Influence of Concrete Material Ductility on Shear Response of Stud Connections. Paper by Shunzhi...

(ProQuest Information and Learning: ... denotes formulae omitted.)

Discussion by Shiming Chen

Professor, School of Civil Engineering, Tongji University, Shanghai, China.

The discusser appreciates the authors' comprehensive work to investigate the potential application of engineered

cementitious composite (ECC) in shear stud connections for steel-concrete composite beams with a desired ductile slip capacity. Some test findings that are interesting to the discusser, however, were not well clarified when the test results were compared with the predictions based on the design method (AASHTO LRFD). Discussions are drawn as follows:

Compressive strength of concrete

Accordingly, it is understood that f'^sub c^, the compressive strength of concrete in Table 2, should be the cylinder compressive strength. For a comparison, the measured and predicted strength per stud based on AASHTO equations were drawn in Fig. A.

It is found that the predicted strengths based on AASHTO LRFD method were all greater than the measured strengths for concrete and RC stud connections so that the method would be unsafe, which is argued by the discusser. As being noted that a steel reinforcement ratio of 0.86% had been used for transverse reinforcement in RC specimens that could prevent earlier longitudinal splitting shear failure in the connections, is f'^sub c^ adopted in the paper a cube compressive strength rather than the cylinder compressive strength?

Strength and failure modes

In design practice, the shear stud connections are normally classified as ductile as far as ratio h/d > 4, where h and d are the overall height and diameter of a stud, respectively. Ductile behaviors were observed in RC, SFRC, and ECC specimens, except for concrete specimens.

It is likely that an ECC connection with a higher strength and less stiffer modulus in ECC properties will possess higher shear connection strength and greater slip capacity. This, however, does not occur for concrete specimens, and the main reason may be owing to the longitudinal splitting shear failure. It is still not clear in the authors' investigation that the longitudinal splitting shear failure can be prevented by ECC material itself other than transverse steel reinforcement.

It is agreed that the direct adoption of the AASHTO equation for ECC material will be very conservative, as the actual failure mechanism of ECC specimens is fracturing of the stud shank near the welds as observed in the tests. However, this should not be as simple as to use A^sub sc^F^sub u^ to predict the load capacity, because for Specimen ECC1, the measured load capacity per stud is greater than A^sub sc^F^sub u^ and for Specimen ECC2, the measured load capacity of each stud is much smaller than A^sub sc^F^sub u^, as shown in Fig. A.

A Comparison with EC4 method

In Eurocode 4 (ECCS Technical Committee 11 1993), the shear resistance of a headed stud in a solid slab is determined by

... (1)

... (2)

whichever is smaller. Where ?v is the partial safety factor taken as 1.25, f^sub u^ is the ultimate tensile strength of the stud, f^sub ck^ and E^sub c^ are the characteristic cylinder compressive strength and modulus of elasticity of concrete, respectively; d is the diameter of stud; and a is coefficient a =1.0 for h/d > 4.

Figure B test results and design formulas

To assess the load capacity of test specimens, let ?^sub v^ equal to 1.0. Design curves based on EC4 (solid line) and AASHTO (dashed line) equations are drawn in Fig. B. The measured load capacity of each test specimen is drawn against [the square root of]f^sub c^E^sub c^, a parameter scaling strength and modulus of concrete. Similar test results of push-out tests on studs (d = 19 mm) in high-strength and normal-strength concrete carried out by Li and Krister (1996) are also drawn in Fig. B.

It is demonstrated that the predicted load capacity by AASHTO is greater than that predicted by EC4. Concrete specimens (Concrete 1 and Concrete 2) fail on the margin of EC4 curve due to longitudinal splitting in concrete. The longitudinal splitting failure should be obstructed with transverse steel reinforcement in RC specimens, whereas the test results lie above the EC4 curve, as in Specimen RC1, steel reinforcement ratios are 0.56 and 0.86% in longitudinal and transverse direction, respectively; and in specimens tested by Li and Krister, the reinforcement ratios are 0.67 and 0.69% in longitudinal and transverse direction, respectively. There appears to be good agreement in tendency between the test results and the EC4 prediction for RC specimens for both normal- and high-strength concrete, whereas the hatched triangles plotted above the right side of the curve refer to the test results of high-strength concrete RC connection, and those on the lower and left positions refer to the normal strength concrete RC connection results.

In Fig. B, the measured results for ECC connections are close to that of SFRC connection as the strength of shear connections is concerned. AASHTO (dashed line) likely gives unsafe predictions of load capacity for RC shear stud connections and should also not be appropriate for ECC connections, though the measured values of the load capacity are greater than the AASHTO prediction.

REFERENCES

EECS Technical Committee 11, 1993, "Composite Beams and Columns to Eurocode 4," Composite Construction.

Li, A., and Krister, C., 1996, "Push-Out Tests on Studs in High-Strength and Normal-Strength Concrete," Journal of Constructional Steel Research, V. 36, No.1, pp. 15-29.

AUTHORS' CLOSURE

The authors appreciate the discusser's interest in our paper.

In the first section of the discussion, Compressive strength of concrete, the discusser states that transverse steel reinforcement in reinforced concrete (RC) could prevent earlier longitudinal splitting shear failure in the connections and questions whether f'^sub c^ in the original paper is cube compressive strength or cylinder compressive strength. First, no longitudinal splitting shear failure was observed in the RC specimens, at least for the geometry tested, as shown clearly in Fig. C. In the RC specimens, macrocracks initiated from the shear stud near peak load and developed upwards to the surface of the specimens. Second, the steel reinforcement was unable to prevent the fracture of concrete. Rather, it served as a confinement measure after the concrete was fractured, as demonstrated by the less catastrophic softening behavior after the peak load, in comparison with the plain concrete specimens. Finally, the f'^sub c^ used in the paper is cylinder compressive strength.

In the second portion of the discussion, Strength and failure modes, the discusser mentioned that the occurrence of longitudinal splitting shear failure of concrete results in its low performance. As is also the case for RC, longitudinal splitting shear failure was not observed in our concrete specimens. Instead, a radiating fracture initiated from the stud head in the concrete specimens. Furthermore, the experiments clearly demonstrated that this type of fracture was suppressed in ECC due to its intrinsic high ductility. No steel reinforcement (longitudinal or transverse) was used in the ECC specimens investigated. It should be emphasized that the ECC material used has a tensile ductility of 2.5%, approximately 250 times that of normal concrete (Table 2 of original paper).

For both ECC 1 and 2 specimens, four typical stages were involved in the load-slip curves (Fig. 8(b) and 9(b) in the original paper). In the first stage, the load increased linearly with no microcrack observed. All material behaved elastically during this stage. The beginning of the nonlinear stage (second stage) in the load-slip curve was associated with the initiation of many microcracks from the stud head. These microcracks developed outwards towards the specimen surface. In the third stage, the load-slip curve gradually bent over and reached the peak load. During this stage, localization of microcracks into a fracture was observed, while extensive microcracks continued to develop in the other parts of the ECC specimen, likely due to load redistribution among the different shear studs. In the final stage, fracture of the steel shank eventually led to a sudden and drastic load-drop. These observations suggest that the ductility of the ECC and the steel stud play active roles in delaying the final failure of the ECC specimens.

It would be most desirable if a design equation for load capacity accounting for the ductility and three-dimensional microcrack development of ECC is derived. This requires more detailed numerical and/or experimental investigations beyond the scope of this paper. Because the ultimate failure was observed to be associated with fracturing of the steel shank of the stud, near the welded site, A^sub sc^F^sub u^ was suggested as a tentative simple means to predict the load capacity. Admittedly, however, variation of workmanship for welding, which affects the properties of the adjacent steel, can be expected to influence the connection shear capacity. Use of A^sub sc^F^sub u^ should be accompanied by high-quality workmanship. More research is needed to develop a good prediction expression for connection using ECC. At this point in time, it is definite that the AASHTO and EC4 design curves underestimate the load capacity for ECC stud connection (refer to Fig. D).

In the last portion of the discussion, a comparison of test data with both AASHTO code and Eurocode 4 (ECCS Technical Committee 11 1998) was made. Figure D of the discussion is replotted herein to show the correct data points for Concrete 2 and SFRC (refer to Table 2 of the original paper). The authors agree with the discusser that the AASHTO design formula would lead to overprediction of the strength of shear connections in most cases for concrete and RC. Whereas the EC4 method may conservatively predict the test results for RC specimens of the authors and those from An and Cederwall (1996) (the same as Li and Krister [1996]); however, it significantly underestimates the load capacity of the ECC specimens. The focus of the original paper has been on experimental demonstration of the ability of ductile ECC material in suppressing brittle fracture failure as typically observed in normal concrete and RC interacting with steel studs, and translation of ECC material ductility into improvement in connection performance in terms of load capacity, deformation, and damage tolerance. As previously noted, more research is necessary to develop reliable design guidelines for shear stud connections with ECC material.