Back to search

University of Patras

Stochastic analysis of structures made of composite materials

Abstract

dc:description

Wind turbine rotor blades are large composite structures operating in a completely stochastic environment. Hence, the applied wind loads and further the developed stress resultants in the rotor blade airfoils are stochastic themselves. Moreover, stochastic behaviour is also exhibited by composite materials showing great scatter both in their fatigue and static mechanical properties. A rational way to quantify the variability in the basic variables and take into account these uncertainties in the final design of the structure is provided by probabilistic methods.Therefore, in the present thesis, a probabilistic methodology was proposed for the evaluation of the reliability level of a composite rotor blade under ultimate static loading at the ply level. Reliability calculations were performed both for the strength and the instability analysis. An integrated probabilistic tool was developed to carry out the necessary computations and prove valuable for the probabilistic design of the composite rotor blades.The task was multifold with several issues to be addressed. Firstly, uncertainty (physical, statistical and model) related to the basic variables (material, loads etc.) has to be quantified. Appropriate probabilistic models should be selected to represent the basic variables based on data measurements. Therefore, a number of tests for the mechanical characterization of the composite material as well as 10 min aero-elastic response simulations should be provided to start the reliability analysis procedure. In the frame of the UPWIND research project, a comprehensive experimental work was delivered for a Glass/Epoxy fiber reinforced plastic material. Many tests were performed and a database was created. The probabilistic models should quantify both physical and statistical uncertainties arisen to the material properties mainly due to the inhomogeneous nature of the fiber reinforced plastic material, the manufacturing process and the limited number of experimental tests. Statistical treatment of the measured material properties was thus performed following international reliability standards. Several probabilistic models were suggested. Correlation between the material properties was estimated based on the derived experimental data. Procedures for including statistical uncertainty to the final distribution models of the strength and stiffness properties were presented. General probabilistic models considering also model uncertainties related to the strength properties were proposed and for the first time quantified with respect to appropriate experimental data. As far as the loads were concerned, the 10 min long term extreme stress resultant distributions were estimated implementing load extrapolation techniques according to IEC 61400-1 ed.3 standard. The necessary load simulations were provided performing beam aero-elastic calculations for the 65m UPWIND reference rotor blade. Various peak methods were examined while several parametric distributions were fitted to the extracted maxima. Dependencies between the sectional stress resultants were estimated directly by the derived data. A convergence study directly on the long term exceedance probability of the extreme stress resultant was implemented to determine the necessary number of 10 min aero-elastic simulations. The statistical uncertainty due to the limited number of load simulations as well as a variety of model uncertainties related to the load evaluation by means of the aero-elastic code was further considered.Second step in the proposed methodology comprises the identification of the most important failure modes and their expression in terms of limit state functions. In the context of the ply level reliability approach, considered herein, various failure criteria were adopted. Depending on the formulation of the failure criterion, one or several failure functions corresponding to the different failure modes of the composite material were considered for every lamina in the structure. Taking into account that a typical laminate of a blade structure contains dozens of plies, a large number of state functions were thus accumulated.To efficiently deal with the problem at hand, convenient reliability methods should be selected and implemented. In the last step of the probabilistic methodology, an appropriate implementation of the response surface method combined with direct Monte Carlo simulation (RSM/MC) was proposed. That approach was validated implementing an extensive numerical experimentation program.The aforementioned methodology was implemented for two real rotor blade designs, namely a 30m Glass/Polyester and the 65m UPWIND reference rotor blade. Initially, calculations were performed for the first case study using a 3D shell finite element (FE) formulation in a commercial probabilistic code. An efficient procedure was introduced to define the stochastic character of the concentrated loads acting on the 3D FE model starting from load time series of sectional stress resultants from aero-elastic beam simulations. For the first time such a detailed model was analyzed and assessed in a probabilistic base. It should be highlighted that the finite element model of the rotor blade consisted of several thousands of layered shell elements, ca. 15,250, each with a stacking sequence composed of numerous plies. The analysis thus was performed for approximately 600,000 limit state functions. Reliability level was determined using the proposed RSM/MC method.The exhaustive reliability exercise of the detailed 3D shell finite element blade model indicated that the approach was very time consuming. The problem was persistent even with optimized reliability methods such as the response surface method. To circumvent the difficulty, an efficient probabilistic tool was developed based on thin wall multi-cellular Euler-Bernoulli beam theory and the finite strip method for the strength and buckling analysis respectively. The tool, PROBUST (Probabilistic ROtor Blade analysis in Ultimate STatic loading), was capable to perform consecutive reliability analyses at the ply level of the composite rotor blade structure. The probabilistic analyses were executed in very short time exhibiting a fair accuracy of the estimated failure probabilities.To demonstrate the efficiency of the developed tool, the impact of various probabilistic modelling assumptions directly on the β-index value of a rotor blade design was studied. The investigation concerned (i) the probabilistic model selection for the strength and stiffness properties (ii) the effect of considering correlation between the mechanical properties as well as (iii) the influence of the model uncertainties related to the material properties. The analysis was performed examining both rotor blade designs; the 30m Glass/Polyester and the 65m Glass/Epoxy UPWIND reference rotor blade.

Degree

thesis:*
Grantor dc:publisher
University of Patras
Year dc:date
2014

Author and committee

dc:creator, dc:contributor.*
Authors dc:creator
  • Bacharoudis, Konstantinos
  • Μπαχαρούδης, Κωνσταντίνος

Subjects

dc:subject × 19

Rights

Language dc:language
eng

Identifiers

dc:identifier.*
Identifier
10.12681/eadd/39848

Chain of custody

source
Harvested from
Greek National Archive of PhD Theses
Base URL
phdtheses.ekt.gr/eadd_oai/request
Last updated
2026-11-18
Source record
OAI-PMH GetRecord
citation

Bacharoudis, Konstantinos; Μπαχαρούδης, Κωνσταντίνος. Stochastic analysis of structures made of composite materials. University of Patras, 2014. http://hdl.handle.net/10442/hedi/39848