(Received 9 February 2009; accepted 26 January 2010)
Published Online: 2010
| ||Format||Pages||Price|| |
|17||$25||  ADD TO CART|
Statistical models for prediction of potential compressive strength up to 28 days of Portland clinker from production conditions in a cement kiln have been evaluated. The potential compressive strength of clinker was predicted from the mineralogy part of the microstructure of cement. The mineralogy of the clinker was characterized by X-ray diffraction analysis (XRDA) and the potential compressive strength was predicted from XRD-patterns of two selected 2θ ranges. The statistical method applied for evaluating the prediction models was partial least square regression (PLS). The influence of the production variables was examined by sensitivity analysis using simulation by optimization of variation, and prediction. Significant minimum and maximum strength at 1 and 28 days, respectively, was achieved by optimizing the production condition. The influence of production conditions alone on the strength at 28 days was higher than on the strength at 1 day. The methodology demonstrated is not limited to strength (which is easily measured directly), but it is also applicable to other more difficult/expensive achievable performance parameters that may beneficially be optimized via prediction from cement characteristics before actually documenting it for the most promising final products.
Process and Environmental Dept., NORCEM A/S, Brevik,
Centre for Advanced Data Analysis, Kgs Lyngby,
Stock #: JAI102370