You can build it, but can you finish it?

Many parts produced by additive manufacturing (AM) require some post-process machining to generate precision surfaces and interfaces. But finish machining of AM parts can be challenging due to their light-weight and complex forms.  This can lead to problems with work-holding and vibration and can result in poor process yields. Furthermore, there is the additional problem of aligning complex components when they lack precise geometric datums in the 'as built' state. 

Scrapping additive parts at the finish machining stage is costly and highly frustrating!

Image right - many additive parts require finish machining to create precision interface surfaces.

This article will look at the challenge of making light-weight parts stiff enough for effective finish machining. We will explore how to design and make effective work-holding solutions to make non-rigid additive parts machinable. And we will demonstrate how machine tool probing can be used to perform sophisticated alignments of AM parts, enabling us to 'find the good part' within the shape that we have built and produce the critical datum surfaces correctly.

Thanks to my colleagues Mark Kirby and Mark Buckingham for their contributions to this article.

AM has a dark side

Additive manufacturing enables us to design and manufacture products that we just can’t make any other way. This flexibility allows us to deliver product performance benefits that make AM parts highly efficient. Performance benefits are available in areas such as light-weighting and thermal efficiency, whilst we can also integrate separate components into consolidated designs, often with very complex forms. The build process itself is often highly automated and highly efficient it its use of materials, producing very little waste.

Image right - finish machining of precision bearing features in a mountain-bike bottom bracket.

However, these benefits should be offset against some very real challenges with post processing, which can be painful if we take the wrong approach. Additive cannot produce features with the very finest of tolerances, so post-process machining is often needed to produce precise round holes and smooth, flat surfaces for interfacing with other parts. Light-weighting often has the effect of reducing the stiffness of our AM parts, which can mean that they do not stand up well to the machining process. Their complex forms also make them hard to grip securely without causing damage. And finally, it is common to produce the datum features on additive parts after the build itself, and so setting up components for finishing can be tricky.

Image above - pros and cons of AM

There are a lot of similarities here with the challenges that manufacturers of composite and super-plastically formed parts also face – namely complex shapes that may have some distortion, onto which precision features must be machined. Additive manufacturers can learn from the best practices in these other sectors, whilst adding an AM twist of their own.

Finish machining challenges

If our additive part requires finish machining, we are faced with a number of challenges. The first consideration is whether it is likely to be stiff enough to cope with the loads that it must bear during machining. Will the part deflect away from the cutter, and will it vibrate such that we get tool chatter and poor finish on machined surfaces?

If we conclude that our light-weight design is not strong enough to withstand machining, what can we do about it? Can we design it differently to make it stiffer? Or if that’s not an option, then how can we hold the part to support it sufficiently that it will not deflect or vibrate excessively? We will use a case study to explore these questions.

Image right - work-holding and alignment of complex AM parts need careful up-front consideration.

Another important challenge, supposing that we now have a part that is designed or supported in a way that provides sufficient rigidity, is how to datum and align it on our machine tool. The complex form of AM parts, the fact that there may be some distortion in the build process, and the lack of crisp datums, all mean that we have to somehow 'find the good part' inside the shape that we have made. Getting an optimal 5-axis alignment of our AM part is critical. We will come back to this point a little later.

Case study - microwave guide

We will look at a microwave guide designed for a telecoms satellite application. The critical performance factors here are the weight of the wave-guide, the efficiency with which microwaves are transmitted through it, and the space claim that it makes on the satellite payload.

Image above - a good example of light-weighting and part consolidation, as the integral wave-guide (right) replaces a jointed assembly of short wave-guide sections (shown left). We have eliminated lots of joints – which benefits transmission efficiency by reducing back-reflections, and we have shed around half of the weight of the original part thanks to fewer flanges and no joining bolts. The manufacturing and assembly process is also greatly simplified.

These are attractive benefits. But we are still going to need to finish machine the flanges at each end of the guide – how easy will this be?

Step #1 - establish expected cutting forces

First we need to know if our part is going to be rigid enough to handle the cutting forces that we will apply to it. We can run experiments to assess these forces using a scrap build plate of the appropriate material attached to a dynamometer:

All machining is carried out in the XY plane, with cutting under flood coolant. The dyno data shows us the loads under repeated passes. We see peak forces as we enter and leave the cutting zone which are roughly twice the mid-cut values. We can also experiment with different depths of cut to see how this affects the loads on the part.

So now we know the peak and typical cutting forces – is the part stiff enough?

Step #2 - simulate cutting forces on the part

Next we apply these forces to our part, in this case held in a basic fixture along the parallel section of the exterior of the wave-guide. Machining around the edge of the flange at the free end of the component leads to significant deflection – more than 150 microns. The finite element analysis (FEA) also shows significant twisting, which is likely to lead to uneven cutting. 

Image above - FEA analysis of deflection of the wave-guide under anticipated cutting forces when held in a simple fixture.

Step #3 - initial cutting trials

When we machine under these conditions we encounter predictable problems. The part deflects away from the tool during cutting and also springs back, leading to resonant vibration, tool chatter and gouging of the component surface. The result is that the periphery of the flange is machined under-sized and we see very poor surface finish.

We need to improve the stiffness of the part during cutting. There are two possibilities: change the part design, or change the way that we support it during machining. We'll look at part design first.

Step #4 - improvement by design?

One approach is to consider whether we can make some detail design changes to our AM part to make it stiffer. Here we have added some braces to connect the two ends of the component to reduce the defections that we see during cutting. 

Image above - When we simulate a braced design (in SolidWorks) we do see an improvement, reducing deflection by around 50%.

A more sophisticated approach could be to design a truss structure between the two arms of the wave-guide:

Image above - a bracing truss, designed and simulated by Ryan Schmidt at gradientspace, using 3DS Simulia software to simulate the deflections and resonant frequencies of the component. Once again, we can stiffen the part appreciably, and increase the resonant frequency to reduce the likelihood of tool chatter.

The problem with this approach is that we have increased to footprint of the part, impinging on space that could be occupied by other components, making the design less efficient overall. Plus, the extra stiffness that we achieve may still not be sufficient to make machining of the part viable using the basic work-holding method that we have chosen.

Step #5 - re-thinking work-holding

If we cannot easily change the part design without losing some of the benefits that AM has brought us, then we can look at other ways to hold the component during metal cutting. We need to find a way to distribute support across the component to minimise deflection and vibration, whilst not causing damage to the part as we clamp it against a hard fixture. A range of methods are available:

We can also consider an additive alternative, in which we create some encapsulating 3D-printed jaws. These distribute the clamping force across the metal part, reducing the risk of part distortion and surface damage. These also support the metal part closer to the machined features, reducing deflection and vibration:

Step #6 - modelling the new work-holding solution

When we run an FE analysis of the part held in the new jaws, we see some worthwhile improvements in rigidity:

There is not much more that we can do about the curved end of the wave-guide (analysed on the right, above), but the straight section (analysed on the left) can be better supported, as in this second iteration of the 3D-printed jaws:

Step #7 - getting ready to machine

So now we have our part fixtured firmly and securely, we are ready to start cutting metal. Or are we?

One of the major causes of machining scrap is poor machine tool geometric performance, in both absolute terms and relative drift over time. Errors arise when the linear and rotary movement of the machine's axes fall outside the tolerances necessary to manufacture an accurate part. This becomes critical where we need to hold geometric tolerances between datum features that are machined in different orientations, as is the case with our wave-guide.

We can use metrology on our machine to characterise its geometric accuracy, using a touch probe to check the linear and rotary movements. NC-Checker by metrology software products performs these checks and combines the results in one report, allowing us to confirm how accurate our 5-axis machine tool actually is.

Image above - benchmark report that combines the results of a series of checks, plotted against our specified machine tolerances. This can then be used to assess the machine in much more detail, or compare it against previous performance. In this case, one of the rotary axes on the machine is performing outside of specification, which could lead to geometric misalignment between features produced in different orientations.

A benchmark of our machine tool, taken before we start machining, gives us a reference point to help identify machine issues that could affect its performance. Over time this report can also show machine drift, and alert us when it reaches a critical point where maintenance may be required.

Step #8 - part set-up

If everything is OK with our machine tool, then we can start cutting, right? Not quite...

AM turns many aspects of manufacturing on their head, and one example of this is the way in which datums are generated. In conventional machining, we tend to create our datums first, and then use these features to align and position the part for subsequent machining operations. We cannot do this with AM, as precision datums have to be added in a final machining operation after all the other surfaces have been generated.

The challenge with setting AM parts is therefore to set up the part by taking account of the actual shape that has been built, so that we can finish it successfully. Essentially this involves understanding the material condition of the part in all the regions where we plan to cut precision features, taking account of planned stock allowance as well as unplanned part distortion. We are seeking an alignment of the part that leaves enough material in all of these locations to allow consistent and effective cutting.

Once again, we can use probing to achieve this. Metrology software products' NC-PerfectPart software provides a range of multi-point alignment options, taking account of the actual material condition to find a 'best fit' set-up for finish machining.

Image right - probing points for a 'best fit' setting cycle for the end flange of our wave-guide (shown in red). This involves an initial alignment using the top face of the flange and points around its edges. An alternative alignment could be performed using the interior surfaces of the wave-guide channel.

The initial alignment process is iterative, using the material condition of all of the measured points to find a multi-axis alignment and position offset that allows for the most consistent cutting conditions. The image below illustrates the material condition as initially measured (shown left), and the optimised stock once the best fitting process has been performed:

The second stage of the setting process involves probing the six holes in the end flange to find the optimal position offset to enable cutting of the hole pattern with the most consistent cutting conditions:

Another way to perform complex part setting for finish machining is to use a shop floor programmable gauge to measure the part and perform the alignment. This approach is best suited to higher volume applications, where automation of the machining process is advantageous.

Image right - a topologically optimised AM door hinge component is measured on an flexible gauge to generate a 5-axis alignment for subsequent machining. The alignment is automatically transferred to the machine tool controller before cutting begins.

Step #9 - machining

At last! With the part optimally fixtured and aligned on our 5-axis machining centre, we are ready to cut the datum features.

The resulting component has critical dimensions that are within tolerance and shows good surface finish. Tool chatter and wear are much reduced compared to our earlier machining trial.

Summary

Precision machining is often the 'finishing touch' in an additive manufacturing process chain. It is a high stakes process - if we get it wrong we might scrap a valuable component. And it is challenging because functionally optimised, light-weight AM parts may not be very rigid and often need supportive work-holding during finish machining. Their complex shapes can require equally complex fixtures - 3D-printed encapsulating jaws can provide a good solution.

Metrology is essential when machining to tight tolerances, particularly where part distortion is a factor and when geometric tolerances must be met. Probing on a machine tool or on a shop floor gauge enables complex alignments to ‘find the good part inside’, catering for part distortion and making finish cutting conditions more consistent.