a b s t r a c t  ：The lack of in-depth understanding of the seismic behavior and ductility of precast concrete structures makes it difficult to reach to ductility demand which could be exhibited during an earthquake. The lim- itations are mainly related to the beam-to-column connections as the main load transfer paths. Two dis- tinct exterior beam–column connections made of normal-strength concrete are investigated experimentally. Both dry and wet type installment techniques are used in the industrial type joints while the residential type joints are wet connections. The specimens are subjected cyclic displacement reversals in order to obtain information on strength, stiffness and ductility characteristics of the connection details. The preliminary design of the joints has been updated during the tests based on the damages observed, thus a set of improved specimens have also been built and tested, and a relatively better performance is obtained expectedly. The industrial and residential types of connections showed stable load–displacement cycles with high energy dissipation up to structural drift of 2%, though a significant level of pinching and deterioration of the critical section have occurred at around 3% drift level. The tested specimens have been numerically modeled to calibrate the analytical tools, and a satisfactory approximation has been obtained between experimental and numerical  results.

© 2015 Elsevier Ltd. All rights reserved.

1. Introduction

One of the major challenges in the design of precast structures in earthquake prone areas is the proper design of the connections, primarily beam-to-column ones. Various types of beam-to-column connections such as monolithic, emulative, bolted, and dry pinned are being used in the practice. Details of different types of connec- tions, in terms of design and behavior, are discussed extensively by Park [1] and fib Bulletin 43 [2].

Bhatt and Kirk [3] and Seckin and Fu [4] developed some welded connections for use in precast concrete  structures. Although the behavior of the connections is acceptable, the details require welding of the beam and column reinforcement,  which may cause some problems on site. French et al. [5,6] tested various types of beam–column connections; some of them developed plas- tic hinges outside the connection region. It was revealed that the threaded reinforcing bar connections with tapered,  threaded splices proved to be the most favorable solution in terms of perfor- mance,  fabrication,  and  economy.  Ersoy  and  Tankut  [7]    tested

precast concrete beams with dry joints under reversed cyclic load- ing. The original beam consisted of two steel plates one at top, the other at the bottom, welded to the anchored steel plates in the col- umn bracket and the beam. The design was later revised by adding side plates. The strength, stiffness and energy dissipation of the member with side plates are comparable to those of monolithic member. However, application of such a detail on site is quite dif- ficult and requires a careful quality control mechanism. Priestley and Tao [8] proposed the use of lateral load resisting systems built incorporating unbonded prestressing for use in earthquake prone areas. Nakaki et al. [9] proposed a precast concrete ductile frame that takes advantage of the inherent discrete nature of precast con- crete by providing ductile links in the connections. These ductile connectors which are also eliminating the need for corbels contain a rod that yields at a well-defined strength, effectively limiting the load that can be transferred to less ductile components of the frame. PRESSS project [10,11] is the most notable effort on the experimental investigation of the earthquake response of precast structures. Ductile connections for precast concrete frame system, ductile connections for precast concrete panel systems, precast frames with unbounded tendons, high performance fiber-reinforced-concrete   energy   absorbing   joints   for precast