Assessment of Long-Term Losses in Prestressed Concrete Structures - Application for Nuclear Reactor Containments

Abstract: Most nuclear reactors, both in Sweden and worldwide, are enclosed by a prestressed concrete containment. The main purpose of the containment is to prevent any radioactive discharge to the environment in the event of a major internal accident. The performance and structural integrity of the containment depends on the compressive stresses induced in the concrete by the prestressing system. However, due to shrinkage and creep in the concrete and relaxation in the tendons the prestress will decrease with time, significantly reducing the safety and accidental performance of the containment. The corrosion protection of tendons are for Swedish containments arranged in two different ways, either by cement grouting (bonded tendons) or e.g. by grease injection (unbonded tendons). The disadvantage with bonded tendons is that no possibility of assessing their status, e.g. measuring tendon forces, is possible as is the case for containments with unbonded tendons. The objective of this thesis is to investigate different methods for determining the remaining tendon forces in prestressed concrete structures with an emphasis on the conditions inside nuclear reactor containments. The work has been divided into three different parts, the first in which prestress losses have been modeled using existing prediction models for creep and shrinkage of concrete and relaxation of the prestressing steel and compared to measured prestress losses in both test beams and losses measured in-situ in reactor containments. The test beams stored inside reactor containments showed very high losses compared to bridge beams of similar age, this difference was attributed to the climate in which the test beams were stored, i.e. around 30°C to 50°C and low relative humidity. For the test beams the prediction models significantly underestimated the measured losses. However, when applied for Swedish containments with unbonded tendons the models were in relatively good agreement with the measured prestress losses, a slight tendency for overestimating the losses was observed. In addition, it was shown that the accuracy of the models increased when modified by taking the actual drying conditions into account. In the second part a first step in the development of a nondestructive method for monitoring the prestress losses based on resonant ultrasound spectroscopy in the context of acoustoelasticity is presented. Several previous studies have shown that the resonance frequencies of concrete structures are stress dependent, i.e. they increase with the applied compressive stress. However, based on linear elasticity this behavior has been difficult to explain. In this work short-term dynamic measurements on prestressed concrete beams have confirmed this stress dependency and also explained this behavior through the theory of acoustoelasticity, which in short states that the elastic constants of a material, and hence the resonance frequencies, are stress dependent. A finite element model based on Murnaghan’s third order elastic theory confirmed this stress dependency. Long-term measurements on the same beams showed that the resonance frequencies can be measured continuously over a longer period of time and that by taking the development of the modulus of elasticity with time into account the decrease in resonance frequencies follows the loss of tendon forces. These results show that a change in the state of stress in a simple concrete structure can possibly be monitored by measuring the resonance frequencies of the structure. The third part consists of an investigation on the influence of an elevated temperature on the prestress losses in prestressed concrete test beams. Six of these beams were subjected to a climate similar to that of reactor containments and two were subjected to a normal indoor climate for a period of almost three years. The prestress losses in the beams subjected to the elevated temperature were approximately 25 % higher than the ones subjected to the normal indoor climate. In addition, there was no significant difference between the prestress losses in beams with bonded and unbonded tendons. The prestress losses in the beams were determined using the co-called crack re-opening method, in which the load required to reopen a flexural crack in the bottom of the beam is determined. Since the stress in the bottom fiber of the beam is zero at the instant the crack reopens, the remaining tendon force can easily be calculated. It was found that the normal procedure for determining the crack reopening load underestimates the tendon forces and that the accuracy for determining the crack re-opening load can be greatly increased by using a simple two-dimensional finite element model of the testing procedure.