Richard A. Hardin

Fatigue of 8630 cast steel in the presence of porosity

Abstract Fatigue and monotonic test specimens having porosity ranging from micro- to macroscopic levels were cast from 8630 steel. Monotonic and fatigue properties were obtained to determine the effect of porosity on the mechanical performance of the cast steel. Axial fatigue tests were conducted under fully reversed conditions in both strain and load control on specimens containing microporosity, and in load control for specimens containing macropores. Monotonic tests revealed that specimens containing microporosity had strength properties comparable to sound material, but with substantially reduced ductility (76% less reduction in area). At stress amplitudes of 126 MPa, microporosity specimens were found to have lives greater than 5 million cycles (run-out) whereas macroporosity specimens had fatigue lives in the 102–104 cycle range at the same stress level. Fatigue lives for macroporosity specimens were in a range from 104 to 106 cycles when tested at the lowest stress amplitude, 53 MPa. The measured specimen elastic modulus was found to vary with porosity volume. Specimens with higher measured modulus outperformed the lower modulus specimens. Fatigue lives of the cast steel specimens were calculated using conventional models of fatigue behaviour, the strain–life and linear elastic fracture mechanics (LEFM) approaches. Life calculations made using the strain–life approach gave good agreement with measurements for specimens having microporosity, but this approach gave non-conservative results for macroporosity. LEFM modelling gave non-conservative results for both micro- and macroporosity specimens. For specimens with macroporosity, the calculations are difficult because of the porositys complex shape and large size relative to the specimen, and the inability to determine the specific macropores responsible for fatigue failure of the specimens which is necessary for direct model-measurement comparisons.

Analysis of ASTM X-ray shrinkage rating for steel castings

This paper presents the results of two different studies that examined the ASTM x-ray shrinkage rating system for radiographs of steel castings. The first study evaluated the repeatability and reproducibility of x-ray shrinkage ratings through a statistical study of 128 x-rays that were each rated seven different times. It was found that the seven ratings for each x-ray were in unanimous agreement on both shrinkage type and level for 12.5% of the x-rays. All of the x-rays that had unanimous agreement for both type and level were either completely sound or very unsound (Level 5). The largest variance was found to occur in Level 2 and 3 x-rays, which had 95% confidence intervals of about ±2 x-ray levels. The average 95% confidence interval for all 128 x-rays was ±1.4. The second study involved an effort to determine the shrinkage severity level of x-rays through digital analysis of scanned radiographs. It was found that defect area and circumference correlated reasonably well with x-ray level, but only if the shrinkage type was correctly determined first.

Reliability based casting process design optimisation

Abstract Deterministic optimum designs are unreliable without consideration of the statistical and physical uncertainties in the casting process. In the present research, casting simulation is integrated with a general purpose reliability based design optimisation (RBDO) software tool that considers uncertainties in both the input variables as well as in the model itself. The output consists of an optimal design that meets a specified reliability. An example casting process design is presented where the shape of a riser is optimised while considering uncertainties in the fill level and riser diameter. It is shown that RBDO provides a much different optimum design than a traditional deterministic approach. The deterministic optimal solution offers a 12% increase in casting yield over typical safety margin design practice, but has an unacceptable 61% probability of failure. The RBDO design has a 7% increase in casting yield over the safety margin approach and a probability of failure of 4·6%.

Integrated design of castings: effect of porosity on mechanical performance

Porosity can significantly reduce the strength and durability of castings in service. An integrated design approach has been developed where casting simulation is combined with mechanical performance simulations. Predictions of the porosity distribution from the casting process simulation are transferred to and used in stress and fatigue life simulations. Thus, the effect of casting quality on service performance can be evaluated. Results of a study are presented where the measured porosity distribution in cast steel specimens is transferred to an elasto-plastic finite-element stress analysis model. Methods are developed to locally reduce the mechanical properties according to the porosity present, without having to resolve individual pores. Plastic deformation is modeled using porous metal plasticity theory. The predictions are compared to tensile measurements performed on the specimens. The complex deformations and the reductions in the ductility of the specimens due to porosity are predicted well. The predicted stresses are transferred to a fatigue analysis code that takes the porosity distribution into account as well. The measured and predicted fatigue lives are also in good agreement. Finally, the results of a case study are presented that illustrate the utility of the present integrated approach in optimizing the design of a steel casting.

Prediction of porosity characteristics of aluminium castings based on X-ray CT measurements

ABSTRACT Porosity is a main factor limiting the fatigue performance of aluminium castings. Using micro X-ray computed tomography, size and morphology characteristics of porosity distributions are analysed for material from a cast Al–8Si–3Cu–(Sr) crankcase as well as from cast Al–8Si–3Cu–(Sr), Al–7Si–0·5Cu–Mg–(Sr) and Al–7Si–0·5Cu–Mg–(Na) cylinder heads. Correlations are developed between the porosity volume percentage and mean and maximum pore sizes. Two characteristic size measures of the porosity distribution are identified: the volume weighted spherical mean diameter and the volume weighted mean envelope diameter. Both correlate linearly with the corresponding diameters of the largest pore. The pore morphology is described by a volume weighted mean sphericity. This mean sphericity and the local amount of porosity are used to predict the mean and maximum pore sizes of the porosity distributions. These correlations will find applications in integrated computational materials engineering.