MODELING TECHNIQUES
Steady State Initialization Effects
Material and thermal properties affect how temperature gradients react to ambient changes. A very massive part will react very slowly to changing ambient temperatures. For this reason, the start of an analysis should begin long before the time where results should be trusted.

A thermal analysis is always initialized to the steady state condition. The following plot shows the effect that start time has on the solution. The two analyses depicted in this plot differ only by start time. The analysis for the black line was initialized at 10:00am while the blue line was at 5:00am. Because of different amounts of solar energy present at those times, the steady state initialization produced very different initial temperatures.

Only after two days did the plots converge. This example shows how the ambient boundary conditions and start time will impact any model.

There are two major factors to consider when obtaining transient results. The first factor is choosing when to do the steady state initialization. Different times of day will produce steady state results that are more accurate than others. Knowing when to start a model depends on the type of model, weather conditions, and the accuracy required from the model.

The second factor is how far in advance to do the steady state initialization. It may take anywhere from 6 to 48 hours past the steady state initialization to obtain results that are accurate. This depends on the material properties, weather conditions, and accuracy required from the model.

Assigning Temperatures vs. Applying Imposed Heat
The following four pictures show the problems that can result when using the wrong type of heat source.

Presently, there are two choices available for applying heat to an object. An assigned temperature can be used. This keeps the temperature of the part constant, regardless of the surrounding environment. Imposed Heat can also be used. This applies a constant or fluctuating heat source to the part.

These pictures show the results of four scenarios for a simple revolving wheel model. The wheel bearings in Cases 1 and 2 were assigned a constant temperature, and placed in both a cold and warm environment. In Cases 3 and 4, the wheel bearings had imposed heat, and were also placed in cold and warm environments. The wheel was placed on asphalt in July for the warm environment, and placed on snow in November for the cold environment.

We would usually expect the temperature of the wheel bearings to be warmer than the rest of the wheel, independent of weather conditions. But this was not the result for Case 1; the temperature of the bearings were in fact cooler than the tire. This is because a temperature of 100^oF was assigned to the wheel bearings. In July, the calculated temperature of the tire itself exceeded 100^oF.

Cases 3 and 4 produced much better results. The temperature of the bearings was warmer than the rest of the wheel, and the bearing temperature was greater in July than in November. In this example, an imposed heat source was the correct modeling device, while an assigned temperature resulted in modeling errors. From this we see that the type of heat source used greatly affects the accuracy of the analysis.

Connecting vs. Disjoining Mesh Edges
The following geometry is an enclosure containing a cylindrical heat source. The heat source has an assigned temperature and radiates energy to the enclosure.

Welding mesh edges will be a determining factor in the overall accuracy of the model. Welding mesh edges will cause conduction to occur between two parts. Sometimes there is no conduction due to gaskets or insulation. In these cases, the model must be produced in a way to depict this.

The picture on the left is the case where all part edges are welded together. Notice that there is heat transfer from the access panel to the side surface. In a "real world" case, there would be no conduction because of gaskets.

The picture on the right has corrected the seam around the access panel. There is no conduction around the panel; this accurately depicts the "real world" case.

The following two pictures show the access panel without the mesh lines. This shows more precisely the difference between the two scenarios. Notice in the image on the left, it is harder to depict the border of the access panel. It is very easy to see this border in the image on the right.  In "real life," we would expect to see a temperature difference between the access panel and the side surface. The image on the right shows more accurately what we would expect in real life.

The pattern for temperature distribution is important for many heat transfer problems. Knowing where the maximum temperature is and how it propagates to surrounding surfaces is often a requirement for accurate analysis.

Geometry Heat Source vs. Non Geometry Heat Source
This example demonstrates the importance of geometry with regards to heat sources and their effects on their  surroundings. In both cases, a heat source is radiating energy to the enclosure. The heat source is emitting 800W of energy. The first case uses geometry that has 800W of energy applied to it. The second case uses no geometry, and instead applies a fractional imposed heat to the back surfaces of the enclosure.

The average temperature of all surfaces, in both cases, is the same. In both cases, all of the energy from the heat source goes to heating the other surfaces.

The reason for using geometry for heat sources is that the distribution of heat can be found. The first case was easier to analyze and produced results that are more accurate. We are also able to determine the maximum temperature of the system.