TL;DR: A small satellite's thermal design must move internally generated heat to a surface that can reject it while keeping vulnerable components warm during low-power and eclipse conditions. Average temperature is insufficient: local gradients, transient loads, attitude, contact conductance, and control failures determine whether hardware stays within its qualified limits. The calculations here are simplified design exercises, not a flight-ready thermal model.
A proposed imaging satellite, fictional Kestrel, has enough electrical power for its camera and radio. Its designers nevertheless face two apparently contradictory problems. The processor becomes too hot during a long image-processing session, and the battery becomes too cold during a quiet eclipse. Adding a larger radiator appears to help the first problem while worsening the second.
The contradiction disappears when heat is treated as a flow through a particular structure over time. Kestrel is not one temperature in a cold environment. It is a collection of components that produce, store, exchange, and radiate energy. A useful design notebook follows those pathways and identifies which assumptions make each simplified calculation valid.
NASA's Small Spacecraft Technology thermal-control chapter describes the core distinction: vacuum removes ordinary external convection, leaving conduction through connected material and radiation between surfaces as the relevant heat-transfer mechanisms. Surface absorptivity and emissivity affect different parts of the exchange. The rest of this article develops original examples around those principles rather than reproducing a catalogue of spacecraft hardware.
Notebook page: define the component limits before the radiator
Kestrel's team begins with the allowable temperature ranges supplied for each component and assembly. Operational limits can differ from survival limits. A device may tolerate a cold storage condition while unpowered but require a narrower range to operate correctly. The thermal design must identify which condition applies during each mission mode.
The limits also need a physical reference. A processor junction limit is not the same as a case-temperature limit or the temperature measured on a nearby board. If the sensor sits several centimeters away, the model needs a defensible relationship between its reading and the component of concern. A comforting telemetry number can be misleading when the hottest location is elsewhere.
For this exercise, Kestrel's team invents a processor requirement and a battery requirement solely to structure the trade study. It does not borrow generic limits from an unrelated product and treat them as qualification evidence. Flight hardware would need its actual manufacturer data, mission requirements, and test results.
The team lists mission modes beside those limits: launch, initial acquisition, normal imaging, processing, downlink, eclipse, safe mode, and recovery. Each mode has a power profile and attitude assumption. The thermal problem is already broader than the nominal science sequence because hardware must survive the transitions that make that sequence possible.
A requirement also names duration. A brief temperature excursion and a sustained exposure can have different implications, depending on the component and qualification basis. The notebook does not invent a universal permitted excursion. It records the applicable evidence and asks the responsible engineer to resolve any gap before the operating plan depends on it.
Notebook page: follow twelve watts through a mounting interface
Suppose a fictional electronics box dissipates twelve watts in a mode of interest. Its mounting interface has an effective thermal resistance of two kelvin per watt under the assumed assembly condition. In a simple steady-state model, the temperature difference across that interface is twenty-four kelvin. This is a calculated gradient, not the absolute box temperature.
If the supporting panel is at twenty degrees Celsius, the corresponding box-side temperature would be forty-four degrees Celsius under that simplified model. If the panel warms by ten degrees, the box warms by the same amount unless another heat path or operating condition changes. The box's temperature cannot be assessed from its own dissipation alone.
Now imagine that an assembly change doubles the interface resistance to four kelvin per watt. The gradient becomes forty-eight kelvin at the same twelve watts. A design that appeared comfortable can fail without any increase in electrical load. Contact quality, mounting pressure, surface condition, and interface materials are therefore part of the thermal design rather than minor mechanical details.
The example uses one resistance to make the dependency visible. A real assembly can have several parallel and series paths, temperature-dependent properties, and nonuniform contact. The engineering model should reflect the level of detail needed to predict the relevant temperatures, then be checked against test data.
Kestrel's interface drawing consequently specifies more than the bolt pattern. It identifies the intended conductive path and the assumptions that support its modeled conductance. If manufacturing substitutes a material or changes a mounting arrangement, the thermal owner reviews the effect. Mechanical compatibility does not automatically preserve heat-transfer performance.
Notebook page: a radiator rejects power at a temperature
A radiator's capacity is not a fixed number independent of temperature. In an idealized view to a very cold sink, emitted thermal power is proportional to emissivity, area, and the fourth power of absolute temperature. The Stefan–Boltzmann constant is approximately 5.67 times ten to the minus eight watts per square meter per kelvin to the fourth.
Use a fictional radiator area of one tenth of a square meter, emissivity of eight tenths, and temperature of three hundred kelvin. The ideal emitted power is about 36.7 watts. At two hundred seventy kelvin, the same expression gives about 24.1 watts. A lower operating temperature reduces the power that the same surface emits.
These numbers omit absorbed sunlight, planetary infrared radiation, reflected sunlight, and view blockage. They are not net rejection estimates for an Earth-orbiting spacecraft. Their purpose is to show why a radiator sized with an unspecified temperature is incomplete. The surface must both receive heat from the equipment and reject it under the actual environment.
If another spacecraft panel fills part of the radiator's view, the radiator exchanges energy with that panel rather than an ideal cold sink over that portion. If attitude turns the surface toward sunlight, absorbed power can rise. A geometric area on a drawing is therefore not equivalent to an unobstructed effective area in every mission mode.
Kestrel's team records radiator orientation and surrounding geometry alongside its surface properties. The design review asks which attitudes are permitted during high-power operation and what happens if the spacecraft cannot maintain them. That question connects thermal performance directly to attitude control and fault recovery.
Notebook page: absorption and emission are different properties
A surface that looks dark to the eye is not fully described by that visual impression. Solar absorption concerns incoming radiation over the relevant solar spectrum, while thermal emission concerns outgoing radiation at the surface's operating temperature. Treating the two as a single generic reflectivity can produce an incorrect balance.
NASA's thermal chapter discusses the separate optical properties used in spacecraft analysis. Kestrel's design notebook uses those documented concepts without assuming that any named paint has the required performance. The selected surface needs measured or qualified properties appropriate to the mission, including credible changes over its lifetime.
For an arithmetic illustration, suppose a surface receives two hundred watts of incident power in a particular geometry. An absorptivity of one quarter corresponds to fifty watts absorbed. An absorptivity of one half corresponds to one hundred watts. That difference can exceed the internal load the radiator was intended to remove.
The incident power itself depends on geometry and environment. A tilted surface intercepts a different amount of direct radiation from a normally illuminated surface. Reflections and shadowing can complicate the picture further. The simplified multiplication is useful only after the incident quantity has been defined correctly.
Kestrel therefore avoids selecting a coating solely from a room-temperature appearance sample. The choice is tied to optical properties, durability, contamination sensitivity, electrical requirements where applicable, and manufacturing constraints. The thermal solution must survive the actual mission rather than only a clean laboratory drawing.
Notebook page: thermal mass buys time, not permanent capacity
During a short processing burst, Kestrel's electronics may generate more heat than the radiator rejects. The excess can temporarily raise the temperature of connected material. A lumped model expresses the rate of temperature change as net power divided by effective heat capacity.
Assume a fictional effective heat capacity of nine hundred joules per kelvin and a net heat input of fifteen watts. Over sixty seconds, the added energy is nine hundred joules, corresponding to a one-kelvin rise in the simplified isolated model. Over ten minutes, the same constant imbalance would add ten kelvin if the changing temperature did not alter heat flow.
That last condition is usually too simple for a long interval. Radiation increases with temperature, conduction responds to changing gradients, and different parts of the assembly do not necessarily remain isothermal. The calculation is a first estimate that identifies a possible transient issue; it is not a substitute for a time-dependent thermal model.
Thermal mass can make a burst feasible even when its instantaneous dissipation exceeds the steady rejection capacity. It cannot support that imbalance indefinitely. A duty-cycle plan must allow sufficient recovery and must account for the temperature at the start of the next burst. Repeating a short operation before the system cools can accumulate heat across cycles.
The operations team therefore tracks a thermal state, not merely a timer. A fixed cooldown interval might be acceptable under tested conditions, but a warm initial state or altered attitude can invalidate it. The notebook identifies the measurements and margins needed to decide when another high-power sequence is allowed.
Notebook page: eclipse is a different balance
When direct sunlight disappears in eclipse, the spacecraft's external heating changes. Internal equipment may also enter a lower-power mode. A surface designed to reject a large payload load can continue radiating while that load is absent, and a vulnerable component can cool below its permitted range.
Kestrel's cold-case analysis uses a specific eclipse duration, attitude, initial temperature, and equipment state. It does not infer the answer from the average power over a whole orbit. Averaging a hot illuminated interval and a cold eclipse can hide both extrema, which are precisely the conditions the design must control.
Heaters can supply heat where needed, but they draw electrical energy. A fictional six-watt heater operating for half an hour consumes three watt-hours before accounting for distribution losses. If several heaters run together, the power system must support their combined demand while also maintaining essential avionics.
NASA's power-subsystem overview treats energy storage and power management as spacecraft design concerns. Kestrel's original trade study connects those concerns to thermal operation: battery temperature can affect the conditions under which the battery is permitted to charge or discharge, so the thermal and electrical plans cannot be reviewed independently.
The cold-case design may combine insulation, controlled conductive paths, heater placement, and operating constraints. The objective is to protect the components that need warmth without unnecessarily heating every structure. A local requirement often calls for a local thermal solution rather than an indiscriminate increase in spacecraft power consumption.
Notebook page: insulation changes both directions of the problem
Insulation reduces unwanted heat exchange, but it does not create heat or remove it. Wrapping a dissipating electronics box can make it warmer by restricting radiation. The same treatment can help a low-power component retain heat in a cold condition. Its usefulness depends on the heat paths the design intends to preserve.
Kestrel marks radiating surfaces, insulated surfaces, and conductive attachment points separately. A blanket added for one component must not cover a radiator needed by another. Likewise, a structural bracket can bypass an intended isolation path. Thermal ownership crosses mechanical interfaces because energy follows the physical assembly rather than the organizational chart.
Openings, seams, fasteners, and compression can affect real insulation performance. An idealized surface property in a model should not be treated as proof that the manufactured assembly will behave identically. The test plan examines the integrated configuration and documents differences between the model and the article under test.
A thermal strap can deliberately carry heat from a crowded electronics region to a useful rejection surface. That may solve a local gradient without increasing total radiator area. But the strap's interfaces, routing, mechanical loads, and effect on the cold case require review. Improving one path can couple components that were previously thermally independent.
The notebook records why each passive element exists. This prevents a later packaging change from removing a seemingly optional feature whose purpose is not obvious in a mechanical rendering. Clear interface intent is often more durable than a long list of unexplained material choices.
Notebook page: safe mode must be survivable
A fault sends Kestrel into safe mode. The payload shuts down, reducing heat production. The spacecraft adopts a simpler attitude that improves power generation but changes radiator illumination. The thermal state can become either hotter or colder in particular locations despite the reduction in total equipment activity.
This is why “all nonessential systems off” is not automatically a safe thermal prescription. Some components may rely on nearby dissipation or controlled heaters. Others may be exposed to a worse external view under the recovery attitude. The safe-mode definition needs a coupled power, attitude, communication, and thermal analysis.
The team asks what happens if a temperature sensor fails high or low, a heater remains on, or a control command is delayed. It identifies which protections operate autonomously and which depend on ground contact. These are bounded failure cases for design analysis, not a claim that every imaginable fault can be tolerated.
A heater stuck on can threaten the hot case; a heater unavailable can threaten the cold case. Redundancy and control choices should follow the consequence and likelihood analysis appropriate to the mission. Simply adding another heater does not resolve a common control failure if both share the same vulnerable path.
Recovery operations are then checked from the temperatures the spacecraft could actually reach. A plan that assumes an already warm battery or already cool processor may be impossible precisely when needed. The safe-mode test should establish a feasible path back to operation, not merely survival at one static condition.
Notebook page: correlate a model with the hardware
A thermal model is a structured prediction. Testing supplies evidence about how well its conductances, heat capacities, dissipation estimates, and surface assumptions describe the assembled spacecraft. Correlation means adjusting justified model parameters to explain observed behavior, with the changes documented rather than hidden.
Kestrel's test plan specifies sensor locations and expected gradients before the test. It records heater and equipment power as well as temperatures. Without knowing the applied energy, a temperature trace can be difficult to interpret. A sensor offset, unexpected contact resistance, and incorrect dissipation estimate can produce superficially similar discrepancies.
The team exercises more than one steady condition. Transient responses can reveal a wrong heat capacity or an unmodeled path that a single equilibrium point would not expose. Hot and cold cases test different parts of the model. The objective is to constrain the physics relevant to the mission, not simply make one curve match one test.
A ground test has its own environment and support hardware. Cables, fixtures, and chamber conditions can add heat paths that do not exist in flight. The analysis distinguishes the test configuration from the flight configuration and explains how evidence transfers between them. Ignoring the fixture can produce a beautifully correlated but misleading flight prediction.
Uncertainty remains after correlation. The release analysis includes credible ranges and margins for the quantities that matter. It does not replace unknowns with exact values merely because the simulation requires a number. Sensitivity studies identify which assumptions deserve more measurement and which have little effect on the decision.
Notebook page: rank uncertainties before buying more hardware
Kestrel's team has limited time for additional testing. It can measure a mounting interface more carefully, improve the estimate of processor dissipation, or characterize a coating sample. The best next measurement is the one that can materially change the design decision, which requires sensitivity analysis rather than a general preference for more data.
Return to the twelve-watt box. If its interface resistance could plausibly be between one and three kelvin per watt, the predicted gradient ranges from twelve to thirty-six kelvin. That twenty-four-kelvin spread is large in relation to many component temperature margins. A better interface measurement may therefore be more valuable than refining a small uncertainty elsewhere in the model.
Now suppose the dissipation is known only to lie between eleven and thirteen watts while the resistance is fixed at two kelvin per watt. The corresponding gradient ranges from twenty-two to twenty-six kelvin. The four-kelvin spread still matters, but it is smaller than the interface uncertainty in the previous scenario. These ranges are invented to demonstrate prioritization; they are not confidence intervals derived from measurements.
The team also checks combinations. The highest plausible dissipation and resistance together give a different bound from varying one parameter while holding the other at its nominal value. Correlations can matter: a change in operating mode may alter both power consumption and the attitude that determines external heating. A collection of individually harmless uncertainties can align into a difficult case.
Sensitivity analysis does not justify stacking every imaginable worst case into a physically impossible scenario. The cases should be coherent. An attitude, eclipse condition, payload state, and initial temperature must be able to occur together under the mission's rules. Otherwise the design may acquire unnecessary mass and complexity in response to a situation the spacecraft cannot encounter.
Kestrel records the dominant uncertainties and the evidence needed to reduce them. Some are resolved by measurement, some by a design change that makes performance less sensitive, and some by an operating constraint. This gives the project a practical decision tree. A larger radiator is one possible answer, but a better conductive interface or a shorter processing burst may address the actual limiting mechanism more directly.
The exercise also protects the cold case. A change chosen solely to reduce the hot-case temperature is rerun through low-power conditions before acceptance. Thermal engineering is a coupled trade: confidence comes from checking the relevant cases together, not from optimizing one attractive number in isolation.
Notebook page: translate temperatures into an operating envelope
The completed thermal work becomes an operating envelope: which activities can run, for how long, from which initial states, under which attitudes and environmental conditions. Kestrel's payload team needs that envelope more than a single declaration that the spacecraft “passes thermal.”
For a hypothetical processing sequence, the envelope might require a measured starting temperature below a specified mission limit, an approved radiator orientation, and a maximum duration established by analysis and test. The actual limits would come from Kestrel's qualified design; the article intentionally supplies no reusable flight thresholds.
Operations can then schedule imaging, processing, and downlink with thermal recovery in mind. A high-power radio session immediately after a processor burst may be worse than either activity alone. Conversely, some dissipation may help a cold component during another phase. The schedule becomes part of the thermal solution when its constraints are explicit and enforceable.
The design review closes with a traceable chain: component requirement, physical heat path, modeled case, test evidence, uncertainty allowance, and operational rule. That chain explains why the radiator has its chosen area and why the battery has a heater. The spacecraft can be warm and cool in the right places because the team designed the exchanges, rather than relying on an average temperature to describe an entire orbit.