Академический Документы
Профессиональный Документы
Культура Документы
FE Analysis
Stress Response
Material Data
SN Curves
Stress Hypothesis
Safety Concept
Damage Accumulation:
Life, Strength Reserves,
Degree of Utilisation
Page 1 of 4
The diagram shows a typical fatigue analysis using the two above mentioned parts. The life of a component is
a function of these two parts. The stresses on the component are calculated using appropriate stress hypothesis
and then cycle counted to enable a comparison with applicable material data such as Stress-Life curves (SN
curves). Different safety and knock down factors are considered in these calculations before damage
accumulation is carried out to determine the component life and the degree of utilisation.
Component
Material Strength
50% P
97.5% P
5%
Safety Factors
50%
Probability Density
Component Stress
due to Loading
Stress
Based on FKM Guidelines [2], the concept of safety for fatigue is accordingly represented in the above
diagram. It vividly tells that even component stress due to loading (with a low probability of occurrence), needs to
be below a material strength value with high probability of survival, after consideration of all applicable knock
down and correction factors. The difference between these stress values is the safety margin and should at least
cover the fatigue safety factor requirements determined by the consequences of failure and the probability of its
occurrence.
3 Determination of material properties
Complexly varying loads call for fatigue analysis methods to consider the different phase relationships
between the loads acting on the components. These can cause variations in stress directions, usually termed as
multiaxiality. The problem is that the limitations of brittle materials, such as cast iron, to withstand multi-axial
stresses are a little known problem outside the research centres.
Fatigue strength of materials, such as SN curves for spheroidal graphite cast iron, is experimentally
determined on specially prepared laboratory test specimens and usually defined only for uniaxial stress states.
The uncertainties involved in quantifying this property are covered using statistical parameters such as probability
of failure, scatter, slope & knee of SN curve, level of confidence and the number of tests. Influence of surface
roughness, technical defects, wall thickness, mean stress and residual stress, notch effects are also appropriately
considered as knock down factors to determine the component material strength under fatigue loads for uniaxial
stress states. These factors cover all related known risks and a careful combination of these factors is necessary
to theoretically optimise load carrying capacity of the component.
This is the generally accepted procedure to derive component SN curves since it has been financially unviable
to conduct accelerated tests on full size components to gain the necessary material data.
4 Determination of component stresses
Because of the increased complexity of the component geometry and varying load scenarios, the local stress
approach in combination with detailed FE models and load time histories is the most widely used method in this
industry.
The finite element calculations are based on CAD models of component geometries with linear, isotropic
material properties and multiple loading points. The load histories for the components are extracted from multibody simulations of a wind turbine model for various fatigue loading scenarios. Linear superposition of the loadcoefficients (unit load stresses or transfer functions) determined from FE computations are multiplied with load
histories to get local stress tensor histories at critical locations. This is the generally accepted procedure for linear
calculations but if non-linear boundary conditions are to be observed, for e.g. contact surfaces of bearings etc.,
then modified approaches are necessary.
It is to be noted that reliable fatigue material properties exist predominantly for uniaxial stress states. Several
hypotheses also exist to enable conversion of component stresses to an equivalent uniaxial stress. There also
Germanischer Lloyd Industrial Services GmbH, Business Segment Wind Energy
Page 2 of 4
exist separate methods such as the critical plane approach to deal with multiaxial stress states caused by non
proportional loading conditions.
To help understand this, structural components of a 2 MW wind turbine are used as an example in this study.
Fatigue loads in combination with FE models of the components were used to compute the stress histories at
critical locations. Based on this, a local stress state analysis was performed to analyse the multi-axial behaviour
of the stresses at critical locations and investigate the degree of multiaxiality (i.e. change in principal stress
directions during a loading sequence).
M
U
M
U
U: Uniaxial Stress during loading
M: Multiaxial Stress during loading
The results show that at several locations in the hub there exist multiaxial stress states. At these locations, it
was observed that though the directions of principal stresses change greatly, the magnitude of the stresses
remain low. The main frame on the other hand, has several locations with uniaxial stress states. Following are
plots showing samples of typical uniaxiality and multiaxiality.
The procedure for fatigue analysis needs to consider the influence of these changing principal stress
directions by applying appropriate stress hypothesis and quantify its detrimental effect on fatigue strength of the
chosen material. This is critical since the used materials show an additional cyclical stress hardening effect with
varying principal stress directions. Fatemi has even indicated that out of phase stresses are more damaging than
equivalent in-phase stresses in the low cycle region [7].
The appropriate stress hypothesis is to be selected based on several factors:
- Uniaxial or multiaxial stress state
Germanischer Lloyd Industrial Services GmbH, Business Segment Wind Energy
Page 3 of 4
Page 4 of 4