Determination of Ductile Material Properties by Means of the Small Punch Test and Neural Networks

Abstract

The ductile material behavior in structural components is changing due to in service loading, aging, irradiation, embrittlement a.o., which requires an in situ monitoring of the material state. In order to determine material parameters at various locations e.g. in weldments or gradient materials, the size of the material taken out for a test specimen should be very small but representative.In the small punch test (SPT), a disk like specimen of AE 8 0.5 mm size is deformed in a miniaturized deep drawing experiment. The measurable output is the load displacement curve (LDC) of the punch, which contains information about the elasto-plastic deformation behavior and about the strength properties of the material.The SPT was introduced to determine post irradiation mechanical properties of materials used in the nuclear industries. [1±3] Baik et al. [4] defined the area under the LDC as small punch fracture energy and found correlations between results from Charpy-V-notch and small punch experiments determining the ductile to brittle transition temperature. Some researchers used the SPT to predict the elastic plastic properties [5,6] and the ductile fracture toughness J Ic [7,8] or the brittle fracture toughness K Ic . [9] Huber et al. [10] showed that neural networks (NN) are suitable for the determination of constitutive properties from spherical indentation tests. By Huber and Abendroth [11,12] deformation and ductile damage properties are determined by using NNs as inverse approximations of the finite element (FE) solution of the SPT.In the present paper, two different approaches are tested to identify the material parameters of ductile hardening. In both cases the LDC is transferred to a NN, which has been trained with a data base of LDCs generated by FE simulations of the SPT with systematically varied hardening parameters. The NN can either approximate the FE solution directly and the parameter identification is done by a conjugate directions root finding algorithm (case II) or the NN approximates the inverse problem of the FE solution and gives the material parameters as answer directly (case I).These identification procedures were performed for the materials StE-690, 18Ch2MFA and GGG-40 that are widely used in mechanical engineering. The advantages and restrictions for using both approaches are discussed.Experimental Methods: The SPT is performed using the device as seen in Figure 1. The specimen (8) is clamped between die (9) and down-holder (4), which are supported by the bottom housing part (7). The screw able upper housing part (3) is used to provide the clamping force. The punch (2) driven by the cross head punch (1) of the testing machine deforms the specimen centrically. The punch displacement is measured (5) parallel to the punch and close to the specimen to prevent errors due to the bending of the cross head of the testing machine and other elastic deformations of the experimental setup. A load cell between cross head and punch measures the force acting on the punch.The result of this experiment is the...

Mechanical characterization of PM aluminum composites by small punch test

“…However this intersection P Y MAO falls out from P vs.  curve. Following a similar procedure suggested by the European Code [12] a vertical projection from P Y MAO was made to define a P Y MAO* [4].GARCIA et al [6] proposed an offset method defined by a parallel line to regime I shifted in displacement t/100. The intersection is called P Y t/100 is equivalent to engineering definition of 0.2 % strain to calculate the yield stress.…”

“…That is, due to a complex stress field along the test where the disk is subjected to multiple combination of loading modes, including indentation, elastic and plastic bending, and membrane stretching at different level of deformation [4]. The literature has attempted model by numerical methods in order to reproduce the elastoplastic processes in order to identify which parameters controls the deformation in different stages along the test [4,5,6].…”

“…SPT has been applied in many engineering materials including metallic alloys and composites [5,6,7] to correlate the SPT results with the traditional mechanical properties like tensile, fracture or creep. For that purpose some semiempirical correlations between SPT parameters extracted from load-displacement curve and tensile properties like yield strength, tensile strength and elongation are often reported [4,7,3].…”

Development and validation of small punch testing at NIST

“…The model used in this case is a Ramberg-Osgood model with a possible modification to accommodate the discontinuous yield observed in several low-alloy steels [6]. Other analytical methods have also been proposed, involving the use of Neural Networks and Finite Element simulations [7][8][9].…”

Neural network utilization for evaluation of the steel material properties

“…The standard process of identifying yield stress, tensile strength is to perform tensile test and based on this test, the material properties can be evaluated [5,6]. Moreover, fracture toughness [7] needs to be also evaluated from the individual test. However, tensile test as well as fracture toughness assessment requires large specimen of the material and it could be financially and time demanding task.…”