Modified Split Cantilever Beam Test

2.2.4.1. Test Description

As discussed in Section 2.2.2, the SCB specimen could not be used for mode III testing due to a significant mode II component across the delamination front (Martin, 1991). Robinson

and Song (1994) proposed a solution to the mode II issue in the SCB. They presented an “improved” SCB test which was different from that originally introduced by Donaldson (1988) in two ways. First, they bonded and inset the composite specimen into thick steel bars. Loads were applied to the thick bars rather than the specimen itself in order to reduce the undesirable deformation associated with mode II loading. Second, rather than a single pin load in each cracked leg, they applied two sets of pins. The first set, identical to those of Donaldson (1988), applies the anti-plane shear load. The second set, which is oppositely oriented, applies a restoring moment to the crack tip that significantly reduces the mode II component. This geometry has not been used beyond the work of Robinson and Song (1994) due to issues with accurately bonding the steel bars to the specimens and difficulties aligning the specimen in the fixture (Sharif et al., 1995).

At around the same time as the improved SCB was introduced, the more commonly used modified split cantilever beam (MSCB) test (Figure 2.5) was also introduced (Sharif et al., 1995). The MSCB does not include the thick steel bars of the improved SCB, as it was found that the bars have only a small benefit, but a number of drawbacks. The MSCB, however, does take advantage of the shear load and restoring moment configuration created by two sets of pin loads applied to each cracked leg (Robinson and Song, 1994). Extensive three dimensional FE analysis have showed that the MSCB is also essentially a pure mode III test (Sharif et al., 1995). The MSCB fixture is generally configured with a number of different holes through which pins can be set, in order that specimens of different geometries can be tested with the same fixture. This setup has been used extensively for mode III toughness testing (Cicci et al., 1995; Sharif et al., 1995; Trakas and Kortschot, 1997; Rizov et al., 2006; Szekrényes, 2009; Szekrényes, 2011; Khoshravan and Moslemi, 2014).

Figure 2.5. Modified split cantilever beam (MSCB) geometry (Khoshravan and Moslemi, 2014).

2.2.4.2. Critical Load for Delamination Growth Determination

In MSCB specimens, a visual observation can sometimes be used to determine the load corresponding to delamination growth, which is typically called Pvis. For transparent specimens,

such as the MSCB specimens used by Szekrényes (2009; 2011) and Rizov et al. (2006), this technique is relatively straight forward. The entire crack front can be seen, and the accuracy of determining Pvis is only limited by the resolution of the imaging technique. Typically, a

magnifying glass is used to achieve high accuracy. For opaque specimens, the delamination is only visible at the specimen edges, and visual observation of delamination growth initiation is not as straight forward. If delamination growth initiates locally at the specimen center, an edge observation will over-estimate the critical load. Szekrényes (2009) and Martin (1991) used FE ERR distributions to conclude that the delamination front would advance relatively uniformly across the specimen width, with only a slight amount of growth in the center of the specimen before growth reached the edges. From this, they rationalize that Pvis obtained from an edge-

observation of delamination growth would be acceptably close to the load at which localized delamination advance actually took place. However, Szekrényes (2011) observed crack initiation in the center of transparent specimens and used the associated load as the critical values. This

work did not comment on how close an edge-observation of Pvis was to a center-observation of

Pvis.

As in the ECT, Pmax is sometimes used as the critical load for MSCB specimens.

Robinson and Song (1994)’s improved SCB work show linear load-deflection plots up to the peak load, and concludes that the peak load corresponds to the critical load for delamination growth. However, the high-stiffness steel blocks adhered to the specimen make it impossible to tell whether there truly is no nonlinearity in the specimen response, or whether it is being masked by the blocks. Trakas and Kortschot (1997) used Pmax as the critical load from their MSCB tests

even though there was clear non-linearity, implying there may have been sub-peak load localized delamination growth.

Also similar to that used with the ECT, the critical load in MSCB tests may be based off of specimen non-linearity. Sharif et al. (1995) qualitatively selected a PNL on the load-deflection

plot to use as the critical load. Khoshravan and Moslemi (2014) used a 5% offset-type method where a line with a slope of 5% less than the experimental load-deflection slope (determined in a linear region) is superimposed on the load-deflection plot to determine the critical load (shown in Figure 2.4).

Four different determinations of critical load have been used with MSCB specimens. There are three general categories of critical load used for the onset of delamination growth in MSCB specimens. As with ECT specimens, Pmax, PNL, and 5% offset techniques have been used.

Additionally, the Pvis technique has been used for MSCB specimens. For transparent specimens,

Pvis is certainly the most accurate technique, as determination of delamination growth is

unambiguous. For opaque specimens, however, it is not clear whether Pvis can be accurately used

is not appropriate for MSCB specimens due to the likelihood of pre-peak load crack growth. PNL

and the 5% offset technique are both likely appropriate for use as critical loads.

2.2.4.3. GIIIc Determination

As described when discussing the ECT geometry, a number of different methods have been used to calculate GIIIc. Compliance calibration, a theoretical approach that uses LPT, or a

numerical approach that uses FE results may be used. Note that for MSCB specimen geometries, LPT can be reduced to Euler beam theory.

Beam theory is often used to calculate GIIIc for MSCB specimens. Trakas and Kortschot

(1997) used Euler Beam Theory (EBT) for their MSCB specimens, which considers

contributions to ERR due to bending, but nothing else. As split-beam specimens are usually composed of relatively short, thick “beams,” and because shear deformation can be a significant form of deformation in composite laminates, it is not appropriate to use solely EBT to analyze the MSCB. Cicci et al. (1995) used a Timoshenko Beam Theory (TBT) formulation, which also considers the contribution due to transverse shear. Szekrényes (2009) used an “improved” beam theory (IBT) which considered TBT, Saint-Venant effects, and effects from the free torsion of orthotropic beam. Szekrényes (2011) and Khoshravan and Moslemi (2014) use these same equations. However, although the formulations of Cicci et al. (1995) or Szekrényes (2009) achieve improved accuracy, the results are still heavily dependent on knowledge of material properties.

GIIIc determined via VCCT has been used by a few authors (Sharif et al., 1995; Rizov et

al., 2006). However, this technique has many already-discussed drawbacks, and has not been used with any regularity for MSCB specimens.

Compliance calibration techniques can also be used to determine GIIIc for MSCB

specimens. A multi-specimen compliance calibration, like that for the ECT, can be used with MSCB specimens, whereby specimens with several different crack lengths are tested

(Szekrényes, 2009). Additionally, a single-specimen compliance calibration can be used, where one specimen is shifted around in the fixture several times and sub-critically loaded in order to experimentally calculate a relationship between compliance and crack length (Cicci et al., 1995; Trakas and Kortschot, 1997). This eliminates the specimen-to-specimen variation concerns that exist with a multi-specimen CC. While a single specimen compliance calibration method is the preferred technique for obtaining GIIIc, as it contains the fewest assumptions, there are still

unresolved issues associated with MSCB testing. MSCB specimens fabricated from

carbon/epoxy have very low compliance, and because of this it can be difficult to develop a C(a) trend for both single-specimen CC (Cicci et al., 1995; Trakas and Kortschot, 1997) and multi- specimen CC (Szekrényes, 2009). There is also issue with determining the appropriately linear region over which to conduct the compliance calibration (see Figure 2.4), especially when specimens with large crack lengths are used.

In summary, beam theory, FE, and compliance calibration techniques have been used to determine GIIIc for MSCB specimens. The beam theory and FE formulations suffer from

dependence on material properties. The single-specimen CC would be the preferred technique for GIIIc measurement, although there are known issues with extracting C(a) for MSCB

specimens. The multi-specimen compliance calibration technique has this same issue with C(a) determination, but also has added uncertainty from specimen-to-specimen variation. Issues with all of these techniques has contributed to MSCB not moving forward as a generally accepted test (especially for carbon/epoxy). Nevertheless, it is likely the next best existing approach (after

ECT) for determining GIIIc, and may be used to assess possible geometry dependence of the

apparent toughness.

2.2.4.4. Mode III Toughness Testing

Several studies of geometry dependence have been presented using MSCB geometries. However, the results for the dependence of toughness on crack length for MSCB are not all in agreement. For glass/epoxy composites, it has been found that toughness is higher for specimens with larger delamination lengths (Szekrényes, 2009; Khoshravan and Moslemi, 2014).

Conversely, for carbon/epoxy composites, toughness has been found to be lower for specimens with larger delamination lengths (Szekrényes, 2007; Szekrényes, 2009; Szekrényes, 2011). It is not clear why the trends are different for the two different materials.

The above results are based on data acquired from MSCB tests using several definitions of the critical load (Pvis, PNL, 5% offset), as well as data reduction methods that do not rely on

material properties (IBT, multi-specimen CC). Even with the variety of techniques utilized, all of these studies find that mode III toughness depends on geometry. Thus, it seems likely that the observed trends are not simply artifacts of the particular techniques.