【正文】 on of RC beams strengthened with FRP (for example, Charkas et ), these models do not account explicitly for creep of epoxy adhesives. Herein, we use two different approaches to determine if creep in the epoxy can account for the different behaviors observed in the beams: a stepbystep intime analysis allowing incremental creep of concrete and epoxy in each time step and enforcing equilibrium at the end of the time step, and finite element (FE) modeling with shear flow allowed in the epoxy adhesive layer.RESEARCH SIGNIFICANCEThe potential effects of creep on RC beams strengthened with externally applied FRP strips are considered. It was thought that creep in the epoxy resin might relieve stress in the FRP, making the FRP less effective from a serviceability point of view under sustained loads. Thus, FRP strips used to strengthen a beam, which was then subject to increased sustained load, might end up with the extra sustained load being carried by the original concrete and steel reinforcement, not the FRP. The experimental and analytical work performed revealed that the situation is more plex. Nevertheless, creep deflections are greater than predicted from the creep of concrete alone, indicating contributions from creep of the epoxy. The reported experimental program was designed to identify the existence of epoxy creep rather than replicate a practical retrofit scenario. The results highlight the potential for epoxy creep to affect the longterm performance of FRP retrofits in practice.EXPERIMENTAL PROGRAMTest specimens and materialsTwo similar RC beams were cast from the same concrete batch (Fig. 1). Each beam was 3500 mm ( in.) long, 280 mm ( in.) wide, and 180 mm ( in.) high, reinforced with four longitudinal bars (Canadian mm [ in.] diameter, 100 mm2 [ ] area) at an effective depth of 135 mm ( in.) from the top surface of the beam. Seven 10M stirrups were spaced uniformly in each shear span of each beam. The 28day pressive strength of the concrete, as determined from 100 mm (4 in.) diameter, 200 mm (8 in.) high cylinders—cast from the same batch of concrete as the test beams—was 177。 therefore, rapid and large variations in force in the strip are not possible, as there is no mechanism to acmodate such stress variation. Hence, the concrete in the constantmoment section has to crack with a large number of very thin cracks so that the strain in the FRP remains constant. Several months after loading, a thin flexural crack could be seen extending from the tension face of Beam 2 close to the beam midspan. The CFRP strips significantly increased the cracking moment capacity of Beam 2. Although not tested, the additional FRP reinforcement was also expected to increase the ultimate moment capacity of the beam.The longterm midspan deflections (total deflection minus the initial deflection) of both beams are shown in Fig. 2, plotted versus time after loading up to 2470 days of loading. The relative slip movements versus time at each end of one of the CFRP strips bonded to Beam 2 are shown in Fig. 3. Several observations can be made:1. The longterm deflection of Beam 2 is significantly less than that of Beam 1. Longterm deflection due to concrete creep is a function of the immediate deflection, which depends on crosssectional stiffness and cracking status. Beam 1 cracked much sooner than Beam 2.2. The longterm deflection of Beam 2 constitutes a larger proportion of its immediate deflection pared with Beam 1. Plevris and Triantafillou2 observed a similar response in beams externally reinforced with FRP strips pared to their control specimen (no FRP).3. The midspan deflection for Beam 2 does not appear to be “catching” that of Beam 1. That is, at first appearance, there is no indication that the CFRP reinforcement is unloading, thereby being ineffective against the sustained load.4. The rate of increase of deflection changes with time. This is particularly evident when pared to the smooth curves of predicted longterm deflections discussed and presented in the following sections (Fig. 2, 4, and 5). Similar trends occur for both beams. The periods of reduced rate of increase of deflection coincide with the summer months and those of increased creep coincide with the winter months. In the calculation of creep coefficient using CEBFIP,10 the creep coefficient reduces as the relative humidity increases, implying a reduced rate of creep during periods of higher relative humidity. The relative humidity was not recordedduring the current tests. However, relative humidity was recorded during creep tests by Hall and Ghali,6 conducted in the same laboratory as these tests. Their results showed relative humidity varying between 5 and 50