THERMAL OPTIMIZATION OF LED HEAT SINK

It is an age-old problem – probably as old as R&D itself: You must produce the best design, as fast as you can, with the least amount of risk of design failure. No biggie, right? But how will you:

Our best strategy would be hand calculations, manually running simulations on a few designs and air on the side of caution when choosing my final design. Still, you would have no indication of how optimal or failure prone the design will be.

That is, if you don’t have design exploration and data analysis tools at your disposal. To show you what we mean let’s take a look at a heat sink for an LED light (if you recognize it, it is from the SOLIDWORKS Simulations Professional training manual), using CORTIME. For high volume designs like an LED light, small improvements can have significant value. Here, we have decided to reduce the mass of the fins by using the fin length, height, thickness, fillet and number as variables. In this case, the objective is to reduce the mass as much as possible while keeping the temperature below 76°C.

Sensitivity Analysis

We start out by running a quick Design of Experiments (DoE) algorithm. This will give me an indication of where my optimal design might be located, but more importantly, it will give me a sensitivity analysis of my design parameters.

In the sensitivity analysis, every variable is represented by a bar and the colors represent each objective – the mass and the temperature. Thereby, it becomes clear that we can leave out the fillet out as a variable which will reduce the run-time of the design exploration.

Design Optimization

After simplifying the problem, we can explore the solution space using an advanced optimization algorithm that will initially randomize its search for a solution, but gradually hone in on an optimal design. As a result, the mass was optimized by 60% (from 1,560g to 948.1g) with a temperature of 75.99°C.

Even if we want to accept a lower or higher temperature, we can simply locate the data analysis tool and search the pareto front for a more satisfying result. As you can see a clear front is formed which indicates that we have hit a pareto optimal – if the mass is lowered, the temperature will rise. This is good, because then we are assured that an optimal solution has been reached.

Robust Design

Concluding the thermal optimization, we run a robust design analysis to predict how error-prone our design will be once it is manufactured. We do this by applying accepted tolerances to each variable. CORTIME then simulates small variations which naturally occur in production, to evaluate which designs have a low probability of failure. In this case, the histogram shows you the expected changes in mass and temperature once the design is produced in large quantities.

Let data drive your decisions

Follow the design exploration road map

Because they are technically called ‘optimization algorithms’ it is natural to link the use of the algorithms exclusively to design optimization. But as this example shows, you gain insight about so much more by applying them in your design process. Making the same predictions about your design’s parameter simplification, performance optimization and failure probability strictly through manual calculations would either be impossible or way too time-consuming. Or, in other words, use your ingenuity where it matters the most and automate what would be pointless to spend your precious time on. Feel free to explore the process more in depth below.

CORTIME can be used in many other cases and with the majority of the simulations studies included in Solidworks Simulation. So download a free trial to try out CORTIME on one of your own designs. If you want help to get started you can also book a meeting with an Application Engineer.