【正文】 beams have been subjected to sustained loading for over 61/2 years. The objective of the experiments was to assess the significance of creep in the epoxy adhesive and whether such creep might allow the FRP strips to unload over time. Slip movements at the ends of the FRP strips were also monitored.The experimental deflections have been pared to deflection predictions using ACI 209R92 and CEBFIP MC 90. The creep deformations of the FRPstrengthened beam are not as predicted from the control beam. Two analytical approaches are used: a stepbystep intime analysis and finite element (FE) modeling. Both techniques demonstrate that creep of the adhesive layer can account for the differences observed between the predicted and actual behaviors of the beams.Keywords: creep。 deflection。 fiberreinforced polymer(FRP)。 MPa (4900 177。specimen. Two CFRP strips were bonded to the tension face of the second beam (Beam 2) using an epoxy adhesive. The strips are 100 mm ( in.) wide, mm ( in.) thick, and 2970 mm ( in.) long. Over the shear spans at each end of Beam 2, GFRP sheets were wrapped in a Ushape to cover the two side faces and the tension face of the beam. The CFRP strips are unidirectional with the fibers aligned along the length of the beam. The strips have a modulus of elasticity of 165 GPa (23,571 ksi) and a tensile strength of 2800 MPa (400 ksi) in the direction of the fibers (manufacturer’s data). Test setup and procedureThe beams were simply supported (pinroller) over a span of 3200 mm ( in.) (Fig. 1). The midspan deflection due to selfweight was recorded using a dial gauge (least count mm [ in.]) mounted on a lightweight steel frame over the 3200 mm ( in.) span. For Beam 2, electronic spring gauges were mounted on the concrete adjacent to each end of one of the CFRP strips. The deflected tip of the spring gauge was carefully positioned to touch the exposed end of the CFRP strip. These gauges were designed to record the longitudinal slip of the ends of the CFRP strip relative to the concrete to which the strip is bonded. Each beam was then loaded in fourpoint bending by applying twopoint loads (each kN [ kips]) at a distance of 930 mm ( in.) from each support. This level of loading was designed to push both beams into their working range. For Beam 1, the load was expected to result in significant cracking of concrete in the tensile zone, but stresses in the concrete in pression and steel in tension would remain elastic. In Beam 2, the load was expected to cause far less damage but still result in significant deflections over the long term. The loads were provided by hanging concrete blocks on load hangers bearing on the upper surface of each beam. The loads were transferred to the beams by gradually lowering the blocks onto the beams using hydraulic jacks. The hydraulic jacks were then removed. The midspan deflection of each beam relative to the selfweight value as well as the relative slip movements for the CFRP strip were recorded immediately after loading and at regular intervals thereafter. Data were recorded several times during the first 24 hours of loading, daily for the first month after loading, increasing to every 3 days, weekly, biweekly and, finally, greater intervals. The loaded beams are located in the airconditioned basement level of a laboratory building where the average temperature is expected to remain relatively constant with time.Experimental results and discussionUpon loading, the immediate midspan deflections of the beams relative to the measured selfweight deflections were and mm ( and in.) for the control and CFRPstrengthened beams, respectively. Numerous flexural cracks were observed extending from the tension face of the control beam, predominantly over the region of constant bending moment between the point loads. No flexural cracks were observed in Beam 2 at first loading. Such cracking would be hard to see unless significantly wide, due to the existence of the FRP strips. This is because the tensile strain in the FRP will be constant in the constantmoment section. If there were a few wide flexural cracks, the strain, and thus the stress, in the FRP would dramatically vary between the FRP still bonded to the concrete and the FRP stretching across such cracks. Equilibrium of the FRP strip has to be maintained。 = ) and Beam 2 (reinforced with steel and FRP), including effects of concrete creep, tension stiffening of concrete (), and stress relaxation of FRPepoxy posite (anddays).Results of analytical deflection predictions using CEBFIP10 and ACI11The CEBFIP10 approach was first used to predict the immediate midspan deflections of both beams. The model predictions were calibrated agai