The evolution in polymer technologies has ensured that there is virtually no application where polymers do not provide some benefit. Polymers are increasingly being turned to in areas that require high-stress resistance combined with light weight. As such, metal replacement is being done across industries, facilitating the need for extensive testing in order to prove the polymer can suitably sustain itself over the long term.

The rise of 3D prototyping has had many believe that conventional methods of sample development would soon be phased out. In fact, many who do not fully understand the limitations of 3D printing have even suggested that conventional production methods could soon be done away with, as the speed of 3D printing is augmented to match the throughput of existing processes.

While 3D printing is no doubt a useful tool, it suffers from limitations. For one, the layering process – while very effective in making certain kinds of shapes – does not lend itself of creating a polymer structure with adequate tensile properties. Especially in the direction of the layering, the part will always be inferior to a polymer that was either cast whole from its molten form or compressed at high loads and then cured under heat. Another issue is that many polymers – such as PEEK – are subjected to internal stresses that do not fully release until the heat on the part crosses the glass crystallinity point. Conventional machining processes make allowance for these stresses. Parts are often annealed mid-way through the machining operation, allowing the stresses to be removed so that the part itself does not suffer deformation while under application.

Another limitation of 3D printing relates to the rise of 3D software technologies. While earlier it was assumed that 3D prototyping allows a better understanding of the fitment of the part into the assembly where it would be used, we can now use 3D software to get the same idea. 3D software not only allows for the fitment to be assessed to a very high degree of tolerance, but also allows the stress and load parameters to be applied, giving a full understanding of whether or not the part will deform during use. This, in effect, negates the need for 3D printing as a whole and allows for a part to be fully proven before moving into the sample development and bulk production phases.

Given that the conventional production processes are here to stay, it is worth exploring them further to understand the strengths and weaknesses of each. Primarily, polymer parts can be made either by machining or by injection moulding.

Machining involves

As we will see, each has its benefits and limitations. We will compare these processes along some of the key parameters.

1. Complexity
While machining does allow for some degree of complexity, more often than not, very complex parts need to be injection moulded. Typically, contours, fine radii and a high-gloss finish can only be attained using injection moulding. While machining programmes employing CAM software can be used to achieve intricate dimensions, the equipment needed and the extra time it would take to attain this would make the process costly.

2. Cycle time
In general, the cycle time needed during machining will increase along with part complexity. More dimensions imply more operations and parts can sometime take in excess of an hour to machine, if the dimensions are varied enough.

In comparison, injection moulded parts would rarely vary in cycle time as complexity increases. Usually, an injection moulded component would take only under a minute to mould, and this time would only increase if the part weight was increased.

3. Volume
While machining can be done at high volumes, truly large-scale parts usually rely on injection moulding. As described above, a cycle time of only a minute may be needed for an injection moulded part, as compared with machining that can take as little as under a minute but go all the way up to over an hour.  As such, the cost differential between injection moulding and machining can be as much as 5-10X – considering the cost per minute to run the machines are somewhat comparable.

4. Tolerance
Tolerance is one of the areas where machining truly gains an advantage. Machining tolerances go down to as close as 10 microns and ensure that this tolerance is held over a large batch of components. In comparison, injection moulding would usually not commit to such close tolerances. One key difference between the processes is that with machining, the drawing is King. Any dimensions given on the drawing (within reason) are set in stone and must be achieved if the part is to be approved. In contrast, injection moulding will try to come to as close as possible to the drawing dimensions. Any minor deviations are then submitted for review and the drawing is re-created accordingly. Usually, this is because the cost of re-making/ re-working the tool may be high and/or infeasible. So the end-user takes a call based on fitment and the part can be approved.

5. Polymer
Other than cycle time, the key cost comparison between moulding and matching rests on the amount of polymer used. Machining is a process that starts with a stock shape (a rod or a sheet) and removes that which is unnecessary to make a final part. In contrast, moulding is additive. We create an empty mould and fill it with just the right amount of molten polymer needed to make the part.

In the case of more expensive polymers, this cost saving from lower consumption can be quite significant – especially for high-volume parts.

However, there are some polymers that simply cannot be injection moulded. PTFE and UHMWPE – to name just two such polymers – can only be made from compression moulding. These polymers have no melt-flow, meaning that even at their melting point, they do not become liquids, so that putting them into a mould would be infeasible.

6. Development

While injection moulding is usually the cheapest way to make a part (owing to a lower cycle time and a more efficient use of raw materials), where it is expensive is in the development.

A machined component can be developed for practically nothing more than the time and effort it takes to programme the part into a CNC machine. Injection moulding requires a die, which is expensive to make and can sometimes take weeks to complete. It is mainly for this reason that most prototype applications prefer to get the parts machined and only look towards moulding when the volumes increase. Even with increased volumes, the cost of the mould needs to be amortized over the expected volumes the mould can produce. This is then added to the part cost to compare whether or not there is any saving from moulding the component.

Once all these factors are weight against one another, an informed decision can be taken on the process to follow.

PTFE sliding bearings have become a mainstay for any application that requires a combination of loads and movement. The key feature of the bearing is that it can accommodate high vertical loads, while offering a low coefficient of friction. As such, the standard model of PTFE against polished Stainless Steel allows for coefficients of friction as low as 0.03 (near rolling friction) while being able to sustain vertical loads of up to 400Bar.

The applications of this sliding arrangement include pipe support systems, boiler plants, bridges, and walkways. The plain PTFE sliding bearing does not lend itself to loads exceeding 150 Tonnes, hence the application does not extend to areas such as flyovers and railway bridges, where the loads are in excess of this value.

Design

Designing a plain PTFE sliding bearing involves an understanding of the various loads involved and ensuring that the PTFE does not succumb to these loads, even under their most extreme applications. For the most part, while PTFE is rated to a compressive load of up to 400Bar, we apply a 50-70% safety factor in design, keeping the design load within 120-200Bar.

Other loads such as lateral loads, uplift loads, and even rotation, can be factored into the bearing using metal bolts that can accommodate these loads. Care, however, must be taken in ensuring that PTFE is used in any area where there would be sliding. So, if metal bolts are used to prevent movement in one of the lateral directions, a PTFE element is employed around the metal bolt, such that the bolt itself does not rub directly against the sliding plate thereby increasing the friction.

One of the main design arguments that result with PTFE bearing is around the thickness of the PTFE pad employed. The thickness of the pad has long been specified as being between 3mm and 5mm as per the BS 5400 standard – which remains one of the only standards that pertain specifically to the design of PTFE sliding bearings. The standard recommends that the PTFE sheet should be 5mm in the care of a recessed bearing (a bearing where the PTFE pad is bonded within a recess on the metal plate) and 3mm in the case of an un-recessed bearing. The case for making the recess is quite sound. Bonding the PTFE within a pocket milled onto the surface of the metal plate ensures that the PTFE does not come loose of the metal over time. Barring cases where the shear load on the PTFE is likely to be low, a recessed option is generally preferred by designers.

Nonetheless, a lot of debate does arise as to why the values of 3mm and 5mm are found acceptable.

Shear Loads

Recently, we were asked to supply 8mm thick polypropylene sheets for a bearing project. Among the technical requirements was a shear load confirmation. Against a given horizontal load, we needed to confirm that the sheets would not deform, causing failure.

Given the Shear Modulus (G) of the material, the above formula can be used to calculate the value of F needed to displace the sheet. Since this was an infrastructure project, we assumed Δx to be 0.1mm (a very small displacement) and given the other properties, the force was calculated.

In our particular example, a force of 750Tonnes would be needed to displace the polypropylene sheet by 0.1mm. Considering the actual loads expected were within 10 Tonnes, this safely allowed us to state that the bearing sheet would not be affected by the shear load.

The more interesting aspect of this, however, is the function that l (the thickness of the sheet), plays in the calculation. For the same shear strain, a higher value of l would imply a higher value of Δx. In other words, increasing the thickness increases the risk of deformity.

Thus, the sheet thickness – whether it is polypropylene or PTFE – needs to be minimized in order to minimize the effects of shear. However, too thin a sheet would risk damage from tearing and/or wear out. Furthermore, grit that finds its way between the bearing plates would be more likely to cause damage. PTFE normally swallows the grit, allowing it to stay embedded just below the sliding surface, so that it does not compromise the sliding movement. If the sheet was under 3mm thick, larger pieces of grit would cause problems.

Because PTFE has such a low coefficient of friction, shear loads are not normally considered an issue. Whatever the thickness of the PTFE, it would require a tremendous load to deform the sheet due to shearing. Nonetheless, the shear load is worth keeping in mind, especially when a design engineer argues that 3mm or 5mm is too thin and that a higher thickness of sheet is needed to accommodate the load.

When manufacturing machined components, the key challenge is to ensure the consistency and repeatability of the dimensions and tolerances. With metals, this challenge is not so acute. There is a wealth of industry experience around the machining, punching and shaping of metals and therefore enough data to support how the material will behave when subjected to external forces. Metals also adhere to much closer tolerances. Precision metal part manufacturers will frequently cite tolerances of as low as one micron (1µm), which are unheard of in the polymer space. In addition to this, polymers are not as well traversed as metals and manufacturers who develop specific methods to ensure close tolerances will usually keep these as proprietary. Furthermore, each polymer behaves in its own manner and as such throws up its own limitations to what can be allowed by way of dimensional accuracy. A few such examples are given below:

1. PEEK – tends to absorb stress during machining and can deform post machining (sometimes after days of being kept on the shelf), if not treated properly
2. PTFE – does not absorb heat from the tool during machining, but can experience thermal expansion and its softness, can make it difficult to achieve close tolerances
3. Nylons – can be affected by the heat from the tool, causing melting at the part’s surface. At the same time, coolant is not an option as the material tends to absorb moisture and swell post-machining. Methods need to be developed to balance the machining RPM with the feed rate to ensure tolerance and part finish are both achieved

The above examples are only a few to illustrate how each polymer behaves. There are literally thousands of polymers grades and each would come with its own set of properties to influence how it must be handled.

Non-contact inspection

One feature of most polymers is their softness when compared with metals. This is especially true when considering a thin cross section, where the part will deform even when held by a human hand.

For such thin sections, traditional inspection tools – such as callipers, micrometres and probes – do not work. The mere act of holding the part and applying the pressure of the tool will throw out the dimension, rendering the process useless. To remedy this, non-contact methods, such as profile inspection and vision inspection were developed.

Vision inspection is not only useful for non-contact inspection, but can also be used to inspect radii, chamfers, angles and other unique profiles that gauges and tools cannot be easily developed for. In addition to inspection, the process also allows for the dimensions to be captured on the part and a visual to be created to be send to the client or end-user as proof of the dimensional accuracy.

Inspection of complex profiles

More recently, we encountered a set of parts which can only be inspected by sectioning the final component and using the vision inspection system to capture the dimensions. Obviously, once the part is sectioned (cut through), the part itself is destroyed.

Given the size of the parts, this kind of destructive testing can be cost prohibitive. Furthermore, as the parts were made from an especially expensive PTFE-based compound, the challenge to inspect and approve the FAI parts before beginning manufacture was proving to be an expensive one. To be absolutely sure that the dimensions did not deviate during bulk manufacture, it would be optimal to perform this destructive test at the beginning, the middle and the end of the machining cycle. As such, a batch of say 50 such parts would require the destruction of 3 parts to ensure the dimensions were within tolerance. This represents a wastage of 6% - which is high and very costly.

To remedy this, we made three suggestions:

1. We machine the parts needed for destructive testing from a polymer that is inexpensive. When making both PTFE and PEEK components, we usually machine a few parts in POM (Delrin) first. POM is dimensionally very stable and mirrors the behaviour of PTFE and PEEK to a high extent. Once we are happy with the dimension, we shift over to the more expensive polymer
    
2. The second option was to machine the FAI part in pure virgin PTFE, which would most closely match the behaviour under machining of the PTFE compounded material. This would work out more expensive than option 1, but still cheaper than destroying the compounded parts
    
3. The final option was to ask the client to specify the percentage of parts they needed tested and work this back into the commercials. In such a scenario, we would need to define a minimum batch quantity – say 40 parts – and come to an agreement that 1 out of the 40 would be checked, leading to a 2.5% increase in the costing

Eventually, because of the sensitivity of the part and the stringent need for quality, it was decided that option 3 would be used.

The client was not comfortable with option 1. This is understandable, since as we have described above, no two polymers behave exactly alike. While we may be comfortable using POM as an approximation for PTFE and PEEK for internal dimensional verification, the same cannot always be considered accurate for bulk manufacture.

In order to better map the behaviour of POM with PTFE and PEEK, we will be undertaking a correlation study to be done over 3 months across 200-300 parts. Once the data is collected, it would put us in a better position to recommend option 1 to clients in the event that cost restrictions limit the level of destructive testing allowed.