In this contribution, we present and evaluate a systematic framework for comprehensive uncertainty and sensitivity analysis of a model used for design of Rotary Kiln processes. We consider two sources of uncertainties, namely operational (such as measurement errors, feedstock composition, etc.) and model (key assumptions in the model used for design equations) parameter uncertainty. As model outputs for evaluation we considered the impacts of these uncertainties on key process design metrics, specifically the minimum required rotary kiln length and the conversion degree of limestone. The results revealed that the operational sources of uncertainty lead to a higher uncertainty in process design metric (e.g. standard deviation of the computed length is 6.1 m) compared to the model parameter uncertainty (standard deviation of the computed length is 4.5 m). In order to achieve a robust process design, one needs to dimension the length of the reactor with 187 m so that all particles will be converted with 100 % efficiency with 95 % confidence. Ignoring these sources of uncertainty will lead to suboptimal process performance with the degree of conversion of limestone reduced to 97 %. Among input uncertainty considered, the global sensitivity analysis revealed measurement errors of temperature sensors as the most influential parameters. Overall the results encourage application of comprehensive sensitivity and uncertainty analysis methods for robust design of rotary kiln processes.
|Conference||29th European Symposium on Computer Aided Process Engineering |
|Period||16/06/2019 → 19/06/2019|
|Series||Computer Aided Chemical Engineering|
- Rotary Kiln
- Monte Carlo simulation
- Morris Screening
- Sobol’s method