3D Strain Based Fatigue Analysis

3D Strain Based Fatigue Analysis

Designing your products to survive the maximum load they see in service may not be enough, even if you include a significant safety factor. If the loads vary during usage, your product may become subject to fatigue damage over time, failing in a completely unexpected way. As metallic components are loaded, there is a small microscopic amount of plasticity. This may occur due to base material imperfections, welds, or at discontinuities caused by the machining or manufacturing process. 

Predicting fatigue failure is possible through analytical modeling. Fatigue analysis can be divided into two main categories: stress based, or strain based. With stress based fatigue, a Stress-Life (SN) curve for the material provides a mapping for the amount of damage caused by individual load cycles. Using a special technique called Rainflow Cycle Counting, a load sequence can be broken down into stress cycles, each causing its own damage. By accumulating the damage, it is possible to make a prediction of the approximate life of the component. Stress based fatigue analysis is essentially a linear technique, and only provides accurate results when a component has a long life (otherwise known as High Cycle Fatigue). Most CAD companies sell FEA packages that include Fatigue Analysis. However, in most cases, they are only stress-based models.

Strain based fatigue analysis is a non-linear method, where the elastic and plastic strain components are individually accounted for. The elastic and plastic strains are calculated at critical locations where failure will eventually occur, based on material properties in as-machined, as forged etc. conditions. Damage is then calculated based on the hysteresis loops on a stress-strain curve, and the life of the component can be accurately predicted.

Fatigue has traditionally been calculated using two dimensional methods. Nominal stress or strain is used to calculate local stress or strain using concentration factors. For strain-life, a method known as Neuber Analysis is used to calculate the elastic and plastic strain at the root of the failure location. For this type of analysis, it is necessary to have an a-priori knowledge of the location of failure, so that the appropriate concentration factor can be applied.

Nowadays, we can use Finite Element Analysis (FEA) to calculate the stress and strains for an entire component. This removes all the guesswork, and provides a full-field picture of possible failure locations, and can give a a 3 dimensional damage plot, in much the same way that an FEA can plot von-misses stress. We can also take it a step further, and use non-linear FEA to calculate the elastic and plastic strains for strain-based analysis, completely eliminating the need for a separate Neuber Analysis.

Fatigue analysis is very complex, and accurately predicting life takes a lot of experience. For more information, refer to: https://re-test.com/fatigue-analysis-consulting














Pedro Morales

Multidisciplinary Stress Engineer

6 年

Agree. That is why a thorough understanding of this life estimate theory is necessary. I am still doing this by hand because I don't trust a model to calculate the stress severity factor accurately enough or the model representing the geometry/mesh refinement where failure may occur. Unfortunately I see many young engineers relay too much on models to even obtain data that can easily be calculated by hand.?

Simon Stedman

FEA Design Stress Engineer and Physical Testing. See Short PP on My Profile for Overview. NALtd Contract Engineer

6 年

I agree, also,? No 1, please ask purchasing not to buy dodgy bolts.... like the ones that stretch like bananas when torqued correctly with a thread lubricant. No mill uts, yield, hardness nor chemistry certificate.? No 2. Bad idea to use mech/fatigue props off the internet for strain/stress FEA/fatigue analyses. No 3. Fatigue scatter and statistics, min 12 samples for staircase fatigue, really 20 samples better, upper and lower bounds, etc. No 4. Any vibration/harmonic response in the real application? Stresses go thro? the roof. No 5. Any real physical test data at all? Is the complex FEA simulation even near the real product on the road/etc, in use? No 6. Just one element thro' the thickness in the FEA model??? whoops, starting point should be min 3 to 5 elements, hot spots need more mesh refinement. No 7. FEA simulation over constraint 'FIXED' -asking for trouble, FEA results all wrong, needs always appropriate constraint away from 'hot zones' to avoid masking real hot spots. etc, etc. Liked the Test Inc web site a lot, loads of real testing plus non linear FEA, great stuff. Simon

要查看或添加评论,请登录

Nigel Linden的更多文章

社区洞察

其他会员也浏览了