Prediction of major pressure losses pumping non-Newtonian foods and beverages – The role of uncertainty propagation for power law fluids
Year: 2025
Type: article
Abstract: To predict major pressure losses from pumping shear thinning liquid foods and beverage, rheometer-measured flow behaviour ( n) and flow consistency ( K) indices are needed. These are always obtained with some measurement uncertainty. This contribution develops a tool that allows for quantitatively estimating how a given measurement uncertainty in the rheological parameters translate to prediction uncertainty in major pressure losses, and how much measurement uncertainty can be allowed while ensuing that the prediction uncertainty is within a given tolerance. Methodologically, the study combines the general uncertainty management framework, Monte Carlo simulations, and an artificial neural network regression to predict uncertainty propagation. Experimental data on a range of liquid foods and beverages are used. Results show that relative measurement uncertainties of ~2-5% in n and K results in a prediction uncertainty of less than 10% in the major pressure losses. Moreover, uncertainty in n influences the prediction quality more severely than uncertainty in K , and prediction uncertainty is larger the closer n is to 1. This is the first contribution providing a quantitative relationship between measurement uncertainty and prediction uncertainty for this application. It provides measurement guidelines and helps identify the role of uncertainty propagation in predicting pressure losses when pumping complex liquid foods. • Uncertainty in measured n and K propagates to uncertainty in pressure loss • This is a challenge in process design and in setting up rheological measurements • MC simulations are used to train an ANN for predicting uncertainty propagation • A spreadsheet tool is supplied, predicting requirements on measurement of n and K • Guidelines are provided for ensuring a low prediction uncertainty.
Source: Food and Bioproducts Processing
Authors: Andreas Håkansson, Hanna Ekelund, Hanna Isendahl, Hannah Viklund, Helena Wijkander, Dragana Arlov
Institutions: Lund University, Tetra Pak (Sweden)
Language: English
Subfield: Fluid Flow and Transfer Processes
Field: Chemical Engineering
Domain: Physical Sciences
Open Access status: hybrid
Estimated APC paid: $3,970
Funder: Vetenskapsrådet
OpenAlex ID: W4415532660