Assignment 05 — Transfers in Real Missions

This assignment takes the transfers of the course into six situations that real missions meet. A space station has to be kept up against drag. A dead satellite has to be brought down safely. A working satellite has to dodge a piece of debris. A probe has to reach the Sun, and a capsule has to come home from the Moon. The last problem asks whether a few point masses can stand in for the Earth’s flattening. Each problem should take about half an hour, so the whole assignment takes about three hours.

The notes you need are The Impulsive Model, The Hohmann Transfer, The Bi-elliptic Transfer, Phasing, Escape, Patched Conics and The N-body Problem. Everything that the last problem needs about the shape of the Earth is given in its statement.

Every problem has three kinds of part. An analytical part asks you to derive a formula or prove a property of it, with symbols only. A numerical part asks you to evaluate the result once for a concrete case. A discussion part asks what the result means for the mission. Show every step of a derivation, and say which result of the notes you use at each step. Keep \(\mu\), \(r\) and the speeds as symbols in the algebra, and do not set any of them equal to \(1\). Give numerical answers to three or four significant figures, with units.

Some data are marked assumed. They are reasonable values chosen for the problem, not flown values. All burns are impulsive unless a part says otherwise. Use the following constants.

Quantity Value
\(\mu_\oplus\); \(R_\oplus\) \(398\,600.4418\) km\(^3\)/s\(^2\); \(6378.137\) km
\(\mu_\odot\); \(R_\odot\); astronomical unit \(1.327\,124\,400\,18\times10^{11}\) km\(^3\)/s\(^2\); \(6.957\times10^{5}\) km; \(1.495\,978\,707\times10^{8}\) km
\(\mu_{\rm Moon}\); \(R_{\rm Moon}\); Earth–Moon distance \(D\) \(4902.8\) km\(^3\)/s\(^2\); \(1737.4\) km; \(384\,400\) km
\(g_0\) \(9.80665\) m/s\(^2\)
one year \(365.25\) days

Problem 1 — Keeping a space station up

The Bharatiya Antariksh Station is reported to be planned for a low circular orbit, at roughly \(400\) to \(450\) km. Take a station of mass \(50\) t (assumed) on a circular orbit at \(400\) km. The thin air at this height slowly lowers the orbit. Take the rate of decay as \(2.0\) km in \(30\) days (assumed; it depends strongly on the activity of the Sun). From time to time the station is raised back by its own thrusters, with \(I_{sp}=300\) s (assumed). This is called a reboost.

Let \(a_D\) be the deceleration that drag gives the station. It acts against the velocity and is so small that the orbit stays nearly circular while it shrinks.

  1. Analytical. Use the energy per unit mass, \(\varepsilon=-\mu/(2r)\), and the rate at which drag removes it, \(d\varepsilon/dt=-v\,a_D\). Show that the radius falls at the rate \(dr/dt=-(2r/v)\,a_D\). Then show that the speed rises at the rate \(dv/dt=+a_D\). Drag slows the station down, and yet the station speeds up. Explain where the extra speed comes from.
  2. Analytical. The reboost is a Hohmann transfer from the circle of radius \(r-\Delta r\) back to the circle of radius \(r\), with \(\Delta r\ll r\). Show that, to first order in \(\Delta r/r\), each of the two burns is \(v\,\Delta r/(4r)\) and the total is \(v\,\Delta r/(2r)\). Show that this total equals the velocity that drag removed, \(\int a_D\,dt\), while the orbit decayed by \(\Delta r\).
  3. Numerical. Find the drag deceleration \(a_D\) at \(400\) km, the reboost \(\Delta v\) needed in one year, and the propellant it uses.
  4. Analytical. The density of the air, and with it the drag, falls with height as \(e^{-(h-h_0)/H}\), with the scale height \(H=60\) km (assumed). The station must never go below \(h_0=400\) km. It is raised to \(h_0+A\), allowed to decay to \(h_0\), raised again, and so on. Show that the reboost \(\Delta v\) per year is the value of part (c) multiplied by \[ f(A)=\frac{A/H}{e^{A/H}-1}, \] and that \(f\approx1-A/(2H)\) when \(A\ll H\). (Hint: find the time the station takes to decay from \(h_0+A\) to \(h_0\).) Evaluate \(f\) for \(A=2\) km and \(A=20\) km.
  5. Discussion. Part (d) says that large, rare reboosts are cheaper. What does a station give up by flying a large \(A\)? Think of the vehicles that visit it, of what happens if the Sun becomes more active, and of the experiments on board.

Problem 2 — Bringing a dead satellite down safely

Megha-Tropiques, an Indo-French satellite that studied the tropical atmosphere, was launched in 2011 into a circular orbit at \(867\) km altitude. Its mission ended in 2022 with about \(125\) kg of propellant left. ISRO used it to bring the satellite down in a controlled way. From August 2022 the perigee was lowered in \(20\) manoeuvres. On 7 March 2023 two final de-boost burns, each made by four \(11\) N thrusters firing for about \(20\) minutes, lowered the perigee below about \(80\) km. The satellite re-entered over the Pacific Ocean.

Take the mass at the start of the de-orbit as \(1000\) kg and \(I_{sp}=220\) s (both assumed). The atmosphere is taken to begin at the entry interface, an altitude of \(120\) km, with radius \(r_e\).

  1. Analytical. One tangential retrograde burn on the circle of radius \(r_1\) puts the satellite on an ellipse with perigee radius \(r_p<r_1\). Show that the burn costs \(\Delta v=v_{c1}\bigl(1-\sqrt{2r_p/(r_1+r_p)}\bigr)\). The satellite crosses the entry interface \(r_e\) (with \(r_p<r_e<r_1\)) at the flight-path angle \(\gamma_e\) below the horizontal. Show that \[ \cos^2\gamma_e=\frac{r_1\,r_p}{r_e\,(r_1+r_p-r_e)} . \] Check that \(\gamma_e=0\) when \(r_p=r_e\), and show that \(\gamma_e\) grows as \(r_p\) is lowered.
  2. Numerical. Find \(\Delta v\) and \(\gamma_e\) for a perigee altitude of \(80\) km and of \(0\) km. For the \(80\) km perigee, find the propellant used and how long the burn would last with the four \(11\) N thrusters. Compare the burn time with the period of the orbit. Is the impulsive model good here?
  3. Analytical. Instead of one burn, the satellite makes several smaller retrograde burns, each at the apogee of its current orbit, until the perigee reaches \(r_p\). Show that the total cost is the same as that of the single burn. Assignment 04 found that rotating the apse line in several steps costs more than in one step. Why is there no such penalty here?
  4. Discussion. Why does a controlled re-entry need a low perigee, and not just one inside the atmosphere? Why did ISRO bring Megha-Tropiques down while geostationary satellites such as INSAT-4B are instead raised a few hundred kilometres above the geostationary orbit at the end of their lives? Use Exercise 1 of the Hohmann note.

Problem 3 — Dodging a piece of debris

ISRO’s satellites make collision avoidance manoeuvres every year: \(23\) in 2023 and \(10\) in 2024, according to ISRO’s space situational assessment reports. Consider an Earth-observation satellite on a circular orbit at \(500\) km altitude. A close approach with a piece of debris is predicted. The satellite must be moved by \(s_0=1\) km from where it would otherwise be at the time of closest approach.

The phasing note gives two first-order facts for a small tangential burn \(\delta v\) on a circle of radius \(a\) and speed \(v\). The semi-major axis changes by \(\delta a/a=2\,\delta v/v\), and the period by \(\delta T/T=\tfrac32\,\delta a/a\).

  1. Analytical. The burn is made a whole number \(k\) of revolutions before the closest approach, at the time \(t=kT\) before it. Show that at the closest approach the satellite is displaced along its orbit by \(3\,\delta v\,t\). Which way, for a prograde burn? If the burn is made instead half a revolution before the closest approach, show that the satellite is then displaced radially by \(4a\,\delta v/v\) and along the orbit by \(\tfrac32\,T\,\delta v\).
  2. Numerical. Find \(\delta v\) to move the satellite by a total distance \(s_0\) with a burn half a revolution before, and with a burn one day before. How many times cheaper is the early burn?
  3. Analytical. The early burn has a weakness. The drag on the satellite is not known exactly, so its predicted position drifts away from the truth. Let \(\epsilon\) be the uncertainty of the drag deceleration, a small tangential acceleration that we do not know. Use part (a) to show that after a time \(t\) it moves the satellite along its orbit by \(\tfrac32\,\epsilon\,t^2\). To be safe, the planned displacement must cover the miss distance and this error, \[ 3\,\delta v\,t\ \ge\ s_0+\tfrac32\,\epsilon\,t^2 . \] Find the lead time \(t^*\) that needs the smallest \(\delta v\), and that \(\delta v^*\). Show that at \(t^*\) the two terms on the right are equal.
  4. Numerical. Take \(\epsilon=5\times10^{-8}\) m/s\(^2\) (assumed). Find \(t^*\) in hours, \(\delta v^*\) in mm/s, and \(\delta v\) for a burn five days before the closest approach.
  5. Discussion. ISRO’s report for 2024 says that fewer separate avoidance burns were needed partly because avoidance was combined with the routine burns that keep the orbit up. Why does a satellite in low orbit need such routine burns, and which direction of avoidance burn fits with them? What else must be checked before the burn is made?

Problem 4 — Reaching the Sun the long way round

Problem 7 of Tutorial 1 showed that it is much harder to fall into the Sun than to leave the solar system. Here we look for a cheaper way in. The target is a perihelion of \(r_p=10\,R_\odot\), about as deep as the Parker Solar Probe goes. Treat the Earth’s orbit as a circle of radius \(1\) AU with speed \(v_E\), and ignore the other planets except where a part says so. Write \(x=r_b/(1\ \text{AU})\) and \(\rho=r_p/(1\ \text{AU})\).

  1. Analytical. The direct route is one burn at \(1\) AU that lowers the perihelion to \(r_p\). Write its cost in terms of \(v_E\) and \(\rho\). The bi-elliptic route first raises the aphelion to \(r_b\), and then, at that aphelion, lowers the perihelion to \(r_p\). Show that its two burns are \[ \Delta v_1=v_E\Bigl(\sqrt{\tfrac{2x}{1+x}}-1\Bigr),\qquad \Delta v_2=v_E\Bigl(\sqrt{\tfrac{2}{x(1+x)}}-\sqrt{\tfrac{2\rho}{x(x+\rho)}}\Bigr). \]
  2. Analytical. Let \(\rho\to0\), a fall into the centre of the Sun. Show that the bi-elliptic route is then cheaper than the direct route for every \(r_b>1\) AU. Find the limit of its cost as \(x\to\infty\), and compare it with the cost of leaving the solar system.
  3. Numerical. For \(r_p=10\,R_\odot\), find the cost of the direct route. Take \(r_b=5.2026\) AU, the distance of Jupiter. Find \(\Delta v_1\), \(\Delta v_2\), their total, and the time from the Earth to the perihelion, in years.
  4. Numerical. The burn at \(1\) AU is made by the launcher, which leaves a parking orbit at \(300\) km altitude with the hyperbolic excess speed \(v_\infty=\Delta v_1\) (or the direct cost). Use the escape note to find the burn from the parking orbit for both routes. The burn \(\Delta v_2\) must be made by the spacecraft itself, far from the Earth. With \(I_{sp}=320\) s, what fraction of its mass is left after that burn?
  5. Discussion. Real missions do not make \(\Delta v_2\) with an engine. Ulysses used a flyby of Jupiter to leave the plane of the planets, and the Parker Solar Probe uses repeated flybys of Venus. Using the patched-conics note, explain what a flyby at aphelion can do in place of \(\Delta v_2\). Why did India’s Aditya-L1 not go near the Sun at all?

Problem 5 — Bringing a sample home from the Moon

ISRO’s Chandrayaan-4, approved in 2024, is to bring a sample of the Moon’s surface back to the Earth. Consider the last part of such a mission. A capsule leaves a circular lunar orbit at \(100\) km altitude with one burn, the trans-Earth injection, and enters the Earth’s atmosphere directly. The Moon moves on a circle of radius \(D\), and everything happens in its plane.

Use the patched-conic method in its simplest form. The capsule’s velocity relative to the Moon on leaving the Moon’s sphere of influence is the difference between its geocentric velocity at the distance \(D\) and the Moon’s velocity, and the size of the sphere is ignored. The entry interface is at \(120\) km altitude, with radius \(r_e\). The return orbit is an ellipse with its apogee at \(D\) and its perigee at \(r_p<r_e\). Its perigee lies inside the atmosphere, and is called the virtual perigee.

  1. Analytical. Show that the capsule must leave the Moon with \[ v_\infty=\sqrt{\frac{\mu_\oplus}{D}}-\sqrt{\frac{2\mu_\oplus r_p}{D(D+r_p)}}, \] and say in which direction, relative to the Moon’s motion. Write the cost of the trans-Earth injection from a circular lunar orbit of radius \(r_L\).
  2. Analytical. Show that the capsule reaches the entry interface with a speed \(v_e\) such that \[ \frac{v_e}{v_{\rm esc}(r_e)}=\sqrt{1-\frac{r_e}{D+r_p}}\approx1-\frac{r_e}{2D}, \] where \(v_{\rm esc}(r_e)\) is the escape speed at \(r_e\). The flight-path angle at the entry interface obeys the formula of Problem 2(a) with \(r_1\) replaced by \(D\). Solve it for \(r_p\) in terms of \(\gamma_e\).
  3. Numerical. A capsule of this kind must enter at a flight-path angle between \(5.5^\circ\) and \(7.5^\circ\) below the horizontal (assumed). Steeper and the heating and deceleration are too severe; shallower and the capsule may skip back out of the atmosphere. Find the virtual perigee altitudes at the two edges of this corridor. Aim at \(\gamma_e=6.5^\circ\), and find \(v_\infty\), the trans-Earth injection, the entry speed and the flight time from the Moon to the entry interface.
  4. Analytical and numerical. Show that a small error \(\delta v_a\) in the capsule’s geocentric speed at the Moon’s distance moves the virtual perigee by \(\delta r_p=2r_p\,\dfrac{D+r_p}{D}\,\dfrac{\delta v_a}{v_a}\). How far does an error of \(1\) m/s move it? How wide is the whole corridor of part (c) in terms of \(v_a\), in m/s?
  5. Discussion. What does part (d) tell the mission designers? The capsule could instead brake into a low Earth orbit and come down from there at the lower speed of Gaganyaan’s crew module. Suppose the return ellipse is aimed at a perigee of \(400\) km altitude instead, and the capsule brakes there into a circular orbit. Estimate the cost of that burn, and say why sample-return capsules enter directly.

Problem 6 — An Earth made of point masses

The Earth is not a sphere. It bulges at the equator, and this is measured by the coefficient \(J_2\). In this problem we try to imitate the bulge with a few point masses, and ask what such a model gets right and what it gets wrong. A small piece of the n-body problem thus stands in for the shape of the Earth.

What you need. A point mass \(m'\) at the position \(\mathbf s\) has the potential \(-Gm'/|\mathbf r-\mathbf s|\) at \(\mathbf r\). Outside it, with \(|\mathbf s|=s<r\), \[ \frac{1}{|\mathbf r-\mathbf s|}=\frac1r\Bigl[1+\frac sr P_1(\cos\psi)+\Bigl(\frac sr\Bigr)^2P_2(\cos\psi) +\Bigl(\frac sr\Bigr)^3P_3(\cos\psi)+\Bigl(\frac sr\Bigr)^4P_4(\cos\psi)+\dots\Bigr], \] where \(\psi\) is the angle between \(\mathbf r\) and \(\mathbf s\), \(P_2(x)=\tfrac12(3x^2-1)\) and \(P_4(x)=\tfrac18(35x^4-30x^2+3)\). Outside the Earth, at latitude \(\varphi\) and longitude \(\lambda\), the potential of the Earth of mass \(M\) (\(\mu=GM\)) is \[ U=-\frac{\mu}{r}\Bigl[1-J_2\Bigl(\frac Rr\Bigr)^2P_2(\sin\varphi)-J_4\Bigl(\frac Rr\Bigr)^4P_4(\sin\varphi)\Bigr] -\frac{\mu}{r}\Bigl(\frac Rr\Bigr)^2\,3\cos^2\varphi\,\bigl(C_{22}\cos2\lambda+S_{22}\sin2\lambda\bigr)+\dots, \] with \(J_2=1.08263\times10^{-3}\), \(J_4=-1.6196\times10^{-6}\), \(C_{22}=1.5745\times10^{-6}\) and \(S_{22}=-9.0387\times10^{-7}\). In every model below the central mass is \(M-m\), so that the total mass is \(M\), and the extra masses add up to \(m\).

  1. Analytical. Put two equal masses \(m/2\) on the polar axis at \(z=\pm b\). Show that this model has \(J_2=-(m/M)(b/R)^2\). What shape of Earth does it describe? Where must the extra mass go instead?
  2. Analytical. Put \(N\) equal masses \(m/N\) on a circle of radius \(b\) in the equatorial plane, at the longitudes \(\lambda_k=\lambda_1+2\pi(k-1)/N\). Use \(\cos\psi_k=\cos\varphi\cos(\lambda-\lambda_k)\) and the fact that \(\sum_k\cos2(\lambda-\lambda_k)=0\) when \(N\ge3\). Show that the degree-two term then does not depend on \(\lambda\), and that the model has \(J_2=\tfrac12(m/M)(b/R)^2\). Find \(m/M\) for a ring at the Earth’s surface, \(b=R\).
  3. Analytical and numerical. For such a ring the part of the degree-four term that does not depend on longitude is \(m\,P_4(0)\,P_4(\sin\varphi)\) times \(-G b^4/r^5\), with \(P_4(0)=\tfrac38\) (you may use this). Show that the ring’s \(J_4\) is \(-\tfrac34J_2(b/R)^2\). Each further term of the series carries one more factor \((b/r)^2\). Explain why the ring of part (b) at \(b=R\) is a poor model for a satellite at \(700\) km, and why a smaller ring is better. Since \(m\) cannot exceed \(M\), how small can \(b\) be? Find this \(b\) in kilometres and the \(J_4\) of that model, and compare it with the Earth’s.
  4. Analytical and discussion. Now take \(N=2\), so that the Earth is modelled by three bodies, the central mass and two masses on the equator at \(\lambda=0\) and \(\lambda=\pi\). Show that the degree-two term now has an extra part of the form of the \(C_{22}\) term, with \(C_{22}=\tfrac14(m/M)(b/R)^2=J_2/2\). A geostationary satellite turns with the Earth, so it sits at a fixed longitude. Find the sign of the along-track acceleration this term gives a satellite at a longitude between \(0\) and \(90^\circ\). Using what you know about tangential burns from the phasing note, decide which way the satellite then drifts. At which longitudes can a geostationary satellite stay without control? Finally, use the Earth’s \(C_{22}\) and \(S_{22}\) to find the longitude of the long axis of the Earth’s equator and the longitudes where uncontrolled geostationary objects collect.