Typical construction errors and how to avoid them: Error #3

This is a continuation of my blog about typical construction errors.

Screw Connections

Screw Connections


Error #3:
Screws on bending load

Why do screw connections fail?

The following generally applies for a wide variety of standards:

The strength verification of a component is divided into two steps:

  1. Static proof:
    Consideration of maximum stresses
  2. Dynamic proof:
    Consideration of stress changes (stress amplitude) during load changes.

Maximum stress and stress amplitude are compared with respective permissive values.

Static permissible stress value arises from the yield strength or tensile strength multiplied by a reduction factor.

The comparison value is dynamically the permissible stress amplitude, e.g. the bending fatigue strength multiplied by a reduction factor.

Screw connections fail statically, when the force leads to an excess static stress. This is the case when for instance, the screw is tightened excessively or the load is too high for the screw.

Dynamically, a screw fails when the stress amplitude is too high.
A rule of thumb here is that the stress amplitude should be less than 40 N / mm².

A wrong screw design often reveals itself after a certain period of operation due to screw breaks.

A screw breakage may be caused, due to the fact the screw has been designed too weak for the acting loads. More often however, the reason for screw breaks is that the preload force has been reduced. This increases the strain amplitude.

Among other things, possible causes can be:

  • Setting the screw
  • Loosening
  • Creep
  • Too small clamping length
  • Vibrations
  • Temperature changes

The individual points are described in the relevant standards and solutions are also proposed. For example, see http://www.schrauben-lexikon.de/td3-werkstoffe-stahl.asp(German) for basic considerations on screw connections.

However, I often encountered the following construction error in cases of damage:

Although the screw connection is designed to withstand tensile and shear forces, they will be additionally loaded by bending due to the design.

These additional bending stresses are however not taken into account.

The following image should illustrate this case:

Schraubverbindungen mit und ohne Biegung

Screw connection with and without bending

The flange is very stiff, the screw hardly experiences any bending, but the thinner the flanges become, the higher the bending stresses.

Usually a screw connection is made so that the pure pretension force causes a stress of 70% – 90% of the yield point.

Applying a bolt of strength class 8.8, the yield strength is about 640 N/mm².

However, if the flanges bend, the yield point is quickly exceeded by the additional bending stress in the usual design that only sees the normal force of the bolt. The screw deforms plastically at single or repeated exposure, the preload drops and the screw sees higher stress amplitude due to the reduced preload force. Under varying load, the creep strength is reached very quickly and the screw breaks.

These effects are only visible on closer inspection E.g. a finite element analysis. However, when this is considered in the framework of a damage analysis, it’s too late.

For standard flange connections, it is often not possible to conduct the proof of strength on a closer inspection of all relevant effects. Why there are less problems in reality, lies on the usually limited number of load cycles.

However, if this construction principle is adopted for dynamically heavily loaded components, such as a pump cover, you often see unpleasant surprises.

The principle that the screw should not be subjected to bending still applies.

If this is not otherwise possible due to its design, you should look more closely for instance at the stress by way of an FEM calculation.

I am looking forward to your feedback 🙂

Stefan Merkle

PS: In the following video you can see the FEM simulation of a screwed connection in which no bending occurs because the flanges are superimposed.

➞ Read part 4 of this article