Boosting AM adoption - the next phase of market growth

Metal additive manufacturing (AM) is taking off. Early adopters are increasingly deploying it to manufacture end-use components, many of which make use of AM's unique capabilities to deliver exceptional product performance. But even with the current high market growth rates, AM occupies only a tiny fraction of the total manufacturing landscape.

Three 'boosters' are needed to power the next phase of market growth, developments that will accelerate adoption and bring in a wider range of users. More manufacturers will get on board if we can 1) reduce the cost of AM parts, 2) streamline process validation and quality assurance, and 3) extend the capability of AM equipment. Meeting these challenges is vital if the benefits of AM are to be enjoyed more widely.

This article explores the current status of industrial AM and the drivers behind its current and future growth. We will consider each of the three boosters - examining how current performance limitations can be overcome. We will try to avoid grandiose long-term predictions, focusing instead on the practical steps that we can take today to move forwards.

Rapid growth in AM adoption has only just started

In recent years we have witnessed a steady advance in metal AM use, but 2017 represented a dramatic change of gear. According to the latest Wohlers Report, system sales jumped by around 80%, with numerous new machine manufacturers joining the fray. Clearly many new users are also now buying machines, whilst established users are scaling up their operations. Expectations of further growth seem well founded.

But we need a sense of perspective here. Even with the recent advances, AM will only represent around 0.1% of the total manufacturing market by 2020. Growth may be rapid, but the absolute numbers are still rather small. Whilst early adopters are pushing ahead, many manufacturing sectors have yet to really join the party.

Despite recent rapid growth, AM is less than 0.1% of total manufacturing

Where this will all end up is anyone's guess. Predictions for the eventual penetration of AM by the middle of the century vary from a few percent to as much as half of all manufacturing. BCG's predictions (below) are at the more modest end of the scale, but still require a strong compound growth rate and a fundamental shift in the way that products are designed and produced in many manufacturing sectors.

Whatever the future holds, market growth will be driven by a combination of market demand: using AM to create new value - and supply: technology and process innovation to make AM more competitive.

Enterprise applications are most mature

Some of the more outlandish claims of the 3D printing mania have proved to be wide of the mark. How many of us are using a home 3D printer to produce our own shoes or spare parts for our household appliances? Or did we make a few trinkets, realise that they were not much more than novelty items, and put the printer on the shelf to gather dust? I'm in the latter camp. Indeed, consumer 3D printing is firmly in the 'trough of disillusionment' of the Gartner hype cycle.

Image above - hype cycle for 3D printing and additive manufacturing, adapted from Chung, Niezgoda, and Beissmann, 2016

But it's a very different story for enterprise AM. Here there are now plenty of companies that are working their way up the 'slope of enlightenment', taking pragmatic steps towards production, working with the capabilities of today's equipment and steadily advancing the state of the art. They are developing business cases, executing implementation plans and making real progress.

Underpinning these early successes have been important advances in the safety and usability of AM processes. Integral powder handling, safe-change filters, intuitive human-machine-interfaces and increasing automation have made AM equipment suitable for factory use. Meanwhile, advanced AM design and engineering tools are streamlining product and process development, making it easier to develop high-performance AM components.

AM business impact

In Additive impact - how AM could disrupt your market I discussed the different types of benefit that can flow from deploying AM: production benefits that are accrued whilst you are making the product, and product benefits that accrue once that product is in use. It is the larger product benefits that underpin the demand side of the AM growth equation.

Moving up through the deployment levels requires increased commitment to design and qualify radical new products as well as a new manufacturing process, but the benefits tend to rise too.

Where AM is competitive today

The current state of the additive manufacturing art is best suited to high-value manufacturing applications. AM is most at home with complex, high-performance products, produced from difficult-to-process, expensive materials. Successful applications are often found where weight is critical, and where extensive quality assurance is expected. Products are sometimes customised, or made in low-to-medium volumes only. Markets that tick these boxes are aerospace and healthcare, whilst tooling is often the entry-point into other market sectors.

To move beyond these niches and into a wider range of higher volume applications, metal AM must make three significant improvements: increase productivity to reduce part costs, improve quality validation to provide assured quality, and enhance process capability by tackling process limitations*. We will look at each of these in turn and consider current developments that aim to address these issues.

* In the case of laser powder-bed fusion (LPBF), the most widely used metal AM technology, the principal process limitation is residual stress. Other AM processes have different capabilities and limitations. For instance, e-beam melting is limited in terms of feature definition and surface finish, binder jetting in terms of density and material properties.

Booster #1 - lower cost per part

In conventional manufacturing, we are used to economies of scale. High fixed costs associated with the deployment of tooling and capital equipment can make low volume production comparatively expensive, whilst unit costs fall progressively as manufacturing volumes rise and these costs are amortised.

AM is somewhat different. Here the cost of low-volume production is often lower than conventional techniques since no tooling is needed. The dis-economies of small scale are largely eliminated. However, economies of scale are less significant - the cost of building additive parts in larger production volumes may not reduce all that much. This tends to confine AM to applications where volumes are relatively low.

Of course, it's not fair to make a straight cost comparison as a well-designed AM part will exhibit higher performance than a conventional one. However, the point still stands - if an AM part is too expensive, then it will not gain the acceptance it deserves.

But what if we can move the AM cost curve downwards? As we reduce the cost per part, AM becomes cost-competitive in more situations, strengthening the AM business case. Higher productivity is critical if AM is to penetrate more cost-sensitive market sectors.

What drives AM part costs?

So, what are the drivers for AM part costs? The short answer is: 'it depends'. It depends on the nature of the build and the process chain that is required to manufacture the finished component. The choice of material is also a factor as these costs vary significantly. And, of course, it depends on the capabilities of the AM equipment that is used.

It is helpful to break these costs down into four elements - build costs, powder costs, post-processing costs, and labour. The example shown here is a lug-set from a custom mountain bike manufactured from titanium, produced on a single-laser LPBF system. These parts are heat treated, bead blasted, removed from the build plate and some are finish machined.

For a typical build, the largest component is build costs - i.e. the cost of owning and running the AM machine. These can range from 40% to 60% of the total part costs. The primary factor here is depreciation of the capital asset, with smaller contributions coming from facility overheads, consumables, power and gases. Powder costs can also be significant if the material is costly, typically in the range of 10% to 30% of total costs. Post-processing (such as support removal, heat treatment, surface treatment, machining, metrology and non-destructive testing) can vary significantly depending on the needs of the application, but can comprise as much as half of the total costs in some aerospace applications. Finally, labour is typically the smallest cost component since the build process is automated once the machine is running, with activities mostly focusing on powder management and machine turnaround.

Reducing build costs with multiple lasers

A key factor in increasing machine productivity is deploying multiple lasers. We cannot keep increasing the laser power and speed indefinitely as the metal powder cannot accept such energy inputs (see X marks the spot for more details) without compromising feature detail, so to melt faster we need more lasers.

Early multi-laser machines were larger and, unsurprisingly, focused on making larger parts in sensible build times. Their increased size resulted in higher machine running costs and typically two to three times higher capital cost, so the 'cost per cc of built part' did not reduce much, if at all. More recently, multi-laser machines such as Renishaw's quad-laser RenAM 500Q, have focused on offering more lasers in a mid-sized machine, with the goal of reducing cost per part in this popular size range.

A compact machine with four lasers, each of which can address the whole build plate, provides maximum flexibility in laser task assignment, irrespective of the build layout. This efficiency boosts productivity such that build times are typically reduced by a factor of 3 to 4 in most materials and geometries. Whilst the machine cost rises by around 50%, productivity rises by 200-300% (sometimes more) and so the capital depreciation element of the AM part cost is typically halved.

A good example of this is the galvo mounting from the RenAM 500Q itself. This substantial component is used to mount the steering mirrors for the four lasers and features conformal cooling channels to maintain optical stability. The build time for this 4 kg aluminium part has reduced from over 60 hours on a single laser machine to just 19 hours using four lasers.

Image above - galvo mounting build progress in 19 hours using 4, 2 and 1 laser. For more details see this blog post by Peter Zelinski, Additive Manufacturing Magazine.

Reducing powder costs

As build costs fall, powder costs become a more significant proportion of AM part costs, and may even become the dominant factor. Whilst the cost per kg is likely to fall as more production capacity comes online and atomiser yields rise, it is still vital to use this costly raw material efficiently. Re-cycling is critical to keep powder usage to a minimum, whilst AM process yields also contribute to eliminating waste.

Powder re-cycling

Effective powder re-cycling involves re-use of as much un-melted material as possible for subsequent builds. Many materials used in additive manufacturing readily absorb oxygen, especially when they are hot. This change in chemistry can have an unwanted impact on mechanical properties, reducing ductility and fatigue performance. If our powder feed-stock is to be re-used repeatedly, then it is critical that it does not pick up oxygen either during the build process or during powder handling.

To prevent oxidation, an inert atmosphere is created inside the build chamber. This is most effectively done by firstly creating a vacuum, followed by a purge with inert gas such as argon. If necessary, this process can be repeated several times to further reduce the starting oxygen level. A vacuum-sealed machine will not draw in air from outside during the build, so the oxygen level inside the machine cannot rise. By contrast, an unsealed machine is likely to draw in oxygen during the build, requiring periodic purging. For more details, refer to Oxygen algebra - minimising oxygen pick-up.

Another key factor is the powder management cycle. Ideally, we want to keep our powder feed-stock away from oxygen and moisture. Sieving is essential to remove large spatter particles and maintain the particle size distribution in the feed-stock, but removal of powder from the machine for off-line sieving creates opportunities for powder to be exposed to an oxygen-rich and humid atmosphere, so extra care and control is required.

Renishaw's RenAM industrial AM machines feature an integral powder management circuit. The bulk of the powder is stored in a large hopper positioned outside of the chamber. When more powder is needed, it is drawn from the hopper through an ultrasonic sieve and up via an argon transport system and into the dosing silo. The powder is then deposited in front of the dosing wiper, with any overflow passing back into the hopper. The entire circuit operates under argon, so the powder is never exposed to oxygen or moisture throughout the build cycle.

These machine features enable continuous recycling of powder feed-stock, with periodic topping up with new powder to maintain feed-stock levels and consistent powder chemistry.

Higher AM process yields

Process yield is another key factor that drives powder consumption, as well as contributing to post-processing costs. An AM process that exhibits consistently high material properties may enable more efficient designs to be realised, whilst post-build testing may also be scaled back. By contrast, a process that is more variable will drive the adoption of heavier designs with a higher safety margin, as well as costly 100% post-process inspection.

The design of the inert gas flow system is critical here. The purpose of gas flow is to prevent oxidation of the alloy, to carry process emissions away from the melt pool, and to keep the inside of the chamber clean. Failing to clear heavier process emissions from the build volume can lead to these particles falling on the powder bed where they can be incorporated into the component. This can result in 'spatter shielding' where the large spatter particle prevents full melting of the powder layer, or disrupts the re-coating process, reducing the fatigue performance of the component. In multi-laser machines, this is even more important since there are more emissions to contend with. For more details, refer to Gone with the wind - how gas flow governs LPBF performance.

Image above - the inert gas flow on the RenAM 500Q quad-laser AM machine features consistent planar flow across the powder bed (left), combined with substantial top-down flow to suppress re-circulation (right).

The result of good gas flow is consistent mechanical properties. Ductility is particularly sensitive to part defects, so we are looking for a low level of variation in our test results. The table below shows test data from 80 ASTM E8 vertical tensile specimens in Ti6Al4V produced on a RenAM 500Q using four lasers, following stress relief heat treatment and hot isostatic pressing. The samples show consistent properties, including a high mean and low coefficient of variation (CoV) in 'elongation at break' and 'reduction of area':

We can see a similar benefit in more ductile materials like Inconel 625. In this instance we have can see the stress-strain curves for 16 'as built' vertical tensile samples, once again built with four lasers and distributed across the bed. The mean elongation at break is more than 60% with a low level of variation.

Confidence in material ductility enables AM parts to be designed with less mass whilst still achieving the required safety margins in fatigue and creep. This benefits both build times and material costs.

So, we are making tangible progress in reducing part costs by boosting productivity, preserving our feed-stock and increasing consistency and process yields.

Booster #2 - assured quality

LPBF gives us great design freedom, but process development and qualification can be challenging. As we have seen, process anomalies can produce defects that affect strength and fatigue performance. This can lead to heavy reliance on post-build testing and costly production process control. However, new technologies give us the opportunity to detect and identify defects during the AM build process. For more details, refer to Real-time AM monitoring opens up new process control opportunities.

AM inspection challenges

Metal AM is both foundry and forming in one process and the process operates at a very small scale and in very short time periods. Metallurgical integrity is difficult to measure non-destructively and internal features are often hard to inspect. As a result, post-process measurement can be complex, expensive and slow.

Ideally, we want our solidified metal to exhibit 100% density, with no pores or defects that reduce its strength and durability. This requires stability of the melting process, with consistent melting conditions and thermal history across all regions of the part.

Of course, we try to design our build process to achieve this, but there are several failure modes that can lead to minor defects. These include variable local processing conditions due to part geometry and scanning sequence, laser guiding precision, powder condition and dosing consistency, welding ‘spatter’, and macro-scale variation due to optic cleanliness and gas flow.

Real-time process data capture

AM's rapid melting and cooling phenomena require high frequency data. Renishaw's InfiniAM Spectral real-time process monitoring technology enables sensing of delivered laser power and melt pool response at a 100 kHz sample rate. The sensors are entirely passive, so process parameters are unaffected. The sensor data is synchronised with actual galvo mirror positions so that a precise 3D model of the energy input and melting behaviour can be established.

Process data visualisation and analysis

We can now collect and view process data live as the build progresses. 3D visualisation allows us to understand process behaviour in context of a particular build. We can view the whole part or zoom into regions of interest. We can also set thresholds to highlight anomalies in laser power delivery or melt-pool response, highlighting hidden 'hot spots' inside the component that may indicate heat build-up and keyhole pore formation. We can also compare the different sensor signals to identify correlations.

When we spot an anomaly, we will want to investigate further. This is where 2D analysis is helpful, looking at the data from a single layer, or scrolling up and down through successive layers to understand defect propagation. In this image we are looking at delivered laser power across a layer at a resolution of 40 microns, highlighting stripe overlaps in our hatching scans.

Another key analytical tool is comparison with other data sets, such as those from the other machine sensors for laser energy input, temperature, pressure, oxygen content and other machine events. Such detailed process data provides insight into process behaviour, as well as a traceable record of process execution.

Future process control possibilities

Comparison of data with known-good builds enables us to spot inconsistencies that experience may tell us correspond to defects in the build. We can potentially use the locations of these anomalies to direct micro-CT scanning to check for porosity only in those regions, reducing time-consuming 100% inspection.

A further future development could be to identify and fix minor anomalies by responding to process feedback signals, but this could also affect the thermal history of the part and influence processing consistency. Effective closed loop control such as this will need to be well proven to be accepted by industry.

By instrumenting our AM machines and developing sophisticated data analytics, we have the opportunity to further improve process consistency and reduce reliance on costly post-process inspection.

Booster #3 - process capability

All AM processes have their capabilities and their limitations, affecting the geometry and precision of the part, as well as its performance in use. In the case of LPBF, the rapid cooling rates that generate attractive material properties, also give rise to residual stresses that accumulate within the component, which can lead to distortion and failed builds.

Residual stress varies by material, driven by the properties of the alloy. This graph shows different materials distort due to residual stress build-up. Inconel and titanium produce high levels of strain, as measured using cantilever artefacts, whilst aluminium is less prone to distortion. For more details refer to Want to build accurate AM parts? No stress!

Minimising residual stress

Applications engineers use various techniques to reduce residual stress by adapting the processing conditions using scan strategies. These are largely based on experience and are often a compromise between achieving acceptable build rates, consistent quality processing and feature definition. Each geometry demands a slightly different approach, and the process can demand an unwelcome level of skill and judgement.

Image above - impact of hatching scan alignment relative to the major axis of a cantilever artefact. Longitudinal scans (along the major axis) produce more distortion, whereas lateral scans and rotating the scan direction after each layer both reduce part distortion.

Another way to overcome this is to modify the processing conditions. Elevating the build temperature prevents stress accumulation firstly by reducing the temperature differential between new melt tracks and the substrate, and secondly by temporarily reducing the yield stress of the alloy so that it cannot build up such high stresses.

Image above - Renishaw's new high temperature build volume enables the substrate to be heated up to 495 °C (± 10 °C). This can reduce the residual stress in Ti6Al4V by more than 70%, as shown in this graph.

Mitigating residual stress

Whilst it may not be possible to eliminate residual stress altogether, there are increasingly sophisticated ways to mitigate its effects. One technique is to measure the distortion in a component and then 'reverse distort' the CAD model design so that it 'pulls itself straight'.

CAD and CAE software vendors are developing simulation tools that model the stress build-up in the process to predict distortion in flexible components, enabling compensation to be applied to substantially reduce dimensional errors.

So, through careful design and engineering we can minimise the impact of AM process limitations, boosting process capability and improving AM product performance.

Summary

AM’s unique capabilities are disrupting product markets and the pace of adoption is accelerating. Metal AM is already taking off in key manufacturing sectors.

The next phase of AM market growth will be propelled by improvements in productivity, quality assurance and capability. These developments will deliver lower part costs, faster process development and higher-performance AM products.

Real progress is being made on all of these fronts, as new multi-laser machines equipped with real-time process monitoring and high-temperature processing capability are introduced.

Next steps

Visit www.renishaw.com/amguide for more education resources and to access downloadable versions of this and other LinkedIn articles by Renishaw authors.