ENBIS-11 in Coimbra

4 – 8 September 2011 Abstract submission: 1 January – 25 June 2011

My abstracts


The following abstracts have been accepted for this event:

  • Optimal strategies for quality improvement programs - a discrete time Mathematical Programming formulation

    Authors: Belmiro P.M. Duarte Nuno M.C. Oliveira Lino O. Santos
    Affiliation: Department of Chemical and Biological Engineering, ISEC; Dep. of Chemical Engineering; Faculty of Sciences and Technology; University of Coimbra
    Primary area of focus / application: Quality
    Keywords: Quality management , Iimprovement policy , Mathematical programming , PAF model
    Submitted at 28-May-2011 00:40 by Belmiro Duarte
    Accepted (view paper)
    5-Sep-2011 17:05 Optimal strategies for quality improvement programs - a discrete time Mathematical Programming formulation
    Organizations that improve the performance of processes, goods and services with Quality Management Systems (QMS) feature major benefits in various areas, ranging from a clear and functional chain of responsibility to market position. To follow the pace set by the Quality paradigms, an organization needs to improve continuously its QMS. Very often the adopted improvement strategy requires a continuous set of active projects to achieve gains in the quality level of the goods/services delivered. Because of the resources limitations, the Quality improvement programs should be traded off with the quality gains within the specified time horizon of the improvement programs. To address this problem it is of great interest to develop an optimal policy for the quality improvement programs that take into account the most relevant costs and the desired quality level.
    Here we propose a mathematical programming strategy to find an a priori optimal quality improvement policy that determines the best sequence of quality investments over a given time horizon. We assume that the model representing the quality level depends on the improvement investment, and on the costs of quality accounting for prevention-appraisal and failure (PAF). The resulting quality improvement policy problem involves the discretization of the time horizon, and includes a set of constraints on the quality level profile at each time instant. Each time interval corresponds to a time slot with potential actuations (increase/decrease of the investment) occurring only at its limits. The resulting mathematical formulation leads to a Nonlinear Program that is solved with a robust solver based on the Conjugate Gradient algorithm. The dynamics of the quality level are described by a lagged nonlinear model, where the quality level at each time instant includes the effect of the investment decisions on all subsequent time instants. To account for the uncertainty associated to the PAF costs and to the quality level, a set of scenarios is built with an independent stochastic component (of white noise nature) being added to the quality level and failure cost predictions. This uncertainty representation is commonly found in the modeling of Ito processes. The proposed discrete time formulation is illustrated employing a benchmark example.
  • Using Six Sigma as a Fastrack to CMMI level 5 Appraisal

    Authors: José Gonçalo Alves Oliveira Basto da Silva
    Affiliation: Quality Manager at Critical Software
    Primary area of focus / application: Quality
    Keywords: Six Sigma , CMMI , Critical Software , High Maturity , Software Development , DMAIC , Defect Density
    Submitted at 30-May-2011 11:21 by José Gonçalo Basto da Silva
    Accepted (view paper)
    5-Sep-2011 10:10 Using Six Sigma as a Fastrack to CMMI level 5 Appraisal
    Critical Software has been and endeavors continuing at the forefront of quality in software development in Portugal. Our previous goal was to reach the first CMMI-Dev [1] level 5 certification in Portugal. Taking into account the global financial crises and the especially difficult situation that Portugal was already experiencing, this objective was of the utmost importance for the company’s strategy.
    Using the Six Sigma [2] philosophy and it’s Define, Measure, Analyze, Improve, Control (DMAIC) [3] methodology, Critical Software was able to increase efficiency in the definition and effectiveness in the execution of a metrics program which was based in the measurement of the Defect Density [4] of the software that we deliver to customers.
    Main benefits of the Six Sigma philosophy and DMAIC methodology is the efficient and effective definition of metrics program and the consistent inter-relationship of the indicators in a way that provides process performance baselines and, through the correlation of the control charts, the identification of process performance models. After the first completion of the DMAIC cycle, our Defect Density was reduced from over 10 def/KLOC to approximately 0,06 def/KLOC.
    The major importance of this approach is a faster and more comprehensive approach to a metrics program that is coherent in all of its life cycle, hence, the use of the Six Sigma philosophy and the DMAIC methodology not only increased the chances of success in the CMMI pursuit but also reduced the time to appraisal.
  • Performance of Neonatal Intensive Care

    Authors: Filipa Mourão, Ana Cristina Braga, Pedro Oliveira
    Affiliation: IPVC, UM, ICBAS
    Primary area of focus / application: Design and analysis of experiments
    Keywords: CRIB , Curva ROC , Area Under the Curve , Accuracy
    Submitted at 30-May-2011 12:54 by Filipa Mourão
    Accepted (view paper)
    5-Sep-2011 16:53 Performance of Neonatal Intensive Care
    Mortality rates are considered one of the most important indicators for assessing the performance of neonatal care. Increasingly, infants born with very low birth weight (<1500g at birth) or less than 37 weeks of gestation, contribute significantly to the neonatal mortality and morbidity. However, the mortality rates have to be adjusted in accordance to the severity of illness and, therefore, several severity indexes have been developed.
    The ROC (Receiver Operating Characteristic) analysis is often used to evaluate the accuracy of the performance measures in medical diagnosis. This technique allows the study of the variation of sensitivity and specificity for different cutoff values.
    One major advantage of using ROC curves is the possibility to compare different severity indexes, as is the case in medical diagnosis.
    In this work, we use the ROC analysis to evaluate the performance of the CRIB (Clinical Risk Index for Babies) in classifying the health status of newborns with very low birth weight in two Intensive Care Units of Neonatology and thus, to compare the performance of these two units.
  • A Comparison of Two-way and Three-way Methodologies for the Prediction of Wine Age

    Authors: Ana C. Pereira, Véronique M. Gomes,Pedro M. Saraiva, Marco S. Reis
    Affiliation: Department of Chemical Engineering, University of Coimbra, Portugal
    Primary area of focus / application: Modelling
    Keywords: Prediction , Three-mode data arrays , PLS , PARAFAC , TUCKER , N-PLS
    Submitted at 30-May-2011 19:45 by Véronique Gomes
    Accepted (view paper)
    7-Sep-2011 12:40 A cCmparison of Two-way and Three-way Methodologies for the Prediction of Wine Age
    Data generated by modern hyphenated instruments, i.e. instruments that combine two or more measurement principles (such as, GC-MS, HPLC-DAD, Excitation-Emission spectroscopy, among others), have been raising relevant challenges to analysts, given the large amount of information conveyed and the structure in which it is organized, usually consisting of three-mode data arrays or cubes of data. Usually, two of these modes regard the two analytical measurement principles considered in the technique, while the third one is relative to the various samples under analysis. The analysis of such type of data can be implemented either by unfolding the cube of data in a two-way matrix, followed by the application of a classical technique (such as PLS or PCR) or it can be analyzed directly using three-way chemometric methods, already proposed and developed. The goal of this work is to compare these two approaches for empirical modeling, namely the two-way unfolded methodology, for which the PLS methods is adopted, and the three-way approach, where the PARAFAC, N-PLS and Tucker3 methods, were contemplated. The real world application scenario involves the analysis of data regarding the use of the HPLC-DAD technique, applied to Madeira wine samples with different ageing periods (1 year up to 19 years). The two and three-way chemometric methods applied to such datasets, allowed for the development of useful predictive models for the wine age, and we assess their individual merits in this case study. The root mean square error for wine age prediction varies from 1 up to 2 years, being the best results achieved for PLS approach. We verify that PARAFAC, N-PLS and Tucker3 have poorer performances when all samples are considered for modeling. However, when the first 10 years aged samples were taken into account, the model's performance improves, namely the same obtained using PLS approach. For aged wines, the three-way models present lower prediction ability, suggesting an additional and adequate data pretreatment or even non-linear three-way methodologies.
  • Quality Improvement in the Eastern Europe: Some Cases from PVC Profiles Production for Windows and Doors Industry

    Authors: Hasan Sagol
    Primary area of focus / application: Process
    Submitted at 31-May-2011 14:41 by HASAN SAGOL
    5-Sep-2011 10:30 Quality Improvement in the Eastern Europe: Some Cases from PVC Profiles Production for Windows and Doors Industry
    In the globalized world, integration of quality into social and economic aspects of industrial culture offers the potential for success of organizations. With the collapse of the USSR, the quality problems in the Ukraine has been more apparent than ever and the quest for quality improvement increased with the foreign competition. Applying the principles of quality improvement as well as the modern statistical tools in industry can help to realize the goal of gaining and sustaining a market share from the global economy. In this talk, we discuss our experiences and difficulties through several cases in improving the quality of PVC profiles used in windows and doors production in Ukraine. A discussion of various areas where quality improvement tools may find useful applications in the PVC manufacturing are provided from a practical point of view.
  • New Frontiers of Robust Design to Human Variation. Application to Motorcycle Concept

    Authors: Giovanni Lo Iacono & Stefano Barone
    Affiliation: Università degli Studi di Palermo
    Primary area of focus / application: Quality
    Keywords: Robust Design , Loss Function , Kansei Engineering , Anthropometrical variation , Six Sigma , Human-machine interface , Ergonomics
    Submitted at 31-May-2011 16:31 by Stefano Barone
    Accepted (view paper)
    6-Sep-2011 12:50 New Frontiers of Robust Design to Human Variation. Application to Motorcycle Concept
    Most of the literature on Robust Design has so far focused on making technical performances of products and processes as much insensitive as possible to the action of noise factors, often representing physical variables.
    While studying human-machine interaction, we can try to achieve system robustness to “human” noise factors, by considering both body and cognitive variation. These are new frontiers of Robust Design.
    This work started from two research lines, namely Robust Ergonomic Design and Kansei Engineering, the former addressing human body variation, the latter involving cognitive and psychological aspects.
    Here the aim is to show how to integrate these two methodologies in early concept design phases of a new motorcycle model. A complete procedure and a software tool have been purposely developed. As a result, it is possible to make valuable evaluations on the dynamic ergonomic performances and on the cognitive feelings characterising the vehicle model under development.
    The dynamic ergonomic evaluations turn around the central concept of loss function, while the kansei engineering aspects mostly rely on the adoption of DOE. In addition to this, a Six Sigma framework has been followed to better organize the flow of ideas.
    Main bibliographic references
    Andersson, P. 1997. ‘On Robust Design in the Conceptual Design Phase: A Qualitative Approach’. Journal of Engineering Design, 8, pp.75-89.
    Barone S., Lanzotti A. 2009. ‘Robust Ergonomic Virtual Design’. In: Statistics for Innovation - Statistical Design of "continuous" product innovation. Ed. P. Erto . Springer.
    Barone S., Lo Iacono G., Pampinella S. 2010, ‘Progettazione emozionale statistica: esempio di applicazione a moto di media cilindrata’, Ingegneria dell’Autoveicolo, vol. 63 n.7/8.
    Barone S., Lo Iacono G. 2011. ‘Robust dynamic comfort modelling for motorcycle riding’. Submitted.
    Nagamachi, M. 1989, Kansei Engineering. Kaibundo, Tokyo.