The dialogical integrated system for a material strength database (DIMS) was newly developed. In this study, the prediction procedures for fatigue strength properties were investigated using DIMS. It was found that the high-cycle fatigue strengths were accurately estimated by a multivariate analysis on the basic material factors of heat treated materials such as chemical composition and heat treatment temperature, under the limited conditions which were classified by materials used, heat treatment conditions and/or the fatigue fracture mode. From these results, the fatigue strength prediction procedure was proposed and examined for several published data sets, and it was found that the proposed prediction procedure, which used the multicorrelations among several material properties in the database system, was quite suitable for the prediction of high-cycle fatigue strength.

View full abstract

The influences of fillers on fatigue crack propagation (FCP) of advanced Nylon 46 were studied. In particular, with a focus on the fatigue reliability of filler-reinforced Nylon 46 in practical usage, FCP was estimated in parallel with the injection direction of the specimen controlled for the equilibrium moisture content under 23°C and 50% R. H. The three kinds of tested materials Were short E-glass fiber-, Miker mineral- and EPM rubber-reinforced Nylon 46, which are reterred to as G, M and R material respectively. The threshold of FCP, ΔK_th for G, M and R material was 1.4, 1.2 and 1. 0 MPa√(m) respectively. So, the resistance of G material against FCP was found to be superior in all materials. Fractography using SEM showed that the fatigue crack near threshold in G material propagated through the matrix in a zigzag line. Therefore, it could be interpreted that the variation in the property of FCP of each tested material was caused by the difference in the roughness of the fracture surface.

View full abstract

The fatigue life expressed as an S-N curve exhibits quite a large difference depending on the type of loading ; cyclic bending fatigue life is larger than the push-pull fatigue life. This is mainly caused by the difference in cyclic deformation occurring under two loading conditions. The present authors aimed to estimate the cyclic strain and the life in cyclic bending fatigue, based on the results of push-pull loading in JIS S45C steel. The cyclic softening occurring in cyclic bending fatigue was successfully analyzed by introducing the simplified push-pull cyclic softening behavior into the numerical elastic-plastic calculation, and the difference in cyclic deformation between the two types of loading was quantitatively clarified. Furthermore, the fatigue life curve under cyclic bending load was satisfactorily estimated by combining the push-pull fatigue life curve and the above calculated cyclic strains.

View full abstract

In order to investigate the effect of matrix on the fatigue behavior of ductile cast iron, both cyclic stress- and strain-amplitude-controlled low-cycle fatigue tests are conducted on ductile cast irons with different microstructures. Three kinds of ductile cast irons, i. e., ferritic, (FDI), pearlitic, (PDI), and bainitic matrix, (ADI), with the same chemical compositions, are prepared by using different heat treatments. Each material shows cyclic hardening behavior at an early stage of fatigue life. The fatigue life satisfies the Manson-Coffin relationship and differs depending on matrix. In particular, the fatigue life trends observed in two test conditions are apparently opposite. However, a definite trend of the fatigue life is obtained when the relationship between the stress amplitude and strain amplitude is correlated by means of the cyclic stress-strain relationship. The microscopic fatigue process is dominated by initiation and propagation of the microcrack. It is recognized that the microcrack initiation originates not only in spheroidal graphite, but also in microshrinkage in PDI and ADI. The relationship between the crack growth rate of a leading crack and cyclic J integral value indicates no unique linear relationship, which suggests the effect of the crack closure behavior of spheroidal graphite.

View full abstract

The change in the ductile-brittle transition behavior of sintered powder chromium (99.8 mass % purity) explosively shock loaded in water was examined by means of a tension test. The present results can be summarized as follows. ( 1 ) The ductile-brittle transition temperature (DBTT) of an as-annealed specimen was found to be 543 and 573 K in tests at strain rates of 1.7×10^-5 and 1.7×10^-4 s^-1, respectively. These findings show that the DBTT of the chromium is very sensitive to the deformation speed. ( 2 ) Due to the shock loading at 930 and 1420 MPa, the sharp yield observed in the as-annealed specimen was completely eliminated and the yield stress value was considerably decreased. ( 3 ) The transition temperature of the specimen shock loaded at 930 and/or 1420 MPa was markedly lower than that of the as-annealed specimen. For example, the DBTT of the specimen treated at 930 MPa was 290 and 403 K in the tension tests at1.7×10^-5 and 1.7×10^-4 s^-1, respectively.( 4 ) Deformation twins, a kind of plastic deformation, were not formed in the present chromium subjected to shock pressures up to 1420 MPa, though they were observed in the steel shock loaded at about 1000 MPa. ( 5 ) The decrease of DBTT in the shock-loaded specimen can be explained in terms of a relationship between the yield stressσr and the brittle fracture stressσc. Since the yield stress of the chromium is lowered by the shock pressure treatment which induces free dislocations around the e1astic discontinuities, such as interfaces between the matrix and the second phases, the condition, σr = σc, for the occurrence of the ductile-brittle transition is satisfied at a lower temperature in the shock-loaded specimen than in the as-annealed specimen.

View full abstract

The growth characteristics of small fatigue cracks in 0.1N-NaOH solution have been investigated on a high-strength steel, SNCM439. The results obtained were discussed on the basis of the stress corrosion cracking (SCC) behaviour, and were compared with those in distilled water and in 1%NaCl solution already obtained in a previous study. Long cracks in 0.1N-NaOH solution grew faster than in room air atΔK >10 MPa√(m), and this behaviour was the same as the results in distil1ed water and in 1%NaCl solution. Small cracks also showed higher crack growth rates than those in room air atΔK >15 MPa√(m), which was larger than theΔK value for long cracks mentioned above. Although, in distilled water and in 1%NaCl solution, small cracks grew faster than in room air, theΔK values at which the crack growth rates were enhanced were much lower than that in 0.1N-Na, OH solution. These differences in the acceleration behaviour were found to be strongly related to the SCC behaviour, because the K_1scc values increased with decreasing crack length in 0.1N-NaOH solution, while they decreased with decreasing crack length in distilled water and in 1%NaCl solution. Fractographic observations on SCC fracture surfaces revealed predominantly intergranular fractures, and there was no evidence of anodic dissolution on intergranular facets in a small crack region.

View full abstract

Fracture behavior of Si₃N₄ and SiC ceramics was investigated in corrosive environments at high temperature. Corrosion tests were made under various stress conditions using specimens coated with molten salts at l000°C in air. Flexural strengths of corroded specimens were investigated at room temperature. Stress-time-dependent and stress-cycle-dependent fracture tests were also performed in air at l000°C using specimens coated with a moltan salt. The following results were obtained by analyzing the chemical reaction and fracture process. ( 1 ) Molten salts of sodium corroded Si3N4 at l000°C in air. The major mode of corrosion attack was the dissolution of the grain boundary phase. The formation of large pits resulted in strength degradation. Strength degradation of SiC was only 5%. ( 2 ) Applied stress affected the morphology of bubbles in layers corroded by molten salts. A model is proposed for the defect formation under tensile and compressive stress. ( 3 ) Stress-cycle-dependent strength of Si₃N₄ was lower than stress-time-dependent strength under a sodium salt condition at 1000°C.

View full abstract

Crack propagation behaviour under static loading condition is divided into three regions. Region 3 cracking is governed by mechanical phenomena, independent of the environment. Its upper limit is the fracture toughness of the material, K_1c. Region 1 cracking can be expressed by the parallel curves shifted toward the higher K region with a lower relative humidity environment, which can be explained by the environmental fracture toughness, K_1e. The acceleration of cracking caused by cyclic loading is observed in the relatively lowΔK region. In this region, a so-called process-zone-like area appeared as a roughened grain boundary, which may explain the acceleration of racking.

View full abstract

Rotating bending fatigue tests of annealed 0.21% C steel plain specimens were carried out for three stress ranges. Sixteen to 18 specimens were fatigued at each stress range, and the behavior of cracks which led to the fracture of specimens was observed successively by the plastic replica method. The scatter characteristics of fatigue life were studied on the basis of the behavior of microcracks. The results show that the scatter in fatigue life is controlled mainly by that in N_0.3, 0.3mm crack initiation life. The coefficient of variation, CV, for N_0.3 is especially large at stress right near the fatigue limit. On the other hand, the scatter in the propagation life from 0.3 mm up to fracture, N_0.3→f, is not so large. Significant dependence ofσa on CV for N_0.3→f is not observed.

View full abstract

Through the recent development of fracture mechanics methodology in piping analysis, the LBB concept has been recognized as applicable to the safety design of LWR primary circuit piping. An important subject still remaining in the LBB analysis is the consideration of multiple fatigue crack growth and development in crack geometry in piping. In the evaluation procedure currently employed, a simple crack geometry is assumed to judge whether the crack is leak-detectable or not. The paper describes a LBB evaluation procedure in which the growth of multiple fatigue cracks is considered. Two criteria, i.e., critical condition of crack coalescence preceding the penetration of wall thickness and evaluation of crack configuration at the penetration, are introduced to account for cracks that coalesce and develop to the largest possible through-wall single crack at the onset of penetration. The paper also presents the results of a case study of PLR piping of BWR using the above procedure. It is shown that LBB can be justified in PLR piping of 4 inches or greater in diameter.

View full abstract

A higher-order theory of caustics for mixed-mode running cracks is developed using the general asymptotic solutions for a dynamically propagating crack-tip. The analytical expressions are obtained for the Jacobian equation that determines the initial curve, and for the image equations on a screen. With the use of the higher-order coefficients determined by the finite-element simulations of actual dynamic fracture experiments, the effects of the higher-order terms on the caustic curve are investigated on the basis of the present theory. It was found that the r^1/2 stress field plays an important role in the formation of caustic patterns.

View full abstract

n linear fracture mechanics, stress intensity factors controlling the fracture behavior of materials are calculated from stress distributions acting on the crack surfaces, with the aid of superposition even for remote applied stresses. A method of evaluating such stress distributions has been proposed by the authors. It is assessed from each mode-discontinuous displacement along a crack with mode I and II components. In this study, the previous calculation method restricted within constant sress fields was developed to nonuniform stress fields so as to be applied to fatigue cracks with residual stresses in the wake of the propagation. By using this method, local stress distributions controlling the crack deformation and stress intensity factors K_I and K_II for the fatigue crack were predicted under a mixed-mode stress condition, and effects of residual stresses near the fatigue crack on the mode II crack deformation were discussed.

View full abstract

Recently, several authors have examined the problems of bending of beam made from bimodulus materials. In the present paper, we apply the Levinson-type theory of beams which includes shear deformation and warping of the cross-section, to bending analysis of thick rectangular beams made from bimodulus materials. Numerical results are obtained by use of the transfer-matrix approach and are compared with results of the methods of Classical theory, Timoshenko-type theory and Levinson-type theories.

View full abstract

In this paper, an inverse analytical technique to obtain material properties of inclusion is proposed. The technique uses strain data. on the matrix distant from the inclusion. Relationships between strain and material properties are formulated and strain sensitivity is analyzed on a thick cylinder model. The finite-element method and the numerical optimizing technique (DFP method) are applied to inverse analysis. By using the inverse analytical method, satisfactory material properties are obtained from the strain data measured in a tensile test.

View full abstract

A closed-form solution for calculating strains produced by temperature change in semiconductor chip bonding structures is obtained on the basis of assumption of the beam theory and elasto-perfect plastic bonding layers. This solution shows good agreement with three-dimensional FEM results and experimental results. New parameters governing the strains produced in the bonding structures are obtained on the basis of this solution. Charts using these parameters to obtain the maximum shearing strains of the bonding layers and the maximum normal stresses of the chips are presented.

View full abstract

The boundary element method (BEM) does not require a domain integral in steady thermoelastic problems without heat generation, but with heat generation, the domain integral is necessary. This paper shows that the problem of steady thermoelasticity under nonuniform heat generation over the region can be easily solved without a region integral by means of the boundary e1ement method. This method can also be applied to steady thermal stress problems under general complicated heat generation. However, for general heat generation the domain must be divided into small areas, where distributions of heat generation satisfy the Laplace equation.

View full abstract

An analytical solution is presented for a micropolar plane-strain thermoelastic problem in a confocal hollow elliptical cylinder subjected to symmetric heating with respect to the x and y axes on the outer elliptical boundary. The problem is formulated in elliptical coordinates based on micropolar theory for an elastic solid initiated by Eringen and Suhubi, and on the extension to a thermoelastic solid with stress functions by Nowacki. The formulation includes an assurance of the single-valuedness of the rotation component in the confocal hollow elliptical cylinder of a micropolar elastic solid by Michell's condition expressed in elliptical coordinates. In order to demonstrate the effect of couple stress upon the relaxation of thermal-stress concentration in the vicinity of an elliptical hole in the cylinder, numerical illustration of the proposed theory is given for three cases of aspect ratios of the elliptical hole and various values of the new material constants of Eringen's micropolar elastic solid.

View full abstract

This paper presents a unified analysis of plate bending problems due to classical plate theory containing an elliptic hole or elliptic rigid inclusion under singular point moments and point discrepancies of deflection and slope as well as applied moments and torque at infinity. The analogies existing between the mathematical formulations of the transverse bending of elastic plates and two-dimensional stress problems are used for analytical treatments.

View full abstract

The cyclic deformation behaviour of TiNi shape memory alloy under various thermomechanical paths was investigated experimentally. The main results are sunlmarized as follows. For the stress-induced martensitic transformation and the reverse transformation under cyclic loading-unloading or heating-cooling, the transformation stress decreases and the transformation temperature rises with an increase in the number of cycles. The amount of variation in the stress and the temperature for the stress-induced martensitic transformation is larger than that for the reverse transformation. These variations are significant if the deformation is repeated in the region of high stress. The nonrecoverable strain which appears with cyclic deformation is large for high stress. The amount of variation in the transformation stress, the transformation temperature and the nonrecoverable strain is large in the early cycles but decreases after a certain number of cycles.

View full abstract

In the present paper, the stress distribution around the apex of dissimilar materials for two-and three-dimensions is calculated using an improved boundary element method (BEM), and the order of the singularity for the stress distribution is obtained from the numerical results. It is shown that the numerical solution is very accurate even at the internal points near the boundary or apex. For the sample structure bonded with ceramics (Al₂O₃) and Cu, it is proven that the order of the stress singularity obtained in this way is in good agreement with that of Bogy's theory. For three-dimensional bonded structure, it is found that the stress singularity is very strong around the apex. This indicates that the possibility of crack generation in this part is very high. Through detailed investigation of the stress field, a simple method to moderate the stress concentration is suggested.

View full abstract

Quenching mediums of low boiling point such as water or oil, are generally used in thermal shock tests. However, these mediums are boiled in the course of the cooling of the specimen. Therefore, the boiling heat transfer coefficients vary largely according to the surface temperature of the specimen, the subcooled temperature of the medium, and the falling speed of the specimen. In this paper, transient boiling curves of water and silicon oil are obtained by experiments and using these curves, thermal stresses of ceramics at quenching in water are analysed numerically by the finite-element method. As a result, it is clarified that the transient boiling heat flux is larger than that of pool boiling, and the maximum thermal stress is affected by transition or film boiling. In the case of specimen with a small diameter, maximum thermal stress decreases in spite of the increase of temperature difference.

View full abstract

Recently, sensitivity analysis of properties such as stiffness and vibration has been applied to various design stages of vehicle structures. However, it is difficult to apply sensitivity analysis to a crash problem because it involves material and geometric nonlinearity. In the first report, we demonstrated a new approach to dynamic buckling problems. As a simple study, we investigated the dynamic properties of elastic buckling of a long beam under impact in an axial direction. The modification of the beam based on a sensitivity for the dynamic buckling mode has a large effect on buckling strength. The dynamic buckling mode of the beam is regarded as high-order static buckling in the sensitivity analysis. As a second report, we investigate in detail the dynamic properties of elastic buckling of a beam in detail. The relationship between impact velocity and buckling strength is shown for various lengths of beams. The relationship between dynamic and static buckling modes is also shown and discussed.

View full abstract

A method for evaluating superplastic properties of a material with different stress/strain-rate relations when deformed in tension and in compression was studied by means of a bend test. To begin with, four-point bending of a beam under a constant deflection rate for a case K_t≠K_c and m_t = m_c in the constitutive equation of superplasticityσ=K_ε^m was analyzed, where σ, ε, m and K are stress, strain-rate, strain-rate sensitivity index and constant, respectively, and the subscripts t and c refer to tension and compression sides of the beam. Then a compression test of Y-TZP was conducted and the result was discussed and compared as a whole with the previous results of the bend and tension tests on the same material. The normal stress and normal strain-rate on the mean strain-rate points in the tension and compression sides of the beam were calculated by substituting the bend test result into equations derived in this study. The normal stress/strain-rate relations calculated on the tension and compression sides of the beam agreed fairly well with those relations obtained directly from the tension and compression tests, respectively.

View full abstract

Recently, in order to replace experiments by computer simulations, the required scale of the model to be analyzed has become extremely large. To solve such a large-scale problem, various parallel FEM algorithms have been investigated to date. Among them, the domain decomposition method (DDM) is suitable for parallel computing because of the small memory requirement, arbitrary decomposition of the domain, and easy implementation on parallel systems. In this study, we applied the DDM algorithm to dynamic FEM analyses, and implemented it on a single CPU system. It is well demonstrated through some numerical examples that the present DDM-based dynamic FEM code is very efficient and superior in speed and memory requirements.

View full abstract

This paper concerns a strategy for using the boundary element shape optimization process for a post-process of finite element solver. In this strategy, firstly, from the finite element analysis result of the initial assumed problem, the subregion to be optimized is determined and separated from the remaining region. The boundary element mesh of the subregion is automatically generated from the initial finite element mesh and employed as the initial data of the shape optimization. The present method is applied to the local shape optimization of a two-dimensional elastic problem in order to confirm its validity.

View full abstract

This paper proposes an approach for design optimization aiming at reduction of mechanical vibration under complicated conditions. Machine structures in operation generally need to reduce the vibrational responses, which must be evaluated at many response and excitation points. Moreover, multiple vibrational natural modes of the machines must also be taken into account. The design optimization problem is first formulated on the basis of a multi-objective optimization method. Then, the weighted mini-max method with the Tchebychev norm is presented to solve the optimization problem. This approach is then demonstrated on a motorcycle frame design in which the frame is subjected to engine excitation with many directional components and the vibrational response is evaluated at three points: a handle-bar end, a front footrest, and a rear footrest.

View full abstract

The method and procedures are proposed for full experimental identification of the Biot model of fluid-saturated porous elastic media by spectrum analysis. The frequency response function is derived for incident and transmitted ultrasonic waves through an inclined porous elastic layer immersed in fluid. It is shown that amplitude ratios of three terms in that function and three travel time differences of the first and second compressional waves and the shear wave from the sound wave in the fluid can be determined by curve fitting in the Bode and/or Nyquist diagram and that the data obtained from the curve fitting and the five equations derived enable full identification of-the Biot model, that is, estimation of four independent elastic moduli and a dynamic tortuosity. The proposed method can be applied to the case of negligible fluid viscosity.

View full abstract

It is important to estimate dimensions of large deformation and stress for flexible material in order to develop highly reliable sheet handling devices. In this paper, a spring-mass beam model is introduced to evaluate deformation behavior of the sheet. This model consists of rotational springs, shear springs and masses. First, the beam model for large deformation of the sheet is presented. Then the calculation results of the deformation configuration and acting force the spring-mass beam model are compared with the theoretical and experimental results. It is concluded that the simulation using the beam model proposed here gives good agreement with experimental results.

View full abstract

Material characterization by ultrasonic microscopy is presented. Velocities of leaky surface waves and leaky skimming compressional waves propagating on an aluminum plate are measured using transducers with lens apertures properly designed for detecting these waves. From these velocity values, the Young's modulus and Poisson ratio of the aluminum can be calculated. Surface velocity changes in porous ceramics are measured and the velocity values are clearly found to be closely related to the hardness of the ceramics. The velocities are also related to porosities of the ceramics. However, in metals, the values of surface velocity are related not to the hardness but to the metal structures. In a layered structure, characterization of materials can be analyzed using velocity dispersion properties measured by ultrasonic microscopy.

View full abstract

In order to study the relationship between cavitation bubble collapse pressures and deformations on materials, a new technique was developed using a pressure sensor which can measure bubble collapse pressures and surface damage at the detecting area simultaneously. Impact loads occurring in a vibratory device were measured and compared with pit or crater sizes observed with a microscope at an early stage on the surfaces of pure Al, Cu, Fe and Ni. A linear relationship was obtained between impact load and the area of pit or crater. The hardnesses of various metals under impact loads at collapse in cavitation bubbles are about one order larger than those under static loads. Critical impact load necessary to form a pit or crater was found to differ among the materials.

View full abstract

In the previous paper, micromechanical analysis was performed on a discontinuous fiber-reinfor-ced composite material showing interfacial sliding behavior between a fiber and a matrix, and analytical expressions for energy release rate and elastic modulus, etc., were derived. In the present paper, evaluation of mechanical properties of a composite with such an interfacial sliding characteristics, like a ceramic-based composite, is conducted based on the theoretical analysis of the previous paper. As a result, the calculated elastic modulus is in good agreement with the experimental one. Furthermore, fracture toughness of such a composite can be evaluated in terms of the critical size of the crack which travels across the material in an unstable manner. The effect of material parameters such as the volume fraction of the fiber, friction stress between the fiber and matrix, and the magnitude of sliding, on toughness can be explained by the interfacial phenomenon, whose results are consistent with common experimental facts. It should also be particularly emphasized that there exists an optimal value of friction stress between a fiber and a matrix for obtaining a tougher composite

View full abstract

Cavitation erosion behaviors of silicon nitride, zirconium oxide and silicon carbide were investigated by a vibratory test. The erosion of the ceramics originated at pores on specimen surface. Particles with a diameter of 1 to 2 microns were removed around pores, due to the cavitation. The erosion resistance largely depends on the density of pores, ρ. Also, the growth rate of erosion-pit diameter varies from material to material, which is inversely proportional to the fracture toughness, K_Ic. Therefore, the erosion resistance can be evaluated by the value ofρ·K_Ic^-2 for these ceramics.

View full abstract

An experiment to examine the possibility of identification of coating defect was performed on a model of pipe-line in sea water. The pipe-line model which is made of SS steel pipe is set in resin made vessel which was filled with 0.001% NaC1 solution. A small current was impressed to the model. In order to keep the temperature of the solution at 30±0.5°C, the vessel was placed in an isothermal chamber filled with water. The distribution of the potential on the surface of the solution was measured using an electrometer with the glass capillary tip. A platinum counter electrode was used with a reference electrode ( SCE : Saturated Calomel Electrode ). The Poisson's equation, which describes the potential in the electrolyte, was so1ved by the boundary element method. The detection was performed by minimizing the difference between actual and calculated values of potential. By using the data of the above experiment, locatons of coating defects were succesfully detected, and the corrosion rates were correctly estimated.

View full abstract

The effect of plate thickness on the propagation rate of a mode I surface crack under biaxial stresses by the simultaneous action of plane bending and cyclic torsion was examined, and the crack opening displacement COD was measured by magnifying a scanning electron micrograph 15000 times. The relationship between the propagation rate da/dN and the stress intensity factor rangeΔK depends on both the biaxial stress ratio C and the maximum principal stress amplitudeσ₁. COD measured under the same conditions ofσ₁ and crack length increases with the decrease in C. A unique relationship is obtained between da/dN and crack tip opening displacement CTOD, irrespective of C and σ₁. Thus, it can be concluded that the main factor which controls da/dN is CTOD.

View full abstract

The measurement of the crack opening displacement during a hydraulic fracturing test is examined to estimate the crack propagation behavior. Hydraulic fracturing tests in granite were carried out under no confining pressure in the laboratory. During the test, the borehole deformation was measured by the clip-gauge-type sensor installed on the center of the injection unit. Three sensors were used to measure the deformation of three different orientations aroulld a borehole at the same time. The pressure-displacement curve indicates the proportional relation initially, and abrupt increase in the deformation is observed at the breakdown. The elastic modulus is estimated from the initial tangent of the pressure-displacement curve. The crack opening displacement, when the fluid pressure starts to act to the fracture surface, is determined from the deformation behavior. The results of the measurements of the displacements in different orientations around the borehole indicate the direction of the crack propagation induced by the hydraulic pressure. Finally, using the measured crack opening displacement, the final crack size is estimated approximately based on the linear elastic fracture mechanics.

View full abstract

This paper describes an application of a hierarchical neural network (HNN) that has excellent nonlinear projection ability to inverse analysis of elastic contact problems. HNN is used to estimate contact stress distribution from some displacements or strain data of elastic contact body. The learning data of HNN, which are the loads and displacements (or strains) of a contact body, are calculated by using the finite element method (FEM). That is, HNN simulates the stiffness matrix relation between displacement and contact load. The effectiveness of HNN on the inverse problems is demonstrated through several numerical examples.

View full abstract