In my last blog post Studying the Next Generation of LED Light Bulbs, I talked about how these light bulbs look like regular light bulbs and how they don’t have the ugly heat sinks. Now it’s time for the fun stuff. Since this bulb uses convection to cool the circuit boards, I want to see how that works and how effective this process is depending on how I mount the bulb.


Some of my fixtures are side-mounted, some have the bulb pointing up, and some have the bulb pointing down.

Using SOLIDWORKS Flow Simulation, I was able to virtually simulate each of these scenarios and see which mounting scenario was most appropriate for this type of cooling.

The set up was pretty straight forward. Set up Materials (Plexiglass for shell, Aluminum for the base, and PCB for the boards), and Heat/Radiation sources on the LED blocks themselves. Then Clone the study and just change the direction of gravity to “rotate” the bulb for each study. I also added Global Goals to make sure the solver focused on the maximum temperature of the solid and fluid (air).

More Googling led me to a Wikipedia page that shows the heat and radiation output of standard LEDs, called Luminous Efficacy.


The bulb I am studying is a 60W equivalent which is an 11W LED. Here are the numbers I used:

60W equivalent :: 11W

Heat (Conv+Cond) 80% = 11W * 0.8 :: 8.8W

Radiation/Visible 20% = 11W * 0.2 :: 2.2W

I used a Volume Heat Source on each of the 8 LED “cubes” of 8.8W (total), and a Diffuse Radiation Source on the same bodies of 2.2W (total).


At this point, I think it is important to discuss some assumptions I have made along the way. If you recall from my last post, I used sketch pictures and a lot of Google to come up with my design. I also do not work for Cree, or any other bulb manufacturer for that matter. All my dimensions are guesses, including the LED size and circuit board thickness and shape. I also conveniently left out the connection between the board and the aluminum base. Any results I get are strictly for comparison purposes. I am running these studies to check out the convection flow and how the bulb orientation affects cooling.

I ran one study (y up) with a variety of meshes to make sure my values were acceptable. Once I was happy with the mesh and input variables, I cloned “y up” twice to “x up” and “y down.” The only difference between studies now is the direction of gravity (as mentioned above).

Using the Batch Run command in Flow, I was able to run all the simulations at once (overnight).


Needless to say, I came in early the next morning to see the results (I told you I was a geek).

Here is a Cut Plot of the “y up” study showing temperatures. Remember, I can’t guarantee temperature accuracies due to my limited knowledge of the actual components. These studies are for comparison purposes only.


In studying the Flow Trajectories, we see that there is plenty of flow across the boards and LEDs. It even creates some swirling, which will help cool the circuits.


As I expected, the horizontally mounted bulb gets some heat trapped inside.


While pretty pictures and cool animations are pleasing to the eye, in the real world we need numbers.

Using the Compare Results tool in Flow Simulation, I was able to compare all three of my studies side by side. I can display any plots I have in all three studies, including Goal Plots.

bulbs bulbs

We see that the “x up” study (where the bulb is horizontal) is significantly warmer both in air temperature and solid temperature. All the temperatures seem a bit high. This might be because I made the LED cubes a bit too big.

In conclusion, it looks like the bulb will stay coolest pointing straight up. But pointing straight down isn’t bad either. I’m sure the engineers at Cree ran longer and more detailed studies than this before the bulb was released. And, with a three-year warranty, on an $8 bulb, you bet I ran out and bought a couple!! Since my post light is on all night and it points the bulb straight up, that is where I installed my first one.

Thanks for reading my series on this new generation of LED lights. As always, Happy Modeling (and Simulating)! 🙂

Tags: , ,