Material Selection for Plastic Design
Correct material selection is a key factor in eliminating warranty and recall risk from delayed cracking in plastic products and components
Today there is significant commercial pressure to design, manufacture and introduce products into the market place in the shortest possible time frame giving the maximum level of margin and a longer lead time before the competition erodes margins.
Finite Element Analysis (FEA) in conjunction with rapid prototyping, rapid tooling and flow analysis is rapidly decreasing time to market of products by enabling more “right first time” products and eliminating the need for expensive and, more importantly, time consuming tool modifications.
There is an increasing pressure on engineering functions to improve on development times in order to beat the competition to market. All this being said, a plastic material should be selected for robust design and a structured approach to selecting the right material is recommended.
Failures due to hasty material selection are very common in the plastics industry and appear to be related to a lack of awareness and understanding plastic properties.
There are over 90 generic classes of plastic, this can be broken down into approximately 400 sub-generic modifications and finally ~ 50,000 commercial grades from a host of suppliers.
In order to approach the material selection logically we need to consider the basics of polymer structure and polymer properties…
Polymers can be broken down into three main groups including Thermosetting Plastics, Thermoplastic Plastics and Elastomers (Rubbers). Since we are only considering plastics products, we have two basic choices, Thermoplastic or Thermoset. However, there are also two basic types of thermoplastic – amorphous and semi-crystalline.
Amorphous plastics, as the name suggests, have no ordered structure whereas semi-crystalline plastics have areas of order which form crystallites. The level of crystallinity in these semi-crystalline plastics can vary between ~30% to 96% crystalline. Anything less than 30% is considered amorphous. As a general rule, amorphous plastics exhibit better creep resistance (constant static load over time) whereas semi-crystalline materials are better in fatigue than amorphous plastics. Amorphous plastics are best used where the following properties are required…
Good transparency, low mechanical abuse and low/no chemical contact.
Semi-crystalline materials are best used where chemical contact, mechanical abuse and repeated/ cyclic loading is required.
The problem is that approximately 70% of plastics fail before their design lifetime!! Why? Because historically, designers have been accustomed to working with metals and other materials, which exhibit a linear elastic stress-strain relationship, which is predictable.
The problem with all polymers is that they are visco-elastic materials. visco-elasticity is defined as the tendency of polymers to respond to stress as if they were a combination of elastic solids and viscous fluids. Consequently they exhibit a non-linear stress-strain relationship and their properties depend on the time under load, temperature, environment and the stress or strain level applied.
When using FEA for elastic materials such as metals, the system asks for the modulus (stiffness) of the material, it’s easy, you just put in the tensile modulus of the material (depending on the loading scenario) and the FEA does the rest. When designing with elastic materials, designers can rely on instantaneous stress-strain properties, and for most applicationbs can disregard the effect of temperature, environment and long term effect of load (creep). However, most plastics products are designed to last more than 20 seconds (the average time it takes to generate a stress/ strain curve with a tensile machine) and you may want your product to last 20 years under load at 40°C in a chemical environment). Putting the tensile modulus in the FEA programme at 20°C will guarantee your product is under-designed, as the apparent modulus at 20 years will be much lower than the instantaneous modulus value. In addition, the stiffness at 40°C may be significantly lower than that at 20°C You may be looking at premature failure and the costs associated with that, product recall, product replacement, loss of brand credibility, and re-tooling etc.
What is needed is the creep modulus at the temperature, in the environment at the stress level initially calculated.
All unreinforced thermoplastics exhibit significant creep properties. Creep is the continued deformation of a part, over time under a static load. It is a function of the viscous nature of the polymer. (think of polymers having a treacle portion to their properties and you’ll get the drift)
As can be seen from the graph above, the creep of the plastic is not linear over time and doubling the loading does not double the level of creep. (It is non-linear and time dependent therefore it is double and “then some”, and the “then some” increases over time). This is because polymers are visco-elastic and time dependency needs to be taken into account, take this polypropylene for example, the creep modulus decays rapidly under this load to less than half of what you calculate from a tensile modulus short term curve.
(86,000 seconds = 1 day)
This is the modulus value that the FEA package is looking for. The luxury with metals is that, to all intents and purposes, this value will not change over time. With polymers we have to measure it. The other problem with FEA is that the level of stress may look to be low compared to the short term tensile strength of the material, things are obviously looking good for the product. Don’t be fooled! As with creep, dynamic fatigue/ cyclic loading will cause the polymer to fail significantly below it’s short term tensile strength and a ductile to brittle transition will occur.
With amorphous polymers this is even more important as shown by the following graph for polycarbonate below.
What this graph tells us is that in fatigue, the strength of polycarbonate at 1 million cycles is only ~10 MPa. This has fallen from 60 MPa reported from our short term tensile strength; a drop off of 83% !! With a design safety factor of two in our stress calculation we need to be operating at 5 MPa, one twelfth of it’s short term tensile strength!!
The polypropylene, which is semi-crystalline, has a 1 million cycle failure stress of 12 MPa, half of that of it’s short term stress. And a design safety factor of two makes it better than PC for fatigue.
This reduction in the properties may be off-putting. However, as long as you know what the material will be like in the time frame you want it to last, you can design around it.
The only unfortunate thing is that many designers assume that “it’s just plastic” and the performance is based on short term test data which is readily available in the manufacturer’s datasheets.
The main issue that we see at Rapra Technology is that many manufacturers/users are unaware of the long term properties of polymers. This results in the one thousand or so product failures we see every year because products have been designed without time, temperature and chemical environment being taken into account.
Designers must be aware that data sheets are only useful for comparing property values of different plastic materials such as the tensile strength of acetal versus acrylonitrile butadiene styrene. Data sheets should only be used for initial screenings of various materials. Data sheets are not produced to provide information for engineering design and final or ultimate material selections. Their information is merely derived from short term tests which do not take into account time and temperature. The data is also a single point measurement and does not take into consideration the effect of time, temperature, environment and chemicals etc. The test pieces are also simple in shape and moulded under ideal conditions. This rarely applies to moulded products.
When talking on the subject of chemical environment, we also must consider environmental stress cracking (ESC). ESC is premature initiation of failure and apparent embrittlement of a polymer under the simultaneous action of stress / strain and the environment. ESC differs from chemical resistance in that chemical resistance testing uses specimens immersed in the chemical for a time period with a measurement of properties before and after exposure (typically tensile and impact properties). However, at no point does the specimen come under any stress except after it has been cleaned ready for test. With ESC, the presence of both a stress and the chemical environment can lead to dramatic effects, mainly catastrophic brittle fracture of even “tough” materials.
Amorphous plastics are, in general, more susceptible to ESC than semi-crystalline materials. But we would always recommend testing in the chemical environment. For more information on ESC, there is a free website devoted to the subject, www.esc-plasticsmold.wiki which covers many plastics and environment combinations.
Many plastics products are stressed in service due to moulded-in stresses from the manufacturing process alone (injection moulding, extrusion , thermoforming etc), in addition to thermal stresses and applied loads. It is critical that the potential for ESC it not ignored by the designer!!
Rapra has been working with many companies to characterize their materials for long term design (out to fifty years using time-temperature superposition) and, when you consider the cost of mould tooling or tool redesign, let alone product recall it is a small price to pay for peace of mind.
Rapra have fully equipped long term creep and fatigue laboratories to operate at extremes of temperature and in aggressive chemical environments.
So when the design engineer asks for the mechanical characteristics of the material, you know to ask “After what time frame, at what temperature and in what environment?”
As a final note, you need to remember the following points…
1. DYNAMIC FATIGUE
3. TIME and TEMPERATURE
4. ENVIRONMENTAL STRESS CRACKING
5. COMPARISON OF EQUIVALENT DATA
Make sure you understand that the data presented is comparable between suppliers and relevant to your product’s application.
If in doubt, generate real test data at the temperature, in the environment that your product will see in service. Typically a 1,000,000 cycle fatigue S-N curve or a 1000 hour creep curve will take six weeks, either/both are worthwhile planning into the development phase of your new product development plan in order that you have no surprises in service.
Rapra can assist with your design development to ensure you’re your product performs in its given application and does not require the services of our failure diagnosis department. Rapra can provide material selection, FEA, mould flow, long term property generation and lifetime prediction services. For further information please contact:
Dr Chris O’Connor – Technical Solutions Manager
(Plastics, Rubber & Design)