Conduction, fluid motion, and radiation explain transfer rates. Heat capacity explains storage.
By the end: use a steady conduction model and identify when it omits the dominant heat-transfer mechanism.
Identify the mechanism
Conduction transfers energy through microscopic interactions without requiring bulk movement of material. In convection, moving fluid carries energy; conduction still matters near a surface. Radiation transfers energy through electromagnetic waves and needs no material between the objects. These mechanisms often operate together. A hot object in a vacuum can still cool by radiation.
A watt is one joule per second. Power describes a rate, while heat Q describes an energy transfer. A steady 50 W loss for 60 seconds transfers 3,000 J. A large heat capacity can slow temperature change without reducing a steady heat leak. Thick insulation can reduce the leak without supplying much heat capacity.
Track energy through a stationary layer
Choose a thin slice of a solid as the system. Energy can enter its left face and leave its right face even though no solid crosses either face. Atoms in a solid move around their local positions and interact with neighboring atoms. In metals, mobile electrons also carry energy. Individual exchanges occur in both directions. A temperature difference produces a net transfer toward the colder side. Conduction therefore does not mean that a particular hot atom travels all the way through the layer carrying its original energy.
For a slice doing no work, its internal-energy change equals net heat entering. Suppose 60 J enters from the left during one second and 40 J leaves on the right. The slice gains 20 J. In this course's signs its net Q is positive 20 J, not 100 J. If these rates continue, its state must change. A steady temperature profile requires equal incoming and outgoing powers for each slice, provided there is no internal energy source. Its internal energy then remains constant despite the transfers. Steady transfer is compatible with continual energy crossing the solid's boundaries.
A slice stores the difference between two transfers
Schematic boundaries and arrows, not scaled energy widths. During one second, 60 J enters and 40 J leaves. With no work, the slice gains 20 J. At steady state, its incoming and outgoing powers would instead match.
Now give that slice a whole-sample heat capacity of 10 J/K. The 20 J gain would raise its temperature by 2 K under a constant-capacity approximation with negligible work and no phase change. A different slice with capacity 40 J/K would warm by only 0.5 K for the same gain. These capacities describe the chosen slices, not their conductivity. Temperature tells you about the local thermal state; internal energy also depends on how much material is present and which energy-storage mechanisms it has.
Convection changes the boundary account because moving fluid carries energy with the matter. Imagine a warm stream entering a chosen region and cooler fluid leaving elsewhere. The region can receive energy without having every joule cross its boundary as conduction. At the solid surface, however, the fluid immediately adjacent to it still exchanges energy through microscopic interactions. Fluid circulation continually replaces that nearby fluid. This explains why bulk motion and conduction near a surface can cooperate rather than being mutually exclusive explanations of the same cooling process.
Radiation supplies a different boundary crossing. Electromagnetic energy can leave a surface and reach another surface without a material connection. A surface also absorbs radiation arriving from its surroundings. Equal incoming and outgoing radiative powers produce zero net radiative heating, not an absence of emission. Before choosing a transfer formula, identify the boundary, the possible carriers, and whether the question asks for gross emission or net energy gain. A vacuum gap removes bulk fluid transport across the gap but does not remove the electromagnetic route or solid connections around it.
Different carriers can cross the same outer boundary
Schematic mechanism comparison, without distance or rate scales. Conduction passes energy through a stationary solid. Fluid motion carries energy with matter. Radiation crosses the empty gap; it does not need gas in that gap.
Follow a steady gradient through a layer
Consider a uniform flat layer of thickness L, area A, and thermal conductivity k. Hold its two faces at fixed temperatures. After its internal temperature profile settles, the same power passes through every cross-section. For constant k and one-dimensional conduction, temperature falls linearly through the layer. The heat rate is proportional to area and face-temperature difference, and inversely proportional to thickness.
The model uses area 1 m², conductivity 0.05 W per metre per kelvin, and a 20 K face difference. A 1 cm layer passes 100 W. Doubling thickness halves the gradient and gives 50 W. These are stipulated material properties and surface temperatures, not a measured building wall. The diagrams keep face temperatures and area fixed while changing thickness.
Read the gradient before calculating the power
The temperature profile assigns a temperature to each position through the layer. It is not a drawing of particles moving diagonally. For the model's 2 cm layer, a linear profile between 320 K and 300 K passes through 310 K at the midpoint. The temperature falls by 20 K over 0.02 m, a gradient magnitude of 1,000 K/m. Multiplying by conductivity 0.05 W per metre per kelvin gives 50 W/m². Multiplying again by the stipulated area of 1 m² gives the total power of 50 W.
A temperature profile has position and temperature axes
Both axes are linear. The temperature axis starts at 300 K, not absolute zero. A drop of 20 K across 2 cm is 1,000 K/m. The line describes local temperature, not motion.
Check the dimensions rather than treating the formula as a rule about bare numbers. Conductivity multiplied by area and temperature difference has units of watt metres. Dividing by thickness in metres leaves watts. Putting the number 2 into the denominator for a thickness described as 2 cm would silently use 2 m and predict only 0.5 W. That factor of one hundred is a conversion error, not evidence for a different material. The layer drawing uses centimetres for readable labels, while the numerical calculation requires consistent length units.
Area and thickness describe different geometry. At the same thickness, two identical patches side by side each pass the original power, so doubling area doubles total power. Placing an identical layer behind the first instead doubles the distance over which the same overall temperature drop occurs. In that series arrangement the power halves. Neither change requires changing the conductivity. Conductivity is the material parameter in this approximation; conductance, equal to kA/L, belongs to a particular layer geometry and has units W/K. For the 2 cm layer, conductance is 2.5 W/K.
Thermal resistance is the reciprocal quantity, L divided by kA, in K/W. The 2 cm layer has resistance 0.4 K/W, meaning that a steady 50 W requires a 20 K face difference. Now join a 1 cm layer and a 4 cm layer of the same material and area in series. Their resistances are 0.2 and 0.8 K/W. The combined 1 K/W carries 20 W under the same 20 K total difference. The first layer drops 4 K and the second drops 16 K, putting their interface at 316 K.
Series layers carry equal power with unequal temperature drops
Horizontal distance is to scale: the second layer is four times as thick. Temperature is linear within each layer here because both have the same conductivity and area. The 316 K interface makes the power 20 W through both.
The unequal drops are necessary for equal steady power through the two layers. Assigning each layer half the overall temperature drop would predict 50 W through the thin layer but only 12.5 W through the thick one. Energy would then accumulate near their interface, contradicting a steady state. The interface temperature is an outcome of the resistance account, not a third independently fixed input. This series example assumes perfect thermal contact and no side loss. A separate contact resistance would require another temperature drop in the same account.
Work through the mechanism
Steady one-dimensional conduction
P is power in W; k is conductivity in W/(m K); A is area in m²; L is thickness in m. The two temperatures are fixed layer-face temperatures.
Net radiation to a large enclosure
ε is emissivity between zero and one; σ is about 5.67 × 10⁻⁸ W/(m² K⁴). This idealized gray-surface model assumes full view of a large enclosure at T surroundings.
Increase thickness without changing the faces
Selected rate 50 W; energy transferred in 60 seconds 3,000 J.
Horizontal widths use the same thickness scale. The diagonal is the linear temperature profile, not a particle trajectory. Both faces remain at 320 K and 300 K; area is 1 m² and k is 0.05 W/(m K). Controls replace thickness immediately; radiation, convection, contact resistance, and transient storage are excluded.
Thickness
Steady heat rate
Energy in one minute
0.01 m
100 W
6,000 J
0.02 m
50 W
3,000 J
0.04 m
25 W
1,500 J
Only thickness changes. The face temperatures remain 320 K and 300 K.
Use thickness changes to test a prediction
In the conduction comparison, choose the 1 cm preset and predict the effect of selecting 4 cm before changing it. The reference panel remains 1 cm wide on the shared thickness scale, while the selected layer becomes four times as wide. Both profiles still connect the same 320 K and 300 K faces. The temperature drop per metre therefore falls to one quarter, and so does power. The selected readout should be 25 W beside the 100 W reference. The different slopes express different gradients, not different starting and ending temperatures.
Next choose 2 cm and use the thickness slider to select 1.5 cm. In metres this is 0.015, so the predicted power is 0.05 times 1 times 20 divided by 0.015, or about 66.67 W. This is not the arithmetic midpoint of the 100 W and 50 W powers at 1 and 2 cm. Power depends on reciprocal thickness. A midpoint in thickness therefore need not produce a midpoint in power. Check the displayed rate and the selected layer width against both parts of that prediction.
Check inverse thickness and the separate energy conversion
Thickness
Gradient magnitude
Power
Energy in 60 s
1 cm = 0.010 m
2,000 K/m
100 W
6,000 J
1.5 cm = 0.015 m
1,333.33 K/m
66.67 W
4,000 J
2 cm = 0.020 m
1,000 K/m
50 W
3,000 J
4 cm = 0.040 m
500 K/m
25 W
1,500 J
Stipulated k = 0.05 W/(m K), area 1 m², face difference 20 K. Energy assumes the displayed steady power persists for 60 s. Rounded values are displayed after calculation.
The one-minute energy readout makes a separate multiplication. At 1.5 cm, a constant 66.67 W for 60 seconds transfers about 4,000 J. At 4 cm it transfers 1,500 J. The interval here is an arithmetic conversion from power to energy, not a playback duration or a calculated settling time. The fixed face temperatures imply that other systems supply and receive these transfers. Without them, finite bodies on either side would change temperature, altering the gradient and usually the transfer rate as the exchange proceeded.
To distinguish this from storage, consider a separate finite cold body with whole-sample capacity 200 J/K receiving all the layer's outgoing energy. If its temperature rise were small enough to treat the power as fixed at 50 W for 20 seconds, its 1,000 J gain would imply a 5 K rise. But a 5 K rise is substantial compared with the original 20 K face difference. Treating that approximation as an exact coupled history would be inconsistent. The layer model does not update a receiving body's temperature or integrate a changing rate.
Finally return to 4 cm and ask whether the model proves that adding thickness always cuts every heat loss by four. It tests only this layer's conduction while its area, conductivity, and face difference stay fixed. The controls do not change surface air flow, radiation, or parallel conductive paths. If a separate path still passes 30 W, total power would change from 130 W to 55 W when this layer changes from 100 W to 25 W. That is a reduction by about 58%, rather than a fourfold reduction of the combined total.
A second path changes the total reduction
Schematic parallel connections between the same thermal surroundings. Path widths and arrows do not encode power. The independent 30 W route remains fixed in this constructed comparison; it is not an extra model control.
Find the missing resistance or transport
Several layers in series each need a temperature drop to carry the same steady heat rate. Their thermal resistances add, much like summing the temperature-drop-per-watt requirement of each layer. Air-side convection and contact resistance may dominate a thin solid. Moisture and thermal bridges can invalidate the uniform-layer assumption.
Net thermal radiation depends on both the object and its radiative surroundings. For an idealized gray surface exchanging with a large enclosure, the net rate is proportional to emissivity, area, and the difference of fourth powers of kelvin temperatures. This does not say that all warmth transfer is radiation or that doubling Celsius temperature multiplies radiation by sixteen. Real convective rates also depend on flow and geometry, not temperature difference alone.
Combine transfers without confusing their temperatures
Layer-face temperatures need not equal distant fluid temperatures. Consider a constructed series account with a warm fluid at 330 K, a cold fluid at 290 K, and the 2 cm layer between them. Give each fluid-side region a stipulated resistance of 0.1 K/W and retain the layer's 0.4 K/W. Total resistance is 0.6 K/W, so the 40 K overall difference drives about 66.67 W. Each fluid-side region takes about 6.67 K of the drop. The solid faces are therefore about 323.33 K and 296.67 K, not the fluid temperatures.
Using the full 40 K fluid difference directly in the solid-layer formula would instead give 100 W. That calculation omits the temperature differences required to exchange energy between the solid and the fluids. The model itself avoids this ambiguity by fixing the solid faces directly. The extra resistances in the constructed example summarize exchanges outside the solid; they are not new controls in the comparison. Their numerical values are assumptions for this account, not calibrated predictions of convection or a particular wall's behavior under changing air flow.
Fluid-side drops leave less temperature difference for the solid
The horizontal axis is cumulative thermal resistance, not spatial distance. The solid occupies 0.4 K/W of the 0.6 K/W total. At 66.67 W, its face difference is about 26.67 K; all temperatures are rounded after calculation.
Radiative exchange needs another choice of surrounding temperature. For an ideal gray surface fully viewing a large enclosure, take emissivity 0.5 and area 0.2 m². If the surface is at 400 K and the enclosure at 300 K, the fourth-power difference is 17.5 billion K⁴. With the rounded Stefan-Boltzmann constant 5.67 times ten to the power minus eight W per square metre per kelvin to the fourth power, the predicted net outward power is about 99.23 W. The temperatures belong to the emitting surface and its radiative enclosure, not automatically to nearby air.
There are two opposing radiative powers in that result. The surface emits about 145.15 W, while absorption from the enclosure contributes about 45.93 W under the same gray-surface assumptions. Subtracting the unrounded terms gives 99.225 W outward. If both temperatures become 300 K, each term is about 45.93 W and their difference is zero. Keeping only emission would falsely predict cooling even in thermal equilibrium. Doubling a Celsius reading also fails because the fourth powers require an absolute scale. Temperature differences alone do not determine a radiative exchange.
Net radiation is the difference of two powers
Bars share a zero-based linear power scale, 2.5 drawing units per watt. Emissivity 0.5, area 0.2 m², surface 400 K, enclosure 300 K, σ = 5.67 × 10⁻⁸ W/(m² K⁴). Rounding each term before subtraction can change the last displayed digit.
At a surface that is not storing energy, the net transfers must balance, but a sum of formulas is useful only when each describes the right boundary crossing. Heat leaving a solid can divide into convection to nearby fluid and radiation to an enclosure. A fluid temperature used in the convection account may differ from the enclosure temperature used for radiation. Inside the solid, conduction supplies the combined outward power. If the outgoing sum exceeds the incoming supply, the surface region cools instead of maintaining the assumed steady temperature. Conservation connects these mechanisms without making their separate temperature inputs interchangeable.
Exercises
Change two inputs
For the same conductivity and face-temperature difference, a layer’s thickness doubles and its area triples. How does its steady heat rate change?
Hint
Area multiplies the numerator; thickness multiplies the denominator.
Worked answer
Change
Rate multiplier
Area ×3
3
Thickness ×2
1/2
Combined
3/2 = 1.5
Hold the remaining inputs fixed before comparing rates.
The rate becomes 3/2 of its original value. If the original rate was 100 W, the new rate is 150 W. More insulating thickness does not guarantee a smaller total heat leak when the exposed area also changes.
Diagnose a vacuum claim
An advertisement says that removing air around a hot container eliminates every route for heat loss. What does this omit?
Hint
Which transfer mechanism does not need matter in the gap?
Worked answer
Route
Effect of an evacuated gap
Bulk air circulation
Suppressed in the gap
Thermal radiation
Still possible
Solid supports
Can still conduct
A missing transport path is not the same as complete thermal isolation.
Radiation can cross the vacuum. Solid supports and connecting pipes may also conduct heat around the gap. Evacuating a gap suppresses gas conduction and convection there; it does not by itself eliminate radiation or all solid conduction.
Sources and further reading
These optional references support the stated laws and model assumptions. The numerical examples and diagrams are authored teaching models, not experimental measurements or engineering specifications. Selected relevant passages were reviewed, not entire textbooks or a comprehensive research literature.