Solutions for Thermal Simulation & Characterization in Automotive Lighting
"…from what we hear from our customers, where they would usually work for five to seven days on really complex full LED headlights, they are usually running the same simulation or same process within two to three days with our tool."
Let’s get started. What is your background and how did you come to work in your current position with Mentor?
Boris Marovic: Well I studied aerospace engineering in Stuttgart and majored in aircraft design and aerodynamics and gas dynamics. After that, I joined the company as an application engineer and worked in several projects in the automotive industry; however, I became more and more involved in traction and the automotive lighting industry, and after some time I was appointed industry manager for automotive and transportation. So that’s my current status, I have been with the company almost seven years.
In terms of LED technology we’ve seen a lot of presentations that go into brand DNA and how LED technology can really help, so everybody wants it, but what are the biggest challenges facing the newcomers, the ones that haven’t really established themselves yet in this part of the market?
I think the biggest challenge is the thermal design change. In the case of lighting systems with bulbs as a light source, the bulbs themselves were not really thermal critical, but with LEDs, it is more the LED itself than the components around it. In the case of the bulbs, you had plastic components that shouldn’t exceed certain melting temperatures. But here the LED is the most important part and of course the heat flow changes, as you don’t have radiation that much anymore the focus is more about conduction and you have to have convection to get rid of the heat from the heat sink. So the thermal management itself changed from radiation to conduction and convection. The material selection and airflow management are becoming more critical.
And is that maybe something that newer players in this part of the industry might not be familiar with because they’ve never dealt with this before?
As long as they’re familiar with the thermal management technologies or properties like radiation and conduction and so on, I think it is not a big problem to understand it, but with LEDs, the design changes, the headlights and tail lights have become more complicated due to the new designs and have more features. So the biggest challenge there is to still be able to manage the cooling flow in such complex environments, you might get the recirculation of hot air and that is not really cooling the LED anymore. So you have to understand the airflow in more and more detail than you had to do before with the old lighting sources.
That’s usually where simulation comes more and more into play. Before LEDs, it was a light bulb and you had a reflector. Basically the biggest part of it now is to understand the airflow in more complex environments.
Regarding simulation, I know you mentioned thermal characterisation to me before. How is this used during the design process?
The OEMs usually don’t do thermal characterization of LEDs that much because their interest is of course in the newest design technology and so on, but they leave the technology, i.e. which LEDs have the right performance, more to the tier one supplier and the tier one suppliers often do these kind of thermal characterisations. So usually what we see is that they of course want to have the best LEDs with the best performance for more reliability but they have different LEDs from different vendors. So it should of course have a high maximum temperature, but still also be very robust in the optical performance; not to decrease in luminous flux or change too much their colour in the higher temperature range. So the behaviour should be very stable and be reliable for a long time, because we're usually talking about a 20,000-hour lifetime for cars and even more for trucks.
Are there any tools that are traditionally used for thermal characterisation? What do you see normally and where do you see a gap/potential for something new or something better?
In the case of thermal characterisation, there aren’t really tools out there. We encounter a lot of our tier one suppliers that have only just started using the technology; they don’t know anything about it. They usually use the optical characterisations - the integrating spheres - in order to measure the luminous flux, colour and so on, but they don’t consider the thermal part of it, the influence, and especially not the thermal part within the LED and the structure, how that changes during aging and so on, in order to see which LED from which vendor is better in terms of structure or to understand it in more detail.
Could you discuss the simulation technology that your team developed which is revolutionary in a way and in use today?
Well it has some key technologies where we see that it fits our ideas perfectly in this kind of design phase. I mean as I said, the headlights are very design related or very design-heavy. It should look nice and therefore involves complex geometries and you change them a lot during the design iterations.
You do a lot of work with your CAD system, so the traditional CAD tool usually exports it and then imports into the CFD tools. They then have to mesh it and the meshing is often done manually. With our tool you are working within your CAD system. So you work on the latest CAD design and cleaning of badly imported CAD models is not going to happen as you work on the native CAD geometry itself. So the only cleaning up you have to do is if it has to be watertight. We also have features that help you to make it watertight and after that the meshing is done automatically. So you click a button and the mesher runs automatically. If you want to improve the mesh in certain areas, such as a finer mesh on the LED or in some gap areas, you can easily apply some local meshes or adjust some mesh levels, make refinements and then click run again and it meshes automatically.
Where you previously had two to five days of manual meshing time, it is now done automatically within hours. So the user is not really working on it manually anymore and can do other things during that time, which saves a lot of time and especially working hours that can be spent on other things. In addition our mesher doesn’t need to have any simplifications of the geometry such as small ribs for stray light or for the facetted reflector geometries, it meshes simply over it without any issues.
So all in all, the process and time it takes to get from CAD model to CFD result is extremely reduced by very easy geometry handling and being always up to date with the design changes and the automatic meshing with very good result accuracy that can be compared with measurements.
Do you have an idea of how much time a team might actually save during the design process?
Well from what we hear from our customers, where they would usually work for five to seven days on really complex full LED headlights, they are usually running the same simulation or same process within two to three days with our tool. So it saves more than half of the time by saving meshing time, project set-up time and being faster with any geometry changes because you’re working directly on the geometry and can just switch it and update the project.
In an industry where time truly is money, that’s quite significant.
What special features or capabilities does your software have that makes it particularly suitable for the lighting industry?
We worked together a lot recently with our automotive lighting customers and developed certain capabilities, such as the Monte Carlo radiation model that suits that kind of lighting application very well and our product can consider the influence on humidity and its condensation in the typical lighting applications. We also have a special LED feature which has optical properties, as well as thermal properties. So when you simulate, the LED reaches its junction temperature in the simulation and the results not only show that temperature but also the actual hot lumen of the LED at that temperature. The source of that information comes either from our measurement equipment or if it is available from data sheets or other measurements and, you can input this data into our database. You get a general idea of how bright each LED is, because different positions in the headlight mean different junction temperatures as the airflow or the temperature is not the same everywhere.
And this represents a saving in terms of time and labour later on because you reduce the number of physical experiments?
Yes, real experiments. It’s not just the simulation of the headlight itself, there’s more to it. You have for example solar radiation if the sun hits the headlight. There are still some lenses in there for the focusing of the beam and you get hot spots up to 900°C or more and if the focus point of that lies on a plastic component it just melts through. So these are also usually done with experiments. You have a big machine that produces the amount of light that you get from the sun. You can then focus and measure it. You can avoid the costly experiments in terms of time, where you often have to manufacture prototypes. You can create these variations very easily by simply using several projects with different settings. We have parametric studies where you can change the position of the sun to find out at which position you have the focal point, what temperature it reaches and so on in order to really understand where the hottest temperature is, what parameters and what materials you have to choose. Our customers really use it from the beginning of the design stage where they get the specs from the OEM. They have to make a very first draft of the headlights, so that they have some basic components. They then simulate it in order to find out roughly where the temperature lies in order to define which materials they have to use, because if I exceed my melting temperature of the one material I need a more expensive material with a higher melting temperature and the headlight gets more expensive. Customers use it from there to the end of the design cycle where they completely model the full LED headlights with PCBs, fans, most complex geometries without any simplification of the geometry.
So you can add in some more active cooling elements for example?
Oh yes, you can either model the fans in detail or you can apply fan curves, so basically simplifications, in order to save CPU time by not calculating the rotation itself. So there are several technologies that make solving faster, but you can also model in every detail, which just takes more CPU time.
I hadn’t thought of the external factors such as sunlight that actually can really have such an effect on the component and the material surrounding it.
And since the headlight is a design piece you also get things such as condensation on the headlight, which usually the customer sees as a defect because there’s water inside. It’s not really a closed environment so the water will get in at some point through humidity, for example, when you drive during the autumn, when it’s rainy and the air is humid. You drive into your garage and overnight it cools down and of course you get condensation on your headlights. When you drive out of your garage you see you have condensation on your headlights and that should evaporate very quickly and this is also something that they are looking into, evaporation and condensation inside the headlamp looks like bad design in the eyes of the customer so it shouldn’t stay that long. Therefore companies are really looking into how long it takes until it’s resolved so that they avoid these sorts of complaints where the customer comes back to the dealer claiming that the headlight is damaged even though it isn't.
There’s often a very fine line when a customer identifies one small issue as a huge defect due to a difference in perception. This occurs in NVH and other areas as well.
Yes, it costs a lot but in general it isn’t really a problem, it just simply cannot be avoided. There is humidity everywhere so at some point it is bound to get into the areas where you don’t want it.
We have a good overview on the specific uses for these simulation tools that and some of the driving factors pushing their development, but a bit more generally, what changes do you see occurring within the industry in lighting moving forward five years or so? Do you think that any of these changes will be defining for the industry in terms of a technology shift and if so maybe what kind of preparations do you think could be made to prepare?
I think we currently see a trend towards OLEDs. Of course OLEDs more as a surface light source, so anything that doesn’t need to be very bright, such as shining very far in the headlights, more tail lights that you only have to see, but that don’t have to actually illuminate the road. We have laser lighting which is being marketed by BMW. I think these are the biggest trends: laser lighting and OLEDs. I think they are coming to market very soon, some of them are already in the market in some areas of lighting in automotive and I think they are getting stronger and stronger depending partially on what the OEM really wants to do. Some might consider laser lighting interesting, others may feel that there isn’t really a market for it in the near future. But if you look at BMW, they really want to push it. I think that in the way Audi made a big difference with their LED headlights, BMW is hoping to make a big difference with their laser lights.
It comes back to brand DNA?
Exactly and that might then also be like with the LEDs. Now everyone wants to go LED and when BMW brings out the laser lights, probably everyone will want to make laser lights. And we already see a lot of interest from our customers in a more general way than actual applications. Probably some are still hesitating.
But you see LED as a technology that has not finished its run and will be around for a while even if it is no longer on the cutting edge?
Thank you for the interview
Boris Marovic is an Industry Manager with Mentor Graphics