A continuum model of stresses in a vertical silo with a flow channel in the vicinity of the wall, using the principal stress cap surface approach for the bulk solids

• Published in: Chemical engineering research & design Vol. 122; pp. 211 - 225
• Main Authors: Matchett, A.J., Langston, P.A., McGlinchey, D.
• Format: Journal Article
• Language: English
• Published: Rugby Elsevier B.V 01-06-2017; Elsevier Science Ltd
• Subjects: Building codes; Bulk solids; Calibration; Continuum modeling; Data points; Eccentric silo; Eurocode 1; Mathematical models; Silos; Solids; Stress; Stresses; Studies; Yield; Bulk solids; Stresses; Yield; Eccentric silo; Eurocode 1
• ISSN: 0263-8762; 1744-3563
• DOI: 10.1016/j.cherd.2017.04.013

[Display omitted]
•Silos with a flow channel at the wall have been modelled.•3-zones were used, as in Eurocode 1 (2006).•The principal stress cap method was used for the static bulk zone.•The model was successfully fitted to experimental data by Chen et al. (2007).
Eurocode 1 (2006) gives design equations for eccentric stresses in silos, including flow channels adjacent to the wall. This has been modelled using the approach of Matchett et al. (2015, 2016).
A three zone model was developed, consisting of:
• The flow channel.
• The transition zone.
• The bulk of the solids.
The flow channel and the transition zone were modelled by Janssen-type equations. The bulk was modelled by the principal stress cap approach.
The transition zone is a complex region and has several purposes:
1. To shelter the low stress flow channel from the high stresses around.
2. To allow high principal stresses at the transition/bulk interface, within the yield locus.
3. To form a transition between the dynamic flow channel and the static bulk.
4. To allow transition from passive stress in the flow channel to active stress in the bulk.
The model was calibrated against the data of Chen et al. (2007) for a full-scale silo, and described the data reasonably well, scaling axially and azimuthally.
Large experimental data sets are required to calibrate a model. Unfortunate data points cannot be arbitrarily rejected.
Further extensive, experimental data are needed to calibrate models.