A case-based and practical approach for multivariate modeling of grinding process

Grinding process is claimed to be the only possible cost-effective means of shaping engineering part(s) or component(s) into finished products with required surface topology. However, because of the complicated relationship(s) between several phenomena and entities involved in a grinding process, determining functional relationship between relevant input(s), in-process variable(s), and output quality characteristic [or response] remains a critical and difficult task for researchers and practitioners. The complexity of the problem further increases as number of responses considered for modeling is more than one, and the situation is so-called multivariate or multiresponse surface modeling. Although various mechanistic and empirical modeling approaches have been proposed in literature, they are found to be of limited validity, with applied functional approximation and simplistic assumption(s). In addition, empirical models are sometimes based on off-line designed experimental data, which are expensive or uneconomical in many line layout type mass-manufacturing situations. This paper attempts to provide a case-based practical approach for practitioners, with detailed steps involved in multivariate modeling, based on conventional regression technique, and unconventional artificial neural network-based technique, using real-time production data. The two case examples selected to verify the approaches are from an automobile engine manufacturing unit located in eastern India.     

Transformation-Induced, Geometrically Necessary, Dislocation-Based Flow Curve Modeling of Dual-Phase Steels: Effect of Grain Size

The flow behavior of dual-phase (DP) steels is modeled on the finite-element method (FEM) framework on the microscale, considering the effect of the microstructure through the representative volume element (RVE) approach. Two-dimensional RVEs were created from microstructures of experimentally obtained DP steels with various ferrite grain sizes. The flow behavior of single phases was modeled through the dislocation-based work-hardening approach. The volume change during austenite-to-martensite transformation was modeled, and the resultant prestrained areas in the ferrite were considered to be the storage place of transformation-induced, geometrically necessary dislocations (GNDs). The flow curves of DP steels with varying ferrite grain sizes, but constant martensite fractions, were obtained from the literature. The flow curves of simulations that take into account the GND are in better agreement with those of experimental flow curves compared with those of predictions without consideration of the GND. The experimental results obeyed the Hall-Petch relationship between yield stress and flow stress and the simulations predicted this as well.     

Online integrated visual inspection and sorting system for fast reactor fuels

Mixed oxide (MOX) fuel for prototype fast breeder reactor (PFBR) is designed to have initial burn up of 100,000 MWD/T. The major differences from thermal reactor fuel are relatively smaller dimension with central hole and higher plutonium concentration (21% and 28% of PuO₂) MOX pellets which are loaded into 2.5 m long clad tubes with depleted UO₂ blanket pellets at either end of the MOX stack. The relatively smaller dimension of fuel pellets for PFBR results in large volume at fabrication and inspection. To ensure fast and accurate inspection and sorting of as sintered pellets with less radiation exposure to personnel an integrated on line pellet inspection system for remote visual inspection and sorting of pellets based on diameter has been developed. Details of the integrated pellet inspection system developed at Advanced Fuel Fabrication Facility, Bhabha Atomic Research Centre, Tarapur along with the results of the performance trials has been described in this paper.     

Modeling Blockage Failures in Sewer Systems to Support Maintenance Decision Making

The objective of this research is to develop and implement a stochastic method that can be applied to characterize random failures in critical infrastructure systems. We particularly focus on blockage failures in sewer systems that are nonmechanistic and result from combination of external factors, including deterioration in condition. The method was implemented using a data set consisting of sewer blockage failure records from a small municipality. Statistical tests were conducted to: (1) ensure that available data set is representative and (2) estimate parameters of distributions that appropriately characterize failure event arrival pattern. Failure trends were also analyzed to identify the influence of local factors and justify the choice of the distributions used to characterize interarrival times. Based on the analysis, we explored the challenges in developing a reliability model across the life cycle of a sewer system. In addition, specific examples were also presented to illustrate how the method can be applied to support system maintenance decisions. The results of this study illustrate how the memoryless property can be assumed in analyzing failure events, while explicitly considering context specific influences. Finally, the methods described in this paper are extensible and can be applied generally to analyzing random failures in other infrastructure systems as well.     

Sintered 434L ferritic stainless steels

Stainless steel powder metallurgical parts are increasingly used in automotive, aircraft, business machines, chemical, marine, tools and hardware industries. Although some results on sintered austenitic stainless steels have been reported, there is no information on a relatively cheaper ferritic stainless steel series of alloys. In the present paper, the properties of 434L ferritic stainless steel have been comprehensively evaluated and their nature discussed.     

Radiation-Induced graft copolymerization of methacrylic acid onto polypropylene fibers. IV. Thermal behavior

Thermal behavior of polypropylene-g-poly(methacrylic acid) fibers prepared by graft copolymerization of methacrylic acid onto polypropylene fibers, using simultaneous-irradiation technique, was evaluated by thermogravimetric analysis, differential scanning calorimetry, and limiting oxygen index measurements. In general, the thermal properties of polypropylene fiber were markedly improved by the grafting of methacrylic acid. The inherent crystallinity of polypropylene, as deduced from DSC, did not show any change in the grafted fibers, suggesting that the grafting occurs in the amorphous region, without disrupting the crystalline part of the polymer. The LOI of grafted fibers showed an increase over the original fiber.     

A new boundary element method formulation for three dimensional problems in linear elasticity

Summary A new boundary element method (BEM) formulation for planar problems of linear elasticity has been proposed recently [6]. This formulation uses a kernel which has a weaker singularity relative to the corresponding kernel in the standard formulation. The most important advantage of the new formulation, relative to the standard one, is that it delivers stresses accurately at internal points that are extremely close to the boundary of a body. A corresponding BEM formulation for three dimensional problems of linear elasticity is presented in this paper. This formulation is derived through the use of Stokes' theorem and has kernels which are only 1/r singular (wherer is the distance between a source and a field point) for the displacement equation. The standard BEM formulation for three-dimensional elasticity problems has a kernel which is 1/r ² singular.With 2 Figures     

Rapid yield estimation as a computer aid for analog circuit design

A rapid yield estimation methodology that aids the analog circuit designer in making design tradeoffs that improve yield is presented. This methodology is based on using hierarchical evaluation of analysis equations, rather than simulations, to predict circuit performance. The new analog rapid yield estimation (ARYE) method has been used to predict the yield of two-stage operational amplifiers and has been incorporated into the Carnegie Mellon University (CMU) analog design system (ACACIA). An example of how ARYE allows analog designers to quickly explore the impact of design changes on yield is presented. The primary goal of ARYE is to make numerous early predictions of parametric yield economical for the analog circuit designer