Move boundary temperatures and watch how the plate temperature gradient and heat-flow direction change.
Parameters
Left T
°C
Fixed temperature at the left boundary.
Right T
°C
Fixed temperature at the right boundary.
Top-edge temperature
°C
Fixed temperature at the top boundary.
Bottom-edge temperature
°C
Fixed temperature at the bottom boundary.
Thermal conductivity k
W/mK
In the steady solution for a homogeneous plate the temperature field does not depend on k; k is used only for the heat-flux index (W/m²).
Results
—
Center Temperature [°C]
—
Interior Maximum [°C]
—
Interior Minimum [°C]
0
Iterations
—
Convergence
Speed
Low Temperature (Blue)High Temperature (Red)IsothermHeat Flow (Hot → Cold)Grid 48×48
The 2D heat equation $\partial T/\partial t=\alpha\nabla^2T$ is solved by relaxation (Gauss–Seidel). At steady state, $\nabla^2T=0$, and each interior point is $T_{ij}=\tfrac14(T_{i+1,j}+T_{i-1,j}+T_{i,j+1}+T_{i,j-1})$. By the maximum principle, no maxima or minima occur in the interior.
Temperature field
—
Center temperature
—
Max temperature span
—
Heat-flux index
—
Boundary bias
Centerline temperature
Heat-flux direction
Model and equations
Model and equations
$$\nabla\cdot(k\nabla T)=0$$
Here, steady conduction in a rectangular plate is treated as an approximate solution of Laplace's equation. Internal heat generation or thermal contact resistance requires different conditions.
How to read it
The hero heatmap runs from cold (blue) to hot (red). The white isotherms join points of equal temperature; where they are closely spaced the temperature gradient is steep.
The yellow arrows are the heat flux q = −k∇T: they always point from the hot side to the cold side and cross the isotherms at right angles. Dense arrows mark regions of strong local gradient.
The 'Centerline temperature' tab below shows the temperature along a horizontal cut through the middle of the plate, and the 'Heat-flux direction' tab shows the local gradient strength. The center temperature in the results tab is the mean of the four edges, which is the exact center value for a square plate in steady conduction.
Learn 2D conduction temperature by dialogue
🙋
Raising the thermal conductivity k from 15 to 400 W/mK changes neither the heatmap nor the center temperature. Is the tool broken?
🎓
No. In a homogeneous plate at steady state the k in ∇·(k∇T)=0 cancels out, leaving ∇²T=0, so the temperature field depends only on the boundary temperatures. k only sets how much heat flows: the heat-flux index in the results tab, k × √((T_left − T_right)² + (T_top − T_bottom)²), is the only value that changes. With the defaults that is 15 × √(80² + 20²) ≈ 1,236.9 W/m².
🙋
Why is the center temperature 55 °C with the defaults (left 100, right 20, top 60, bottom 40 °C)? The animation says 54.1 °C.
🎓
For a square plate with four edges at fixed temperatures, symmetry and superposition make the center equal to the edge average, (100 + 20 + 60 + 40)/4 = 55 °C. The animation solves on a 48 × 48 grid with Gauss–Seidel relaxation, so grid coarseness and the corner treatment bring it about 1 K lower. The preset with three edges at 0 °C and one at 100 °C gives a center near 25 °C for the same reason.
🙋
What do 'Converged ✓' and the iteration count mean?
🎓
Each iteration updates every interior node to the average of its four neighbours. When the largest change in one sweep falls below 0.001 K the readout shows 'Converged ✓'; with the defaults that takes about 1,000 sweeps. Changing an edge temperature restarts the relaxation, and the speed slider sets how many sweeps are done per animation frame.
🙋
Where do I enter the plate thickness or the cooling conditions?
🎓
Thickness does not affect a two-dimensional steady solution, so there is no input for it. Convective cooling (a heat-transfer coefficient h), internal heat generation, contact resistance and time dependence are outside this tool. Use the related 2D transient conduction simulator for time dependence and the 2D conduction FDM simulator for internal heat sources.
Practical use
Estimating the highest and lowest internal temperatures of an enclosure wall or heat spreader when the four edge temperatures are known. By the maximum principle the interior always stays within the range of the edges, and the center tends to their mean.
Checking boundary-condition sensitivity before a detailed FEM run. The solution is linear in the edge temperatures: moving one edge by 10 K moves the center by 2.5 K, which tells you which edge dominates.
The heat-flux index k × √(ΔT_lr² + ΔT_tb²) treats the plate side as 1 m. For a real plate of side L [m], divide the index by L to get a representative heat flux.
FAQ
In steady conduction through a homogeneous plate the temperature field is fixed by the edge temperatures alone; k drops out of the equation. k is used only for the heat-flux index (W/m²). Compare that index when you want to see the effect of the material.
The results tab reports the edge average (the exact center value); the animation is a 48 × 48 grid relaxation. Discretisation and the corner handling make them differ by about 1 K. Both stay within the edge range and move together when you change a boundary.
k × √((T_left − T_right)² + (T_top − T_bottom)²) in W/m², i.e. a representative heat flux for a plate 1 m on a side. For a plate of side L divide by L. The direction is shown by the yellow flux arrows.
Only a homogeneous rectangular plate, steady state, no internal heat generation, with all four edges at fixed temperatures. Convective boundaries (heat-transfer coefficients), contact resistance, temperature-dependent k and transients are not included. Confirm final decisions with detailed analysis or measurement.
How to Use
Enter the left, right, top and bottom edge temperatures (−50 to 300 °C). The defaults are 100 / 20 / 60 / 40 °C.
Set the thermal conductivity k (0.05–400 W/mK; e.g. aluminium 237, steel 50, concrete 1.4). It only affects the heat-flux index, not the temperature field.
There is no Simulate button: changing an edge temperature restarts the relaxation in the heatmap automatically, and 'Converged ✓' marks the steady field with isotherms and heat-flux vectors (plate thickness and material do not change the distribution). The presets (one hot side / opposite hot–cold / corner heating / three edges 0 °C, one 100 °C) give typical patterns.
Read the results tab: center temperature (L+R+T+B)/4, maximum span max − min, heat-flux index k × √(ΔT_lr² + ΔT_tb²) and boundary bias (L+T−R−B)/2, then use the 'Centerline temperature' and 'Heat-flux direction' tabs for the gradient direction.
Worked Example
Steel (k = 50 W/mK) with left = 100 °C, right = 20 °C, top = 80 °C, bottom = 40 °C: the results tab shows center temperature (100 + 20 + 80 + 40)/4 = 60.0 °C, maximum span 100 − 20 = 80.0 K, heat-flux index 50 × √(80² + 40²) = 50 × 89.44 ≈ 4,472.1 W/m² (an index for a plate 1 m on a side; for a 0.5 m plate divide by 0.5, about 8.9 kW/m²) and boundary bias (100 + 80 − 20 − 40)/2 = 60.0 K. The flux arrows point mainly from the hot left edge toward the cold right edge, tilted slightly downward because the top is hotter than the bottom. Switching to aluminium (k = 237 W/mK) leaves the temperature field and the 60 °C center unchanged and only raises the index to about 21,197.9 W/m².
Practical Notes
The boundary bias (L+T−R−B)/2 is in kelvin: a large positive value means the heat flows mainly from the left/top toward the right/bottom. Use the flux arrows to decide where cooling should be placed.
The center temperature in the results tab is always the arithmetic mean of the four edges, which is the exact steady value for a square plate; asymmetric edges change the shape of the isotherms, not this rule.
Changing k does not change the steady temperature field of a homogeneous plate; materials differ only in the heat-flux index (aluminium passes about 4.7 times the heat of steel for the same edge temperatures). Thermal diffusivity matters only for transients, which this tool does not model.
The heat-flux index scales with k and with the edge temperature differences and assumes a 1 m plate; divide by the real side length before using it for a first heat-sink estimate.