Analysis of what might happen in Industry over the next decade, talks mostly of automation, artificial intelligence (AI) and big data. The picture is of ‘smart factories’ operating 24/7 ‘lights out’, with mobile and autonomous robots achieving high productivity, needing very little human intervention.
In Industries that retain operators, ‘cobots’ (see Figure 1) will work alongside them, assisting, watching and learning. IIoT or Industry 4.0 will become all-pervasive with intelligence at the edge – where sensors and actuators touch the industrial process itself. Data gathered in this way will be passed to the cloud to enable minute to minute decisions to be made across global sites, optimising processes and predicting future needs (from raw material shipments to maintenance scheduling). 5G, with data rates reaching 20Gbps, will play a significant role in this expansion.
Figure 1: Cobots working alongside humans – original artwork by Mouser.
The industrial future may look a little soulless with the addition of so many robots and cobots, but the World Economic Forum has predicted that AI alone will generate 58 million new jobs by 2022. At least some of these will be in companies working on power conversion equipment – looking to provide the exponential improvements needed, both in efficiency and power density performance.
Additionally, designers are likely to be pressed for ‘just another 1%’ efficiency improvement when they are already achieving 98% efficiency in a variable frequency drive – pressures to improve efficiency will only continue to grow.
Rising to the efficiency challenge
Luckily, engineers like a challenge. Many look to new semiconductor technologies and power converter topologies to find incremental improvements. The second instalment of this blog series featured the use of wide band-gap (WBG) semiconductors. Their use has opened up a world of possibilities, not only to improve the efficiency of power conversion but also to switch at higher frequencies – with the added benefits of smaller associated components, particularly magnetics.
Conversion topologies such as phase shift full bridge (PSFB) or inductor/capacitor-oriented LLC arrangements (which switch resonantly at high frequency) are close to optimal efficiency, offering ultra-low losses, enabled by WBG semiconductors. This optimal situation makes footprints smaller and power densities higher, with more converters installed in a given space. The net result of this is lower overhead costs and better space utilisation – both critical benefits in modern Industry. WBG devices also inherently withstand higher temperatures, so increased converter density is not necessarily a thermal problem.
Power conversion efficiency is equally a concern at the edge sensors and actuators, as we discussed in the third blog, getting power to the ‘edge’ is not always easy, and every Watt needs to be conserved. Routing up to about 70W of usable power through Ethernet cables is a great help here, coupling nicely with the need to include fast communications to the intelligence now embedded into the sensor or actuator.
Where Power-over-Ethernet (PoE) is not an option, and wireless communication is necessary, local power becomes an issue, and energy harvesting is a solution (with the various schemes outlined in the fourth blog in this series). There is a move to reduce the energy lost to the environment with ‘lights-out’ operation, low RF levels for limited EMI, plus smoother-running machines with less vibration. It’s likely; however, there will still be ways to harvest energy from the sometimes MW-level loads, if only from thermal gradients.
In the fifth blog, we saw how power system architects have to consider the power requirements of the highly integrated ICs used in process control electronics, located in what can be hostile environments, from either an electrical and mechanical standpoint. Processor core voltages, now below 1V, need to be accurate and noise-free, necessitating use of point-of-load (PoL) converters – which in themselves can be power-processing sub-systems (with embedded microcontrollers to maximise efficiency and performance). As CPU power rails are required to supply more current (reaching hundreds of amps) they need to be closer still to the load, to avoid voltage drops. There are possibilities to further improve the performance of individual GaN and wide band gap devices. However, research into the integration (lateral GaN) of multiple devices on to the same substrate is likely to offer the most significant impact on power conversion efficiency in the future.
Undoubtedly, we require Industry to take a more ‘holistic’ approach to energy use over the coming decade. As mentioned in the introductory blog; Industrial energy demand is sure to increase globally, and we must consider all possible efficiency improvements in order to balance this.
As it sits at the end of the power conversion chain, Industry must play its part in reducing energy loss. Currently, the majority of its energy input is converted into ‘waste’ heat. At the same time, local communities pay for warmth in the form of fuel for their domestic needs. Schemes are in place to match these producers and users with piped hot water from heat exchangers. Closer integration with community energy needs is undoubtedly part of the factory of the future.
The next decade is likely to see continued IIoT expansion. AI, predictive maintenance and cobots are just a few of the drivers that will fuel this expansion. Edge computing will produce fast data processing on the factory floor, and technologies such as 5G will make cloud connections more reliable to enable collation, distribution and processing of big data across global sites. Both technologies can then be used to their strengths.
Research into the integration of Wide Band Gap technologies into monolithic packages to enhance PoL converter design are amongst the exciting possibilities we can expect to see over the next decade.