Chapter 24: Thermal Control

"Heat can never pass from a colder to a hotter body without some other change occurring at the same time." — Rudolf Clausius, 1854 (a statement of the second law of thermodynamics)

Overview

Hold a spacecraft still in your mind for a moment and look at where it sits. On one side is the Sun, a furnace pouring out more than a kilowatt of power onto every square meter that faces it. On the other side is the rest of the universe: a black, empty sink at $2.7\ \text{K}$, about three degrees above absolute zero, colder than anything you will ever touch. The spacecraft is a rock suspended in vacuum between a furnace and a freezer, and inside it are batteries that want to be near room temperature, propellant that must stay liquid, optics that must not warp by the width of a wavelength, and — on a crewed vehicle — people who will die outside a band about $20\ \text{degrees}$ wide. Thermal control is the discipline of holding everything in its band using nothing but the flow of heat you can arrange.

Here is what makes it strange, and why it deserves its own chapter. On Earth, if something runs hot you put a fan on it; if it runs cold you wrap it up. Both tricks move heat by convection — by carrying it away in moving air or fluid — and neither one works in space, because there is no air to move. In Chapter 1 we listed vacuum among the things that make space an unforgiving environment; here is one of the sharpest consequences. In vacuum, the only way a spacecraft can shed heat to its surroundings is by thermal radiation — by glowing in the infrared — and radiated power grows as the fourth power of temperature. You cannot open a window. You cannot run a fan to the outside. Every joule your electronics dissipate, every joule the Sun deposits, must ultimately leave as infrared light or it will pile up as rising temperature until something breaks. This is theme #2 of the book — space is an unforgiving environment — in its purest thermodynamic form.

The good news is that the physics is clean. The whole subject rests on one balance — absorbed power in, radiated power out — governed by the Stefan–Boltzmann law you may already know. From that single balance comes a number, the equilibrium temperature, that tells you what a surface will settle to if left alone; and the entire engineering of thermal control is the art of nudging that number into the band you need, surface by surface, using paint, insulation, heaters, and radiators.

In this chapter, you will learn to:

• Name and size the four heat loads a spacecraft faces: the Sun, sunlight bouncing off a planet (albedo), a planet's own infrared glow, and the cold of deep space.
• Derive the equilibrium temperature of a surface from radiative balance, and see why two numbers — $\alpha/\varepsilon$ and the absorbing-to-radiating area ratio — decide almost everything.
• Compute that a bare sphere at Earth's distance sits near $+5\ ^\circ\text{C}$, a black sunlit plate near $+120\ ^\circ\text{C}$, and a white one far below freezing — all from the same sunlight.
• Choose passive tools (multi-layer insulation, coatings, thermal mass) and active tools (heaters, heat pipes, radiators, louvers), and size a heater and a radiator with real numbers.
• Handle the two hardest regimes: cryogenics, where you fight to reach tens of kelvin, and the eclipse cycle, where the same surface must survive a hot case and a cold case hours apart.

Learning Paths

🚀 Space Enthusiast: Read 24.1 for the environment and 24.2 for the one equation that runs the whole show. The demonstration that a satellite's paint can swing its temperature by more than $250\ ^\circ\text{C}$ (24.3) is the idea that will change how you look at the gold foil and white panels on every spacecraft photo. Skim the algebra; keep the tables.

📐 Engineering Student: Read everything and derive the equilibrium temperature in 24.2 yourself before reading ours. The worked radiator- and heater-sizing calculations in 24.4, the boiloff estimate in 24.5, and the hot/cold bracket in 24.6 are the quantitative core; do the ⭐⭐/⭐⭐⭐ exercises and both case studies. This chapter's balance equation reappears every time you size a power system.

🎮 KSP Player: Stock KSP hides most of this, but if you fly with a thermal or life-support mod you have already watched parts overheat in sunlight and freeze in shadow. Sections 24.3 (radiators, coatings) and 24.6 (eclipse) are the physics behind those gauges; the radiator-sizing rule in 24.4 is exactly what the mod is computing for you.

🛰️ Industry Prep: Thermal is a subsystem that quietly sizes the others — heaters draw power, radiators cost mass and area, and the whole design is set by worst-case hot and cold. Read 24.2, 24.4, and 24.6 closely, and do the Mission Design Checkpoint: you will write your mission's thermal note and hand a heater-power line straight to the power budget of Chapter 25.

24.1 The thermal environment

Before you can control a spacecraft's temperature you have to know what is heating it and what is cooling it. In Earth orbit there are four thermal players, and the same four (with different numbers) follow you anywhere in the solar system. Three of them pour heat in; the fourth is the sink that lets heat out.

1. Direct sunlight. The Sun radiates about $3.8\times10^{26}\ \text{W}$ in all directions. By the time that power has spread over a sphere the size of Earth's orbit, it arrives as a flux — a power per unit area — that is the first number every thermal engineer memorizes.

Definition (solar flux). The solar flux (also called the solar constant when quoted at Earth's distance) is the radiant power from the Sun crossing a unit area held face-on to the Sun, in the absence of any atmosphere. At $1\ \text{AU}$ its value is $$S = 1361\ \text{W/m}^2$$ (Appendix B). Because the same total power spreads over a sphere whose area grows as the square of its radius, the flux falls off as the inverse square of heliocentric distance $d$: $$S(d) = S_{1\,\text{AU}}\left(\frac{1\ \text{AU}}{d}\right)^2.$$

That inverse-square law is the difference between a thermal problem and a different thermal problem as you move around the solar system. Run the numbers on a face-on surface:

A probe at Mars gets less than half the sunlight of one at Earth; a probe at Jupiter gets about one twenty-seventh. This is why the outer planets are a cold problem and the inner solar system is a hot one — and, as we will see in Chapter 25, why solar arrays that are ample at Earth become useless past Jupiter. Sanity check on the Mars value: $1361 / 1.524^2 = 1361 / 2.32 = 587\ \text{W/m}^2$. Good.

2. Albedo — sunlight the planet bounces back. A spacecraft in low orbit does not only see the Sun; it sees a whole planet filling half its sky, and that planet is lit. Part of the sunlight striking Earth is reflected straight back out, and a low-orbiting spacecraft catches some of it as a second source of short-wavelength (visible) heat, coming from below.

Definition (albedo). The albedo $a$ of a planet is the fraction of incident sunlight it reflects. Earth's global average (Bond) albedo is about $a \approx 0.30$ — roughly a third of the sunlight hitting Earth is reflected back to space — though locally it ranges from under $0.1$ over dark ocean to over $0.8$ above bright cloud and snow. The albedo heat load on a spacecraft is $q_{\text{alb}} = a\, S\, F$, where $F$ is a geometric view factor (between 0 and 1) that shrinks as the spacecraft climbs and the planet subtends a smaller angle.

Near the subsolar point in low Earth orbit the albedo load can approach $a\,S \approx 0.30 \times 1361 \approx 400\ \text{W/m}^2$ on a surface facing down — a real fraction of direct sunlight, and one that switches off the moment the spacecraft crosses into the planet's shadow or over its night side. Albedo is the most variable of the loads, which is exactly what makes it a headache: it depends on what is below you, and clouds do not file flight plans.

3. Planetary infrared. Earth also glows in its own right. Everything at a finite temperature radiates, and Earth, sitting near $255\ \text{K}$ as seen from space, emits long-wavelength infrared in all directions, day and night. A spacecraft in low orbit bathes in this planetary IR from below whether or not it is in sunlight.

How much? We can estimate Earth's outgoing IR from the same balance we will formalize in the next section: averaged over the globe, Earth must radiate away as much power as it absorbs from the Sun, or it would heat up without limit. It absorbs a fraction $(1-a)$ of the sunlight intercepted by its disk and re-emits over its whole surface, giving a mean outgoing flux of $$ q_{\text{IR}} \approx \frac{S(1-a)}{4} = \frac{1361 \times 0.70}{4} \approx 238\ \text{W/m}^2. $$ So Earth glows at a couple hundred watts per square meter of infrared — a load that, unlike albedo, does not vanish in eclipse. (The factor of 4 is the ratio of Earth's whole surface, $4\pi R^2$, to the disk $\pi R^2$ that catches sunlight; we meet it again in 24.2.)

4. Deep space — the sink. Finally, the reason any of this is survivable: point away from Sun and planet and you see the cosmic microwave background at $T_{\text{space}} = 2.7\ \text{K}$. Its radiated flux, $\sigma T^4 = 5.67\times10^{-8} \times (2.7)^4 \approx 3\times10^{-6}\ \text{W/m}^2$, is utterly negligible — for engineering purposes, deep space is a perfect cold sink at absolute zero. It is the only place a spacecraft can dump heat, and it is infinitely large and never fills up. Every radiator on every spacecraft is, in the end, aimed at this blackness.

💡 Intuition: Think of the environment as three heat taps and one drain. The Sun is a big tap (over $1\ \text{kW/m}^2$), albedo is a smaller and flickering tap, planetary IR is a steady trickle, and deep space is the drain — always open, but only draining through the narrow pipe of radiation. Thermal design is plumbing: decide which surfaces face the taps and which face the drain, and how wide each pipe (each surface's absorptivity and emissivity) should be.

🔗 Connection: Note how different this heat is from the heat of Chapter 7. There, re-entry dumped $\sim 30\ \text{MJ/kg}$ of kinetic energy into a shock layer at thousands of kelvin in minutes — a violent, transient furnace made by compressing air. Here there is no air at all, the fluxes are a thousand times gentler, and the timescale is the whole mission. Re-entry thermal protection and on-orbit thermal control are both "keeping heat where it belongs," but they are almost opposite problems: one is about surviving a brief inferno, the other about balancing a permanent, lopsided trickle between a furnace and a freezer.

🔄 Check Your Understanding 1. A spacecraft moves from Earth ($1\ \text{AU}$) to Jupiter ($5.2\ \text{AU}$). By what factor does the direct solar flux on a face-on surface change? 2. Which of the four loads switches off when a low-Earth-orbit satellite passes into Earth's shadow, and which two keep going?

Answers

1. It falls by $5.2^2 \approx 27$, from $1361$ to about $50\ \text{W/m}^2$ — a 27-fold drop, which is why Jupiter missions are cold-limited. 2. Direct sunlight and albedo both switch off in shadow (albedo is just reflected sunlight, so no sun means no albedo). Planetary infrared keeps going — Earth glows day and night — and so does the spacecraft's own internal heat from its electronics. The cold sink of deep space is, of course, always there. This is why eclipse is the classic cold case (24.6).

24.2 Radiative heat balance and equilibrium temperature

Everything a thermal engineer computes flows from one idea: in steady state, power in equals power out. If a surface absorbs more than it radiates, it heats up — but as it heats, its radiated power climbs steeply (as $T^4$), until emission catches absorption and the temperature stops rising. That settling point is the equilibrium temperature, and finding it is the first thing you do for any surface.

Strategy first. We will write the absorbed solar power on one side of an equation and the emitted infrared power on the other, insert the two material properties that say how well the surface absorbs sunlight and emits infrared, and solve for the temperature that makes them equal. The whole of passive thermal control is then just the study of how to choose those two properties and the surface's shape.

Two surface properties do the work, and they are the heart of the chapter.

Definition (absorptivity). The absorptivity (or solar absorptance) $\alpha$ of a surface is the fraction of incident solar (short-wavelength) radiation it absorbs, from $0$ (perfect reflector) to $1$ (perfect absorber). The rest is reflected.

Definition (emissivity). The emissivity (or emittance) $\varepsilon$ of a surface is the fraction of blackbody power it radiates at its own (infrared) temperature, again from $0$ to $1$. By Kirchhoff's law a surface's emissivity equals its absorptivity at the same wavelength — but sunlight arrives at short (visible) wavelengths near $0.5\ \mu\text{m}$ while a room-temperature spacecraft radiates in the far infrared near $10\ \mu\text{m}$, so a real surface generally has a solar $\alpha$ and an infrared $\varepsilon$ that are different numbers. That gap is the single most useful lever in thermal design.

The emitted power obeys the Stefan–Boltzmann law: a surface of area $A_{\text{rad}}$ and emissivity $\varepsilon$ at absolute temperature $T$ radiates $$ P_{\text{emit}} = \varepsilon\,\sigma\,A_{\text{rad}}\,T^4, \qquad \sigma = 5.670\times10^{-8}\ \text{W}\,\text{m}^{-2}\text{K}^{-4} $$ (the Stefan–Boltzmann constant, from Appendix B). The absorbed solar power is the flux times the absorbing (sunlit) area times the absorptivity, plus any internal dissipation $Q_{\text{int}}$ the electronics add from inside: $$ P_{\text{abs}} = \alpha\,S\,A_{\text{sun}} + Q_{\text{int}}. $$ Set them equal — that is equilibrium — and solve for $T$: $$ \alpha\,S\,A_{\text{sun}} + Q_{\text{int}} = \varepsilon\,\sigma\,A_{\text{rad}}\,T^4 $$ $$ \boxed{\ T_{\text{eq}} = \left(\frac{\alpha\,S\,A_{\text{sun}} + Q_{\text{int}}} {\varepsilon\,\sigma\,A_{\text{rad}}}\right)^{1/4}\ } $$

Definition (equilibrium temperature). The equilibrium temperature $T_{\text{eq}}$ of a surface or body is the steady temperature at which the power it radiates exactly equals the power it absorbs from all sources (sunlight, albedo, planetary IR, and internal dissipation). A body left alone drifts to this temperature and stays there; thermal design is the business of placing $T_{\text{eq}}$ where you need it.

Look at what the boxed equation is telling you, because it contains the whole philosophy of passive control. Ignore internal power for a moment ($Q_{\text{int}}=0$) and pull the temperature apart: $$ T_{\text{eq}} = \left(\frac{\alpha}{\varepsilon}\right)^{1/4} \left(\frac{A_{\text{sun}}}{A_{\text{rad}}}\right)^{1/4} \left(\frac{S}{\sigma}\right)^{1/4}. $$ Three factors. The last, $(S/\sigma)^{1/4}$, is fixed by where you are (the Sun). The middle, $(A_{\text{sun}}/A_{\text{rad}})^{1/4}$, is fixed by the object's shape. Only the first, $(\alpha/\varepsilon)^{1/4}$, is something you paint on. And because everything sits under a fourth root, even large changes in $\alpha/\varepsilon$ move the temperature only moderately — which is both a mercy (the problem is forgiving) and a frustration (you cannot fix a bad design with paint alone).

🚪 Threshold Concept. In vacuum, radiation is the only channel to the environment, and it scales as $T^4$. Once you internalize this, spacecraft thermal design stops being mysterious. There is no convection to the outside, so a surface's temperature is set almost entirely by the balance of what it absorbs and what it radiates — by $\alpha$, $\varepsilon$, area, and geometry — and nothing else. A hot component cannot be cooled by "better airflow"; it can only be given a path (by conduction) to a surface that faces cold space and radiates. Every radiator, heat pipe, and louver in the rest of this chapter exists because the $T^4$ drain is the only drain there is.

Worked example: a sphere and a plate at 1 AU

Let us find the equilibrium temperature of two simple shapes at Earth's distance, with no internal power, first for an idealized "grey" surface with $\alpha = \varepsilon$ (equal absorptivity and emissivity), so the paint factor is $1$ and geometry does all the talking.

Worked Example: the isothermal sphere. A sphere of radius $r$ intercepts sunlight on its cross-sectional disk, $A_{\text{sun}} = \pi r^2$, but radiates from its entire surface, $A_{\text{rad}} = 4\pi r^2$ (assume it spins fast enough to be one uniform temperature). The area ratio is $A_{\text{sun}}/A_{\text{rad}} = \pi r^2 / 4\pi r^2 = 1/4$, and $r$ cancels — size does not matter. With $\alpha = \varepsilon$, $$ > T_{\text{eq}} = \left(\frac{S}{4\sigma}\right)^{1/4} > = \left(\frac{1361}{4 \times 5.67\times10^{-8}}\right)^{1/4} > = \left(6.00\times10^{9}\right)^{1/4} = 278\ \text{K} = +5\ ^\circ\text{C}. > $$ A bare grey ball at Earth's distance sits at about $+5\ ^\circ\text{C}$ — a chilly room, remarkably close to comfortable. That is not luck: it is why a naked human-scale object near Earth is neither instantly frozen nor instantly cooked, and it anchors every thermal intuition you will build.

Sanity check on that arithmetic: $4 \times 5.67\times10^{-8} = 2.27\times10^{-7}$; $1361 / 2.27\times 10^{-7} = 6.00\times10^{9}$; the fourth root of $6.00\times10^{9}$ is $\sqrt{\sqrt{6.00\times10^9}} = \sqrt{77{,}500} = 278\ \text{K}$. And $278 - 273 = +5\ ^\circ\text{C}$. (If instead the sphere reflected $30\%$ of the sunlight like Earth — absorbing only $70\%$ — the same formula gives $278 \times 0.70^{1/4} = 255\ \text{K}$, exactly Earth's mean radiating temperature. The physics is one physics.)

Now the flat plate, which behaves completely differently because its geometry is different.

Worked Example: the sunlit flat plate. A thin flat plate held face-on to the Sun absorbs on its front face, $A_{\text{sun}} = A$. Suppose it is insulated on the back so it can only radiate from that same front face, $A_{\text{rad}} = A$. Now the area ratio is $A/A = 1$, four times larger than the sphere's, and with $\alpha = \varepsilon = 1$ (a black plate), $$ > T_{\text{eq}} = \left(\frac{S}{\sigma}\right)^{1/4} > = \left(\frac{1361}{5.67\times10^{-8}}\right)^{1/4} > = \left(2.40\times10^{10}\right)^{1/4} = 394\ \text{K} = +121\ ^\circ\text{C}. > $$ The same sunlight, on a differently shaped object, gives a temperature $116\ ^\circ\text{C}$ hotter — hot enough to boil water and to cook most electronics. The difference is entirely geometry: the plate pours all its absorbed sunlight out of the one small face it absorbed it on, while the sphere spreads the same absorption over four times the radiating area.

This $+121\ ^\circ\text{C}$ figure is worth cross-checking against Chapter 1, which quoted a spacecraft temperature range of roughly $-270$ to $+260\ ^\circ\text{C}$. Our black sunlit plate lands at the warm end but not the extreme; the very hottest surfaces are sunlit low-emissivity metals, which we can now explain. If a plate has $\alpha/\varepsilon = 3$ (bare polished aluminum: absorbs sunlight reasonably but radiates infrared poorly) and still radiates from only its front face, $T_{\text{eq}} = (3)^{1/4} \times 394 = 1.32 \times 394 = 518\ \text{K} = +245\ ^\circ\text{C}$ — right at that upper bound. The cold end, $-270\ ^\circ\text{C}$, is simply a well-insulated surface that sees only the $2.7\ \text{K}$ sky. The whole dramatic range in Chapter 1 is nothing but this one equation evaluated at its extremes of $\alpha/\varepsilon$ and geometry.

🐛 Find the Error. A student computes the temperature of a sunlit spherical satellite this way: "The sphere has surface area $4\pi r^2$, so I put $A_{\text{sun}} = 4\pi r^2$ and $A_{\text{rad}} = 4\pi r^2$; they cancel, giving $T = (S/\sigma)^{1/4} = 394\ \text{K}$." They conclude the satellite runs at $+121\ ^\circ\text{C}$. What did they get wrong, and what is the right answer?

Answer

They used the full surface area for the absorbing area. But a sphere only intercepts sunlight on the shadow it casts — its cross-sectional disk $\pi r^2$ — not on its whole surface; the back half is in its own shade. The correct absorbing area is $\pi r^2$, giving an area ratio of $1/4$, not $1$, and $T = (S/4\sigma)^{1/4} = 278\ \text{K} = +5\ ^\circ\text{C}$. The student's error inflated the temperature by $116\ ^\circ\text{C}$. The lesson: $A_{\text{sun}}$ is the projected area facing the Sun; $A_{\text{rad}}$ is the total area that can see cold space. Confusing the two is the most common mistake in a first thermal calculation.

🔄 Check Your Understanding 1. Two surfaces at $1\ \text{AU}$ have the same shape (same area ratio) but one has $\alpha/\varepsilon = 4$ and the other $\alpha/\varepsilon = 0.25$. What is the ratio of their equilibrium temperatures? 2. Why does adding internal electronics power $Q_{\text{int}}$ always raise $T_{\text{eq}}$, and when is its effect largest?

Answers

1. $T \propto (\alpha/\varepsilon)^{1/4}$, so the ratio is $(4/0.25)^{1/4} = 16^{1/4} = 2$. A 16-fold change in $\alpha/\varepsilon$ changes temperature only twofold — the fourth root tames it. 2. Because $Q_{\text{int}}$ adds to the absorbed side of the balance, so the surface must reach a higher $T$ to radiate it away. Its effect is largest when the external loads are small — in eclipse or in the outer solar system — which is exactly when a warm box of electronics can be a spacecraft's salvation, keeping itself alive on its own waste heat.

24.3 Passive thermal control

The cheapest, most reliable thermal hardware is the kind with no moving parts and no power draw. Passive thermal control shapes the equilibrium temperature purely through surfaces, insulation, and mass — and because it never fails and never runs out of electricity, engineers use it for as much of the job as they possibly can, reaching for active systems only where passive cannot cope. There are three main passive levers.

Surface coatings: painting the temperature

We just saw that $T_{\text{eq}} \propto (\alpha/\varepsilon)^{1/4}$. A coating is simply a way to choose $\alpha$ and $\varepsilon$, and the range available is astonishing. Here are representative values (Tier 2 — real coatings vary with thickness, supplier, and age):

Put these on the isothermal sphere of 24.2, which sat at $278\ \text{K}$ when $\alpha = \varepsilon$. Each finish multiplies that by $(\alpha/\varepsilon)^{1/4}$:

Read that table again and let it land: the identical ball, in the identical orbit, can be made to sit anywhere from $+168\ ^\circ\text{C}$ to $-116\ ^\circ\text{C}$ — a range of nearly $290\ ^\circ\text{C}$ — by nothing more than its choice of paint. This is why spacecraft are a patchwork of gold, white, and mirrored surfaces: each is a local thermostat, set by a coating chosen to hold that region in its band.

⚠️ Common Misconception: "Black absorbs, white reflects, so black is always hotter." Not in space. "Black" and "white" are statements about visible light — about $\alpha$. But a surface's temperature depends on $\alpha/\varepsilon$, and most white paints are nearly black in the infrared ($\varepsilon \approx 0.9$), so they radiate their heat away beautifully while reflecting the Sun. That is why white paint runs cold. Meanwhile polished metal is "shiny" (low $\alpha$) yet runs hot, because it is an even worse infrared emitter (tiny $\varepsilon$) — it can barely dump the little it absorbs. Your visible-light intuition is exactly backwards for the metals. Always think in $\alpha/\varepsilon$, not in color.

🧩 Productive Struggle. Before reading on, predict: you want a radiator — a panel whose job is to dump the spacecraft's waste heat to space while sitting in sunlight. Should its surface have a high or low $\alpha/\varepsilon$? And which finish in the table would you pick? Reason it out before continuing.

(Answer, developed in 24.4: you want $\alpha/\varepsilon$ as low as possible, so the panel barely heats up in sunlight yet emits strongly — an OSR, at $\alpha/\varepsilon = 0.10$, is the classic choice. A black radiator would cook itself absorbing the very sunlight it is trying to work in.)

Multi-layer insulation: the space blanket

To keep heat in (or a cruel environment out), spacecraft wear a quilted, shimmering blanket that is one of the most recognizable objects in spaceflight.

Definition (multi-layer insulation). Multi-layer insulation (MLI) is a passive insulator made of many thin radiation-reflecting layers — typically aluminized Mylar or Kapton, a dozen to several dozen of them — separated by low-conductivity spacers (a fine polyester net) and held in near-vacuum. Because there is no air to carry heat by convection and the layers barely touch, essentially the only way heat crosses an MLI blanket is by radiation, and each reflective layer throws most of it back. MLI is quantified by an effective emittance $\varepsilon^{*}$ — the emissivity the whole blanket behaves as if it had — which can be a few thousandths in the ideal case and a few hundredths in practice.

Why layers work is pure radiation bookkeeping. Two parallel surfaces of emissivity $\varepsilon$ exchange a net radiative flux $q = \sigma(T_h^4 - T_c^4)/(2/\varepsilon - 1)$. Slip $N$ reflective shields between them and the flux drops by a factor of $1/(N+1)$, because the heat now has to be re-radiated across $N+1$ gaps in series: $$ q_{\text{MLI}} = \frac{\sigma\,(T_h^4 - T_c^4)}{(N+1)\,(2/\varepsilon - 1)}. $$

Worked Example: how much heat leaks through MLI? Wrap a warm electronics bay at $T_h = 300\ \text{K}$ in a blanket whose outer layer sits at $T_c = 150\ \text{K}$, using $N = 20$ aluminized shields with $\varepsilon = 0.05$ each. First the bare-surface flux if there were no blanket and the bay radiated as a near-blackbody: $$ > \sigma(T_h^4 - T_c^4) = 5.67\times10^{-8}\,(300^4 - 150^4) > = 5.67\times10^{-8}\,(8.10 - 0.51)\times10^{9} = 430\ \text{W/m}^2. > $$ Now with the blanket: $2/\varepsilon - 1 = 2/0.05 - 1 = 39$, and $N+1 = 21$, so $$ > q_{\text{MLI}} = \frac{430}{39 \times 21} = \frac{430}{819} = 0.53\ \text{W/m}^2. > $$ The blanket cuts the heat flow by a factor of over $800$ — from $430$ down to about half a watt per square meter. Equivalently, its ideal effective emittance is $\varepsilon^{*} = 1/819 \approx 0.0012$.

🔧 Engineering Reality: That $\varepsilon^{*} = 0.0012$ is a laboratory ideal almost never seen in flight. Real MLI is stitched, taped, and pierced by struts, wires, and bolts; it gets compressed at seams (where layers touch, they conduct); and it has edges. Every one of these is a thermal short. Flown MLI typically achieves $\varepsilon^{*} \approx 0.01$ to $0.05$ — ten to forty times worse than ideal. A good rule for a first estimate is $\varepsilon^{*} \approx 0.03$, which for the example above gives a more honest $q = 0.03 \times 430 \approx 13\ \text{W/m}^2$. Thermal engineers spend enormous effort on the installation of MLI, because the physics of the material is easy and the reality of the seams is hard. This is theme #2 again: in space the details of workmanship are the difference between a warm spacecraft and a dead one.

Thermal mass: riding out the swings

The third passive lever is inertia. A massive component changes temperature slowly, because it takes $Q = m\,c_p\,\Delta T$ joules to move its temperature by $\Delta T$ (where $c_p$ is specific heat — aluminum's is about $900\ \text{J/(kg·K)}$). Thermal mass does not change the average temperature, but it damps the swings — the difference between sunlight and eclipse. A heavy battery pack coasts through a $35$-minute eclipse barely cooling; a thin, light bracket in the same shadow plunges. But mass is the enemy (theme #4): every kilogram of thermal ballast is a kilogram of payload you did not launch, a trade we weighed in Chapter 23. So thermal mass is used sparingly, and mostly you get it for free from hardware that had to be there anyway.

24.4 Active thermal control

When passive design cannot hold a component in its band — because the heat load is too large, too variable, or must be moved from where it is made to where it can be dumped — you add active control: hardware that uses power, moving fluid, or mechanical motion to manage heat. Four devices cover most of the job.

Heaters: the simplest fix, paid for in watts

If a component runs too cold, warm it with an electrical resistance heater on a thermostat. Heaters are utterly reliable and precise, and they are everywhere on spacecraft — but they are not free: every watt of heat is a watt drawn from the power system, which is why the thermal note you write in this chapter hands a number straight to the power budget of Chapter 25.

Worked Example: heater power to keep a battery alive. Batteries hate the cold — below about $0\ ^\circ\text{C}$ they lose capacity and can be permanently damaged. Suppose a battery box must be held at $T = 273\ \text{K}$ ($0\ ^\circ\text{C}$) through eclipse. It is wrapped in MLI with effective emittance $\varepsilon^{*} = 0.05$, presenting $A = 0.2\ \text{m}^2$ to cold space. The heat it loses by radiation is $$ > Q_{\text{loss}} = \varepsilon^{*}\,\sigma\,A\,T^4 > = 0.05 \times 5.67\times10^{-8} \times 0.2 \times (273)^4. > $$ With $273^4 = 5.55\times10^{9}$, this is $0.05 \times 5.67\times10^{-8} \times 0.2 \times 5.55\times 10^{9} = 3.1\ \text{W}$. A modest $\sim 3\ \text{W}$ heater holds the battery at $0\ ^\circ\text{C}$. Now see what the MLI bought you: strip the blanket off, so the box radiates as bare white paint ($\varepsilon = 0.85$), and the loss becomes $0.85 \times 62.9 = 53\ \text{W}$ — seventeen times more heater power, likely more than the battery is worth. Insulation and heaters are partners: MLI makes the heater small enough to afford.

Radiators: the only way out

Every watt a spacecraft generates internally — from computers, radios, payloads — plus every watt it absorbs from the environment must ultimately leave through a radiator.

Definition (radiator). A radiator is a surface designed to reject waste heat to space by infrared emission. It is given a high emissivity $\varepsilon$ and (crucially, if it must sit in sunlight) a low solar absorptivity $\alpha$, so it emits strongly while absorbing little. The net heat a radiator of area $A$ at temperature $T$ rejects is its emission minus what it absorbs from the environment: $Q_{\text{net}} = \varepsilon\,\sigma\,A\,T^4 - \alpha\,S\,A_{\text{sun}} - (\text{albedo and planetary IR absorbed})$.

Worked Example: sizing a radiator. A spacecraft must reject $Q = 500\ \text{W}$ of waste heat while its radiator runs at $T = 300\ \text{K}$ ($+27\ ^\circ\text{C}$, a typical electronics temperature), with $\varepsilon = 0.85$. First the easy case: the radiator points at deep space, out of the Sun. $$ > A = \frac{Q}{\varepsilon\,\sigma\,T^4} > = \frac{500}{0.85 \times 5.67\times10^{-8} \times (300)^4} > = \frac{500}{0.85 \times 459} = \frac{500}{390} = 1.28\ \text{m}^2. > $$ Now the hard case: the same radiator, painted black ($\alpha = 0.95$), is forced to face the Sun. It now absorbs $\alpha S = 0.95 \times 1361 = 1293\ \text{W/m}^2$ while emitting $390\ \text{W/m}^2$ — it absorbs more than it emits and cannot reject any net heat at all; it would run away hot. Even with a good white radiator ($\alpha = 0.2$) the sunlight adds $0.2 \times 1361 = 272\ \text{W/m}^2$ of load, so the net rejection per square meter drops from $390$ to $390 - 272 = 118\ \text{W/m}^2$, and the area balloons to $500/118 = 4.2\ \text{m}^2$ — more than triple. This is why radiators are mounted edge-on to the Sun or on the shaded side of the vehicle, and finished in low-$\alpha$ OSR. Geometry and coating, again, decide everything.

💡 Intuition: A radiator is a window through which heat leaves as light. Its size is set by the $T^4$ law: because emission climbs so steeply with temperature, a radiator running hotter is dramatically smaller. Doubling the radiator temperature shrinks its area sixteen-fold. This is a deep reason spacecraft try to reject heat at the highest temperature the hardware tolerates, and why cryogenic systems — which must reject heat at low temperature — need such enormous radiators (24.5).

Heat pipes: moving heat without power

Waste heat is usually made in one place (a processor, an amplifier) and must reach a radiator somewhere else. A solid metal strap conducts it, but slowly and heavily. The elegant answer is a heat pipe.

Definition (heat pipe). A heat pipe is a sealed tube containing a working fluid (commonly ammonia for spacecraft) and a capillary wick along its wall. At the hot end the fluid absorbs heat and evaporates; the vapor flows to the cold end and condenses, releasing that heat; and the wick draws the liquid back by capillary action — no pump, no power. Because it moves heat as latent heat of phase change rather than by conduction, a heat pipe carries heat along its length with an effective thermal conductivity hundreds of times that of solid copper, while weighing far less.

Heat pipes are passive-active hybrids — they use no power but actively transport heat — and they lace through nearly every large spacecraft, tying hot boxes to cold radiators. Their one weakness is that a simple heat pipe needs its condenser no lower than its evaporator to return liquid against gravity, which is why they are tested carefully on the ground (where gravity fights the wick) even though in orbit, weightless, they work in any orientation.

Louvers: a thermostat with no power

🔧 Engineering Reality: A louver is a set of hinged, polished metal blades — a venetian blind for heat — mounted over a radiator and driven open or shut by a bimetallic spring that responds to temperature. Shut, the polished blades hide the radiator and present a low effective emittance ($\varepsilon^{*} \approx 0.1$), trapping heat when the spacecraft is cold. Open, they expose the high-emissivity radiator beneath ($\varepsilon^{*} \approx 0.7$), dumping heat when it is hot. Because the bimetal is driven by the temperature itself, a louver is a self-regulating thermostat that draws no power — a beautifully simple way to let a spacecraft radiate hard when busy and warm and insulate itself when idle and cold. Mariner, Voyager, and countless satellites since have flown them.

24.5 Cryogenic thermal management

Some payloads are not content with room temperature; they demand cold. Infrared telescopes must chill their detectors and optics to tens of kelvin or their own thermal glow would blind them (a warm infrared sensor is like a camera with the lens on fire). Superconducting instruments need liquid-helium temperatures. And cryogenic propellants — liquid hydrogen at $20\ \text{K}$, liquid oxygen at $90\ \text{K}$ — must be kept liquid against a constant trickle of heat. Cryogenic thermal management is the art of reaching and holding temperatures far below anything the environment will give you for free.

The enemy is parasitic heat leak: every path by which ambient heat sneaks into the cold zone — through support struts (conduction), through wires, and through radiation from warmer surroundings. Whatever leaks in must be removed, or it boils the cryogen away.

Worked Example: boiloff of liquid hydrogen. A liquid-hydrogen tank suffers a parasitic heat leak of $Q = 5\ \text{W}$ despite its insulation. Liquid hydrogen's latent heat of vaporization is $h_{fg} \approx 446\ \text{kJ/kg}$. Each second, the mass that boils off is $$ > \dot m = \frac{Q}{h_{fg}} = \frac{5\ \text{W}}{446{,}000\ \text{J/kg}} = 1.12\times10^{-5}\ \text{kg/s}. > $$ Over a day that is $1.12\times10^{-5} \times 86{,}400 = 0.97\ \text{kg/day}$ — nearly a kilogram of hydrogen lost per day from a mere $5\ \text{W}$ leak. For a tank holding a hundred kilograms, that is about $1\%$ per day, and it is exactly why cryogenic upper stages have a limited loiter time in orbit — a constraint we met as boiloff in Chapter 22. Reducing that leak from $5\ \text{W}$ to $1\ \text{W}$ — with more MLI, fewer conductive struts, and a vapor-cooled shield — directly multiplies the mission's cryogenic lifetime fivefold.

There are two ways to reach cryogenic temperatures. Passive cooling uses the cold of space itself: shade the payload from the Sun and every warm surface with a sunshield (a giant, staged MLI), then let it radiate its own heat away to the $2.7\ \text{K}$ sky until it settles at tens of kelvin. Active cooling adds a cryocooler — a miniature closed-cycle refrigerator (Stirling, pulse-tube, or Joule–Thomson) that pumps heat "uphill" from the cold stage to a warmer radiator, at the cost of electrical power and a vibrating machine that must not blur the very optics it serves.

🔗 Connection: The James Webb Space Telescope is the showcase. It parks at the Sun–Earth L2 point (which we will reach in Chapter 15), where Sun, Earth, and Moon all lie in one direction, so a single five-layer, tennis-court-sized sunshield can block them all at once. The sunshield's Sun-facing layer runs near $+110\ ^\circ\text{C}$ while its shaded side reaches roughly $-235\ ^\circ\text{C}$ (about $40\ \text{K}$) — a swing of some $345\ ^\circ\text{C}$ across five thin sheets, achieved with no power at all (temperatures widely reported; Tier 2). The telescope then cools passively to about $40\ \text{K}$, cold enough for its near-infrared instruments; its mid-infrared instrument, needing about $7\ \text{K}$, adds an active cryocooler. Passive as far as it can go, active only for the last stretch — the universal pattern of thermal design, taken to its beautiful extreme.

24.6 Eclipse cycling and worst-case hot and cold

A spacecraft's thermal environment is not steady — it cycles. In low Earth orbit a satellite circles every $\sim 90$ minutes, and for up to about $35$ of those minutes it is in Earth's shadow: sunlight and albedo vanish, leaving only planetary IR and its own internal heat. It cools. Then it bursts back into sunlight and heats. Around $5{,}500$ times a year, every year, it swings between these states. In geostationary orbit the pattern is different: for most of the year the satellite is sunlit continuously, but around the equinoxes it passes through nightly eclipses of up to $72$ minutes for a few weeks — a seasonal cold shock. Either way, the design must survive both ends of the swing, and the ends are what you size to.

Worked Example: how far does a component cool in eclipse? Take an aluminum component, $m = 10\ \text{kg}$, $c_p = 900\ \text{J/(kg·K)}$, radiating area $A = 0.3\ \text{m}^2$ with $\varepsilon = 0.8$, starting at $T_0 = 290\ \text{K}$ and dissipating $Q_{\text{int}} = 5\ \text{W}$ internally. In eclipse it radiates to space at (initially) $$ > Q_{\text{emit}} = \varepsilon\,\sigma\,A\,T_0^4 > = 0.8 \times 5.67\times10^{-8} \times 0.3 \times (290)^4 = 96\ \text{W}, > $$ using $290^4 = 7.07\times10^{9}$. Net loss is $96 - 5 = 91\ \text{W}$. If that rate held for a $35$-minute ($2{,}100\ \text{s}$) eclipse, the energy lost would be $91 \times 2{,}100 = 1.9\times10^{5} \ \text{J}$, dropping the temperature by $$ > \Delta T = \frac{Q\,t}{m\,c_p} = \frac{1.9\times10^{5}}{10 \times 900} = 21\ \text{K}. > $$ So the component cools roughly $21\ \text{K}$, from $290\ \text{K}$ ($+17\ ^\circ\text{C}$) to about $269\ \text{K}$ (roughly $-4\ ^\circ\text{C}$), in a single eclipse.

🔧 Engineering Reality: That $21\ \text{K}$ is an over-estimate, and it is worth knowing why. We held the radiated power fixed at its initial $96\ \text{W}$, but as the component cools, its emission falls as $T^4$ — at $270\ \text{K}$ it radiates only $(270/290)^4 = 0.75$ as much. The true cooling curve bends and flattens, so the real drop is somewhat less than $20\ \text{K}$. A proper analysis integrates $m\,c_p\,dT/dt = Q_{\text{int}} - \varepsilon\sigma A T^4$ numerically (or lets a thermal code do it), but the linear estimate is the right first move: it is quick, it is conservative (it over-predicts the cold), and it tells you immediately whether you need a heater. Here, a component reaching $-3\ ^\circ\text{C}$ is fine for electronics but too cold for a battery — so a battery in this bay needs the heater we sized in 24.4.

This cycling has a second cost beyond temperature: fatigue. Every hot-cold swing expands and contracts the structure, and thousands of cycles a year work-harden solder joints, crack coatings, and flex hinges — the thermal-fatigue failure mode that connects straight back to the materials and load discussion of Chapter 23. A part that survives one cycle must survive tens of thousands. Theme #2, once more: it is not enough to work; it must work for the whole mission, through every swing, with no one to fix it.

The two cases that size the design

All of this collapses into a design discipline: you do not design for the average, you design for the extremes. Two bounding cases bracket every thermal design.

• Worst-case hot. Full sunlight plus peak albedo plus planetary IR plus maximum internal power plus end-of-life surfaces — because coatings degrade under ultraviolet and atomic oxygen, and a white paint's $\alpha$ can drift from $0.2$ up toward $0.4$ over years, quietly making an old spacecraft hotter than a new one. The radiators must be big enough to survive this.
• Worst-case cold. Deep eclipse, no sun, no albedo, minimum internal power (safe mode, payload off), beginning-of-life surfaces at their most reflective. The heaters must be strong enough to survive this.

The cruelty is that the same surfaces must pass both tests at once. Make a radiator big enough for the hot case and it over-cools in the cold case, demanding more heater power; add absorptivity to warm the cold case and you cook in the hot case. Thermal design lives in the gap between these two walls, and its central number is the margin between predicted extremes and the hardware's survival limits — a margin we will formalize as a design principle in Chapter 29.

🔄 Check Your Understanding 1. Why is eclipse the worst-case cold condition even though the spacecraft is still receiving Earth's infrared and making its own internal heat? 2. A designer sizes a spacecraft's radiator using beginning-of-life coating properties. Why is this dangerous, and which case does it get wrong?

Answers

1. Because the two largest heat inputs — direct sunlight (over $1\ \text{kW/m}^2$ face-on) and its reflected albedo — both vanish, while only the much smaller planetary IR ($\sim 240\ \text{W/m}^2$) and the modest internal dissipation remain. Inputs collapse, the $T^4$ drain keeps pulling, and the temperature falls. 2. Coatings degrade in orbit: $\alpha$ rises with UV/atomic-oxygen exposure, so end-of-life surfaces absorb more sunlight and run hotter. Sizing the radiator with fresh (low-$\alpha$) properties under-predicts the worst-case hot condition — the spacecraft may be fine at launch and overheat years later. Always run the hot case with degraded, end-of-life surfaces.

Mission Design Checkpoint: your thermal note and thermal.py

Across the book you are designing one real mission and, if you like, building the astrotools package to compute it. In Chapter 23 you built a mass budget; this chapter adds a thermal note — a hot-case and cold-case temperature bracket for your spacecraft's exterior — and the thermal.py module that produces it.

The design. For your chosen mission (Track A comsat, B lunar lander, C Mars orbiter, D asteroid rendezvous), write a short thermal note answering three questions. (1) Where are you? Use the inverse-square law to find your solar flux — $1361\ \text{W/m}^2$ near Earth, but $\sim 586$ at Mars or as little as tens of watts at a distant asteroid. (2) What is your hot case and cold case? Pick a representative external surface (a radiator, say, at $\alpha = 0.2$, $\varepsilon = 0.85$) and compute its equilibrium temperature with full sun plus internal power (hot) and in eclipse on internal power alone (cold). (3) What does that demand? State the heater power your cold case needs and the radiator area your hot case needs. That heater number is the line you hand to your power budget in Chapter 25.

The code. Create astrotools/thermal.py with the balance of this chapter:

import math

SIGMA = 5.670e-8    # Stefan-Boltzmann constant, W m^-2 K^-4
S_1AU = 1361.0      # solar flux at 1 AU, W/m^2

def solar_flux(au):
    """Solar flux (W/m^2) at heliocentric distance `au` in astronomical units."""
    return S_1AU / au**2

def equilibrium_temperature(alpha, eps, a_sun, a_rad, flux=S_1AU, q_internal=0.0):
    """Equilibrium temperature (K) from radiative balance.
    alpha: solar absorptivity; eps: IR emissivity; a_sun: sunlit area (m^2);
    a_rad: radiating area (m^2); flux: solar flux (W/m^2); q_internal: internal power (W)."""
    absorbed = alpha * flux * a_sun + q_internal
    return (absorbed / (eps * SIGMA * a_rad)) ** 0.25

if __name__ == "__main__":
    # MDR thermal note: a body-mounted radiator panel, 0.5 m^2, radiating one face.
    # HOT case: full sun, end-of-life alpha degraded to 0.35, plus 60 W of electronics.
    t_hot = equilibrium_temperature(0.35, 0.85, 0.5, 0.5, flux=S_1AU, q_internal=60.0)
    # COLD case: eclipse (no sun), 60 W internal only, beginning-of-life.
    t_cold = equilibrium_temperature(0.20, 0.85, 0.5, 0.5, flux=0.0, q_internal=60.0)
    print("hot-case panel:", round(t_hot - 273.15, 1), "C")
    print("cold-case panel:", round(t_cold - 273.15, 1), "C")
    # Expected output:
    # hot-case panel: 60.4 C
    # cold-case panel: -49.8 C

The bracket — about $+60\ ^\circ\text{C}$ hot, $-50\ ^\circ\text{C}$ cold — is wider than most hardware tolerates, which is the point: it tells you at a glance that this panel needs heaters for the cold case and confirms it can reject its load in the hot case. (Both numbers are computed by hand in this chapter's code/ folder; the code is never run at build time.) In the capstone (Chapter 40) this note becomes the thermal-subsystem section of your Mission Design Review.

Summary

Thermal control is radiative bookkeeping between a furnace and a freezer. Carry these forward:

Constants worth memorizing: $S = 1361\ \text{W/m}^2$; $\sigma = 5.67\times10^{-8}\ \text{W m}^{-2} \text{K}^{-4}$; a grey sphere at 1 AU sits near $+5\ ^\circ\text{C}$; deep space is $2.7\ \text{K}$; to convert $^\circ\text{C}$ to K add $273$.

Spaced Review

Retrieval strengthens memory. Answer from memory before checking, then look back at the cited chapter.

1. (Ch. 7) Re-entry heating and on-orbit thermal control are both "heat problems," but the source of the heat is completely different. In one sentence, what heats a re-entry vehicle, and why is that not what heats an orbiting satellite?
2. (Ch. 7) Roughly how much kinetic energy per kilogram must a vehicle returning from low Earth orbit shed, and into what does it go?
3. (Ch. 23) Thermal mass helps a component ride out an eclipse — but why does adding it conflict with the central discipline of spacecraft structures?
4. (Ch. 23) Thermal cycling drives fatigue in structures. In your own words, what is fatigue, and why do thousands of eclipse cycles a year make it a design driver?

Answers

1. A re-entry vehicle is heated by compression of the air piling up in the shock layer ahead of it (not friction), reaching thousands of kelvin for minutes; an orbiting satellite has no air at all and is heated only by absorbed sunlight, albedo, planetary IR, and its own electronics — gentle, steady loads balanced by radiation. 2. About $30\ \text{MJ/kg}$ (from $\tfrac12 v^2$ at $\sim 7.8\ \text{km/s}$), and it goes into heat — carried away by the shock layer and a heat shield. 3. Mass is the enemy: every kilogram of thermal ballast is a kilogram that is not payload and that cost fuel to launch, so thermal mass is used sparingly and preferably harvested from hardware that had to be aboard anyway. 4. Fatigue is the growth of cracks under repeated cycles of stress even when each cycle is well below the material's one-time breaking strength; each hot-cold swing thermally expands and contracts the structure, and at $\sim 5{,}500$ cycles a year a joint must endure tens of thousands of loadings over a mission — so it is the number of cycles, not any single load, that governs.

What's Next

We have kept the spacecraft alive between the furnace and the freezer — but every heater we switched on, every cryocooler we ran, and every computer whose waste heat we so carefully routed to a radiator drew its energy from somewhere. That somewhere is the power system, and it is the next link in the chain: the thermal note you just wrote ends with a heater-power number, and that number is an input to the power budget. In Chapter 25 we size the solar arrays, batteries, and (for the outer solar system, where sunlight fails) the radioisotope generators that supply it — and we will find that power and thermal are two views of the same energy flowing through the spacecraft, since nearly every watt generated ends its life as waste heat that thermal control must throw away.