ENBIS-14 in Linz21 – 25 September 2014; Johannes Kepler University, Linz, Austria Abstract submission: 23 January – 22 June 2014
The following abstracts have been accepted for this event:
A Cause and Effect Diagram and AHP Based Methodology for Selection of Quality Improvement Projects
Authors: Ozlem Muge Aydın (Hacettepe University), Amir Shaygan (Hacettepe University), Erdi Dasdemir (Hacettepe University)
Primary area of focus / application: Quality
Secondary area of focus / application: Six Sigma
Keywords: Cause and effect diagram, Analytical Hierarchy Process, Project selection, Six Sigma, Quality improvement
Submitted at 30-May-2014 14:27 by Ozlem Testik
Accepted (view paper)
On the Effect of Prediction Uncertainty in the Design of New Pharmaceutical Products
Authors: Pierantonio Facco (University of Padova)
Primary area of focus / application: Quality
Secondary area of focus / application: Process
Keywords: Pharmaceutical product/process development, Latent variable modeling, Projection on latent structures, Uncertainty
Submitted at 30-May-2014 14:45 by Pierantonio Facco
Accepted (view paper)
One very useful application of multivariate statistics in pharmaceutical product/process development is the inversion of the latent variables models (Jaeckle and MacGregor, 1998). In their direct form, latent variables models are usually utilized to explain the correlation between a set of input variables (e.g., raw materials characteristics, settings, process parameters, etc..) and a set of outputs (e.g., the target product quality profile), often with predictive purposes. In their inverse form, latent variables models are utilized in product and process development to suggest the most appropriate raw materials characteristics, settings, process parameters, etc… (i.e., the design parameters) that are expected to lead to a product of desired quality. Accordingly, these methodologies may drive developers and scientists to identify, within the knowledge space of the historical data, the design space within which experimental campaigns may be carried out (MacGregor and Bruwer, 2008).
However, latent variable models are typically affected by uncertainty (e.g.: on calibration data, Faber and Kowalski, 1997; on the model parameters, Martens and Martens, 2000; on prediction), which should be considered in the exercise of model inversion for product formulation and process design.
In this presentation a methodology is proposed to characterize how the prediction uncertainty (Zhang and Garcia-Munoz, 2009) backpropagates from the quality of a desired new product to the design parameters suggested by the model inversion. In particular, once the knowledge space is identified by means of a latent variable model that correlates the design parameters to the product quality based on a historical dataset of other products already manufactured, the proposed methodology segments the knowledge space to identify a subset of it that can be thought as a reasonable design space within which the experimental campaign may be carried out.
The proposed methodology is tested in a typical pharmaceutical process, namely powder granulation.
Department of Health and Human Services - U.S Food and Drug Administration, 2004. Pharmaceutical CGMPs for the 21st century — A risk-based approach. Final report.
Faber, Kowalski, 1997. J. Chemom. 11, 181–238.
Fernández Pierna, Jin, Wahl, Faber, Massart, 2003. Chemom. Intell. Lab. Syst. 65, 281–291.
Jaeckle, MacGregor, J., 1996. Comput. Chem. Eng., European Symposium on Computer Aided Process Engineering-6 20, Supplement 2, S1047–S1052.
MacGregor, Bruwer, 2008. J. Pharm. Innov. 3, 15–22.
Martens, Martens, 2000. Food Qual. Prefer. 11, 5–16.
Reis, Saraiva, 2005. AIChE J. 51, 3007–3019.
Zhang, Garcia-Munoz, 2009. Chemom. Intell. Lab. Syst. 97, 152–158.
Case Presentation: Screening Experimentation on a Pilot Scale Oven for Factor Identification
Authors: Søren Juhl Pedersen (DTU Food), Murat Kulahci (DTU Compute and Luleå University of Technology), Stina Frosch (DTU Food)
Primary area of focus / application: Design and analysis of experiments
Secondary area of focus / application: Education & Thinking
Keywords: Screening, Statistical thinking, Engineering interpretation, Fractional factorial design
Submitted at 30-May-2014 14:46 by Søren Juhl Pedersen
The specific question posed by the company concerned how the distance between air inlet holes and baking sheet would influence the convective heat transfer. A fractional factorial design was used and a full fold-over performed in the follow-up experimentation. The factors / variables of the experiment were; speed of airflow, temperature, distance between inlet holes and baking sheet, band speed, amount of fresh air intake and restriction on airflow through inlet. The results comply with the theoretical expectations by highlighting airflow as the most important factor and the distance as a factor of minor influence but there was a caveat. The findings suggested a faulty design of the specific oven and also the measurement equipment.
The results raised important questions for further experimentation concerning optimization of convective baking processes. One of the key learning experiences from the experiment was the valuable insight on process improvement gained by introducing statistical thinking in engineering applications and also introducing engineering knowledge into the framework of statistical process improvement. For example, the systematic data collection scheme through designed experiments generated the insight of equipment related issues that would not have been detected otherwise.
Modelling a Fragmentation Process of Printed Circuit Boards
Authors: Antonio Pievatolo (CNR-IMATI)
Primary area of focus / application: Modelling
Secondary area of focus / application: Design and analysis of experiments
Keywords: Electronic waste recovery, Population balance models, Markov process, Planning of experiments
Submitted at 30-May-2014 16:52 by Antonio Pievatolo
Response Surface Approximation for Profile Monitoring in Circular Domains
Authors: Esperan Padonou (Ecole des Mines de Saint-Etienne, STMicroelectronics), Olivier Roustant (Ecole des Mines de Saint-Etienne), Jakey Blue (Ecole des Mines de Saint-Etienne), Hugues Duverneuil (STMicroelectronics)
Primary area of focus / application: Process
Secondary area of focus / application: Quality
Keywords: Profile monitoring, Statistical Process Control, Zernike regression, Gaussian process regression, Circular domains
Submitted at 30-May-2014 17:04 by Esperan Padonou
The ICs are produced on thin slices of semiconductor materials, called wafers. In our study, a wafer is a 300-mm diameter circular domain. To monitor its quality, several types of physical metrology measurements (such as thickness, depth, width, angles and overlay) are collected on a fixed number of preselected locations. However, standard SPC techniques hardly detect defects such as curvature change, which is critical in semiconductors manufacturing, as different wafer “profiles”, related to different process issues, may have the same mean and variance over the measurement points. Furthermore, spatial correlation is not taken into account. To overcome these problems, the two-phase profile monitoring procedure  is often preferred:
1. For each time step, fit a response surface based on the measurement points ;
2. Monitor the response surface parameters over time.
In this work, we focus on step 1. Our contributions are twofold. Firstly, we compare different approximation techniques in circular domains: Zernike regression [1, 4] , and Gaussian process regression [1, 2] with standard and customized covariance kernels. Secondly, we exhibit the link between the model and key process parameters.
 G. Pistone and G. Vicario, Kriging prediction from a circular grid: application to wafer diffusion, A.S.M.B.I., 29(4), 350–361, 2013
 C.E. Rasmussen and C.K.I. Williams, Gaussian Processes for Machine Learning, The MIT Press, 7–30, 2006.
 W.H. Woodall, Current research on profile monitoring, Production, 17(3), 420-425.
 F. Zernike, Diffraction theory of the cut procedure and its improved form, Physica, 1, 689–704, 1934.
Application of Kansei Engineering to Design an Industrial Enclosure
Authors: Lluis Marco-Almagro (UPC Universitat Politécnica de Catalunya, Barcelona Tech), Xavier Tort-Martorell (UPC Universitat Politécnica de Catalunya, Barcelona Tech)
Primary area of focus / application: Business
Secondary area of focus / application: Consulting
Keywords: Kansei engineering, Statistical engineering, Cluster analysis, Factorial designs, Ordinal logistic regression, Data visualization
Submitted at 30-May-2014 18:34 by Xavier Tort-Martorell
Accepted (view paper)
Kansei Engineering studies typically follow a model with three main steps: (1) spanning the semantic space: defining the responses, those emotions that will be studied; (2) spanning the space of properties: deciding on the technical properties of the products that can be freely changed and that might affect the responses (factors in a DOE factorial design) and (3) the synthesis phase, where both spaces are linked (that is, how each factor affects each response is discovered).
In an earlier paper we claimed that KE is a good example of what Roger W. Hoerl and Ron Snee call statistical engineering: focusing not in advancement of statistics – developing new techniques, fine tuning existing ones – but on how current techniques can be best used in a new area. This presentation is a practical application of the ideas exposed there to the design of electrical enclosures.
The presentation will show how well-known statistical methods (DOE, principal component analysis and regression analysis) are used together in conjunction with other non-statistical techniques and in the presence of practical real world restrictions to discover how different technical characteristics of the enclosures affect the selected “emotions”.