The next generation of business and commercial jets, military aircraft and electric hybrid vertical take off and landing vehicles will require avionics systems to do things they have never done before, including most prominently the support of artificial intelligence and machine learning applications.
These next generation platforms are also going to require levels of processing power and power generation that aircraft have never seen onboard before. Here, we analyze how emerging methods such as liquid cooling can help the industry address thermal management and overheating challenges presented by ever-shrinking aircraft computing architectures that must achieve ever-expanding levels of processing and functionality.
Liquid Cooling for Aircraft Embedded Computing
Ivan Straznicky, chief technology officer of advanced packaging for Curtiss Wright describes thermal management and overheating challenges as a two-part problem. First, every generation of silicon devices, processors, field-programmable gate arrays (FPGAs) and chips features a new level of more advanced functionality, resulting in high levels of heat and power dissipation which is difficult to maintain and cannot be capped.
“20 years ago we were developing 6U cards that would produce around 20-watts of power. Now, in some cases with our most advanced products, we’re around 200 watts with forecasts for continued increases,” said Straznicky.
According to Straznicky, thermal management challenges start at embedded computer chips, but also need to be addressed at the board, chassis and system level. All embedded components and systems at some point dissipate heat into the ambient air external to the aircraft and the air that is already circulating within the aircraft.
Historically, embedded systems suppliers have addressed thermal management challenges within avionics designs with methods such as conduction cooling at the card or module level. That type of cooling approach dissipates heat across the conduction cooled interface and into an air-cooled or liquid-cooled chassis.
Another more modern method of automatically cooling avionics systems is the use of computer processing throttling. What designers do under this approach is to use firmware and software that automatically scales back processor frequency once a device reaches a pre-determined temperature threshold. As an example, if a 3 GHz processor has a operating temperature threshold of 100 degrees Celsius, it will scale back to 2.5 or 2.25 GHz when its temperature starts nearing the threshold.
However, as intelligence, surveillance and reconnaissance (ISR) capabilities rapidly expand in functionality and require more robust processing power, they require more exotic cooling methods, such as air flow through cooling and liquid flow through cooling.
“When you consider air flow through and liquid flow through cooling, one of the reasons they’re catching on is that they’re bringing the coolants much closer to the heat generating electronics,” said Straznicky.
“Compared to traditional conduction cooling you have a fair amount of thermal resistance from the devices to the cooling air or the cooling fluid whether it’s air or liquid in the side walls. When you use air flow through or liquid flow through, you’re bringing that air or liquid much closer to the heat and reducing the overall thermal resistance path,” he said.
Liquid cooling methods for avionics designs have been used scarcely in the past, by prime defense manufacturers such as Lockheed Martin and Northrop Grumman, but are expanding in popularity today.
A liquid cooled avionics system within an aircraft can best be defined by the setup featured on the F-22 Raptor. The fighter jet’s liquid cooled avionics system circulates the coolant Polyalphaolefin (PAO) through mission critical cockpit electronics cold plates and then pumps it out to the wings to provide cooling for embedded sensors. At that stage, the liquid becomes warm and passes through an air cycle machine where it absorbs heat generated by a heat exchanger. The heat is then transferred from the liquid to the fuel.
The new VITA standard 48.4 has greatly expanded the popularity of the concept by defining a liquid cooling method for plug-in backplane modules. First published on July 26, 2018, the standard establishes the mechanical design, interface control, outline and mounting requirements to ensure the mechanical intermateability of 6U VPX liquid-flow through cooled plug-in modules within associated sub-rack assemblies. Under 48.4 specifications, the modules are designed to feature an integral heat sink which allows liquid to flow through and cool electronics and circuit boards.
“In the world of open systems architecture, we frequently see requirements for increased processing throughput and the resulting high-power dissipation where standard cooling methods such as conduction and air cooling are no longer effective,” said Christal Sumner, VITA 48.4 working group chair and principal mechanical engineer at Raytheon Company.
Straznicky said he expects the introduction of 48.4 to drive more adoption of liquid cooling for embedded aircraft systems into the future.
“There hasn’t been a significant demand for it up until recently, part of the reason for that is that 48.4 was recently ratified and became available,” he said. “Another reason is customers that have used custom liquid flow through in the past, are wanting to use more COTS [commercial off the shelf] modules so they want COTS vendors to step up and have liquid flow through modules available as they upgrade things like radar processors to the latest and greatest technology. We’re developing liquid flow through modules to meet that need.”
Siemens subsidiary Mentor Graphics is a company actively promoting the use of liquid cooled avionics designs with an emphasis on not only the design of the avionics package and its cooling requirements, but also what to do with the dissipated heat once it has been transferred to the coolant.
The company has proposed the use of characterized three-dimensional computational fluid dynamic simulations to establish initial evaluations of the cooling system long before the physical components become available for bench testing. While model-based design concepts such as this have been in use for years, Mentor is using newer more advanced simulation technology.
“We’re now able to combine simulations over a single solver matrix, to develop a liquid cooled avionics design without the use of separate software necessary to allow the tools to communicate back and forth,” said Michael Croegaert, strategic business development manager for military, aerospace and power at Mentor Graphics. “The problem with traditional co-simulation approaches is that you get a lot of convergence problems and often have to pause one tool to let the other catch up.”
Croegaert said Mentor recommends using a design of experiments (DOE) approach to three-dimensional design simulation that is capable of running a sweep of simulations over changing operational conditions. The company has found that the use of DOE allows the designer to generate a series of response surfaces that would be capable of dealing with the type of heat dissipation necessary to support a new product design.
Mentor’s use of DOE has shown that enhanced heat transfer surfaces can significantly improve the heat dissipation capabilities of liquid cooled cold plates used in avionics packaging.
“Liquid cooling methods for avionics design are continuing to evolve, it’s been used in the F-22 and the F-35. Additionally, as aircraft become increasingly become more electric, there will be thermal challenges to address,” said Croegaert.
Electric Power, Propulsion in Aircraft
Another major trend that is creating thermal management challenges across the entire embedded digital and mechanical footprint of modern aircraft is the use of more electric power.
On Dec. 4, 2018, two Carnegie Mellon University mechanical engineering researchers published an article giving an overview of a computer model they constructed to calculate the amount of power needed for a single passenger eVTOL weighing 2,200 pounds. Their analysis showed that the most power intensive phases of flight would occur during takeoff and landing, which these two phases alone would require between 8,000 and 10,000 watt-hours per trip.
That type of wattage and power output will require specialized cooling systems and the use of more heat-resistant materials in designing the embedded computing architectures that will power future eVTOL aircraft.
The same company that supplies the electrical power systems on the 787 — Collins Aerospace — recently started construction on a new facility, “The Grid,” where it will specifically research and develop a hybrid electric engine for a demonstrator aircraft with the goal of the first flight occurring in 2021. But the same facility will also perform research around the use of hybrid electric propulsion and what type of materials are needed to handle the type of power required for takeoff, cruise and landing within an urban air mobility aircraft as well.
“As we go up in power ratings there’s also a thermal aspect to what we’re doing, getting smaller and single points of efficiencies mean lots and lots of thermal losses,” said Greg Winn, director of program management for the United Technologies Advanced Projects (UTAP) Project 804 team. “How do you do that? How do you thermally manage that on the aircraft as an entire system as opposed to just another compliment level? That’s another challenge. There’s a lot of heat that we will have to deal with.”