Experimental characterisation of random field models for CFRP composite panels

A data driven polynomial chaos based approach for stochastic analysis of CFRP laminated composite plates

“…In reality, this knowledge is not available; instead only data sets obtained from measurements at discrete locations are available. As a first step, one needs to estimate the marginal pdf and the correlation structure of the random fields from these measurements and subsequently develop the OLE shape functions; see [27]. In contrast, the proposed PC based framework enables representing the PC representation of the random fields directly.…”

“…However, studies in the composite literature [6,26,27,43] show that these properties exhibit significant non-Gaussian characteristics. Moreover, the material properties E 2 ; m 12 which have been shown to have spatial fluctuations [27] have been assumed to be deterministic. Considering these fields to be random imply that the elements in the constitutive matrix in Eq.…”

“…Even though experimental measurements have been carried out for random field characterisation in composite structures in an earlier study [27], here the measurement data are synthetically generated to check for the robustness of the above approach. The use of synthetic measurements, which are generated from assumed random fields having known second order characteristics following accepted simulation algorithms, enable checking the accuracy of the PC representations.…”

“…Subsequently, a PC approach was used to represent the response. However, experimental studies [27] reveal that the uncertainties in most of the material properties have significant non-Gaussian characteristics and using simplified Gaussian models lead to erroneous predictions about the response characteristics. PC approach to response analysis for composite plates have been carried out in [29] where the material properties were assumed to be non-Gaussian.…”

“…In [25], the authors consider the elements of the constitutive matrix to be random variables and develop a SSFEM formulation for the response analysis of multi-layered composite structural systems. Since the elements of the constitutive matrix are nonlinear functions of the material properties, this methodology is difficult to apply in real life situations where one often finds it easier to quantify the uncertainties associated with the material properties [26,27] rather than the elements of the constitutive matrix. A subsequent study [28] modelled the material properties as random fields and developed a SSFEM formalism for the response.…”