Data-driven casting defect prediction model for sand casting based on random forest classification algorithm

The complex sand-casting process combined with the interactions between process parameters makes it difficult to control the casting quality,resulting in a high scrap rate.A strategy based on a data-driven model was proposed to reduce casting defects and improve production efficiency,which includes the random forest(RF)classification model,the feature importance analysis,and the process parameters optimization with Monte Carlo simulation.The collected data includes four types of defects and corresponding process parameters were used to construct the RF model.Classification results show a recall rate above 90% for all categories.The Gini Index was used to assess the importance of the process parameters in the formation of various defects in the RF model.Finally,the classification model was applied to different production conditions for quality prediction.In the case of process parameters optimization for gas porosity defects,this model serves as an experimental process in the Monte Carlo method to estimate a better temperature distribution.The prediction model,when applied to the factory,greatly improved the efficiency of defect detection.Results show that the scrap rate decreased from 10.16% to 6.68%.

Extension of Flow Behaviour and Damage Models for Cast Iron Alloys with Strain Rate Effect

Cast iron alloys with low production cost and quite good mechanical properties are widely used in the automotive industry.To study the mechanical behavior of a typical ductile cast iron(GJS-450)with nodular graphite,uni-axial quasi-static and dynamic tensile tests at strain rates of 10^(-4),1,10,100,and 250 s^(-1)were carried out.In order to investigate the influence of stress state on the deformation and fracture parameters,specimens with various geometries were used in the experiments.Stress strain curves and fracture strains of the GJS-450 alloy in the strain rate range of 10^(-4)to 250 s^(-1)were obtained.A strain rate-dependent plastic flow model was proposed to describe the mechanical behavior in the corresponding strain-rate range.The available damage model was extended to take the strain rate into account and calibrated based on the analysis of local fracture strains.Simulations with the proposed plastic flow model and the damage model were conducted to observe the deformation and fracture process.The results show that the strain rate has obviously nonlinear effects on the yield stress and fracture strain of GJS-450 alloys.The predictions with the proposed plastic flow and damage models at various strain rates agree well with the experimental results,which illustrates that the rate-dependent plastic flow and damage models can be used to describe the mechanical behavior of cast iron alloys at elevated strain rates.The proposed plastic flow and damage models can be used to describe the deformation and fracture analysis of materials with similar properties.

Three-dimensional modeling of effect of surface intermetallic phase on surface defects of Al-Fe-Si aluminum foils during twin-roll casting

Scanning electron microscopy and X-ray energy dispersive spectrum analysis show that the clusters of intermetallic AlFeSi particle are distributed on or near the aluminum foil stock surfaces heterogeneously. 3D finite element modeling shows that these clusters of hard particles induce the fracture of the nano-scale lubricant oil film at first and further lead to severe deformation in the nearby aluminum foil substrate along the rolling direction. Consequently, the optical property in this region differs from that in the surroundings, resulting in surface defects.

Cardiovascular Cast Model Fabrication and Casting Effectiveness Evaluation in Fetus with Severe Congenital Heart Disease or Normal Heart

To investigate the application and effectiveness of vascular corrosion technique in preparing fetal cardiovascular cast models, 10 normal fetal heart specimens with other congenital disease（control group） and 18 specimens with severe congenital heart disease（case group） from induced abortions were enrolled in this study from March 2013 to June 2015 in our hospital. Cast models were prepared by injecting casting material into vascular lumen to demonstrate real geometries of fetal cardiovascular system. Casting effectiveness was analyzed in terms of local anatomic structures and different anatomical levels（including overall level, atrioventricular and great vascular system, left-sided and right-sided heart）, as well as different trimesters of pregnancy. In our study, all specimens were successfully casted. Casting effectiveness analysis of local anatomic structures showed a mean score from 1.90±1.45 to 3.60±0.52, without significant differences between case and control groups in most local anatomic structures except left ventricle, which had a higher score in control group（P=0.027）. Inter-group comparison of casting effectiveness in different anatomical levels showed no significant differences between the two groups. Intra-group comparison also revealed undifferentiated casting effectiveness between atrioventricular and great vascular system, or left-sided and right-sided heart in corresponding group. Third-trimester group had a significantly higher perfusion score in great vascular system than second-trimester group（P=0.046）, while the other anatomical levels displayed no such difference. Vascular corrosion technique can be successfully used in fabrication of fetal cardiovascular cast model. It is also a reliable method to demonstrate three-dimensional anatomy of severe congenital heart disease and normal heart in fetus.

Fluid Flow and Solidification Simulation in Beam Blank Continuous Casting Process With 3D Coupled Model

Based on turbulent theory, a 3D coupled model of fluid flow and solidification was built using finite difference method and used to study the influence of superheating degree and casting speed on fluid flow and solidification, analyze the interaction between shell and molten steel, and compare the temperature distribution under different technological conditions. The results indicate that high superheating degree can lengthen the liquid-core depth and make the crack and breakout possible, so suitable superheating should be controlled within 35℃ according to the simulation results. Casting speed which is one of the most important technological parameters of improving production rate, should be controlled between 0. 85 m/min and 1.05 m/min and the caster has great potential in the improvement of blank quality.

Physical modeling of gas-liquid interfacial fluctuation in a thick slab continuous casting mold with argon blowing

Combining with the physical model of level fluctuation in a thick slab continuous casting mold with the cross-section of 1500 mm×280 mm and argon blowing, the rationalities of estimating the level fluctuation by three traditional quantitative approaches were discussed, and the effects of gas flowrate, casting speed, and the immersion depth of submerged entry nozzle （SEN） on the level fluctuation were also investigated. As a result, it seems that three traditional quantitative approaches are not very suitable for estimating the level fluctuation in a mold with argon blowing, so a new approach for estimating level fluctuation in the mold with argon blowing was presented. The experimental results show that the level fluctuation is mainly in the region around the nozzle wall. When the casting speeds are larger than a certain value, there is the escape of large bubbles near the nozzle wall, which causes an obvious increase of level fluctuation. Furthermore, optimal process parameters, viz., the gas flowrate of 6 NL/min, the casting speed of 1.1 m/min, and the immersion depth of 170 mm, are presented to restrain the level fluctuation by a physical model.

Water Modeling of Twin-Roll Strip Casting

Twin-roll strip casting is regarded as a prospective technology of near net shape continuous casting. The fluid flow field and level fluctuation in the pool have a strong influence not only on composition and temperature homogeneity of pool, but also on the strip quality. A 1 ： 1 water model of a twin-roll strip caster was set up based on the criteria of Froude number and Reynold number similarity. The level fluctuation was measured. The influence of pool depth, casting speed and feeding system configuration on level fluctuation in the pool was studied. The experimental results provided a basis for the optimization of feeding system and process parameters.

Mathematical model of heat transfer for bloom continuous casting

A mathematical model for heat transfer during solidification in continuous casting of automobile steel, was established on researching under the influence of the solidifying process of bloom quality of CCM in the EAF steelmaking shop, at Shijiazhuang Iron and Steel Co. Ltd. Several steel grades were chosen to research, such as, 40Cr and 42CrMo. According to the results of the high temperature mechanical property tests of blooms, the respective temperature curves for controlling the solidification of different steels were acquired, and a simulating software was developed. The model was verified using two methods, which were bloom pinshooting and surface strand temperature measuring experiments. The model provided references for research on the solidifying process and optimization of a secondary cooling system for automobile steel. Moreover, it was already applied to real production. The calculated temperature distribution and solidification trend of blooms had offered a reliable theory for optimizing the solidifying process of blooms, increasing withdrawal speed, and improving bloom quality. Meanwhile, a new secondary cooling system was designed to optimize a secondary cooling water distribution, including choice and arrangements of nozzles, calculation of cooling water quantity, and so on. 2008 University of Science and Technology Beijing. All rights reserved.

Numerical Simulation of Shrinkage Cavity Formation in Spheroidal Graphite Iron Castings Based on Micro Modeling

A micro-modeling method (MM) for the quantitative prediction of the shrinkage cavity formation in SGiron castings is proposed. The mathematical models describing the volume changes during the solidification ofspheroidal graphite cast iron are established based on the models of solidification kinetics. The shrinkage cavityformation of T-shaped SG iron castings is calculated with MM method. The calculated results are compared with theexperimental results. It is shown that the predicted size, shape and distribution of shrinkage cavity by MM methodare in good agreement with the measured results.

Surface turbulence in a physical model of a steel thin slab continuous caster

In the thin slab continuous casting （TSCC） of steel, the issue of optimum fluid flow is very important due to higher casting speeds and has direct influence on the formation of solidified shells and the quality of final products. In the current work, a full-scale physical mod- eling of a thin slab easter on the basis of dimensionless Reynolds and Froude similarity criteria was constructed. The flow pattern in the funnel shaped mold with a new tetra-furcated submerged entry nozzle （SEN） was investigated. To determinate optimum operational parameters, some experiments were carried out under various casting conditions. The results show that the tetra-furcated design of the nozzle leads to a special flow pattern in the mold cavity with three-dimensional recirculating flow. It is also shown that the increase of casting speed and gas injection results in surface turbulence. On the other hand, using a higher depth of SEN decreases the vortex in the free surface of the caster. To avoid surface turbulent and related casting problems, it is recommended to use 30-cm and 40-cm SEN depth at the casting speeds of 3.5 and 4.5 m/min, respectively.

Mathematical Modelling of Particle Movement Ahead of the Solid-liquid Interface in Continuous Casting

Whether the particle will be trapped by the solid-liquid interface or not is dependent on its moving behavior ahead of the interface, so a mathematical model has been developed to investigate the movement of the particle ahead of the solid-liquid interface. Based on the theory for the boundary layer, the fluid velocity field near the solid-liquid interface was obtained, and the trajectories of particles were calculated by the equations of motion for particles. In this model, the drag force, the added mass force, the buoyance force, the gravitational force, the Saffman force and the Basset history force are considered. The results show that the behavior of the particle ahead of the solid-liquid interface is affected by the physical property of the particle and fluid flow. And in the continuous casting process, if it moves in the stream directed upward or downward near vertical solid-liquid interface or in the horizontal flow under the solid-liquid interface, the particle with the diameter from 5 um to 60um can reach the solid-liquid interface. But if it moves in horizontal flow above the solid-liquid interface, only the particle with the diameter from 5 um to 10 um can reach the solid-liquid interface.

Water modeling of mold powder entrapment in slab continuous casting mold

The optimal parameters were determined by the water modeling of slab casting. It was found that there are mainly three types of mold powder entrapment in slab continuous casting, i.e., the entrapment caused by the shearing flow near the narrow face of mold, the entrapment caused by vortexes around the submerged entry nozzle （SEN）, and the entrapment caused by the Ar bubbling. Both the velocity of the surface flow and the level fluctuation of the liquids are enlarged with increasing the casting speed, reducing the submersion depth of SEN, decreasing the downward angles of the nozzle outlets, and increasing the Ar flowrate, all of which increase the tendency of mold powder entrapment. Among the four above-mentioned factors, casting speed has the largest effect.

Numerical simulation for macrosegregation in direct-chill casting of 2024 aluminum alloy with an extended continuum mixture model

An extended continuum mixture model for macrosegregation is applied to predicting Cu and Mg segregation in large-size ingot of 2024 aluminum alloy during direct chill casting（DC）. A microsegregation model using the approximate phase diagram data was coupled with macroscopic transport equations for macrosegregation profiles. Then, the impacts of transport mechanisms on the formation of macrosegregation were discussed. It is found that copper and magnesium have a similar segregation configuration from the billet center to surface. Negative segregation is observed in the centerline and subsurface, whereas positive segregation is obtained in the surface and somewhat underestimated positive segregation in the middle radius. Further, the discrepancy between the predicted and experimental results was discussed in detail. The results show that the magnesium to some extent alleviates the copper segregation in ternary alloy, compared with that in binary alloy. The predicted results show good agreement with measured experimental data obtained from literatures.

Online forecasting model of tundish nozzle clogging

A nozzle clogging online forecasting model based on hydrodynamics engineering was developed, in which the actual flow rate was calculated from the mold width, thickness, and casting speed. There is a linear relationship between the theoretical flow rate and the slide gate opening ratio as the molten steel level, argon flow rate, and the top slag weight are kept constant, and the relationship can be obtained by regression of the data collected at the beginning of the first heat in each casting sequence when the nozzle clogging does not occur. Then, during the casting, the theoretical flow rate can be calculated at intervals of one second. Comparing the theoretical flow rate with the actual flow rate, the online nozzle clogging ratio can be obtained at intervals of one second. The computer model based on the conception of the nozzle clogging ratio can display the degree of the nozzle clogging intuitively.

A mathematical model for electrochemical chloride removal from marine cast iron artifacts

The aim of this article was to theoretically study diffusion and migration of chloride ions during electrochemical chloride removal. The proposed model would enable optimization of its application by predicting the optimal treatment time and current combination. A mathematical model for simulating the transport behavior of chloride ions was developed by consideration of diffusion and migration of chloride ions when a constant DC current density was applied through the marine cast iron artifacts. The corresponding tests were conducted to validate the mathematical model. This model predicted the data of the extraction ratio of the chloride ion that correlated satisfactorily with the experimental values. An important issue in electrochemical chloride removal was to understand how chloride ions moved, taking account of diffusion and migration of chloride ions and the release of binding chloride ions. The effects of the treatment time, externally applied current density, chloride diffusion coefficient, and rate constant of release of binding chloride ion on chloride removal are studied. The specific quantitative details applied to one-dimensional model were discussed here. This article has proposed a mathematical model for the first time, which was showed to be a useful tool that can reveal the ionic transport mechanism and optimize the application during electrochemical chloride removal.

Numerical simulation of solidification morphologies of Cu-0.6Cr casting alloy using modified cellular automaton model

The purpose of this study is to predict the morphologies in the solidification process for Cu-0.6Cr(mass fraction,%)alloy by vacuum continuous casting(VCC)and verify its accuracy by the observed experimental results.In numerical simulation aspect, finite difference(FD)method and modified cellular automaton(MCA)model were used to simulate the macro-temperature field, micro-concentration field,nucleation and grain growth of Cu-0.6Cr alloy using real data from actual casting operations.From the observed casting experiment,the preliminary grain morphologies are the directional columnar grains by the VCC process.The solidification morphologies by MCAFD model are in agreement with the result of actual casting experiment well.

MATHEMATICAL AND PHYSICAL MODELING OF INTERFACIALPHENOMENA IN CONTINOUS CASTING MOULD WITH ARGONINJECTION THROUGH SUBMERGED ENTRV NOZZLE

The study on the fluid flow, meniscus oscillation, slag entrapment in continuous casting mould was conducted mathematically and experimentally. The results show that the injection of argon into submerged nozzle enhances the meniscus oscillation, thus increases the probability of slag entrapment, and the critical argon blowing flow rate, which will give rise to slag entrapment, is around 10l/min. The trajectory of bubble is affected by the bubble diameter and the molten steel flow, and the bubble diameter is dominant. The bubble with diameter 1.4mm floats fastest with 0.47m/s terminal velocity.

Multiscale modelling and simulation of single crystal superalloy turbine blade casting during directional solidification process

As the key parts of an aero-engine,single crystal(SX)superalloy turbine blades have been the focus of much attention.However,casting defects often occur during the manufacturing process of the SX turbine blades.Modeling and simulation technology can help to optimize the manufacturing process of SX blades.Multiscale coupled models were proposed and used to simulate the physical phenomena occurring during the directional solidification(DS)process.Coupled with heat transfer(macroscale)and grain growth(meso-scale),3D dendritic grain growth was calculated to show the competitive grain growth at micro-scale.SX grain selection behavior was studied by the simulation and experiments.The results show that the geometrical structure and technical parameters had strong influences on the grain selection effectiveness.Based on the coupled models,heat transfer,grain growth and microstructure evolution of a complex hollow SX blade were simulated.Both the simulated and experimental results show that the stray grain occurred at the platform of the SX blade when a constant withdrawal rate was used in manufacturing process.In order to avoid the formation of the stray crystal,the multi-scale coupled models and the withdrawal rate optimized technique were applied to the same SX turbine blade.The modeling results indicated that the optimized variable withdrawal rate can achieve SX blade castings with no stray grains,which was also proved by the experiments.

Mathematical modeling of thermo-mechanical behavior of strip during twin roll casting of an AZ31 magnesium alloy

The effect of set-back distance on the thermo-mechanical behavior of the strip during twin roll casting(TRC)of an AZ31 magnesium alloy was modeled using finite element method(FEM).Model validation was done by comparing the predicted and measured exit strip surface temperature as well as the secondary dendrite arm spacing(SDAS)through the thickness of the sheet to those measured during experiments.Model results showed as the set-back distance increases,the strip exit temperature decreases and the solidification front moves toward the entry of the roll gap.The cast strip also experiences more plastic deformation and consequently,the normal stress on the strip surface and effective strain at the strip center-line increase.Moreover,higher separating forces were predicted for longer set-back distances.Model predictions showed that changing the set-back distance by varying the final thickness has a more significant effect on the temperature and stress-strain fields than altering the nozzle opening height.

Simulation and control model for interactions among process parameters of directional solidification continuous casting

On the basis of analyzing the principles, equipment and control needs of directional solidification continuous casting (DSCC) process, the building and fulfilling methods of control model of DSCC procedure by neural network control (NNC) method were proposed and discussed. Combining the experimental researches, firstly the computer is used to simulate the effects of those solidification parameters on destination control variable (S/L interface) and the reactions among those parameters during DSCC procedure; secondly many training samples can be obtained. Moreover, after these samples are input into neural network software (NNs) and trained, the control model can be built.