Uncertainties in key transport variables in homogeneous slurry flows in pipelines

Several concentrate and tailing pipelines use leak detection systems to predict system ruptures and minimise response times in case of or before the occurrence of pipeline ruptures. Those of internal type compare a set of model predictions for pressure losses and flows with actual readings from the system. Input data including solids specific gravity, concentration by weight or rheology curves, obtained either from periodic measurement campaigns or from installed instrumentation, are used to define values for rheological parameters such as the viscosity to be considered for such computations of head losses. In the present paper, an analysis of propagation of different error sources on the rheology identification and eventually on the computation of the head losses is proposed to investigate the relevance of system instrumentation accuracy and indirect errors derived by lack of instrument parameter update. A simple model for the calculation of friction losses for homogeneous Bingham fluids in the hydraulically smooth turbulent flow, a common regime found in copper and iron slurries, was considered. Assuming standard models to link the solids volume concentration with viscosity, uncertainties between 7% and 12% on the latter parameter are predicted for various conditions and variants of the referred mixtures. This is predicted to propagate to values between 10% and 20% on unit head loss uncertainties, provided errors between 5% and 10% on flow metering. Results suggest the inconvenience of rheology inference when using indirect leak detection systems, in favour of online or very frequent rheometric measurement during the transport process.

► Leak detection systems are crucial for damage prevention in slurry pipelines.
► Input data in pipelines have direct and indirect measurement and input errors.
► Uncertainties in concentration and solids specific gravity propagate to viscosity.
► Overall errors close to 10 calculations.