Laboratory for Mechanical Reliability
Faculty of Industrial Design Engineering
Delft University of Technology
Leeghwaterstraat 35
2628 CB Delft.
A method for determining fatigue damage in (short) Fibre Reinforced Plastics (FRPs) has been developed and evaluated. Fatigue damage is measured by cutting thin foils from the fatigued specimen, and tensile test these. Both strength and fracture strain of the foils change with fatigue, the latter showing a more distinct change. The foils are cut using a microtome, which influences strongly the material. This renders it impossible to use the method for quantifying damage.
To be able to relate the damage to the fatigue process, the fatigue experiment must be stopped at a known percentage of the lifetime of the specimen. Accuracy of the prediction of the fatigue lifetime can be improved by at least a factor two by measuring the creep (increasing elongation of the specimen) during the fatigue experiment. The creep speed shows a distinct correlation with number of cycles to failure.
An adaptation to existing failure models is presented, based on the creep speed and damage measurements, especially the observation that cracks in the material remain bridged (by fibres and/or drawn matrix material), following from the fact that no foils were found with strength zero.
A characteristic of Fibre Reinforced Plastics (FRPs) is their high degree of anisotropy, caused by the fibre orientation. An injection moulded FRP plate has a layered structure. Orientation in the layers as well as the thickness of the layers vary from location to location in the plate, and therefore the strength of the material varies throughout the plate. Consequence of this is that the strength in a tensile experiment of specimens cut from a plate can vary between 100 and 160 MPa, depending on the location from where they were cut, and the direction of the axis of the specimen relative to the Mould Flow Direction (MFD). Modulus of Elasticity of the specimens vary approximately to the same degree as the strength.
Earlier experiments [1] showed that the fatigue strength (stress at which a certain fatigue lifetime is obtained) of a specimen is directly proportional to its Ultimate Tensile Strength (UTS), determined in a tensile experiment. In Fig.1 Wohler curves are shown for different specimen types. Normalization of the fatigue stress by the UTS of that specific specimen type leads to a coinciding of the curves for the different specimens (Fig.2), to a "master curve". A similar dependence of fatigue strength (in this case maximum stress intensity at which a certain crack growth speed exists) with UTS was found for crack growth experiments. Crack growth curves for specimens with different orientation distributions coincide when a kind of "normalised Paris plots" is used, with the stress intensity divided by UTS. Wyzgoski and Novak [2] used the Strain Energy Release Rate to obtain a similar normalization for cracks parallel and perpendicular to MFD, using Elastic Modulus. Deviations from these "Master curves" occur only when different thicknesses of specimens cause a difference in heat build up in the specimens through hysteretic heating [1].
In general the fatigue strength and the tensile strength (UTS) of a specimen are related. It can be stated that the fatigue and the tensile strength both depend in the same way on the orientation distribution inside the specimen. This despite the different appearance of the fracture surfaces (determined using SEM microscopy), indicating strongly different mechanisms in fatigue and tensile tests. In fatigue the Polyamide matrix is highly deformed, and the matrix is debonded from the fibres, while in a tensile test the matrix is cleaved and matrix material is visible on the fibre surface. For use of the "Master Curves" mentioned before, in predicting fatigue behaviour of structures, a knowledge of the conditions (temperature range, humidity, thickness, fibre fraction, fibre aspect ratio) for which the Master Curves are valid is needed. Goal of the current investigations is to understand the failure mechanism of the material, to be able to determine the parameters that influence the Master Curve, and so assess the validity domain of the Master Curve. The complexity of the material leads us to begin with the determination of the location (in the thickness) in the specimen where damage occurs first. Therefore the changes (decrease in strength and changes in fracture strain) in the material were investigated using a newly devised MicroFoil Tensile Test (MFTT).
The material used was Polyamide 6 containing 30%wt. of glassfibres; Akulon K224-G6, provided by DSM, the Netherlands. Square plates of 100x100mm2 and 5.75mm thickness were injection moulded from this. The mould was injected through a line gate, to obtain a straight flow front. For fatigue and tensile experiments non standard dog-bone type specimens were milled from the plates, using a Roland PNC-3000 Computer Aided Modelling Machine. The location in the plate and identification and tensile strength of the types of specimens used are shown in Fig.3.
The fatigue experiments were carried out on a servo-hydraulic MTS 810 bench. The load frequency used was 1Hz, to avoid unacceptable temperature increase due to hysteretic heating and thermal failure of the specimen. Earlier experiments [1] showed the high sensitivity of the fatigue lifetime of this particular material to the test frequency, caused by hysteretic heating as a consequence of the high damping of the material. During the experiments the temperature at the surface of the specimens was measured using an infrared contactless thermometer.
The minimum to maximum load ratio R was 0.1. Experiments were executed at various stress levels, referred to as a fraction of UTS: A fatigue experiment at 0.7UTS is an experiment with a maximum stress of 70% of the tensile strength. Tests were carried out in an environmental chamber at a temperature of 23degreesC and at a relative air humidity of 50%. Specimens had been conditioned by exposing them to laboratory air for at least 1 year.
During the fatigue experiments the displacement of the grips was monitored, enabling the calculation of the Modulus and Creep. In Fig.4 it is shown how these are calculated from the creep curve. The energy dissipation was calculated from the area of the hysteresis loop. It had already been shown [3] that especially the creep speed Vc (increase in grip-displacement during secondary creep, per cycle) presents a good correlation with lifetime N (number of cycles to failure) and can be used to predict the lifetime of a specimen while it is being fatigued. The experiments presented here were done to prove the correlation on a much more substantial amount of data.
The assessing of damage was done by making strength profiles which were obtained as follows: First miniature test specimens are milled from the specimens subjected to fatigue. From these miniature samples, with a width of 2 mm, 85microns thick slices are cut parallel to the surface, using a Leitz microtome. The procedure is outlined in Fig.5. The fracture strength and fracture strain are measured for each slice, using a miniature tensile test machine. Plotting strength or fracture strain to the distance to surface gives the required profile. To make the profiles more clearly visible each measurement was replaced by the average of the strength at that point and that of four of its neighboring points, which doeWINWORD6INI .! SHED INI π2 M DESCRIPTION" fa!m: TELNET LOG (? AUDIT LOG ^Q" TELNET~1LOG E? STATUS ME (K,? CANYON MID `A[ CHKLIST MS E[n U NET MSG `a' NETH MSG `a SYSTEM OLD 4 PROGMAN OLD Ok HPDSKJETP00 z HPDSKJETP01 j^" ! HPDSKJETP02 B j HPDSKJETP03 v SPART PAR sU FS5EFIG1PCL U {. FS5ELPT2PCL t!4 FS5ENONEPCL m nD FS5ELPT3PCL Ar["w4 NORTON00PIF qR3 ! QBASIC PIF ,P? ! EDIT PIF VV! GWBASIC PIF bP' ! DOSPRMPTPIF N ! HSCRIPT PIF A s0! EG PIF V ! NORTON PIF T3 ! WP PIF X[ CHORD WAV @jMa TADA WAV @jQl RINGOUT WAV `AY\ DING WAV `AKN- NETWORKSWRI `Av X PRINTERSWRI `Af WININI WRI `AsZ SYSINI WRI `Ak README WRI `AZ | EXCEL XLB T y MAIN0 GRP *\" INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! INS33IS_MP fa! FW16 R|? # FW16 R|? # FW16 R|? # FW16 R|? # FW16 R|? # FW16 R|? # FW16 R|? # FW16 R|? # FW16 R|? # FW16 R|? # YSTEM 002 X YSTEM 002 X
