FE Update: CFD Software Turns Down the Heat in LED Lighting

Light-emitting diodes (LEDs) have come a long way since calculators first displayed answers with glowing red dots. Manufacturers, users, and even governments tout the benefits of the technology’s low-energy consumption compared to incandescent bulbs. Surprisingly, LEDs also have a longer lifetime than compact fluorescent lamps (CFLs). What’s not commonly understood is that an LED’s lifetime is closely related to its operating temperature range.

Where’s the heat?
Over the years, designers have developed rules and standards to ensure user safety and estimate bulb lifetime for a wide variety of lighting applications and settings. But with LEDs, the rules have changed.

Incandescent bulbs drive heat outward through the glass enclosure, but LEDs send heat down through the plastic base and into the housing. The hotter the base, the higher the diode junction-temperature and the shorter the operational lifetime. In addition, the LED may get dimmer, and operational color wavelength can shift. Add in the extra heat load from neighboring LEDs or high ambient temperatures, and thermal management becomes a critical design element for meeting specs.

LEDs’ small size along with the push for higher brightnesses (wattages) make it imperative to understand the factors that influence cooling. Unfortunately, the classic build-test-rework approach provides minimal information at a handful of physical locations under specific conditions.

In contrast, thermal simulation speeds design and yields more data points. And higher quality and more innovative products come when CAD designers themselves can perform simulations and easily identify trends across multiple variations of a design.

Consider, for example, how “upfront” computational fluid-dynamics (CFD) software CFdesign from Blue Ridge Numerics Inc., Charlottesville, Va., helped designers at several companies do just that. The developer calls the software “upfront” because it helps drive the conceptual stages of design.

One company, Cooper Lighting and Safety in the U. K., previously made architectural lights using halogen, fluorescent, and metal halide bulbs. However, the demand for more energy-efficient lighting encouraged the company to use LEDs.

Company designers incorporated the diodes in display lights and wall-mounted and ground-recessed fixtures. The LEDs’ brightness and compact size were advantages, but engineers soon realized that some designs had problems in operation, efficiency, and lifetime. Engineers knew the difficulty stemmed from heat build-up but adding larger heat sinks was not feasible. Track-lighting must display an aesthetically minimalist profile and canister lights must fit in both new and existing, cramped ceiling spaces.

Drawing on their experience with traditional finned heat sinks and passive liquid heat pipes, designers generated a hundred possible cooling designs. Each varied by number, size, and spacing of the aluminum fins, so a classic build-test-redesign approach could not be justified.

A good alternative was computer simulation, but Cooper, like most lighting companies, had much to learn, says Rod Dixon, Cooper product manager. In fact, he first ran a heat simulation on a full 3D CAD model, spending too much time on noncritical geometric details. Fortunately, Blue Ridge Numerics’ engineering support showed him the far-more useful approach of focusing on the relative performance between incremental design changes.

“We now use CFdesign as a development tool,” says Dixon. “For example, our Axent RXD downlight has 16 fins on the cooling system, but simulations only use four fins, then scale up. Models don’t even include the heat pipes. We just change the fin spacing and predict the heating. Comparing results points to the optimal design, which gets built as a sample and is validated in a thermal chamber.”

Currently, it takes Cooper designers only 45 min to create critical aspects of a new model, simulate it, and decide if it’s worthy of build-and-test. “Even an additional 5% cooling makes a big difference in the lifetime of an LED,” says Dixon.

Keeping score
In another example, Daktronics, Minneapolis, uses LEDs in the manufacture of electronic scoreboards, digital billboards, and large-screen video displays. The devices must be viewable in direct sunlight, withstand all types of weather, and operate to spec in 120° temperatures.

Almost all Daktronic products have a built-in cooling mechanism, either active, such as fans for outdoor displays, or passive, such as natural ventilation for indoor displays. On older bulb-based units, optimization relied on theoretical calculations, vendor-recommended guidelines, and field experience. However, these methods don’t work well for LED-based displays because of design complexity, vague vendor guidelines, and difficulty with troubleshooting.

“Most of our testing is outdoors during summer because solar gain is a big contributor of heat,” says Daktronics design engineer Sunil Gaddam. “But when the weather becomes cloudy, tests can take from a few days to a few weeks.” In these cases, the company often performs tests indoors and extrapolates results.

It was a costly and cumbersome trial-and-error method to build prototypes.” So five years ago, without any CFD experience, Daktronics staff began using CFdesign software to understand the thermal behavior of LEDs operating in dense clusters.

CFdesign proved valuable because it can handle parasitic heating, where dissipated power adds to the heat, as well as the effects of smaller conduction paths. For simple configurations, designers run basic airflow and ventilation analyses. For more-complicated layouts, they add natural convection. Designers compare different combinations of fans, heat sinks, enclosures, and materials, and then create a decision matrix comparing performance to cost and manufacturability. With these tasks complete, the best design becomes fairly obvious.

Optimizing target illumination
The CFD software even proved useful in the design of an “illumination tool,” a specialized flashlight with an internal microprocessor to maximize output and runtime. Designed and built by SureFire, Fountain Valley, Calif., for the U. S. military and law-enforcement officers, the X400 mounts on handguns and longer firearms and combines LED lighting and a targeting laser. The device is intended to help soldiers decide whether or not to shoot, especially in stressful situations. However, the 170 lumens generated by the LED are potentially harmful to the temperature-sensitive laser, necessitating an effective thermal-management approach.

SureFire Testing and Analysis Manager Deepanjan Mitra began using CFdesign in 2005. For the X400, he did a quick preliminary analysis, followed by work on two major design functions: shaping the LED heat sink to thermally isolate the laser unit, and determining the best settings for the thermal-management device. This uses a microcontroller and temperature feedback to control LED power and thereby the temperature of the entire light assembly.

CFdesign allowed performing various “what-if” scenarios, balancing heat dissipation with the need to minimize the unit’s size and weight. The software helped in selecting the best housing material and insulation thickness, and even predicted laser lifetime for different thermal profiles.

Simulations also revealed how quickly thermal management needed to be activated. And it helped identify the amount of time the LED would shine its brightest. In addition, for each new product, Mitra estimates that using the software early in the design process eliminates the need for three different hardware prototypes.

“When we do build a prototype, we instrument it with thermocouples and measure the temperature history, comparing it against the simulation,” says Mitra. “It takes three to four weeks to make three prototypes. Simulations only take one week or less. We’re also leveraging the benefits of the software’s capability to compare multiple designs at one time, using it as a ‘test bed’ to launch new ideas.”

© 2010 Penton Media, Inc.