Advanced numerical modelling of metallic energy-dissipating devices is an important tool for evaluating the effectiveness of seismic retrofit techniques for precast concrete structures. Past earthquakes have shown that precast concrete portal frames with simple pinned connections, in which the beams rest directly on top of the columns, can suffer total collapse due to loss of seat. To address this problem, low-cost C-shaped yielding devices (C-devices) have been proposed, installed around the beam-to-column connections, with the aim of providing an additional source of cumulative energy dissipation and reducing lateral displacements and residual displacements, with the smallest possible increase in base shear force.
After earlier stages of the study relied on a simplified device model using inelastic springs, a detailed two-dimensional nonlinear finite element model was developed using ADINA software (version 9.2.1), based on solid elements under plane stress conditions. This model aims to evaluate the effect of the circular transition radius (r) and the length of the dissipation zone (Ld) on the performance of the device, and to visualize the stress distribution in the inelastic regions where energy dissipation is expected to occur, in addition to taking into account the friction forces generated between the device plate and the connecting bolts.
The study included nine models combining three circular transition radii (30, 40, and 50 mm) with three dissipation zone lengths (58, 116, and 174 mm). To maintain accuracy while reducing computational effort, only half of the device was modelled, taking advantage of symmetry, with a plate thickness of 10 mm. The steel was assigned a cyclic stress–strain curve with a yield stress of 275 MPa and a linear extension beyond the ultimate stress point, while the high-strength steel region was treated as homogeneous elastic, with a modulus of elasticity of 200 GPa and a Poisson’s ratio of 0.3. The plate was meshed with eight-node quadrilateral elements measuring 2×2 mm in the critical regions.
To represent the interaction between the device and the bolt, the contact surface was defined such that the deformable surface of the device is the contactor body and the bolt is a rigid surface that is fully fixed, with zero gap assumed and penetration of the device hole into the bolt surface prevented. A friction coefficient of 0.4 was assigned to the contact when friction was taken into account. Loading was applied as a gradually increasing horizontal displacement of 40 mm over 800 small time steps to improve numerical convergence, while through-thickness stress was neglected because it is expected to be small. Principal stresses were used to identify the zones of maximum and minimum stress, while the von Mises yield criterion was used to identify stress concentration zones and the margins of potential plastic response.
Before the results were adopted, the adequacy of the mesh and element type was assessed using the Stress Band Method of Sussman and Bathe, which considers a mesh sufficiently accurate when the stress bands are continuous and can be clearly distinguished. Comparing a mesh of 18 elements along the circular transition line with a denser mesh of 48 elements showed that the first mesh suffered from discontinuities in the stress bands in the middle of the dissipation zone and in the transition region, whereas the second mesh gave continuous and acceptable results. The eight-node element also outperformed the four-node element in solution accuracy; therefore, the finer mesh with eight-node elements was adopted in all models.
The monotonic loading results showed good agreement in force and stiffness between the ADINA models and the models implemented in SAP2000 across the displacement range. The strength of the device was found to increase as the length of the dissipation zone decreases: the strength ranged between approximately 44.1 and 47.1 kN when friction was neglected, and between approximately 47.0 and 50.5 kN when it was included, and the maximum force increased by an average of 6.34% when the dissipation zone length was reduced from 174 to 58 mm. The initial stiffness remained almost independent of both the transition radius and the dissipation zone length. The strength was also very similar in tension and compression when P-Delta effects were neglected, whereas these effects become significant when they are taken into account.
Regarding friction, its inclusion led to a slight increase in force, averaging no more than 6.8%, resulting from the interaction between the bolt and the device. The study showed that the friction coefficient affects the strength of the device, which calls for care in estimating it, although low values (less than 0.2) showed no significant sensitivity in the results. The highest von Mises stresses in all models remained within the range of 384–388 MPa, i.e., within the maximum permissible stress for the device steel.
The stress distribution maps showed that the circular transition plays an effective role in confining most of the yielding within the dissipation zone: the most severe inelastic values are concentrated over a large volume of this zone, while only light yielding (stress above 275 MPa) appears in a small area of the transition region, and only a limited amount of stress is concentrated around the bolt hole. For example, a von Mises stress of 373 MPa was recorded in the dissipation zone against a yield stress of 275 MPa, with most of this zone yielding, and the yielded volume was larger in the shorter devices. Conversely, using a right-angle (90°) internal corner instead of the circular transition reduces the volume of yielded material in the dissipation zone, causes yielding in the arms of the device, and produces a stress concentration at the corner that could lead to brittle failure, particularly under cyclic loading.
Analysis of the longitudinal stress (SYY) showed that the device behaves in a bending mode, with the upper fibres in compression and the lower fibres in tension, with principal stress values of approximately ±383 MPa, and the zero-stress zone coincides with the location of the neutral axis of bending of the dissipation zone. The strain maps confirmed that the greatest inelastic deformations are concentrated in the dissipation zone, and the device displacements at maximum strain and at critical strain for failure were determined to be approximately 25 mm and 35 mm, respectively. They were also used to determine the yield displacement in order to build a database to be used in subsequent laboratory work.
The study concludes that the developed finite element models are valid and reproduce the monotonic response of the device efficiently, and that they form a basis for calibrating an equivalent one-dimensional spring model that simulates the cyclic response of the entire device, and for comparison with laboratory results. The results also confirm that the details of the dissipation zone and the circular transition should receive great care when designing these devices, supporting the development of economical and effective solutions for reducing seismic damage and preventing collapse in precast concrete industrial buildings.
Prepared by: Dr. Oday Al-Mamoori
Specialization: Civil Engineering – Structural Engineering and Earthquake Engineering
Source:
Al-Mamoori, O. H. H. (2019). Seismic Redesign of Precast Portal RC Frames using Yield C-devices (Chapter 6: Finite element modelling of the C-device). PhD thesis, Civil Engineering Department, School of Environment and Technology, University of Brighton, United Kingdom.