To measure the axial stress generated by the expansion of a specimen, the following test was developed. Figure 1 shows a sketch of the apparatus. The specimen, a mortar or concrete cylinder, is placed in the stainless steel frame and connected to a load cell. Usually some small load is initially placed on the specimen by screwing down bolts set in the frame. The load cell, connected to a computer, monitors the expansive force generated by the specimen in the axial direction only. The specimen is free to expand in the radial direction, subject, of course, to friction at the ends due to the bottom of the frame and the loading disc on the top of the specimen.
If the load cell was infinitely stiff, the sample would be constrained to stay at the prescribed initial strain, which is determined by the load that is initially placed on the sample. This experiment would then be a pure stress relaxation experiment, at a fixed strain level. Because the load cell is not of infinite stiffness, this test is not a pure axial stress relaxation test. It does allow some deformation to take place in the axial direction.
The frame holding the specimen is immersed in a container with limewater. The container is then placed in a water bath with a controlled temperature of 50 ± 3 º C. All length and stress measurements reported in this paper were made at this controlled temperature, so that the effects of thermal expansion stresses were eliminated (Ferraris et al. 1995, 1996).
To extract the stresses due to ASR from stresses generated by other phenomena (water absorption, autogenous shrinkage, etc.), the expansions and the stresses generated in companion samples with non-reactive aggregates were measured. Of course, it is assumed that the stresses due to the ASR are not completely compensated by the stress relaxation of the specimen, otherwise we will observe no stresses at all. All expansions were measured using a digital comparator with a resolution of 0.002 mm (0.0001 in) with the uncertainty in the measured strain, for the size of the specimen used, being approximately 0.001 %.
Figure 1: Schematic of the device used to measure stress due to ASR.