Why do components fail? If you were to think why a bridge, an aircraft turbine blade or even a bicycle pedal might fail you would probably think of an unexpectedly large force that pushed a component beyond its yield stress or even its ultimate tensile strength. Whilst these extreme events can happen components often fail when they seemed to be operating well within their design limits. Two things that can lead to unexpected failure are metal fatigue and creep. Both of these are gradual processes that can take a great deal of time to show themselves. They can however both end in catastrophic failure.
Creep isn’t really a failure in itself, rather it is a permanent, time dependant deformation of the material. Unfortunately in many applications a permanent deformation of a component will lead to failure of the total structure; a turbine blade that has become distorted is not going to help an aircraft engine work any better!
We tend to think of solids as having a fixed shape and a sharp melting point. When they reach that point they become liquids and are then able to flow. The distinction however is not as clear cut as you might think.
If a material (including metals and ceramics) is subjected to a large stress that is none the less below its yield stress then instead of remaining at a fixed strain it will gradually stretch further. The closer the stress is to the yield stress and the higher the temperature, the greater the effect will be. You can check this with the following experiment:
Displacement Sensors or Metre Sticks.
Set one test wire up as shown in the diagram (the power supply need not be connected yet):
Add the masses one by one and measure the extension. Continue this until the material has clearly exceeded it yield stress. Use this information to construct a stress – strain diagram.
Repeat the set up with several identical wires. If you do not have enough space or equipment for this then you can run the experiment several times. Connect each wire to its power supply but do not yet turn it on. Each should be set to a different current including one that has no supply, just the connecting wires (this is the control). Load each wire up to 90 % of its yield stress. Measure the extensions and then turn on the power supplies. Measure the extensions again regularly, every minute to start with. After a while the frequency can be reduced. Ideally the measurements would be made automatically with data logging equipment since this experiment can last for many hours. If you have the necessary equipment available, measure the temperature of the wires. A thermocouple would be ideal for this.
As an extension to the experiment you could investigate the effect of stress on creep. For this experiment you could still make use of the power supplies but all the samples would need to carry the same current and so be held at the same temperature. An elevated temperature would allow stress dependant creep to be measured more easily.
When you analyse the results of the principal experiment you should find that the greater the current, the greater the extension became. This is because the currents heated the wires and this produced a greater rate of creep. For each wire the creep should follow the pattern shown below. There is a short ‘settling in’ period, during this time existing defects in the grains allow quite a lot of movement. Then the sample undergoes a slow and steady increase in its extension. This is the main section of creep that engineers consider most when they are designing a component or a structure, any remaining defects are ‘pinned’ against each other or against impurities and any new movement depends on random, thermal motion in the grains freeing these defects up. Eventually the rate of creep increases again; you may not have seen this section of the graph with your samples but when it occurs the sample is not far from breaking.
If you were able to measure the temperatures of the wires you would be able to plot the graph of creep rate against temperature. Even if you did manage this it would be hard to identify a precise mathematical pattern as the relationship is a little complex. The actual relationship is given by:
There are several terms in here that depend on the material and the specific type of creep that is happening (there is more than one). The most important terms are σ and T, the stress on the sample and its temperature. Clearly, increasing either temperature or stress will lead to increased creep rate.
Experience has shown that creep can be detected in many metals when reach just a third of their melting point temperature. As the temperature needs to be measured in Kelvin this means that metals can be susceptible to creep at surprisingly low temperatures. Generally speaking an engineer may be concerned about the level of creep when the operating temperature reaches 50 % of the melting point of the material. The greater the stress the material will be subjected to the lower the temperature of concern will be. Ceramics do not normally start to show any signs of creep until they exceed 50 % of their melting points.