Eccentricity Approximations 24 6. Computation Technique 32 7. Conclusions 41 Appendix A: Optimum Single Impulse Transfers From 43 Circular Orbits Appendix B: Computer Listings 46 Appendix C: Examples Of Exact Transfers 72 Figures 78 References 92 Biography 94 ONE CHAPTER 1 Introduction bolic to orbit circular only in Edelbaum', to the excellent paper of 50's been the made in 3-dimensions, out a with 1976 efficient very and were not (Lawden) spiral another survey no had 3- symmetric solutions almost Small4 the in in solutions from that t. Gobetz and Doll 2 had In refer should active coaxial, nodal, that except For further very written by arising pointed solutions 2-dimensional was speculated are readers the transfers. limited Marchal3 by written impulse free. , 2-dimensional singularities the symmetric and limited paper to due three and two transfers circle, The transfers. quasi-circular complete to circle i.e. dimensions; paper special some only that out pointed survey 1969 a In 601s. and involve How many impulses ? ' transfers impulse optimum in The para- these will rendezvous, impulses of on the optimum number Research since to be non-optimal when the time is by many experts discussion free. and transfers orbit for time fixed the time a from Finite four impulse solutions which exist considered. are solutions transfers Finite infinity. at impulses impulse excluded states are end with orbit elliptic an and hyperbolic infinitesimal optimum investigates thesis This had progress all generalized computer program optimum impulse in 1972. 5 transfers the orbit in problem Marec elements; i. become 7 a of complicated first due orbit circular semi-major The pioneered Breakwell and vicinity the inclination, 6 Nae , - Eebu Edelbaum axis, and to a, second the t with small changes e, eccentricity, order boundary solutions of in and this effects of the pericenter of layert linearization: 1. When the changes in a and e are small, of the transfer orbit is not well defined. 2 small, i are e and in When the changes 2. the line of nodes of the transfer orbit is not well defined. Marchal of in only changes thus 8 found p Marchal as A will 0, the first order solution does not agree with a -w is not well defined). w Ai (because the matching of other linearized solutions. then the defined, and a, If one calculates the limiting solution allowed. changes If the variation of in W arbitrary i become e and in We impose difficulties also find that the boundary layer effects is well small of i such that arbitrary variations in Aa, Ae and that of small the case for solution order first e and , were in r a and this is the general problem we wish to solve. 2 Chapter solution of circular orbit the describes allow some and simplifications order first from an Transfers transfers. impulse optimum the of analysis basic assumptions initial with no penalty on generality. The contribution theoretical of this to extend is thesis the basic analysis to obtain the following analytic approximations: Small inclination approximations in chapter 3. Small eccentricity approximations in Chapter 4. Large inclination and eccentricity approximations in chapter 5. 6 Chapter approximations to an combines to compute elliptic of estimate orbit. optimal the the formulation of and Small the analytic from a circular orbit the general transfers An analytic approximation trajectory, and Small's provides a first formulation allows of single rapid convergence to an exact numerical solution. Appendix impulse transfers A describes the basic analysis from a circular orbit in state space space. This is an extension of Der's describes the logic of the computer program. M.S. thesis 9. optimum and parametric Appendix B The listing is included 3 for reference. transfers. Appendix C consists of ten examples of exact 4 2 CHAPTER TWO the time free 0 essentially General Analysis An of mination 14 the S transfer requires the deterstate adjoint and the true anomalies of depart- of the transfer orbit, v2 and vi optimum impulse dimensional five p of the transfer orbit, and arrival ure and 1 parameters: and x vectors 0. orbits two elliptic between of solution analytic and the respective true The primer of the initial and final orbits. analies v. and v s 1 10 vector of Lawden can be defined as a vector formed of components which are Lagrange the optimum referred to rocket as constitute duced in the , normal which is a components arbitrary point adjoint primer with vector, the conic of in the variational problem. velocity association 0 and trajectory the which orbit, introduced multipliers The vector are velocity primer because vector. the vector the Along is and of often multipliers, intro- the transfer circumferential radial, are given by the classical T (Fig.1) vector adjoint variables the trajectory, the primer formulation its derivative at an expressions p1esin v + p cos v p1p + p 2 / *P 3 sin v (1+1/$ + p 5 cos p4 sin v /* p V2 + p 1 /3 - where * = e = 2 -p 1+ecos v and 3 the sin v esin v + p 3 (e+cos v) v) + p 5 sin v p (e+cos eccentricity, v/i -pi 2 ) = P = parameter, primer vector 1 = gravitational constants p 1,... constant, , p5 are 5 can which constants integration be identified the as Lagrange multipliers associated with orbit element state variables. points ( v X. = by developed conditions optimum The Lawden at the junction v 1, v 2 ) are = 1 X (1) , 0 The impulsive terms the of velocities at elements of the points may be expressed transfer and final junction the initial, follows: e sin v esin v- AV - (1+ecosv) - JI (+e 'Ti cos v) cos i (1+e cos v) sin i L e sin v - esin v S - AV 2 (l+e cos v) cos(i -i) - I (1+ecos S - where E(1+e cos v) sin(i -i) = inclination of the transfer orbit plane, i i S = inclination of the final orbit plane, e = eccentricity of the initial orbit, e = S eccentricity of the final orbit, v) in orbits as 6 P. = parameter of the initial orbit, P and = parameter of the final orbit, S i > is- i 0, The dynamic relations at the junction 0 points are AV k= k (3) k AV k where The IAVkI = V , k=1,2 geometric relations obtained from the equation of orbit and 2 2 a.(1-e.)/(1+e.cos v.) = a(l-e )/(l+ecos v ) the latitudes at the junction points are (4) s s + v ( s s 2 ) a (1-e 2 )/(1+e cos v ) = a(1-e 2 )/(1+ecos v =0, sin i sin( w + v 2 ) = sin i where w + v sin( ) (5) a = semi-major axis of the transfer orbit, a. = a = semi-major axis of w = argument of the perigee of the transfer orbit, Ws = argument of the perigee of the final orbit. Although transfer, semi-major axis of the initial orbit, Long11 lists 70 Eckel 12r,13reduced the final orbit, equations to define a the system of equations 2 impulse optimum to 4n-5 equations 7 to be solved However, sufficient the 14 (1) equations optimum linearized to determine by described equations transfer. n impulse optimum simultaneously to define a to from transfers (5) are a circular orbit. We will assume for the rest of this thesis that: 1. 9 the argument of the node, the initial circular orbit, v. The plane , reducing the state space to 4 In addition the true anomaly of departure on dimensions. 2. In this case we may neglect The initial orbit is circular. of the initial , becomes arbitrary. circular reference is the orbit plane. 3. orbit is elliptic. The final the initial circular incides with The line of intersection of orbit co- orbit and the final elliptic a reference line such that S. 1 = S = 0. The reference line is the line of the nodes as in Fig.l. The transfer itial is in the outward direction. lies circular orbit orbit, with a, 1 N a S . 4. That is, the in- entirely inside the final elliptic 8 THREE CHAPTER Small Inclination Approximations Zero order solution ( i = 0 S To investigate optimum transfers ) 3.1 consider only is the that S transfer orbit is the final orbit The zero order solution ( is the i = 0 S (Fig.2). transfer The coplanar )a +a.)/2 S 1 = ((1+e e = ((1+e W is arbitrary. S <1. of we first orbit, specified by a S inclination Hohmann elliptic the of i element, orbit small the that problem from a circular ) 3 )a -a.)/((1+e )a +a.) I S S S 1 p1 = 1/2 , The primer vector constants can be easily obtained as 2 = p = 3 (1-e p The characteristic AVER 3.2 4 )/2 = p 5 = 0. velocity of the elliptic Hohmann transfer is (fl - - 1 + 7( ) ) . p2 (6) Study of the primer vector When the inclination of the final orbit is small but not zero, 9 the first order term in the of axis semi-major the of (Fig.3), the and vector primer plane. is impossible for the i = 90 . for except as in Ref.4, derivative its orbit initial the transfer) of impulses optimum number study the to orbit (1800 problem we are considering 2-body depends only on the inclination of AV final transfer The elliptic Hohmann of i To the dominent terms is (when is small but are odd nonzero) can be expressed as p esin v + p2 1$ p sin v / 4 iv 2 2sn 1 - The radial functions and normal v and of the p4 (e+cos v) components derivation may be three impulse transfers. of the primer vector from the elliptic Hohmann transfer The first impulse can be at v1 = 0 and the other two close to the apocenter of the transfer orbit. optimum conditions v2 = v (1) and (2) give a family of The transfer orbits with p 1/2 2 = (2+e 2 (1-2cos v-cos v)+2(1+e) JF* (1-2e-ecos v)) p2 = p (1+e) lI (1-2e-ecos , as an independent variable, v) (7) pp4 = e 2 p 2 1 cos v ( (3+2e+ecos v) (1-2e)/e 10 The limiting solution of = 1/2 p = (1-e2 )/2 gives , p v -w (8) p 2 = e 2 (3+e)(1-e)/4 4 plot the orbit circular the initial are symmetrical three equal maxima on the transfer impulse exists plane two types a (1-e S S orbit three impulses Actually the three distance of the final orbit a (1-e impulse <<l transfers (sections S 3.3, range in which the pericenter distance is neither very close to the radius of the initial a nor too large where the bi-parabolic transfer dominates, impulse first order solution requires to determine , vi i, W Since v e 0 1 derivative , v 2 , pl S 3.4) for larger pericenter distance. ) e, a, third solution. transfer In the practical is coplanar to of the theoretical First order solution ( i 3.3 and v 2 against Then the primer vector has axis. transfer optimal with the bi-parabolic second physical at the vicinity three of the and transfer orbit. first The variation of the pericenter indicates the by (8) transfer orbit first the the semi-major to and the primer vector magnitude of Theoretically (Fig.4,5). (7) ) and transfer of solution the illustrate can ) We .... p5 circular the 11 unknowns: of the first transfer orbit. M 0 , the primer vector and its r , p 2, p ~ v 3 5 2 S v 2 can and the nondimensional velocity vector at v 1 and , v be expressed approximately as 11 p esin v + p 1 3 ( p (1+e)2+ p )/(1+e) I ( p sin v + p )/(1+e) 4 1 5 p esin v 2- p ( p 1 (1-e) 2+ P2 /le ( p4 sin v 2 - p5 )(-e) I p (1+e) 2 + p (1+e) 2 1 1 1= - 3 Fp3 - 2 L - 2 esin v 1 + p3 (1+e) p 4 (1+e) p1 (1-e) 2 + p2 (1-e) - p3-e) p2 esin v2 P4(1-e) esin v/b AV 1 1 AV b 1- cos i s in i Oe sin v /b3 - esin AV 2 1 AV 2 b3 cos(is-i) - b 2 b sin(i -i) S 3 v2/b2 b1 1 b2 F where b3 + 1 es' b1- 1, C= c , 12 . = b 3- b 23 2 The optimum conditions (2) X sin v + Y p3 p 4 give = 0 (9) = X e = 0 X2 Y = (4+5e-e3 )/2 Y = (-4+5e-e )/2 The dynamic relations (3) (p1(+e)2+ p )/(l+e) give =1 (10) 2 (p (1-e) (p 4 + p2 sin v + p)/(1+e) 1 p5 )1e =1 = sin i /c1 (11) (p sin v - 4 2 p5 )/(1-e) = - b sin(i -i)/c s 2' 3 , where 4 , x 2sin v 2+ Y2 3 13 W sin v + p = 0 1 1 3 (12) v- p W sin 2 3 = R 2 = e(p - 1/(c b 1 1 1 )) W = e(p + 1/(c b )) R = W 1 2 2 1 2 2 e sin v /(c b S S 2 3 ) where 2 The geometric relations (4) and (5) give a(1-e) = a. (13) A = +v a (1+e S S ) a(1+e) = 0, + v2 ) sin i sin( We notice that anomaly of impulse case. the (9) sin( sin i = the second Once w transfer (9) - 11 unknowns of the first transfer orbit. to justify our estimate formulations of v . 12 equations for v of the 'final* S orbit at arrival is designated, vS (15) + v) (15) actually consist of - The additional one is 12 unknowns. true (14) orbit or the in the three (14) uniquely determine Then (15) can be used Remember that we will have similar S in chapter 4 and 5. Incidentally the optimum conditions (1) are the same as the dynamic relations (10). Now the approximate solution of the three impulse transfer a = (a S (1+e S )+a.)/2 1 , a circular orbit to an elliptic orbit is specified by from 14 e (a (1+e )-a.)/(a (1+e )+a.) S S 1 S S 1 = p1 = 1/2 , V1 2 p2 = e (3+e)(1-e)/4 , = (1-e 2 )/2 4 sin i = p ) 0, sin v sin v c b 2 b sin i /(c b 2 b + c b 2 1 2 3 s 1 2 3 2 1 = = 0 R /W 2 2 2 p5 = 0. velocity of the three impulse transfer AV = AV + EH 1 a(1-e 2 ) AVEH is given by (6) where 4 s = eI(3+e)(1-e) /2 (16) p 4 is, at this time, chosen such that and = - eI(3+e)(1-e) /2 0 , ,r given by s order effect of inclination in the first p p i sin is ( The characteristic AV. is positive, i.e. 1 7r < w 2 15 S (l-e S ) ) of Numerical (3.3) section the where region shaded is transfer the , 1 solution the in a. = line in Fig.6. order a.i a (1-e S ) : S E and denoted by i S S first pericenter distance a a (1-e ) : a valid be cannot 31) by the that shown have (denoted << 1, by the solid and is indicated impulse studies S pericenter distance the If one ( i First order solution 3.4 Incidentally (denoted by 311). are respectively the regions of 8 5 The validity of the first order solutions of Edelbaum and Marchal . the regions layer locate these dotted lines is impractical because it To effects. since these are not exact, cannot be uniformly valid due to boundary order solutions first regions these separate which lines dotted M will require extremely long computation. as considered the total first transfer that to similar approximation solution of orbit is of still S coplanar three impulses are third impulses can small compared be located to away AV1 from family of three impulse transfers anomaly, v, as an independent variable. S (Fig.7b) the elliptic at a three impulse Theoretically initial the the elliptic 6V 3 increase the AV . All the , therefore the second and the contribute to the inclination term in the total a orbit and and 3.3. to and shall be dominant in we arrive section 31. initial 1 final orbit 0 The 0 the S 6V and second and third impulses 2 1 * apocenter distance, therefore 6V must be opposite to orbit to coplanar to the transfer is actually impossible, Hohmann inner transfer (Fig.7c), actual the of AV the single impulse approximation the of AV Since the became the initial Suppose that this nominal orbit 0 we wish to find the optimum transfers study a,= (a (1+e )+a. )/2 that such order first to 0 orbit nominal 0.11 (Fig.7a), orbit circular interested are be a choose may we First can 311 which solution we section this In specified apocenter, AV by . (7) and they Then we have with the true 16 The coplanar transfer orbit is specified by = a.(ta cos v -a.)/(ta (1+cos v) -2a.) a 1 = - e where = t 5 S 1 (ta -a.)/(ta cos v -a.) s 1 S 1 2 )/(l+e cos v ) S S S closely approximated (1-e transfers is study perturbed the solution. Since this family of three impulse by single impulses, The characteristic we will velocity not of the three impulse transfer family is given by where 1.< f p - T. ( 1 ) + -1 1 T 1- 2 a(1-e p i sin w S P4 S ) Av. 1p = e p (3+2e+ecos v) , 0 4 e p +2e+ecos v) , r < WS w (17) 17 FOUR CHAPTER Small eccentricity approximations Zero order solution ( 4.1 If the zero the then e = S inclination of the (Fig.8). chapter 2, = S Hohmann I tilted or transfer Hohmann final ( e solution order 0 ) 4 The transfer must the transfer is be 0 ) is nodal. as defined that transfer coaxial also is orbit and 0 of the between with The 0 < i <rr, generalized two circles assumption transfer 3 orbit of is specified by a = (a +a.)/2 S 1 e = (a -a.)/(a +a.) S 1 S 1 0 i . Wi= can be determined b22 2 b1 sin i from b -2b cos i +1)/(1-2b cos(i -i)+b )) 2 S 2 1 sin(i -i) S (18) = b2 and constants vector obtained The primer and (2) and dynamic relations (3) are P1 = p2 p 5 where the optimum 2 2 (c2b (b -cos i)-c b (cos(i -i)-b 2))/(4ec1 c conditions (1) b (b1-cos i)/c1- p1b, = b 2 sin i /c 1 c = 1 p = 0 . = 1 b 2 -2blcos 1 1 have p from ' b where i +1 and 2 2= 1-2b cos(i -i)+b We also 18 The characteristic velocity of the generalized Hohmann transfer AV 4.2 2 Fb - = b cos i +1 + b - 2 csi )i+ (19) Study of the primer vector When the eccentricity of the final the 12 is derivation from the generalized three impulse transfers. orbit is small but not zero, Hohmann transfer may be two or The dominant terms of the primer vector and its derivative can be expressed as an p 1 esin v p1 * + p 2 /* J1 p cos v /* ,,5 2 1 - +2 p esin v p sin v If we assume that the first impulse for a three impulse transfer, we obtain is p 1 /( L + 2 =FI 2 ,( = M) 121 p2 5 -1 k1+e) -p 1 (1+eJ 2+2 +p2 )2 at v = 0 as in section 3.2 19 L = 2+e 2 (1-2cos v - where cos 2 v) M = and However, when we consider the limiting solution of imaginary (since = We of type that conclude discussed transfers are only transfers perturbed in of p 5 becomes with the value e p1 -(3+e)(1-e) therefore the 1) v--.Pw, . p e 4 v-2e Cos V). V(-ecos chapter the from there is no three 3. The impulse other three coaxial type. coaxial and nodal type are the transfer Therefore two impulse impulse dominant and will be discussed in the next section. If nor the too impulse large a, change order ), and the the final orbit is neither bi-parabolic e, i, first impulse we may put cos(i -i) t:; , p p^ 0. is cos expressed as p esin v + p = 1 3 ( D (1+e) 2 + p )/(l+e) p /(1+e) dominates, the two pP 5 of the transfer orbit. W S 0. at Since the ( less than 6* in zero small i : . Also we have The primer vector and its non-dimensional velocity vector 1 close to zero ) requires determination of 11 to the case of the v the at transfer , v1 , v 2 ' p1 . ') . W solution is restricted plane v2 where of order solution ( es<<l first unknowns: This inclination 1 ) First order solution ( e s< 4.3 v1 and v "" 0, derivative v2 can be 20 2 1 2 n v - p p esin 3 Lo L 1 p (1-e2+ p )/( ~ -e) p /(1-e) rp (1+e) 2 + p2 (1e - p esin v + p (1+e) - p (1+e) + p sin v 2 1 3 1 p (1-e) 2 + p(1-e) P73T - p esin v 2 Lp (e) O - 2 p (1-e) 3 + p5 sin esin v /b A 1 ~c1 AV1 Cos i 1 sin i Fe 2 AV 2 2 2 b 3 cos i - b 3 s 2 - where b = 1+e b2 b3 es' sin v s/b3- esin v2/b2 b sin(i -i) 3 S 21 C= c2 b-1 /b - 2b2 b3 cos i+ b. The optimum conditions (2) give 0 X sin v + Y p - p p 1 2 13 45 (20) 2 2 3 2 where = 0 p p , X sin v + Y p - 4 5 X = p e( p (1+e) 1 1 1 2 + p (1+e)) 2 - ep + p2 /(1+e) 2 5 X = p e( p (1-e) 2 1 1 2 + p (1-e)) 2 - Bep - p2 /(I-e) 2 5 Y = (1+e)( p(l+e) + p + 1) 1 1 (1-e)( p1 (l-e) + p B = (b3 B) , = - + Y osi- b )/c The dynamic relations (3) give ( p,(1+e) 2 + p2 )/(1+e) = 1 (21) ( p,(1-e)2 + p2 )/(-e) p5/(l+e) = B = sin i /c, (22) p /(1-e) 53 W sin 1 = b sin(i -i) s v + p 1 3 = /c 2 0 (23) W sin v - 2 2 p 3 R 2 22 where W = e 1 ( p - 1/(c b )) W = e 2 ( p + 1/(c b )) 1 1 1 1 22 ) R = e sin v /(c b 2 S S 2 3 The geometric relations (4) and (5) give a(l-e) = a., (24) ) a(l+e) = a (1+e S S sin i The = (25) 0 sin( w + v2 ) of solution = sin iS - (20) (25) sin( w + v ) (26) , + V , W gives the two impulse transfer specified by = (a e = (a sin i (i = S S = (1+e ) + a.)/2 I S (1+e S ) - a.)/(a (1+e ) + a.) I S S I c b2 b sin i S 1 2 3 - V, v1 p =b 2 sin i /c 1 1 5 = (1+e-B(1-e))/(4e) p = 1+e- p1(1+e) , p / (c b2 b cos i + c b 2 1 2 3 S 2 1 ) a 23 p 3 =WX R / (W W (Y1 2 2 1 2 1 sin v= sin v p4 2 Y )-X W - X W) 2 1 2 2 1 -p3 /W1 = ( R + p ) / W 3 2 = ( X sin 41 v + Y p 1 1 3 2 p 5 The characteristic velocity of the two impulse transfer is given by A b2 -2b cos i + 1 1 J + b2b b cos(i -i)+b 2 s 2 23 (27) 24 CHAPTER 5 FIVE Large Inclination And Eccentricity Approximations from transfers impulse is valid for to is objective Our large a circular inclination approximation an determine orbit to an elliptic and eccentricity. to obtain uniformly valid range of inclination and eccentricity. make orbit, which and eccentricity approximations over a larger These improved approximations insight into the nature of the transfers which is needed yield further to optimum To achieve this, we will extend the analyses of the small inclination approximations to for assumptions reasonable arbitrary inclination large and eccentricity. small The from order at the coplanar may be extended to second . The two impulse Horner14 approximation solution H to a fixed point from transfer inclination of K (Fig.9), with plane change only K gives the characteristic velocity AV = I AV 7(x AV 1 = where + 1 -1) (f 2+res -F AV 2 x = r = s=-l (28) AV, 2 (1+i2 /2) s -r)/(1-r (( )) (1-r2)/(-r/ (29) , a.(1+e cos v )/P S S 1 S = a. e sin v S / ( a. (1+e cos v) +J S = x2 - 1 2 2 a = a x /(1-e , e ) The transfer orbit is specified by )j -) i = 0 , 25 = 0 V , is arbitrary W v2= cos-'((rx2 -1)/e) The primer vector constants can be easily obtained as p1 = 2 1/ (x (1+r)) P2= r x / (1+r) In Horner1 1, ' 0 and . p 5=0 5= p 3=p 4 4 p3 AV 1 + A V2 is minimum in (29), be applied in the regions and A V. Equating AV 1 of the inclination 2 1 = a(l-e p4 = = - The regions obtained WS and (28), we have < w P4 - X.sin i e/(3+e)(1-e) /2 , the limiting value s - 2( A V + AV 2 1 validity of p4 0 s AV EH )AV) 2 WS have 7r/2 :.e: 31r/2 been reduced * line of nodes and the in-plane firing angle The stated half is associated with the short transfers the to half of that The remaining half are transfers symmetrical to the while 3w/2 and , in chapter 3. transfers) w i s = e(3+e)(1-e) /2 of and w/2 ) where < of (16) I 1 ( .sino s 1 AV2 = L i L 0 AV can only remaining half is has opposite sign. associated (less than 1800 with the long 1 b323Sb3 26 Local minimum solutions of the transfers (more than 1800 transfers). the remaining half this illustrate AV > AV. 1 is AV for i order solution; the w > S We will of examples i L approximation final orbit, wS important when is eccentricity becomes effective. where numerical and half L eccentricity of stated AV. larger four first the we now have for argument they have the AV I in respectively but in symmetry small pericenter first regions, As in Fig.10, Appendix C. The exist, transfers With A be may extended the if in the , can be included the second order term in 2 5 = a (l-e )/(l+e cos v S S S S ) short long and v is defined by S v + S v 5 W.^: , if sin w , if sin S + W' 0 S 0, W S 0. .< The geometric relations require = a ( AS + a.) / 2 1 = ( A e 5 - a.) / ( A S + a.) 1 1 Except that the transfer we obtain using the . (30) orbit elements a and e are defined by (30), a further extension of the small eccentricity first 4.3. order solution of section approximation The characteristic velocity, similar to (27), is given by AV = b2 -2b cos + i + 2 b -2b b cos(i -i) +b 2 _ 1 (31) 27 where bi , b2 and b3 are calculated from the new AV and AV can be compared to determine E e becomes significant. are valid The extensions cannot be sufficient orbit. However, the limiting only within second order, provide e of (30). e s when a two impulse w s and therefore transfers to a general elliptic to describe they a and transfer mode final (Fig.11) with the following properties: 1. The first impulse is close to the pericenter of the transfer orbit, such that 0, and (32) i , such that , and so <i cos(i -i) S .- cos is S The second impulse is close to the line of nodes, v S + W VOW S 7w 4. p = The first plane change is small compared with i 3. 0, and so 1 1 2. v 13 such that as in Fig.lla, (33) as in Fig.llb. (34) Horner's solution can be used to estimate v2 of the trans- fer orbit. a give = a (A cos v -a )/(A (1+cos 2 i s i s v ) -2a ) i 2 , The geometric relations (4) e = (a -A )/(A cos v 2-a ) i s s 2i . (35) and vector primer The velocity vector at its and derivative v2 can be expressed as v1 and p esin v + p 3 1 1 p (1+e) + p2/(1+e) = (p sin v + p )/(1+e) 5 1 4 p esin v + p cos v2 3 2 1 P 4 + p /$ 2 2 1 2 = - p sin v (1+1/ 2 3 2 * (p sin v + p cos v )/ 2 5 2 4 4 ) 2 2 p (1+e) 2 + p2 (1+e) P - p esin v + p (1+e) 3 1 2 3 - p (1+e) + p sin v 1 5 4 3 =- P2 - p esin v + p (e+cos v 2 3 2 - p (e+cos v 4 = 1 J7 Aici sin i 2 e esin v --A 1 2 sinv ) 2 3 2 2 p1 2 - cos i ) + p sin v 5 2. the 28 nondimensional 29 e sin v * Cos i S s C2 -* where S 1 2 s- =2 2 S - = c 2 sin(i -i) 2 $2=1+eosv 2 s = 1+ecos v $) is + 2cos $ -1 AV2 ' 2 v2 , AV 2 esin are known quantities. The optimum conditions (2) give 5 sin v 1 )/(1+e) X sin v + Y p + (p sin v +p cos v )(-p (e+cos 4 2 5 2 2 3 4 2 2 = 0 (36) , X1sin v1+ Y 1 p3 + (p4 sin v 1 +p5 )(-p 4 (1+e)+p v )+p sin v )/ 5 2 2 = 0, , 2 (37) 2 - p 2e X, = p 1 e(p 1(1+e) + p 2 (1+e)) x =p (p 1 X2 * 2 V ) -p e(p +p 2 1 2 2 Y, = p1 (1+e) 2 + p 2 (1+e) y = 2 The dynamic relations (3) p1 (1+e) + p2 /(1+e) + 1+e ) p Cos v (p* 2 + 2 1 2 2 1p 1 +p 2 ~J 2 2 , where ' + (e+cos v )(p 2 2 1 + P /i) 2 2 2 give = 1 (38) p1 2 + p2 p3 sin v (1+1/ = C) is (fscos )/c, p esin v + p = esin v /(c fl7 1 31 1 1 ) , 30 (39) = (essin vs- (p sin v + p )/(I+e) 4 1 5 = esin sin i / c v2)/(c2 , p esin v + p cos v (40) (p sin v + p cos v )/ $' Since v2 is estimated by = - I sin(i -i) Hornerts solution, / c. p1 , p2 ' p3 and v1 can be solved by (38) and (39), giving p1etan v 2 +(e ssin vs s esin v2)/(c 2 cos v2 r (42) ) p1 = (tf+g-(1+e))/( *2 +tm-(1+e)2 2 = sin v1 where g (43) (1+e)(1-p (1+e)) = = p3 cr~iT / (e-p 1 ec1 '2 (PsCos is - = (e ssin vs - t = m = e tan (44) / C2 t 2Sesin v2 Mc 2 cos v 2 Ws ) f )i ) p2 (41) , = - p v2 * 2sin v 2(1+1/ i2 p = n 1sin i + n p = n sin i + n , We can now solve (40) for p 4 and p5 in terms of sin i, as (45) (46) 31 2= - 2 sin i / (c cos 2 is' '2 2 n3= cos i -(1+e)c cos v )/(c c cos v (tan v -sin s 2 2 1 2 2 2 (1+e-cln sin v )c, 1 n4= - 1 v (tan v -sin v 1 v1 n2 sin We can now solve sin i from (36), A (sin i) A = where (n2 1 B = + B sin sin v 3 and (46); (n n + n n /(1+e)) - n n 1 3 - sin v 3 4 1 sin v - n n + Y p - n n n n 2 3 1 4 1 2 4 1 3 p 7 unknowns solution of the i.e. (47) 1 C = (X + n2 + n2 /(l+e)) 1 2 4 The (45) = 0 , + C + n2 /(1+e)) 1 2 2 2 2 v 2 ) n = (c 1 1 12 ) where p , i obtained from the 7 equations (41) ,sin .5 guess arily satisfy (37) because of our initial D , , sin v p , p (47) does not necessv . Theoretically the However, is satisfied. such that (37) 2 solution obtained from (41) - (47) is accurate enough to initialize a we v can recompute good set velocity of of input the two for the computer impulse transfer program. for a The general characteristic elliptic final s 2 (48) orbit is given by 2 ) 2 V-2 Iwcos(i -i)+ AV =E (t2+e-2q7 cos i + 2 1+e s ]s(2 32 SIX CHAPTER Computation Technique 6 Optimal control theory 6.1 This He formulated only a to section set a condition of necessary necessary He further found positive definite. If conditions characteristic explicit the th Small's thesis primer vector a quartic to relationships when this independent in the concept, variable rate 4 to have and reduced these conditions for to another velocity, for orbits, Using the analysis basic conditions from one impulse developed. the the single maximum on elliptic definite. be the explores quartic a is for be the just computer chosen of change of positive switching ceases program was proportional state and to to adjoint x d.(x., u). = J , variables are respectively J i (49) =, . . . ,5) aj. J S where _ 3H , (ij =1, the , unit vector in the direction of thrust, and u, the angular position of the radius to the vehicle measured in the instantorbit plane, are controls. A.d. = x. is an arbitrary set The Hamiltonian defined by of orbit elements. H The state . aneous JJ is constant X.d. the e h If is summed optimal as the A. (j along J over J u is 1, j = 1, ... and may be normalized , 5. that which maximizes radial = an extremal, ... and , jJJ adjoint variables X. J angular momentum 5) are functions The optimal - ) 2 . 8 to unity. is XA.d R R unit vectors and e= eh L h x. of the state variables (2. d. Jj We define And , and and J eR. R* and X.d. A = A ) e R33 sin (u + 1 si 2 [ A 3 + (1+$) + A 1 cos (u+A 2 eL (50) -1 +4i [ A4 sin (u+A5 eh H requires that The optimal u which maximizes 2 - [X.d.(u)] J J -2 Au other for all 0 [.d.(u+Au)] JJ - 0, than 21 (51) ,...... This maximizing test can be written in the form 2 a 0 for + a t + a2 t 1 all t functions tan( = 3 4 + a 3 t + a 40t Au/2 ) and of the eight parameters 0 ai(i A. = , 0, ... e, v, and , 4) u. are explicit # a0 If 0 and let Z, = 4a a4 - a1 a3 z 2 CZ +a 2 Z2 2 = ay 1 a4 + a9o a3 3 then, the test for the a a1 a2 a 2 a33 positive quartic definite can be expressed completely by a > 0 (52a) 3 2 2 Q(a.) The left = Z1 [(a 2 member of Z1 )2 +9a 2 Z2 ] -a2 Z2 (52c), Q(a. ), switching condition which represents 2 27 Z equals zero /4 > (52c) 0. corresponds a double maximum of to the H, and will be violated before (52a) or (52b). The case a0 >0 corresponds case corresponds to impulsive extremals. to either impulsive firing included in The the a = 0 testing 34 conditions (52) or non-optimal by Marchal condition switching The firing continuous 15 be can was which reduced a to It is desired to determine independent parameters. From (51), we have 222 2 =4cos 2t (ao+a1t+a2 proved function be to of five a. from a set of five independent parameters. = 4 3 +a3t +a4t ) -Xd(u)] 2 X] [X.d.(u)] -I[X.d.(u+Au)] 02 (1+t 2)[*-2te sin v + (2-,)t2 2 (53) and the particular set of a t a s is 2 2 2 [1-4k -(2+ *k )tan * = a, = 44 tan ] - .2 j (2k-cos y)-2esin v (1-2 tan 2 a3= a1 +8(-1)tan The = [(3-4)) 2 cos y-2k][2k-cos y] - (*-1) sin y firing angles having the limits of 0 4 y.4 , 0 accord with the geometry and cos 4 ] 5 . a 2 2 (cos y-k)+4esinv [(cos 1-k) -tan 2 ] , a 2= a +a +8esin v tan *(cos y-k)-4($-1)[(cos y-k) 2-tan 2 and 2w W/2 4lr/2 The thrust vector in (53). is defined by and the parameters k and cosy j eL + cos4 sin y eh (55) are defined by cos (u + A2)]/ k 5 [A 3 + A j S {A 4 sin (u+A5 ) sin $ + A * cos * ) sin 4 eR + cos4 = cos (u+A5 ) cos * cos Y + [A1 cos (u+A 2 ) esin v - A1 sin (u+A 2 ) ecos v] cos 4 sin y} / cos 2 [A4 cos (u+ A 5 )/cos *+ esin v sin i ] cos y + [tan 4 - k e sin v] sin y . = (56) 35 On using H = 0, we have the functional relationship U j sin y + k e sin v tan * (57) 1 + k $cos y can be chosen as A desirable set of independent parameters ( e, v, = 1/2 set an initial Having . , y, k 1/2 ) q q1 now define we , nondimensional the * independent variable h T= - AV which is desired to yield after impulse an The radius ( e axes an impulse, during for determined , therefore e h /* given T h* and h is a R XV. to the Figure 12 remain constant are constants. terms in = firing with respect f and 2 set . per unit mass ) in 2 , expressed increment velocity a R and the direction of vector orthogonal fixed be value of the angular momentum reference the can T(r ) so that the independent = of the Equating instantaneous elements and that of the impulse, three integrals result [f$' cos y ~ sin h AT ] cos (58) * A = * e sin v [F sin These determine in terms of during AT y ] = 0 the three independent . A . h an impulse. The remaining two Thus, we have parameters 1/2 k*1 and q = ( e, v, Y ) of q .- 1/2 j41 are constants q( AT ) which leads to the 36 important functional relationship (59) a.i(AT) = ( a. F Finally, (52c) and (59) give the switching condition in the form > Q(Ar) (60) 0 where the left hand in h 6.2 side can be expressed as a 12th degree polynomial 1/2 AT /h* Two point boundary value problem An optimum orbit transfer which may the initial adjoint variables be solved if the general i relates = ( p an input variables initial A. adjoint , 1 adjoint the vector are known. p are In arbitrary A linearized solution establishes an estimated independent variables. set of of elements five is a two point boundary value problem ... ), p , _5 set of q defined in ( e, v, Y = (49). A. variables the optimal control while Therefore are formulation 1/2 . -1/2 ) to the adjoint ,,k$1 , jiP known if an the estimated set of input can be q J_ calculated from the estimated p V, A First we define elliptic orbit (Fig.13). R, , AR , K, 4, 3, of the first transfer 37 = P/) eR esin v 0 + p cos .3 p P1 + /$ + p 2 si - p sin v (1+1/ V/ + p5 cos v / * Lsi 2 p 4, + p2 p3, sin v - p esin v + p (e+cos v) - p (e+cos v) + p5sin p sin v + p cos v p cos v 4 AKF - p e 3 ,. 4= e 0 3= 4, ' 1 $ V v ) p esin V - sin i L cos i 0 p sin 5 v v 38 i is the same as the velocity adjoint x If we define the state vector = - 1. ~R . I - * = V .V cos v /* V'(sin h1 v eR+ cos v eL ' Xe .R- R , =2 + S o ) where S , then by contact transform, we have h, L W (61) h' , L = log ( L, e, i, = . where the primer vector = K 4 = K. 3 and N I sin v e V cos v /e (62) )cos v )+ x(1+ * )sin v /e ~ (L+ X (e+(+ / X~4.cos u +( ) -(x +x)cos i)sin u A i It is = d. 0 - X sin ij that Small proved in U ) h X V / h* V and so (50) and (62) give Q = 1 -A 2 0= O '3 cos A A 1 = 112 A 3 = h ( X 2 sin A h ( X cos sin 03/ e) / h* e = L h ( X sin e + e e )/ 03+ h*, X cos oi/ e) / h*, (63) 39 Q 5 =h( X A 4 cos A 5 Q4 . / h* =h = A sin A 5 4 cos i )/ (h* sin i) - 1 k give The definition of the firing angles and cos$ A 4 sin( u+ A 5 siny cos A 4 sin(u+A 5 = k =( A3 + A cos( sin $ = cos $ 1/2sin y Q1 sin u - Q 2 cos u u + A2 = ( Q 4 sin u + Q 5 cos u ) / e Now with an initial circular orbit, constants -1/2 (65) = ( Q3 + Q 1cos U + Q 2 sin u ) / *-1/2 * 1/2k q= ( is reduced to (64) ) Finally (63) and (64) give on using (50), (55) and (56). cos $ )' / = , sin an during , k. impulse , j. ). and u = v. 0 ,4)=1 $ Since A + V, ,p /2s = 0 , , the input are ,1/2k (65) can rewritten as 1. Q2 = - ' sin cos *. sin Y. 1 1 cos *. k. 11 i3 = = / *-1/2 Q 5 ( 0 + Q 1 ) / *-1/2 Equations (57) and (66) enable us to compute the input (66) be 40 q which in = ( 0 , 0 turn , gives . the k. , j. ) initial point boundary value problem. (67) adjoint variables to solve the two 41 CHAPTER 7 SEVEN Conclusions The ion to general linearized The transfers. of this solutions or analysis of analysis general simplified, but in solution. However, the provides thesis we of approximations transfers still cannot approximations a from impulse optimum circular obtain are formulat- a simple a analytic unique estimates excellent is orbit since the numerical solutions have rapid convergence. An only important exist at the transfers cannot transfers from long theoretical transfers is vicinity of Hohmann provide a result minimum circular as are orbit that 3 transfers and when The solutions. are not illustrated simple in the of Small impulse due first transfers the Hohmann symmetries to the four of short and examples in Appendix C. The extremal condition quartic using defined equation double computed generation by contribution can generally some of approximations this and There are often not compute mations. every approximations of The is solution 8 digits are The practical of scheme of this solutions space. combination of of accuracy numerical state generation the changes matching the of this between characteristic (for exact in the analytic however, trajectory The The the analytical Small to provide solution. combinations possible, the errors extremal different inclination. equation. therefore thesis the consists of a switching be computed up to acceptable the exact numerical and quartic precision, with different a scheme problem any approximations in semi-major these is not axis, approximations necessary obtained achieving the with eccentricity since two families of analytic velocities of analytic associated from same we is can approxi- two analytic final orbit), 42 are local minimum but approximations, some near are trajectories In solutions two families analytic approximations case, this transfers may two to lead different of families if the analytic solutions are identical. the from farther solutions. numerical different boundary different the approximations, analytic of between In for of slightly numerical respective especially cases, boundary the the may be which one is and the better both are local minimum solutions obvious. limits The the validity of define, although to difficult in the changes of analytic different the close of Results compute. are approximations trajectories some inclination and eccentricity extensive not computations are limits obtained. When the eccentricity of the final orbit is less than and/or inclination the of the final orbit is to are the to difficult for less than .9, 60*, the of the numerical solutions usually converge rapidly. This parking orbit guidance to an thesis the improves problem technique. which could Knowledge elliptic orbit in theoretical understanding provide the basis of transfers 3-dimensions could from for a subsequent a circular improve the orbit operational envelope and fuel economy of the space tug. A good transfers in 4, a the project 3 dimensions. parametric simple problem research figure. will be the general ellipse to ellipse Since the number of input parameters is space will no longer be visualized or constructed in A similar approach as that of the circular orbit should give us some theoretical understanding, solution will not be forthcoming in the near future. but the general 43 Appendix Optimum single impulse transfers from circular orbits A This appendix continues basic of analysis the study of Der t s M.S. space parametric and space. considered in intersect, the firing angles and the normalized can be state determined precisely, unique. necessarily but Figure 14 between two intersecting orbits. then the change in h '/h = = D and I = T ), adjoints are not physical transfer T h cos h is + RX oA (1+ D cos y /fr) cos E eh cos i' , ~ - = 3 D sin y /fi) eL But sin E eL eh = cos . and a , orbits If we define (values after a firing are primed). eL (y AV, initial illustrates two When is * C = h'/h the The . from a circular orbit transfers optimum impulse 9 thesis Therefore we have C cos i' = 1 + D cos y /F C sin i' = D sin Since C and D are given, we can determine C I/F = cosy With a fixed have i y/F f T - 2 C cos i' ( C cos i' - + 1 , (Al) 1 )/ D . (A2) and the given initial and final states (x , x'), we 44 * tan 'sin v'/ = - e sin v/ )I D (A3) on using the first equation of (58). From (57) we also have * tan = (j sin y + k e sin v)/(l + k * cos y (A4) . above, we have D * Finally from the definition of ) T $ D /(h cos # ). h = the When = and orbit initial (h/h) IC2 - 2C cos i' D = (C cos i' - tan * = e'sin v' /(DC) tan = j =h a for plot of single values of -1/2j of (*-1/2 , (i) (ii) (iii) 1/2k (A9) , (Alo) , , t T ) ) uniquely from from a circular in Figures (A7), (A8) and orbit, 15a,b reveals are not unique in general. but the that the The solutions can be: * max (A8) . , and *1/2k 1, = i (A7) , impulse transfer 1/2 1/2 (* 0, (A6) +k cos y ) ( y We can determine (AlO) )/D ) D/( cos 1, e = , cos T h = (Al) to (A5) can be simplified to + sin y /(1 then circular, is h'. = (A5) > 02 , no single impulse solution, , a unique solution for (-1/2 j1 = 0 , a family of solution for (*-1/2 2k), 1/2k). shows are the unknowns the non-uniqueness in this equation. Figures 15 a,b also (initial adjoints) multiple impulse within the boundaries For to But solutions since j and j k are adjoints, k thus ki transfers from of of the input variables show the boundaries j* 1/2 and discussion on 9 M.S. thesis. further Der's the adjoints are not unique in general. the initial All of j- 1/2and these 1/2at an out of plane angle a circular k$1/2 boundaries, orbit must originate associated with the the reader y . also should y . (A9) 45 and refer 46 subroutines 3 Appendix PEST, XKUJ, MAINE, subroutines Small 4 and OUTPUT MATX circular orbits. SWITCH, TIME START, are program applicable only are DTDU 2 and to extremal-generation The and four the and RUN developed by The program logic is illutrated in Table Bl. . The extremals at with the nominal derivative and are trajectories will matrix which variables adjoint the beginning of to run the neighboring's at i=2,3,4. required are for the computing partial used in a 3-dimensional Newton method. be for of each are stored at I extremals the the J impulses in the array S(I,J,K). of each and I=1 (I) arranged subroutines extremal-generation neighoring State main The C. from transfers impulse The of results generate to used program computer the is listing following The the Computer listings B Appendix The K variables are identified in Table B2. of computation is The procedure specify in We first 1. I =1, J1 summarized as follows: RUN , IM = 4 JM = 2 or 3, , = an initial state (a circular orbit), = a final state (an elliptic orbit). S RUN calls PEST to estimate the primer vector x j = 1,...,5, and PEST calls XKUJ to compute the constants input vector p., J q of (67). 2. RUN calls MAINE extremal-generation initial calls (52). values SWITCH of to which is scheme. the S(I,J,K) check the the driving MAINE calls array input subroutine START to for K = 1,......,20. satisfies the of the set the START inequalities of 47 3. order first which update the not next 4. the step 3 is JM, impulse J+l. calculate I=1, at the given final state and nominal x xz= to Au, find T, by a check the and calls DTDU the exact If T to J+l of the MAINE returns to RUN. the the to impulse J+1. next repeated to trajectory nominal the J+l equals JM, If OUTPUT RUN calls for array AV, found, MAINE calls is T angle the coast calculates S(I,J,K) equals When the exact (52). inequalities of SWITCH calling and method Newton SWITCH to find the exact normalized TIME calls MAINE final norm of state x(I). For the difference of final state -X(i) is computed. 5. If xz step 2 not is 4 for each increment of the input method. With may ical to be repeated than less the Then I. vector new input until 10 xz solution is acquired. is , we put MATX to compute RUN calls 6q vector less than by ( q 10 I=2,3,4 and repeat a 3-dimensional + 6q the Newton ), step 2 to 5 and the exact numer- 48 Table Bi Diagram Flow I RUN MATX PE ST I - XKUJ MAINE I S TART f ,F I: TIME SWITCH DTDU OUTPUT 49 Table B2 The variables stored in the S(I,J,K) s(I,J,K) K S(I,J,K) 1 T- 11 * 1/2 (cos y - k) 2* AT 12 cos(u+ 1 13 sin(u+ A ) 1/2 tan(Au/2) ) K 3* * array (h/h*) *-1/2 cos 14* cos Au 5 *- 1/2 esin v 15* sin Au 6 k *1/2 16 cos i 7 tan 17 sin i cos u 8 A4 cos(u+ A 5)/cos 18 sin i sin u 9 *1/2 sin y 19 sin i cos 10 'p 20 sin i sin Information for the ' * 4 J th impulse is stored in S(I,J+1,K) 00010 C RUN 00020 C 00030 IMPLICIT REAL*8(A-H,O-Z) 00040 00050 DIMENSION S(4,7,20) ,DP(5) ,X(4,4) ,DY(3) ,P(5) COMMON/KLN/JM 00060 00070 00080 00090 00100 00110 00120 51 00130 81 00140 82 COMMON/STR/S COMMON/BIN/BPOBPTE1,V1,PI,DTFTAUl COMMON/FCOMP/X COMMON/FIN/H2,ECES,SNCV1 COMMON/TIN/ETVTWTPTSTSIT,OM COMMON/ONM/TPHI ,CUPSUP FORMAT(4G10.7,I2,GI0.7) FORMAT(4D10.5,I2,D10.5) FORMAT(/' CASE',13,//' INPUT',/,' H2 00150 00160 83 JM SIN I 1ESINW FORMAT(4(D12.5,3X),15,8X,D10.5) 00170 91 00180 92 00190 FORMAT(' OUTPUT') FORMAT(/' H**2 UP PHI 2 ECOSW DTAU') 00200 IM =4 00210 DP(2) 00220 DP(3) =1.D-6 00230 DP(4) 00240 Ox =100 00250 00260 00270 ZZ =0.DO ONE =1.DO PI =DARCOS(-ONE) 00280 00290 KN =100 El =0.DO 00300 VI 00310 00320 00330 00340 DO 1 I =1,4 DO 2 J =1,7 DO 3 K =1,20 S(I,J,K) =I*4+J*7+K*20 =1.D-6 =1.D-6 =0.DO ESINW ECOSW CLIM') SIN I TAU V 00350 00360 00370 00380 00390 00400 00410 00420 00430 00440 00450 00460 00470 00480 00490 00500 00510 00520 00530 00540 00550 00560 00570 00580 00590 00600 00610 00620 00630 00640 00650 00660 00670 00680 3 2 1 CONTINUE CONTINUE CONTINUE C DO 79 NN =1,10 READ (5,51) H2,ECESSNJMCLIM C 11 12 WPITE (6,82) NN WRITE (6,83) H2,ECES,SN,JMCLIM IF (JM-1) 12,12,11 CALL PEST(P,PS,PUPK,JM) WRITE (6,91) WRITE (6,92) C C 17 MULTIPLE IMPULSE TRANSFERS IF(PS.EQ.0.DO) GO TO 18 PHI =DARSIN(PS) PC =DCOS(PHI) UP =DARSIN(PU/PC) GO TO 19 18 19 PHI =DARSIN(PS)-PI*.001D0/180.D0 PC =DCOS(PHI) UP =DAPSIN(PU/PC)-PI*.lD0/180.D0 SIUP =DSIN(UP) XK =PK/PC IF 21 (UP) 22,21,22 XJ =0.DO GO TO 23 22 XJ =DTAN(PHI)*(l.DO+XK*DCOS(UP))/SIUP 23 DO 76 KL =2,KN 73 DP) CALL MAINE (ElVlXKUPXJONEIM,JM, CALL OUTPUT (1,JM,0,XXH2,0,OX) IF (DABS(X(3,1))-.D-5) 73,73,74 X(3,1) =ES H 00690 00700 74 XZ =DSQRT((EC-X(2,1))**2+(ES-X(3,1))**2+(SN-X(4,1))**2) GO TO 77 IF (OX.GE..179995D3.AND.XZ.LE..5D-4) 75 CALL OUTPUT(2,JM,0,XXH2,0,OX) CALL OUTPUT(3,JM,0,XXH2,0,OX) CALL OUTPUT(4,JM,0,XXH2,0,OX) CALL MATX(ECES,SN,DY) XK =XK+DP(2)*DY(1)*DMIN1(ONECLIM/XZ) UP =UP+DP(3)*DY(2)*DMIN1(ONE,CLIM/XZ) XJ =XJ+DP(4)*DY(3)*DMIN1(ONE,CLIM/XZ) CONTINUE S(1,JM+1,2) =DTF IF(XZ-1.D-7) 77,77,75 00710 00720 00730 00740 00750 00760 00770 00780 00790 00800 76 77 00810 DO 78 J =1,JM 00820 78 00830 00840 79 00850 CALL OUTPUT(1,J,1,ZZXX,1,OX) CALL OUTPUT(1,JM,0,XXH2,2,OX) CONTINUE END U, 00010 00020 SUBROUTINE PEST(P,PS,PU,PK,JM) IMPLICIT REAL*8(A-H,O-Z) 00030 00040 DIMENSION P(5),V(50),X(50),Y(50),Z(50),DF(50),F(50) COMMON/FIN/H2,ECESSN,CV1 COMMON/TIN/ETVTWT,PT,STSITOM 00050 ONE =1.DO 00060 PI =DARCOS(-l.DO) 00070 P12 =.5D0*PI 00080 FIRST ORDER ESTIMATE OF FORMAT (/' 00090 10 INITIAL STATE 00100 C AI =1.DO 00110 EI =0.DO 00120 00130 00140 00150 00160 C 00170 00180 00190 00200 TAU SI =0.DO WI =0.DO XMI =0.DO FINAL STATE E =DSQRT(EC*EC+ES*ES) W =DATAN2(ES,EC) S =DARSIN(SN) A =H2/(1.DO-E*E) 00210 C TRANSFERS, GENERAL 00220 C TWO IMPULSE 00230 00240 00250 00260 11 00270 00280 00290 12 00300 00310 13 CIS =DCOS(S) SIS =SN IF (W) 11,11,12 ECVS =EC ESVS =-ES GO TO 13 ECVS =-EC ESVS =ES OM =W 00320 00330 00340 IF (OM.LE.PI2.AND.OM.GE.-PI2) ONE =-l.DO IF (JM.EQ.3) ONE =1.DO ES =ONE*ES =',D15.10,/) 00350 00360 00370 00380 00390 00400 00410 00420 00430 00440 00450 00460 00470 00480 00490 00500 00510 00520 00530 00540 00550 00560 00570 00580 00590 00600 00610 00620 00630 00640 00650 00660 14 00670 00680 EC =ONE*EC BU =1.DO-E BV =1.DO+E AT =.5DO*(A*BV+1.DO) ET =(A*BV-1.D0)/(A*BV+1.D0) BS =1.DO+ET BT =1.DO-ET Bl =DSQPT(BS) B2 =DSQRT(BT) B3 =DSQRT(BU) Cl =Bl-1.D0 C2 =DSQRT(BT-2.DO*B2*B3*CIS+BU) BB =(B3*CIS-B2)/C2 P(1) =.25DO*(BS-BB*BT)/ET P(2) =BS-P(1)*BS*BS SIT =CI*BT*B3*SIS/(C1*BT*B3*CIS+C2*BS) Gi =Cl/(B4-P(1)*ET*Cl) G2 =C2/(P(1)*ET*C2-B5) RHO =(1.DO+ECVS)/H2 R2 =DSQRT(2.DO/(1.DO+RHO)) S =ESVS/(1.DO+ECVS+P2*DSQRT(H2)) DS2 =DSQRT(1.DO+S*S) Xl =R2*(DS2-RHO)/(1.DO-RHO*DS2) ETI =X1*X1-1.D0 CV2 =(RHO*X1*X-1l.DO)/ET1 SV2 =0.DO IF (ES.NE.0.DO) SV2 =DSQRT(1.DO-CV2*CV2) Wl =ET*(P(l)-l.D0/Cl/Bl) W2 =ET*(P(1)+1.D0/C2/B2) R22 =ESVS/C2/B3 IF (JM-2) 14,14,15 P(5) =-Bl*B1*SIT/Cl Xl =P(1)*ET*BS*(P(1)*BS+P(2))-P(2)*ET+P(5)*P(5)/BS X2 =P(1)*ET*BT*(P(1)*BT+P(2))-BB*ET*P(2)-P(5)*P(5)/BT 00690 Yl =BS*(P(1)*BS+P(2)+1.DO) 00700 Y2 00710 00720 00730 00740 00750 15 P(3) =Wl*X2*R22/(W1*W2*(YI-Y2)-Xl*W2-W1*X2) SV1 =-P(3)/W1 P(4) =(X1*SV1+Y1*P(3))/P(5) GO TO 16 P(5) =0.D0 00760 P(3) 00770 SV1 =0.DO 00780 00790 00800 P(4) P(1) P(2) 00810 16 SV2 =(R22+P(3))/W2 =-BT*(P(1)*BT+P(2)+BB) =0.DO =-.5D0*ONE*ET*B2*DSQRT(3.DO+ET) =.5DO =.5DO*BS*BT 00820 CV2 =-DSQRT(1.DO-SV2*SV2) 00830 VT =DARSIN(SV1) 00840 00850 00860 WT =-VT ST =1.DO+ET*DCOS(VT) PT =AT*(1.D0-ET*ET) 00870 00880 00890 17 00900 S2 =1.DO+ET*CV2 IF(JM-2) 17,17,18 CC =Cl+C2*DSQRT(S2/BS) GO TO 19 00910 18 CC =Cl+B2*(B3-B2)/Bl+DABS(P(4)*DSIN(W))*S/Bl/B2 00920 19 WRITE (6,10) CC 00930 00940 00950 CALL XKUJ RETURN END (PPSPUPKJM) 1-l U,1 00010 C XKUJ 00020 C SUBROUTINE XKUJ (PPSPUPKJM) 00030 IMPLICIT REAL*8(A-H,0-Z) 00040 DIMENSION P(5) 00050 00060 00070 00080 00090 00100 00110 00120 00130 00140 00150 00160 00170 00180 00190 00200 00210 00220 00230 00240 00250 00260 00270 00280 00290 00300 00310 00320 00330 00340 COMMON/FIN/H2,ECESSNCV1 COMMON/TIN/ETVT,WT,PTSTSIT QP =DSQRT(PT) QS =DSQRT(ST) SW =DSIN(WT) CW =DCOS(WT) SV =DSIN(VT) CV =DCOS(VT) UT =WT+VT SU =DSIN(UT) CU =DCOS(UT) R =PT/ST VR =ET*SV/QP VL =ST/QP YR =P(1)*ET*SV+P(3)*CV YL =P(1)*ST+P(2)/ST-P(3)*SV*(1.DO+1.DO/ST) YH =P (4) *SV/ST+P (5) *CV/ST ZR =(P(1)*ST*ST+P(2)*ST-P(3)*ST*SV)/QP/PT ZL =(-P(2)*ET*SV+P(3)*(ET+CV))/QP/PT ZH =(-P(4)*(ET+CV)+P(5)*SV)/QP/PT XKR =(P(4)*SV+P(5)*CV)/QP XKL =(P(4)*CV-P(5)*SV)/QP XKH =-P(3)*ET/QP SX =SIT CX =DSQRT(1.DO-SX*SX) XL =2.DO*R*ZR-(YR*VP+YL*VL) XE =-R*ZR*CV/ST+(SV*YR+CV*YL)/QP XW =XKH XI =XKR k-n 00350 XM =XKL*SX+XKH*CX 00360 C 00370 Ql =(XE*CW-XW*SW/ET)*QP 00380 Q2 =(XE*SW+XW*CW/ET)*QP 00390 00400 00410 11 00420 00430 00440 12 00450 00460 13 00470 00480 00490 00500 Q3 =(XL+ET*XE)*QP IF (SIT) 11,11,12 Q4 =0.DO Q5 =0.DO GO TO 13 Q4 =(XM-XW*CX)/SIT*QP 05 =XI*OP PS =Ql*SU-Q2*CU PU =(Q4*SU+Q5*CU)/QS PK =(Q3+Ql*CU+Q2*SU)/QS RETURN END 00010 C MAINE 00020 C 00030 00040 00050 00060 SUBROUTINE MAINE (EV,XKUPXJHRIMJMDP) IMPLICIT REAL*8(A-H,O-Z) DIMENSION S(4,7,20),T(30) ,DF(30),F(30),ZK(30),ZJ(30) DIMENSION D(4),DP(4) 00070 COMMON/STR/S 00080 00090 00100 91 00110 00120 COMMON/BIN/BPOBPT,E1,V1,PIDTFTAUl COMMON/ONM/TPHICUPSUP FORMAT (6D15.10,' FAILED GENERAL TEST') ZZ =0.DO X =0.DO 00130 C 00140 CALL START 00150 00160 110 00170 111 00180 112 00190 IF (NF) 220,220,110 IF (IM-1) 210,210,111 DO 112 K =2,4 D(K) =0.DO DO 113 IA =2,IM 00200 00210 D(IA) =DP(IA) CALL START (E,V,XK+D(2),UP+D(3),XJ+D(4),HRIA,1,NF) 00220 00230 00240 00250 00260 00270 00280 00290 00300 00310 00320 00330 00340 113 210 211 212 213 214 (E,VXKUPXJ,HR,1,1,NF) D(IA) =0.DO J =0 BPO =(DSQRT(2.DO*(1.DO+E))-l.DO-E)/HR J =J+1 DO 218 I =1,IM IF (I-1) 213,213,212 CALL TIME (I,JJMS(1,J+1,2),S(1,J+1,2)/100.DOS(IJ+1,2)) GO TO 218 IF (JM-1) 214,214,217 DXK =1.D-5 N =1 ZK(1) ZJ(1) =XK =XJ 00350 CALL TIME (IJJM,0.DO,SPANT(l)) 00360 F(l) 00370 ZK(2) 00380 00390 215 ZJ(2) =TPHI*(1.DO+ZK(2)*CUP)/SUP CALL START (E,VZK(N+1),UPZJ(N+1),HR,1,1,NF) 00400 00410 00420 00430 216 00440 CALL TIME (I,JJMT(N),T(N)/100.DOT(N+1)) F(N+1) =T(N+1)-TAU1 GO TO 221 IF (DABS(F(N+1)).LE.1.D-5) DF(N+1) =(T(N+1)-T(N))/(ZK(N+1)-ZK(N)) DZK =F(N+1)/DF(N+1) 00450 N =N+1 00460 00470 ZK(N+1) =ZK(N)-DZK ZJ(N+1) =TPHI*(1.DO+ZK(N+1)*CUP)/SUP 00480 GO TO 215 00490 217 00500 218 CALL TIME (I,JJMZZSPANS(1,J+l,2)) CALL SWITCH (I,J,JMS(IJ+1,2),X,NX,0) 00510 00520 219 IF (J+1-JM) 211,219,219 RETURN 00530 00540 00550 220 221 =T(1)-TAUl =XK+DXK WRITE (6,91) E,VXKUP,XJHR RETURN END U, 00010 C START 00020 C 00030 00040 00050 SUBROUTINE START (EFZKUAZJHR,I,J,NF) IMPLICIT REAL*8 (A-HO-Z) DIMENSION S(4,7,20) 00060 COMMON/STR/S 00070 COMMON/TIN/ETVTWT,PTSTSITOM 00080 C 00090 00100 00110 00120 00130 00140 00150 00160 00170 S(I,J,12)= DCOS(F) S(I,J,13)= DSIN(F) S(IJ,10)= 1.DO+E*S(I,J,12) PAD= DSQRT(S(I,J,10)) S(I,J,5)= E*S(I,J,13)/RAD S(I,J,9)= RAD*DSIN(UA) RAD*DCOS(UA)-ZK S(I,J,11)= S(I,J,7)= (ZJ*S(I,J,9)+ZK*S(I,J,5))/(1.DO+ZK*(ZK+S(I,J,11))) S(I,J,4)= HR/DSQRT((1.DO+S(I,J,7)**2)*S(I,J,10)) 00180 S(I,J,6)= ZK 00190 00200 00210 S(I,J,8)= ZJ*(ZK+S(I,J,11))+(ZK*S(I,J,7)-S(I,J,5))*S(I,J,9) CALL SWITCH(I,1,JM,O.DO,QX,NF,1) IF (NF) 202,202,201 00220 201 00230 S(I,J,1)= 0.DO S(I,J,2)= 0.DO 00240 S(I,J,3)= 0.DO 00250 00260 00270 00280 00290 00300 S(I,J,14)= S(I,J,15)= S(I,J,16)= S(I,J,17)= S(I,J,18)= S(I,J,19)= 00310 S(I,J,20)= 0.DO 00320 202 RETURN 00330 END 1.DO 0.DO 1.DO 0.DO 0.DO 0.DO 00010 C TIME 00020 C (I,J,JMDT1,SPAN,DTS) (A-HO-Z) 00030 00040 SUBROUTINE TIME IMPLICIT REAL*8 00050 00060 C DIMENSION DT(50) ,Q(51) 00070 00080 00090 601 602 603 FORMAT(8D16.8) FORMAT(413,6D16.8) FORMAT(' FAILED KSP') K= 2 00100 00110 IKS= 70 00120 00130 K75= 0 M4= 20 00140 DTNEG= 5.D-10 IF (DT1) 305,500,506 00150 00160 00170 00180 00190 00200 00210 00220 00230 500 503 IF (NF) 306,306,501 DT(2)= 1.D-3 CALL SWITCH(I,J,JMDT(K) ,Q(K) ,NF,2) IF (NF) 505,505,503 IF (K-7) 504,306,306 504 DT(K+1)= DT(K)*5.DO 501 502 K= K+1 GO TO 502 00240 00250 00260 505 (DT(K)-DT(K-1))/5.D0 506 CALL SWITCH(IJ,JMDT1,Q(1),NF,1) 507 DT(1)= DT1 DT(2)= DT(1)+SPAN CALL SWITCH(I,J,JM,DT(2),Q(2),NF,2) IF (NF) 50,50,305 DT(2)= DT1 IF 00290 00300 00310 00320 00330 00340 SPAN= GO TO 79 00270 00280 DT(1)= 1.D-6 CALL SWITCH(I,J,JMDT(1),Q(l),NF,1) 508 (NF) 508,508,507 00350 00360 00370 00380 00390 50 00400 C NOW 00410 00420 51 00430 00440 00450 53 00460 54 55 00470 00480 00490 56 00500 00510 00520 00530 75 76 00540 00550 00560 78 00570 79 00580 00590 80 00600 100 00610 00620 304 00630 305 00640 00650 00660 00670 306 00680 00690 Q(2)= Q(1) DT(1)= DMAX1(1.D-6,DT1-SPAN) CALL SWITCH(I,J,JMDT(1) ,Q(1) ,NF,2) IF (NF) 305,305,50 KS= K-1 GOOD AT DT(KS),BAD AT DT(KS)+SPAN KSP= KS+IKS DO= Q(K)-Q(K-1) DDT= (DT(K-1)-DT(K))/DQ*Q(K) IF (DDT*(DTNEG-DDT)) 54,53,53 IF (NF) 75,75,100 IF (KSP-K) 304,304,55 DT(K+1)= DT(K)+DDT-DTNEG/2.DO IF ((DT(K+1)-DT(KS)+DTNEG)*(DT(KS)+SPAN-DT(K+1))) K= K+1 CALL SWITCH (I,J,JMDT(K),Q(K),NF,2) DQ= Q(K)-Q(K-1) 51,100,51 IF (DQ) IF (K75-M4) 76,76,305 K75= K75+1 K= KS SPAN= SPAN/5.D0 K= K+1 DT(K)= DT(K-1)+SPAN CALL SWITCH(I,J,JM,DT(K) ,Q(K) ,NF,2) IF (NF) 50,50,78 DTS= DT(K)+DTNEG RETURN WRITE (6,603) WRITE (6,602) I,J,K,K75,DTlSPANDT(1),Q(1),ZKZJ WRITE (6,601) (DT(IK) ,Q(IK) , IK=1,K) KK= 51 Q(KK)= 1.DO RETURN END 75,75,56 00010 C SWITCH 00020 C 00030 SUBROUTINE SWITCH(I,J,JM,WW,QlNFKDU) IMPLICIT REAL*8(A-H,O-Z) 00040 00050 DIMENSION S(4,7,20) 00060 COMMON/KLN/KL 00070 COMMON/STR/S 00080 COMMON/FIN/H2,ECESSN,CV1 00090 C 00100 20 21 IF (KDU-1) 27,20,21 ASK= 1.DO-S(I,J,9)*S(I ,J,9)-S(I ,J,6)*S(I,J,6) TF= 1.DO-2.DO*S(I,J,7)*S(I,J,7) QS= S(I,J,5)+(S(I,J,6)-S(IJ,11))*S(I,J,7) QS8= 8.DO*S(I,J,7)*QS ASKO= ASK*(2.DO*TF-l.DO) THO= -3.DO*(2.DO*TF-l.DO) D22= -(2.DO+TF)*S(I,J,7) D21= -D22*S(I,J,6)-QS*TF XE= TF-S(IJ,6)**2-((S(IJ,8)+S(IJ,5)*S(IJ, 9)) **2+ 1 (S(I,J,5)*S(I,J,6)-S(I,J,7))**2)/S(I,J,10) F= S(I,J,11)+WW*S(I,J,4) DFF =WW*S(I,J,4) F2= F*F P= (F+S(I,J,6))*(F+S(IJ,6))+S(I,J,9)*S(I,J,9 XA= P*(ASK+F2)-4.DO*F2 XB= S(I,J,7)*(2.DO*P*(F+S(I,J,6))-4.DO*F)+(S( I,J,5)+S(I,J,7) ) 00110 00120 00130 00140 00150 00160 00170 00180 00190 00200 00210 00220 00230 00240 00250 00260 *WW 00270 00280 00290 00300 00310 00320 00330 00340 00350 1*S(IJ,4))*(2.DO*F2-P) XC= ASKO+QS8*F+THO*F2+P*XE XD= D21+D22*F X4= XA*XE-XB*XD Z4= XE*XB*XB+XA*(XD*XD-XE*XC) 01= (XC*XC+16.DO*X4)*X4*X4-Z4*(XC*(XC*XC+18.DO*X4)+27.DO*Z4) IF (XE) 25,25,22 22 23 IF(Q1)25,23,23 PP= 2.DO*XC-3.DO*XD*XD/XE a' 00360 00370 00380 24 25 IF (PP) 24,26,26 IF (XC*XC+12.DO*X4-PP*PP) 25,25,26 NF= 0 26 NF= 1 00390 00400 00410 00420 00430 00440 00450 00460 00470 RETURN 27 28 00480 PETURN DER= XD*XD/XE/XE TR= DER/2.DO*(XC/XE-3.D0/4.DO*DER)+2.DO*X4/XE/XE OP= XC/XE-3.D0/2.DO*DER 00= (XD*(DER-XC/XE)+2.DO*XB)/XE IF (00) 29,28,29 S(I,J+1,3)= 1.D20 S(IJ+1,14)= -l.DO 00490 S(IJ+1,15)= 0.DO 00500 00510 00520 00530 GO TO 30 PPR= OP+DSORT(XC*XC+12.DO*X4)/XE YS= -DABS(OQ)/OQ*DSQRT(TR/PPR) ZS =XD/2.DO/XE 29 00540 00550 S3 =YS-ZS S(I,J+1,3)= 1.DO/S3 00560 DEN= 1.DO+P*S(IJ+1,3)*S(I,J+1,3) 00570 00580 00590 S(IJ+1,14)= 2.DO/DEN-1.DO S(IJ+1,15) =2.DO*S(IJ+1,3)*DSQRT(P)/DEN CALL DTDU(I,J,JMWWJ+1,1) 00600 00610 30 RETURN END a' 4S 00010 C DTDU 00020 C SUBROUTINE DTDU(I,LJMDTM,ISW) 00030 IMPLICIT REAL*8 (A-HO-Z) 00040 00050 DIMENSION S(4,7,20) 00060 COMMON/STR/S 00070 C 00080 DF= S(I,L,4)*DT VL= 1.DO+DF*(S(I,L,6)+S(I ,L,1l))/S(IL,10) 00090 00100 00110 00120 00130 00140 00150 00160 00170 00180 00190 00200 00210 00220 00230 00240 00250 00260 00270 00280 00290 00300 00310 00320 00330 00340 VH= DF*S(I,L,9)/S(IL,10) RAD2= VL*VL+VH*VH S(I ,M,1)= S(I ,L,1)+DT S(I,M,2)= DT S(I,M,5)= S(I,L,5)+DF*S(I ,L,7) S(I,M,8)= S(I,L,8)*VL-VH* ((S(I ,L,10)-1.DO) *S(I ,L,7)- I S(I,L,11)*S(I,L,5)) S(IM,10)= S(I,L,10)*RAD2 S(IM,11)= S(IL,11)+DF RAD= DSQRT(RAD2) CX= VL/RAD SX= VH/RAD S(IM,16)= CX*S(I ,L,16)-SX*S(I,L,17) S(IM,17)= CX*S(IL,17)+SX*S(IL,16) S(I,M,18)= S(IL,18) TX= (CX+S(I,M,16))/(1.DO+S(I,L,16)) S(I ,M,19)= TX*S(I,L,19)+SX*S(IL,12) 11 S(I,M,20)= TX*S(I,L,20)+SX*S(IL,13) IF (ISW) 12,12,11 CDU= S(IL+1,14) SDU= S(I,L+1,15) SOR= SDU/DSQPT(S(I,M,10)) SPT= (1.DO-CDU)/S(I,M,10) TP= CDU+SRT-SOR*S(I,M,5) TTO= CDU*S(I,L,7)+SOP*S(IM,11) U, 00350 00360 00370 00380 00390 00400 00410 00420 00430 00440 00450 00460 00470 00480 00490 00500 00510 00520 00530 00540 00550 00560 CSOC= DSQRT(1.DO+S(I,L,7)*S(I,L,7)-TTO*TTO) FPR= CSOC/DSQRT(TP) S(I,M,4)= FPR*S(I,L,4) S(I,M,5)= (CDU*S(I,M,5)+(S(I,M,10)-1.DO)*SOR)*FPR/CSOC S(I,M,6)= (S(I ,L,6)-SOR*S(IL,7)-SRT*S(I ,M,11))/FPR/TP S(I,M,7)= TTO/CSOC 12 S(I,M,9)= (CDU*S(I,L,9)+SOR*S(I ,M,8))/FPR/TP S(I,M,8)= (CDU*S(I,M,8)-SOR*S(I,L,9)*S(I,M,10))/CSOC S(IM,11)= (CDU*S(I ,M,11)-SOR*S(I,L,7)*S(I,M,10))/FPP S(IM,10)= TP*S(I,M,10) Tl= S(I,M,17) S(I,M,17)= CDU*S (IM,17) -SDU*S (I ,M,18) S(I,M,18)= CDU*S (I ,M,18) +SDU*T1 SSI= S(I,M,19)*S (I,M, 19) +S(I,M,20)*S(I,M,20) 13 S(I,M,12)= 14 S(I,M,12)= S(I,M,13)= RETURN END IF 15 (SSI) 13,13,1 4 CDU*S (I ,L,12) -SDU*S(I ,L,13) S(I,M,13)= CDU*S (I ,L,13) +SDU*S (IrL,12) GO TO 15 (S(I, M,17)*S(IM,19)-S(IM,18)*S(IM,20))/SSI (S(I, M,18) *S(IM,19)+S (IM,17) *S(IM,20) )/SSI a' 00010 C MATX 00020 C 00030 00040 00050 SUBROUTINE MATX(ECESSNDY) IMPLICIT REAL*8(A-H,O-Z) DIMENSION S(4,7,20),DY(3),X(4,4) 00060 00070 COMMON/STR/S COMMON/FCOMP/X 00080 00090 00100 All =X(2,2)-X(2,1) A12 =X(2,3)-X(2,1) A13 =X(2,4)-X(2,1) 00110 00120 A21 =X(3,2)-X(3,1) A22 =X(3,3)-X(3,1) 00130 A23 =X(3,4)-X(3,1) 00140 00150 A31 =X(4,2)-X(4,1) A32 =X(4,3)-X(4,1) 00160 A33 =X(4,4)-X(4,1) 00170 00180 00190 00200 00210 00220 00230 00240 00250 00260 00270 00280 00290 00300 00310 21 00320 00330 B11 =A22*A33-A23*A32 B12 =A21*A33-A23*A31 B13 =A21*A32-A22*A31 B21 =A12*A33-A13*A32 B22 =All*A33-A13*A31 B23 =A11*A32-Al2*A31 B31 =A12*A23-A13*A22 B32 =A1*A23-A13*A21 B33 =All*A22-Al2*A21 Yl =EC-X(2,1) Y2 =ES-X(3,1) Y3 =SN-X(4,1) S15 =S(1,2,15) IF (S15) 21,22,21 DA =All*Bl1-A21*B21+A31*B31 DY(1) =(Yl*Bll-Y2*B21+Y3*B31)/DA DY(2) =(-Y*B2+Y2*B22-Y3*B23)/DA C' 00340 00350 00360 22 00370 00380 00390 00400 23 00410 DY(3) =(Yl*Bl3-Y2*B23+Y3*B33)/DA GO TO 23 DA =B23 DY(1) =(Yl*A32-Y3*A12)/DA DY(2) =(-Yl*A31+Y3*All)/DA DY(3) =0.DO RETURN END 00010 C OUTPUT 00020 C 00030 00040 00050 00060 00070 00080 00090 00100 00110 61 00120 62 OMEGA 00130 SUBROUTINE OUTPUT(I,J,KKDT,Z,IOUT,OX) IMPLICIT REAL*8(A-H,O-Z) DIMENSION S(4,7,20) ,X(4,7) COMMON/KLN/JM COMMON/STR/S COMMON/BIN/BPOBPTElV1,PIDTFTAUl COMMON/FCOMP/X COMMON/TIN/ETVT,WTPTSTSITOM FORMAT(4(D9.3,1X),D8.3,IX,D1O.4,1X,D9.3,1X,D9.3,1XD8.3) FORMAT(/' TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS 1 =',D14.5//) 00140 00150 OME =1.DO ONE =1.DO 00160 P12 =.5DO*PI 00170 00180 IF(OM.LE.PI2.AND.OM.GE.-PI2) IF (JM.EQ.3) ONE =1.DO 00190 00200 11 IF(KK)11,11,12 P =Z/S(I,J,4)**2/(1.DO+S(I,J,7)**2) 00210 PP =P-S(I,J,9)**2 00220 00230 00240 00250 00260 12 DF DF DT GO DF 00270 PP =(S(I,J,6)+S(I,J,11)+DF)**2 00280 P =PP+S(I,J,9)**2 00290 13 00300 00310 RAD =DSQRT(P) ESF =RAD*(S(I,J,5)+DF*S(I ,J,7)) VL =1.DO+DF*(S(IJ,6)+S(I,J,11))/S(I,J,10) 00320 VH =DF*S(I,J,9)/S(I,J,10) 00330 00340 RAD2 =VL*VL+VH*VH CX =VL/DSQRT(RAD2) ONE =--1.DO =DSQRT(PP)-S (IJ,6)-S (IJ,11) =DF*S (I,J,6)/DABS (S(I ,J,6)) =DF/S(I,J,4) TO 13 =DT*S(I,J,4) 00350 SX=VH/DSQRT(RAD2) 00360 00370 00380 00390 00400 51 00410 S16 =CX*S(I,J,16)-SX*S(IJ,17) S17 =CX*S(I,J,17)+SX*S(I,J,16) S18 =S(I,J,18) IF (S18) 52,51,52 UX =0.DO GO TO 53 00420 00430 00440 00450 UX =DATAN2(S18,S17) TX =(CX+S16)/(1.DO+S(I,J,16)) S19 =TX*S(I,J,19)+SX*S(I,J,12) S20 =TX*S(I,J,20)+SX*S(I,J,13) 00460 00470 00480 00490 00500 00510 00520 00530 00540 52 53 41 42 43 44 45 IF (S20) 44,41,44 IF (S19) 42,43,43 OMS =PI GO TO 45 OMS =0.DO GO TO 45 OMS =DATAN2(S20,S19) SSI =S19*S19+S20*S20 OX =180.DO/PI*OMS 00550 SI 00560 00570 14 IF(SI) 14,14,15 CU =S(I,J,12) =DSQRT(SSI) 00580 SU =S(I,J,13) 00590 00600 00610 15 00620 00630 16 00640 00650 OMS =0.DO GO TO 16 CU =DCOS(UX) SU =DSIN(UX) H2 =P*S(I,J,4)**2*(1.DO+S(I,J,7)**2) X(1,I) =H2 X(2,I) =(P-l.D0)*CU+ESF*SU 00660 X(3,I) 00670 X(4,I) =SI =(P-1.DO)*SU-ESF*CU 00680 TAU =S(I,J,1)+DT 00690 00700 00710 00720 00730 00740 00750 00760 00770 00780 00790 00800 00810 00820 1,2) 00830 00840 00850 00860 00870 00880 00890 00900 00910 DTF =DT X21 =X(2,I) X31 =X(3,I) IF(IOUT-1) 23,21,22 21 PHI =DATAN2(S(I,J,7),1.DO)*180.DO/PI SUP =S(I,J,9)/DSQRT(S(I,J,10)) UP =ONE*(DARSIN(SUP)*180.DO/PI) X(2,I) =ONE*X(2,I) X(3,I) =ONE*X(3,I) 31 VT =0.DO IF (ESF) 32,31,32 GO TO 33 32 33 22 23 VT =DATAN2(ESF,P-1.DO)*180.D0/PI WRITE(6,61) X(1,I),X(2,I),X(3,I),X(4,I),TAU,VTPHI,UPS(IJ+ RETURN X(2,I) =ONE*X(2,I) X(3,I) =ONE*X(3,I) WRITE(6,61) X(1,I),X(2,I),X(3,I),X(4,I),TAU IF (OM.LE.0.DO) OME =-1.DO OMEGA =-90.DO*OME*(ONE-1.DO)-OMS*180.D0/PI WRITE (6,62) OMEGA RETURN END -_ 72 Appendix Examples of exact transfers C 7 the state vector total normalized = TE= AV ( h , ecos w , esin w , sin i ) An exact numerical solution may be described by: TAU location of the impulse defined by the true anomaly = v PHI the firing angles UP AC= magnitude of each impulse DTAU location of the first impulse w.r.t. the reference frame The following 10 examples from a circular orbit TE = = A OMEGA illustrate 2 and 3 impulse transfers ( 1, 0, 0, 0 ). The final states 7 are chosen with no restriction. Case 1 - 4 The transfers. describe final the elliptic symmetrical properties of orbits have large 2 impulse inclinations and eccentricities. Case 5 6 - are 3 impulse transfers. The final elliptic transfers. The final elliptic orbits have small inclinations. Case 7 8 - are 2 impulse orbits have small eccentricities. Case 9 - 10 are elliptic Hohmann transfers. CASE 1 INPUT H2 ECOSW 0.42000D+01 FIRST ORDER ESTIMATE OF I JM CLIM 0.42262D+00 SIN 2 .15000D+00 ESINW 0.34641D+00 TAU 0.20000D+00 =.4708666384D+00 OUTPUT H**2 ECOSW ESINW SIN I TAU V PHI UP DTAU 0.100D+01 0.0 0.0 0.0 .0 0.0 -.513D+00 0.880D+01 .321D+00 0.174D+01 0.738D+00 0.379D-02 0.373D-01 .321D+00 0.1708D+03 0.399D+01 -. 544D+02 .153D+00 0.420D+01 0.346D+00 0.200D+00 0.423D+00 .475D+00 TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS CASE : OMEGA 0.81642D+01 2 INPUT H2 0.42000D+01 FIRST ORDER ECOSW -0.34641D+00 ESTIMATE OF TAU ES INW SIN I 0.20000D+00 CLIM .15000D+00 JM 2 0.42262D+00 =.4708666384D+00 OUTPUT H**2 ECOSW ESINW SIN I TAU PHI V UP DTAU 0.513D+00 -. 880D+01 .321D+00 0.0 .0 0.100D+01 0.0 0.0 0.0 0.174D+01 -. 738D+00 0.379D-02 0.373D-01 .321D+00 -. 1708D+03 -. 399D+01 0.544D+02 .153D+00 0.420D+01 -. 346D+00 0.200D+00 0.423D+00 .475D+00 TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS : OMEGA = 0.17184D+03 CASE 3 INPUT H2 -0.34641D+00 FIRST ORDER ESTIMATE OF SIN I FSINW ECOSW 0.42000D+01 TAU -0.20000D+00 0.42262D+00 JM CLIM 2 .15000D+00 =.4708666384D+00 OUTPUT H**2 ECOSW SIN I ESINW TAU V PHI DTAU UP 0.0 -. 513D+00 -. 880D+01 .0 0.0 0.0 0.100D+01 0.0 0.174D+01 -. 738D+00 -. 379D-02 0.373D-01 .321D+00 0.1708D+03 0.399D+01 0.544D+02 0.420D+01 -. 346D+00 -.200D+00 0.423D+00 .475D+00 TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS CASE : OMEGA = .321D+00 .153D+00 -0.17134D+03 4 INPUT F2 0.42000D+01 ECOSW FIRST ORDER ESTIMATE OF TAU CLIM SIN I ESINW 0.34641D+00 -0.20000D+00 2 0.42262D+00 .15000D+00 =.4708666384D+00 OUTPUT H**2 ECOSW ESINW SIN I TAU PHI V UP DTAU 0.513D+00 0.880D+01 .321D+00 0.0 .0 0.0 0.0 0.100D+01 0.0 0.174D+01 0.738D+00 -. 379D-02 0.373D-01 .321D+00 -. 1708D+03 -. 399D+01 -. 544D+02 .153D+00 0.420D+01 0.346D+00 -.200D+00 0.423D+00 .475D+00 TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS : OMEGA = -0.81642D+01 CASE 5 INPUT H2 0.39600D+01 ECOSW FIRST ORDER ESTIMATE OF SIN I ES INW 0.87155D-02 TAU 0.99619D-01 0.87156D-01 JM CLIM 3 .50000D-01 =.4519980884D+00 OUTPUT H**2 0.100D+01 0.159D+01 0.259D+01 0. 396D+01 ECOSW SIN I ESINW 0.0 -. 586D+00 -. 245D+00 0.872D-02 0.0 0.723D-03 0.325D+00 0.996D-01 0.0 0.299D-02 0.846D-01 0.872D-01 V TAU TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS CASE PHI UP DTAU .0 0.0 0.127D+00 -. 831D+00 .259D+00 .259D+00 -. 1558D+03 -.703D+01 0.209D+02 .109D+00 .369D+00 0.1730D+03 -. 432D+01 -. 256D+02 .953D-01 .464D+00 : OMEGA = 0.93696D+02 6 INPUT H2 0.39600D+01 ECOSW 0.0 FIRST ORDER ESTIMATE OF ESINW 0.10000D+00 TAU SIN I 0.82808D-01 JM CLIM 3 .5000OD-01 =.4520750429D+00 OUTPUT H**2 0.100D+01 0.159D+01 0. 254D+01 0. 396D+01 ECOSW -.139D-16 -. 589D+00 -. 266D+00 0.228D-07 ESINW 0.0 0.637D-03 0.318D+00 0.100D+00 SIN I 0.0 0.261D-02 0.805D-01 0.828D-01 TAU PHI V UP DTAU 0.111D+00 -. 724D+00 .261D+00 .0 0.0 .261D+00 -.1561D+03 -. 697D+01 0.207D+02 .103D+00 .364D+00 0.1722D+03 -. 390D+01 -. 248D+02 .990D-01 .463D+00 TO MATCH ACTUAL FINAL OPBITFIPST IMPULSE SHIFTS : OMEGA = 0 .90170D+02 CASE 7 INPUT H2 ECOSW 0.49875D+01 ES INW 0.45853D-01 FIRST ORDER ESTIMATE OF TAU SIN I 0.19937D-01 CLIM .10000D+00 JM 2 0.64279D+00 =.5807860994D+00 OUTPUT H**2 ECOSW ESINW SIN I TAU V PHI UP DTAU 0.100D+01 0.0 0.0 0.0 .0 0.0 -.675D-01 0.135D+02 .302D+00 0.168D+01 0.679D+00 0.461D-03 0.543D-01 .302D+00 0.1792D+03 0.358D+00 -. 699D+02 .273D+00 0.499D+01 0.459D-01 0.199D-01 0.643D+00 .575D+00 TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS CASE : OMEGA = 0.70359D+00 8 INPUT H2 0.79928D+01 ECOSW 0.15000D-01 FIRST ORDER ESTIMATE OF TAU SIN I ESINW 0.25981D-01 JM 2 0.81915D+00 CLIM .10000D+00 =.6187259196D+00 OUTPUT H**2 ECOSW ESINW SIN I TAU V PHI UP DTAU 0.0 -. 467D-01 0.933D+01 .338D+00 0.0 .0 0.100D+01 0.0 0.0 0.178D+01 0.781D+00 0.368D-03 0.410D-01 .338D+00 0.1793D+03 0.352D+00 -. 804D+02 .281D+00 0.799D+01 0.150D-01 0.260D-01 0.819D+00 .619D+00 TO MATCH ACTUAL FINAL ORBITFIPST IMPULSE SHIFTS : OMEGA = 0.68374D+00 0 CASE 9 INPUT H2 0.19200D+01 ECOSW 0.20000D+00 FIRST ORDER ESTIMATE OF TAU ES INW SIN I 0.0 0.17365D+00 JM CLIM 2 .20000D+00 =.3124921504D+00 OUTPUT H**2 ECOSW ESINW SIN I TAU V PHI UP DTAU 0.100D+01 0.0 0.0 0.0 .0 0.0 -. 642D-04 0.192D+02 .197D+00 0.141D+01 0.412D+00 0.263D-06 0.545D-01 .197D+00 0.1800D+03 0.845D-03 -.415D+02 .104D+00 0.192D+01 0.200D+00 0.273D-04 0.174D+00 .302D+00 TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS : OMEGA = -0.36000D+03 CASE 10 INPUT H2 0.19200D+01 ECOSW -0.20000D+00 FIRST ORDER ESTIMATE OF TAU SIN I ESINW 0.0 0.17365D+00 JM CLIM 2 .20000D+00 =.3124921504D+00 OUTPUT H**2 ECOSW ESINW SIN I TAU V PHI UP DTAU 0.100D+01 0.0 0.0 0.0 .0 0.0 -.642D-04 -.192D+02 .197D+00 0.141D+01 -.412D+00 -.263D-06 0.545D-01 .197D+00 0.1800D+03 0.845D-03 0.415D+02 .104D+00 0.192D+01 -.200D+00 -. 273D-04 0.174D+00 .302D+00 TO MATCH ACTUAL FINAL ORBITFIRST IMPULSE SHIFTS : OMEGA = -0.18000D+03 --j Figures 78 A SO 2 T AA Circular orbit : Final orbit : Transfer orbit : Initial orbit plane : Final orbit plane Fig.1 The coordinate system of the non-coplanar transfer from a circular orbit (Fixed reference axes 2 3 rx 1, A2,3) 79 0 Fig. 2 An elliptic Hohmann transfer, is = 0. 1 - s s S N Fig. 3 . 80 axis inclination relative to tli,. initialorbit-plane 1 First order influence of inclination (with impossible elliptic Hohmann transfer orbit 0 ). 81 09: First transfer orbit : Second transfer orbi 7 0 a 400 A I AV I Fig.4 Illustration of a three impulse transfer. 3 iMgulse transfer family hi; 000 0 4-J C-) 1. *Tr Limiting solution Zt 0 cc VS Fig.5 vs \V of the first (transfer angle) transfer orbit 0 , (Cos 1i 2 ) True anomoly ( e ) Single impulse 01.0 4J toP LH 0I 121 I Ratio of the semi-major ases rig. Fi rsL order so-LUL1 6 UL and 3I1, nl i .3IaLLUL, (i /K00 ). mo (d5 4e41 I Va - 0 0,s 08 O Fig. 7a Single impulse approximation Fig - 7b Three impulse approximation Fig- 7c Actual three impulse approximation 84 - Fig.8 0 Generalised Hohmann transfer between circles, (k= 0 ). 85 o, 060 Fig. 9 A two impulse transfer with plane change at further impulse. I* 86 -Pd -I-) 0 U U) -P0 fd Cd In>nto ofj th fialori U 04' Inclination of the final orbit Determination of the limit of the first order solution of small inclination. ( 4V/g= first order, V = second order ) Fig. 10 ( ) --. 0 18Q)0 1 . ')70 0 87 05 0 / , A (b) 0 -c::: 90 20 Fig.1l a,b 2 impulse transfers visualization with an elliptic final orbit. 88 and 6 in 4 e the plane of ) Fixed orthogonal set ( e 4 4- Fig. 12a. A supporting impulse -qo -< 1 4 o0' e .2 Fig . 12 b. An opposing impulse 4, 2702 v and 89 A S3 (a~) In the plane of e and W.*\ e,4 .A I Omo (b) Iv 4" 1 K ei A .2 Fig. 13 (a, b) Definitions of the transfer parameters e 90 3 I. 1-- AA Fig. 14 A single impulse for two intersecting orbits. 91 0J =constant plane > 0 % constant , I . x 4 0 /5a A supporting impulse from a circular orbit with I = constant and i 0 1 I - constant , Fig- 'O. ~' constant A -~.-- 4I Fig./ /5b Nonuniqueness of at ~J = constant. I 92 References 1. T.N. How Many Impulses? Edelbaum Astronautics and Aeronautics, Vol. 5, no.11, Nov.1967, pp.64-69. 2. F.W. Gobetz J.R. and Doll Survey Of Impulsive Trajectories. Of Trajectories. AIAA J. Vol. 7, no.5, May 1969, pp. 801-834. 3. C. Marchal Survey - Paper Optimization I.A.F. - 27th Congress - Anaheim 4. H.W. Minimum Small Ph.D. Orbits. nautics 5. T.N. Thesis, Marec Congress, 7. J.V. and a Madrid, Nearby Methods for Editor. 8. C. Non-coplanar Elliptic Flight Thesis, in The Near of Aero- Vicinity of Impulsionnels Orbit. Economiques Proc. M. I. T. Advanced Optimization. 17th IAF Problems and B. Fraeijs de de Hohmann as Cas des Veubeke 1969. des Transfers Faiblement Optimum Single M.S. Department Minimum Impulse Transfer Between a Circular Orbit Excentriques Publication ONERA No. 124, Der Elliptic 1966. Planetaires G.J. between non-coplanaires. Generalisation Inclines. University, Infinitesimaux Pergamon Press, Orbites 9. oct. Space Marchal Transfer s.8. J. Astronaut. Sci. 14, 2, 1967. Transferts Breakwell no.76010 1972. Quasi-Circulaires orbites Time-free Minimum Impulse Transfers Edelbaum J.P. 10-16 Oct. 1976 Stanford and Astronautics, a Circular Orbit. 6. Fuel Space Faiblement 1968. Impulse Transfers Department et from Circular of Aeronautics Orbits. and Astronautics, 1975. 10. D.F. Lawden Butterworth, 11. R.S. Long Optimal Eckel Orbits. 13. K.G. Transfer Optimum between N-Impulses. Space Non-coplanar Elliptical Navigation. Optimum Orbits. 6, 1960. Transfers Astronautica Acta, Eckel for London, 1963. Astronautica Acta, Vol. 12. K.G. Trajectories Vol. between 8, Transfer Astronautica Acta, Vol. Fasc. in Non-coplanar 4, 1962. Central 9, Fasc. Elliptical Force 5-6, 1963. Field with 14. J.M. Optimum Horner coplanar terminals. 15. C. Marchal ltOptimalite 1968. Impulse des Transfer between 93 Arbitrary ARS Journal, Jan. 1962. Tridimensionel Generalisation Champ Newtonien), pp.3-13, Two Arcs a Pousee Intermediaire La Recherche Aerospatiale, et de Lawden No. Etude (Dans de un 123, Mars-Avril, 94 Biography Raised 1949. London in University, a with graduated September from B.Sc. degree 1974, he studied He Hong Kong. England, was born in Canton, Der Jew Gim' at 1971 Aeronautical in China, Queen Mary College, to June and 1974, First Engineering, Class Honours. In September, Technology as a graduate He Astronautics. Astronautics Laboratory as in a June, student in obtained a 1975. research to came Massachusetts the Department S.M. He has assistant of orbit transfers from 1975 to 1977. the degree of in N. Aeronautics and Aeronautics and in S. Draper Edelbaum on high thrust been working T. of Institute C.