Opinion I’ve been lucky enough to learn from some very experienced engineers in my career. One of the first tried to teach me technical drawing
It’s 2035, and you’re a mechanic on the Russian steppe holding the mangled end cap from an electric lift motor; the pilot of your 10 year old eVTOL survey aircraft lost an argument with a Ural Owl. The nearest supply depot is 750 km away, and they wouldn’t have the part for such an ancient vehicle anyway. The aircraft software won’t allow the aircraft to fly without the motor. There is no way you can machine a replacement; its complex geometry enables the part to work at the absolute minimum weight, and such weight control is why the aircraft can fly at all.
But you’re not stuck – you switch on your 3D printer and feed it the plans for the end cap. You spend the few hours before the part is ready to be finished and fitted discussing with the pilot how good it would be if they could avoid the local fauna once the vehicle is airworthy again.
Such a scenario demonstrates three key benefits of additive manufacturing: the ability to create complex shapes not possible by other means, to re-create old parts whose templates and tools have been scrapped, and to manufacture at remote locations.
Additive manufacturing (AM) – 3D printing by another name – differs from traditional manufacturing techniques (TMT) in that material is combined to create the part rather than taken away by drilling, milling, lathework etc. There are many different AM techniques (Rafal Wrobel of Newcastle University identified 7 main categories) such as adding layers or sheets and bonding them using glues or heat, extruding material, fusing powders with lasers or heat, amongst others.
AM enthusiasts have been working since the late 1980s to create increasingly complex items in plastic, metal, magnetic and even biological materials. Commercial manufacturing has been more circumspect, due to the lack of understanding of a novel approach feeding an inability to certify that components manufactured using those approaches would perform adequately.
AM promises of lightweight, strong parts with complex shapes are of interest in electrical machine manufacturing, particularly where motors and generators need to deliver high-performance in demanding environments, such as electric aircraft.
The UK’s Advanced Propulsion Centre (APC) 6th Spoke on Electrical Machines organised the “Additive Manufacturing of Electrical Machines” seminar on 4th June 2019 at The Centre for Modelling & Simulation Services Ltd in Bristol. The programme presented the latest progress on research into and commercial use of AM, and outlined future work. Barrie Mecrow of Newcastle University’s PEMC group chaired the agenda of 10 speakers from industry and academia, followed by an open floor panel of all the speakers.
Reassuringly, the speakers largely reinforced or expanded each other’s points. They shared failures and successes, identified strengths and weaknesses of AM, and passed on lessons they’d learned in experimenting with AM and with integrating it into design and manufacturing processes.
Providing context, Robert Scudamore of TWI Ltd estimated the potential global market for AM worldwide at £9.3 bn in 2018 with 18% growth in 2018, and a forecast of £41 bn in 2027; use in automotive was estimated at $1.5 bn in 2018 rising to $12.8bn by 2028.
Why would I use AM?
Various speakers confirmed that you could make the same design with AM as with casting, but then asked: Why you would want to? AM allows unconstrained 3D design with no compromise to manufacturing methods. Both points open new design possibilities which enable
- combination of parts such as integrating subsystem components of different functions, e.g. motor casings and cooling systems;
- integration of fine features not always possible with TMT;
- advanced thermal engineering;
- use of new materials.
All of which should produce better designs which
- improve component performance;
- decrease lead times;
- reduce weight;
- improve packaging;
- optimise topologies.
At the same time, from an operational support viewpoint, because AM is a literal build-to-print technology which depends only on having the correct feedstock, it will enable
- repair solutions for aging components no longer in support;
- local repair at remote sites.
As an example of several of these points in a single piece of work, Siemens took a TMT manufactured motor end cap weighing 10.5 kg, analysed its functions to optimise its geometric design for AM, and made it for 4.9 kg. They then used material optimisation to reduce its weight to 2.3 kg.
Successful use / integration of AM in manufacturing processes
The speakers agreed that AM is the future of manufacturing, just not yet. Given that belief, to successfully use AM, you need to understand its manufacturing value chain, not just have a printer in the corner; i.e. you need to understand the AM process of design, build preparation, build, post-process, and inspection, and how it fits into or alters your existing manufacturing processes.
For any component to perform requires it to have both form and functionality, i.e. its shape and appearance need to be matched by its ability to do what it needs to do. Iain Todd, Director of the MAPP / EPSRC Future Manufacturing Hub, explained good design and manufacturing are both needed to ensure the component
- has the correct geometry and topology;
- will and does have sufficient structural integrity;
- has consistent microstructures;
- results from processes which are well defined before, and in control during, manufacturing.
The quality of each example of each component depends heavily on the quality and understanding of these manufacturing processes. Without careful attention, each run of the same design file can produce drastically different results. Using optical, thermal and X-Ray monitoring of several bonded powder AM experiments, Todd’s experiments showed powder being blown around and hotspots causing random pitting and clumping. They also found that even before a run, storage of raw material could be critical to component quality – something as simple as storing powder badly could allow enough moisture into the powder to cause inconsistencies such as inhomogeneities, stresses and holes in the finished piece.
Finally, Todd stated that if you understand and have control of a specific AM process, then with careful operation you could exploit potential process failures to create a component with better performance, for example, pre-stressed parts.
Dragica Kostic-Perovic and Philip King of NIDEC Group reported that the company has begun to introduce AM into their production work. Although they feel AM techniques are currently too slow and so not suitable for high volume production, and don’t always produce consistent results, they do see a use for AM right now, and that it might go on to play a bigger role.
NIDEC started by buying some AM machines to investigate the technology, discovering the potential for faster turn around times. They developed their own training in design for AM – i.e. how best to prepare a design for AM, and how to take advantage of AM capabilities to create the best design – and have now started using AM in certain applications. Rather than production components, they currently produce parts which will help production, e.g. poke yoke check devices and reflector wheels for final checks of motor shaft rotational speed. They found AM could produce devices like these faster, cheaper and better than TMT. For example, the reflector wheels were originally machined from PTFE and metal, costing £34 each; using AM they were £3.30, and not only cheaper, but gave better data as they have less inertia.
One significant problem NIDEC have encountered is that the speed with which AM allows designs to be iterated means more effort has to be focussed on documenting changes to processes or design drawings. They found that it was too easy to allow the speed of design / AM manufacture iterations to outstrip documentation, meaning a successful component could be developed which couldn’t be qualified because it didn’t match its design drawings.
Bearing this in mind, NIDEC are now testing AM in rapid manufacture for prototypes and in high-value / low volume products, being especially careful to check component performance, e.g. ensuring thermal performance in prototype motor end caps.
Others agreed that easy access to AM was a benefit to the engineering process, but that it could cause issues. NIDEC’s red flag regarding documentation was repeated by others – that AM must not be allowed to ride roughshod over existing processes, but instead be used within standard design process frameworks, and doubly so for functional parts.
Others reported that AM can be seen as a Star Trek replicator, meaning that because it delivers rapid results, the time and cost to manufacture the actual component are underestimated – it’s easy to end up with a working design that can’t be manufactured. For example, no allowance for the injection moulding features needed in mass manufacturing, or a design which relies on materials unique to AM. To counteract this, before a design is completed, the mass-manufacturing and assembly processes for the approved design must be considered.
Al Lambourne of Rolls-Royce reported on a series of experiments with different types of AM and a variety of components, including some multi-material designs. The results were promising, although they did hit some limitations of current technology, which may be overcome by materials development or by careful design of both the components and the manufacturing process.
One experiment combined austenitic non-magnetic materials and ferrite magnetic materials to create a compromise rotor with specific magnetic and mechanical properties. While it worked – it delivered a rotor with variations of properties, it apparently wasn’t a great success.
Another attempt to create power dense machine coils by co-depositing the insulation and the conductor failed due to the difference in material melting temperatures – the insulation at around 350 deg C, and the copper conductor at 1,085 deg C. Laying down the conductor melted the insulation beside it.
They have had more success in detailed component design or integrating normally separate components to drive out weight, such as motor end caps and casings, or combining caging with housing for liquid cooling.
Simon Jones of HiETA Technologies Ltd offered the most sensible advice to the adoption of AM: use AM where it makes good engineering sense.
AM in electrical machines
Some general challenges in the performance of electrical motors with which AM could help were agreed, such as maximising power density, higher rotational speeds, and better thermal performance via targeted cooling of critical components. Various experiments from Newcastle, Nottingham and Bristol universities amongst others were reported, showing that different designs and the integration of functions could provide improvements in performance such as
- motor housings with integrated cooling, increasing mass flow;
- integrated rotor designs providing mass reductions of 25% and moment of inertia reductions of 23%, while increasing rotor magnetic permeability 12 -15 times;
- integrating stator windings with liquid cooling facilities reduced winding temperatures by 44%.
The general opinion of AM printed magnets themselves was that they can provide similar magnetic results to current sintered and bonded neodymium and AlNiCo permanent magnets, but are mechanically inferior. Other experiments showed some promise but require further work to achieve useful performance improvements
- an AM magnetic core gave comparable magnetic properties, but specific power losses were 4.5 times larger;
- some magnetic and mechanical properties of an AM soft magnetic material were comparable with commercially available steel, but resistivity needs further optimisation.
One interesting project is HiETA’s AMPERE, aiming to develop an on-vehicle permanent magnet electric motor, with a prototype around June 2020. Targets for the project are 200 kW at 30 k rpm, weighing less than 10 kg and ultimately costing around £1 k, with a power density of less than 20 kw/kg. If achieved, this would significantly improve on motors fitted to electric vehicles such as the Chevy Bolt.
Other notable AM research projects underway include a complete stator-winding assembly from the Chemnitz University of Technology, and a US DoE funded motor with a conical air-gap. Other AM research targets include
- using electrical copper without degrading its electrical properties;
- insulation of electrical systems;
- soft magnetic materials with correct grain structure;
- AM laminations with effective insulation;
- permanent magnetic powders with correct orientation / structure.
Applications in electrical machines in aerospace
Nour Eid from the ATI stated that increasing electrification in aerospace is driving in two distinct directions: evolutionary more electric aircraft, and revolutionary hybrid / electric propulsion. In turn both these drive innovation in motor design, and the opportunities identified above would be great assistance to the success of both efforts.
Current aerospace applications need a power density of 6 – 7 kW/kg, but future applications will need 15+ kW/kg (achieved non-cryogenically), up to MW level aerospace-qualified machines. Current motors use well established functional materials including SmCo or NdFeB magnets, FeCo / FeSi laminations, copper wire and polymer insulation, and while there have been incremental improvements to key properties in these materials and motor designs in the last 30 years, the aerospace environment presents temperatures and environmental conditions too challenging for these materials.
So aerospace has a particular set of requirements for its motors – high power density, light weight and compact volume, all delivered in a high temperature, high threat environment, with cost a critical factor. AM is seen as significantly helping with at least some of these.
Using AM commercially
AM only has a use commercially when it can deliver reduced cost and provide confidence to the user of product maturity, i.e. that the component will be as reliable and effective in operation as one produced using TMT. This can be done by both proving product maturity and proving design and manufacturing processes.
Processes are discussed above; but on costs, the one incontrovertible benefit AM offers over TMT is that it generates much less waste during fabrication. Although AM components have some flash or roughness which require post processing to create a precise component, ultimately a part is made by adding material required rather than removing waste from a blank or a casting.
Supporting this fact, users reported the majority of AM process costs are in activities other than the AM stage, which now costs around 25% of the total AM production cost. Eid provided ATI survey findings estimating that by 2021 50% of AM manufacturing cost will be in component post-processing (surface finish, etc).
For the manufacturer and the end-user, AM can also deliver other cost-savings. Jones gave an excellent example from HiETA, of a 30 kW recuperator for motorsport. In 2016 the original design was 45 kg; a first AM design reduced the weight to 15 kg with 19 days build time, and by 2019 the design weighed 5 kg and took 40 hours to build.
Currently though, AM production costs give a sales price point suitable for selling to low volume, medium-to-high cost markets such as generation or aerospace, but entirely unsuitable to high volume markets like automotive.
Due to component certification product maturity is a little more difficult, especially of safety critical components. AM materials and the internal and micro-structures created in the end products are just not as understood as those in TMT. But this issue should fade as more work is done and this knowledge matures, as it did in TMT – indeed NIDEC stated they are a few steps away from validating certain AM materials and processes for production.
There are ongoing efforts to resolve this issue, but the ultimate proof are AM components operating as normally as TMT components. Jones gave several examples of AM components HiETA have had in operation
- automotive recuperators with 10 years running in the field;
- heat exchangers in motorsport on 100,000 cycles;
- turbomachinery parts with 17,000 hrs of hot running;
- heat exchangers in the chemical industry operating at >2,000 bar pressure.
Looking to the future
AM is a proven technology starting to be deployed by forward thinking organisations. Like all techniques, each AM approach has its benefits and drawbacks, necessitating in depth understanding of existing manufacturing and AM processes and how the two can fit together. But the potential for more effective, lighter, cheaper components of almost any type makes it well worth the effort.
Looking into the far future, some suggest traditional TMT, factories and stock can be completely removed and replaced with on-site build to print AM facilities – as some plans have suggested for a moon base. Whether or not this comes to pass, additive manufacturing will be a solution in some situations – it would certainly be easier than shipping myriad spare parts to the Russian Steppe.