An Expert Speaks Out on CFD Modeling of Heat Sinks

Chris Aldham of Future Facilities has something to say about CFD modeling of heat sinks. And he should know after 30 years in the business. Chris will present a webinar for ATS on May 24, 2012 “CFD as a Tool to Perform Heat Sink and System Modeling,” that you can attend for free by registering on Qats.com.

https://www2.gotomeeting.com/register/467986842

We asked Chris to share upfront some general knowledge and opinions on the topic ….

What are some of the recent advances in CFD technology and how might they improve heat sink modeling?

The main advance I’ve seen is the increase in computer power and lowering of computer cost that has occurred over the past few years. It is now possible to solve larger (more grid cells) and more detailed (more objects and better geometrical representation) models and more of them very efficiently. So now representing the detailed geometry heat sinks in a CFD model is easy. Importing MCAD heat sink geometry and using that geometry directly in the software ensures an accurate representation of the heat sink.

The other advance is the automation possible in specialized tools such as 6SigmaET. The mesh necessary to represent the heat sink is determined automatically within the software it doesn’t rely on the user creating a good mesh.

These two trends seem set to continue so it will be possible to model increasingly complicated heat sink designs.

Meshing is very core to CFD modeling. What are the do’s and dont’s when it comes to meshing heat sink models?

I think there are two aspects to consider when meshing a heat sink. The solid geometry must be accurately captured to ensure the heat spreading and conduction through the base and up the fins is accurately represented. Then the airflow between the fins must be accurately captured. This invariably requires a fine mesh at least 3 cells between the fins and maybe more depending on the gap size.

What are some of the benefits from developing a high quality CFD model of a heat sink?

At first sight heat sinks seem quite simple in function but their interaction with the components they are cooling and the air flow around them is quite complex. The heat spreading of the heat sink base can subtly change the thermal resistance of the component. The increase in surface area the heat sink provides improves heat transfer but also represents an increased resistance (increased pressure drop) to the airflow. So a good heat sink design must balance heat spreading, heat transfer and pressure drop. As a detailed CFD model can represent all these aspects accurately in the situation in which it will be used it can be the only way to optimize them before the heat sink is manufactured and tested.

Can you cite any examples where your CFD tools led to improved heat sinks solutions?

We have published a couple of examples together with ATS Europe who have used 6SigmaET in a number of projects. One was an unusual heat sink design on an LED replacement for a traditional light bulb where a 14% improvement in lamp performance was produced (as well as a much nicer looking design in my opinion) by changing the heat sink design. This work also showed good agreement between 6SigmaET simulations and measurements performed on the real devices. See images below.

How long does it take a typical engineer to master CFD modeling? Are there any innovations in training?

I’ve been doing CFD for over 30years and I’m not sure I’ve mastered it yet. Fortunately engineers do not have master CFD modeling today as some CFD software products are focused on specific applications and these can really present CFD in a very usable form. Of course it helps if the engineers have some idea of the physics of fluid flow and heat transfer but much of the numerical work in CFD can be preset, automated and hidden away. This has been especially true in the field of electronics cooling where specialized software has been around for decades. These tools can be learned in a few days and users can be proficient in a few weeks.

How is Future Facilities different from its competitors?

Future Facilities is highly focused on a small number of related application areas. We produce software for design, operation and management of data centers which includes CFD modeling of the airflow and temperatures as well as other non-CFD analysis modules. We also use the software in our engineering consultancy group providing services that ensure the software development is focused on exactly what is needed and making it easy and efficient.

6SigmaET is a recent product focused on electronics cooling and integrated into our data center suite. Like the whole software suite it presents the user with a set of specialized intelligent objects which represent the real things encountered in electronics (pcbs, fans, heat sinks, power supply, components, etc.). As every object knows what it is, it knows how to behave and this can make creating a model very intuitive for the users. It also allows us to automate the meshing rules for each object so we can ensure a heat sink, for example, is meshed correctly.

I believe the many years of experience we have in using and developing CFD products alongside a strong focus on particular application areas and a desire to make complex technology available to engineers (expert and beginner, full-time or occasional users) makes us very different from other CFD companies.

Dr. Chris Aldham has worked in computational fluid dynamics (CFD) for over 30 years (starting with PHOENICS at CHAM with Prof. Brian Spalding) and for more than 20 years in the field of electronics cooling. After 16 years at Flomerics, Chris joined Future Facilities as a Product Manager responsible for 6SigmaET electronics cooling simulation software which is part of a suite of integrated software products that tackle head-on the challenges of data center lifecycle engineering (including equipment design analysis) through the Virtual Facility

Leave a Reply

Your email address will not be published. Required fields are marked *

This site uses Akismet to reduce spam. Learn how your comment data is processed.