Size and scale eects in composites: II. Unidirectional laminates L.S. Sutherland a , R.A. Shenoi b, *, S.M. Lewis c a Department of Naval Architecture, Technical University of Lisbon, Lisbon, Portugal b Department of Ship Science, University of Southampton, High ®eld, Southampton S017 1BJ, UK c Department of Mathematics, University of Southampton, High ®eld, Southampton S017 1BJ, UK Received 4 September 1997; received in revised form 29 March 1998; accepted 30 March 1998 Abstract This paper is concerned with the scale and size eects in the strength characterisation of composite materials with particular reference to hand laid-up unidirectional laminates. Attention has been focused on the tensile and ¯exural strengths of glass/epoxy and carbon/epoxy laminates. The experiments have been designed through the use of statistical, factorial-based techniques, whereby a more comprehensive analysis of the inter-relationship between dierent parameters and their in¯uence on strength can be made. The work highlights the importance of fabrication factors and the distinguishing dierence between scale eects and size eects. # 1999 Elsevier Science Ltd. All rights reserved. Keywords: Scale eects; Size eects; FRP; Composites; Dimensional analysis; Weibull analysis; Factorial experiments 1. Introduction The relatively recent introduction of FRP composite technology, together with the large range of materials used and being introduced, mean that a broad design base, such as that available for many metals, has not yet been compiled for FRP materials. Hence much testing of composite components has to be carried out either on full-scale prototypes, or, in order to save both time and expense, on small-scale models by use of the principles of dimensional analysis. It follows therefore, that any discrepancies encountered whilst scaling from model to full size (i.e. any size eects) should be both identi®ed and understood. Similarly, much of the design of com- posite components is based on material properties derived from small laboratory-scale coupons, often leading to a trial and error approach if the properties obtained in the laboratory tests do not correctly predict the component behaviour. It has been thought for some time that a strength `size eect' may exist for some composites, which is usually (but not exclusively) detrimental with increasing size. This is thought to be due to the increased probability of a larger specimen containing a ¯aw large enough to lead to failure. However, an accurate quantitative descrip- tion of such eects, or even concrete evidence of their existence, has proved elusive. In a review of this literature, see Sutherland et al. [1], it was shown that the majority of the existing work in the ®eld concerns high quality, pre-preg carbon/epoxy laminates for use in the aero- space industries. It was demonstrated that a number of authors have come to the conclusion that the scale/size phenomenon exists and that statistical strength theory may be used to quantify it. However, despite the fact that this theory is statistical in nature, very few statis- tical analyses of the results and trends are reported. The scaling problem is especially complex for com- posites on account of the intricate nature of their micro- structure. Also the many possible material properties that may be considered, such as manufacturing techni- que and conditions, and ®bre and matrix materials, further complicate the problem. Further still, the mechanical testing of composite materials is a con- tentious subject, with many variables that may aect the observed material properties. The marine industry gen- erally uses more variable, hand laid-up laminates than the aerospace industry, leading to greater variation in mechanical properties. Thus an investigation of composite materials strength size eects concerns both a number of pertinent vari- ables, and also experimental data subject to consider- able scatter. This type of problem requires an ecient experimental programme and statistical analysis techni- ques in order to separately estimate the eects of each Composites Science and Technology 59 (1999) 221±233 0266-3538/99/$Ðsee front matter # 1999 Elsevier Science Ltd.. All rights reserved. PII: S0266-3538(98)00083-9 * Corresponding author. Tel.: +44 1703 592316; fax: +44 1703 593299