Creep-fatigue interactions in glass fibre/polyester composites

Abstract

Glass-reinforced plastics (GRP) are employed in the cooling water (CW) systems of coastal power stations, where they are exposed to an aggressive environment, sea water at temperatures up to 40°C, and to a complex loading pattern. This latter comprises a background static load, on which are superimposed sinusoidal variations and periodic unloads. The aim of this project was to ascertain the extent of any interaction between static and dynamic loading. Also examined were the effects of environment and of cyclic loading at low frequencies upon fatigue life. The material was modelled on the CEGB-preferred layup of mixed chopped-strand mat (csm) and woven-roving (wr) in polyester. Two loading regimes were used to model aspects of the CW system service conditions. Both used a frequency of 1 cycle per 2 min (0·0083 Hz). Periodic unloads were represented by a ‘rectangular’ load-time variation (R1), with a ratio of time under load to time off-load of 20:1. Tidal pressure variations were represented by a sinusoidal régime with load amplitude equal to 20% of mean load (S1). When the aggregate time at maximum load during the rectangular-wave test R1 was compared to life under static loading at the same level, intermittent unloading seemed to make no difference. This observation was contrary to some published work. Since the different behaviour might arise from the use of different loading frequencies, the effect of frequency of unloading was investigated over a range from 0·0005 Hz to 0·25 Hz. At the higher frequencies, failure occurs after a specific number of cycles, irrespective of the time under load; at low frequencies a time-at-load criterion is satisfied first. A ‘boundary frequency’ must exist between these régimes, and it is calculated to lie between 0·12 and 1·2 cycles/min (cpm). The frequency of R1 (0·5 cpm) lies within this range. The interaction between sinusoidal and static loading was investigated with two ratios of stress amplitude to mean stress; with frequencies of 1·0, 0·25 and 0·0083 Hz; in air and in water at 40°C. Results were assembled in the form of a partial Haigh diagram. From this it was possible to deduce: (a) that in tension, the effect of combined static and cyclic loading conforms to the expected pattern, (b) that the fatigue design limits of the material in water are less than in air for tests at the same frequency, as was expected, and (c) that a further reduction in limits is required in wet environments at low frequencies, relative to those at 1 Hz. Implications for BS4994, the principal GRP design code, are discussed.