On-line steady-state data reconciliation for advanced cost analysis in the pulp and paper industry

The objective of this thesis was to develop a methodology for online manufacturing cost analysis using real-time process and cost data available from the information management systems that would be capable to assess actual product margin costs, and use this information for operational and tactical decision-making. Furthermore, the knowledge from applying this methodology can be explored in the strategic decision-making level for addressing the process-cost impact of retrofit design alternatives. This methodology consists of three major steps. First, a signal processing technique, based on multiscale wavelet transformation and filtering, is applied to simultaneously analyze every segment of the plant-wide instrumentation network. A second step further improves process data quality by confronting the plant-wide set of variables to the underlying fundamental process model using data reconciliation techniques. The plant-wide manufacturing information is updated by coaptation and correction of biased measurements. Third, this operational knowledge is integrated with the financial data in an operations-driven cost model to calculate and analyze production costs of operating regimes for short and long term benefits of the company. The methodology is applied to a case study considering current newsprint mill characteristics and potential retrofit biorefinery implementation.;It was found that using a combination of the wavelet technique with classical data reconciliation technique provides multiple facility-level benefits. An online implementation of this technique was able to provide a significant number of data sets that were extracted from the information management systems as potential candidates to represent plant-wide and near steady-state operation. These data sets have provided sufficient statistical basis for characterising manufacturing operation per different operating regime. By doing this automatically, the methodology was able to enhance the quality of data and highlight the manufacturing region where the uncertainty in measurements is significant. The number of near steady-state candidates that can be detected was increased, when state identification parameters were being relaxed. However, it was shown that the uncertainty in the resulting data sets is increasing with relaxing the steady state assumption. In the analyzed rather simple newsprint operation, the technique was able to acquire multiple near steady-state data sets representing plant-wide operation with satisfactory level of accuracy. Moreover, the on-line implementation in combination with data reconciliation method, helped to improve measurement sensor performances by identifying sensors with systematic errors. This valuable information was then used to tune individual instruments further, and hence improve the overall performance of the methodology.;The manufacturing cost assessment based on these data sets that represent individual operating regime, has provided a new insights into the cost structure of the facility with transparent and visible process-cost interpretation capabilities. The application of the overall methodological framework, in the context of real production processes, has proved the ability to identify favourable and costly operating regimes when producing the same product grade. The characterisation and interpretation of the cost variances between individual regimes as well as within the same operating regime helped to identify process problems. Further application of the methodology for evaluating manufacturing costs of retrofit design scenarios have shown the ability to exploit the current operations-driven manufacturing knowledge based on regimes to enhance strategic decision-making at the facility. The results from this application showed that the operational profitability of new integrated production lines strongly depends on the operational differences in current manufacturing regimes of core business products. These differences in manufacturing costs can be visible from a process perspective and enable assessment of individual future product and mix-product margins. This information is essential for margin-centric supply chain planning of the enterprise and for exploring process flexibility to achieve an optimal production profile according to market conditions.;Future work includes the expansion of this work into strategic investment decision-making at the corporate level in order to enhance tactical and strategic planning. Furthermore, marginal cost analysis based on real-data and operations-performance analysis could be included in the methodological framework in order to obtain more flexible forest biorefinery retrofit designs with good strategic fit. (Abstract shortened by UMI.).