(185a) Predicting Screw Feeder Flow Rates from Powder Properties and Operating Conditions
AIChE Annual Meeting
Monday, November 15, 2021 - 9:20am to 9:40am
Our predictive feeder model is built from three components: a simple volumetric displacement screw feeder flowsheet model, a statistical autoregressive moving average model (ARMA), and a predictive projection to latent spaces (PLS) deterministic model. First, the hybrid deterministic displacement and stochastic ARMA model, as proposed by Johnson et al. , were fit to experimental runs of five powders with varying property values and feeding behaviors using both Coperion K-tron KT20 and MT12 feeders with twin concave screws operated at many different volumetric and gravimetric setpoints. This yielded sets of deterministic and stochastic deviation model parameters for each powder-feeder-setpoint operating configuration. Next, a PLS model was built that maps a database of powder properties , an abstraction of feeder geometries, and operating state to their experimentally estimated deterministic and stochastic parameters, extending the work of Bostijn et al. . Constraints were integrated into the PLS model as needed to ensure deterministic parameters are physically possible and stochastic ARMA parameters meet theoretical requirements for stability and invertibility. The PLS model scores and loadings are presented, providing insight into how different powder properties influence the minute-by-minute and second-by-second dynamic feeding behavior. Last, the predictive capabilities of our PLS model are demonstrated and evaluated by simulating the feeding behavior of out of sample powders and conditions.
 H. Berthiaux, K. Marikh, and C. Gatumel, âContinuous mixing of powder mixtures with pharmaceutical process constraints,â Chemical Engineering and Processing: Process Intensification, vol. 47, pp. 2315â2322, 2008.
 J. W. Fernandez, P. W. Cleary, and W. McBride, âEffect of screw design on hopper drawdown of spherical particles in a horizontal screw feeder,â Chemical Engineering Science, vol. 66, pp. 5585â5601, 2011.
 W. E. Engisch and F. J. Muzzio, âMethod for characterization of loss-in-weight feeder equipment,â Powder Technology, vol. 228, pp. 395â403, 2012.
 L. X. Liu, I. Marziano, A. C. Bentham, J. D. Litster, E. T. White, and T. Howes, âEffect of particle properties on the flowability of ibuprofen powders,â International journal of pharmaceutics, vol. 362, pp. 109â117, 2008.
 M. Capece, K. R. Silva, D. Sunkara, J. Strong, and P. Gao, âOn the relationship of inter-particle cohesiveness and bulk powder behavior: Flowability of pharmaceutical powders,â International Journal of Pharmaceutics, vol. 511, pp. 178â189, 2016.
 Y. Wang, T. Li, F. J. Muzzio, and B. J. Glasser, âPredicting feeder performance based on material flow properties,â Powder Technology, vol. 308, pp. 135â148, 2017.
 U. V. Shah, V. Karde, C. Ghoroi, and J. Y. Y. Heng, âInfluence of particle properties on powder bulk behaviour and processability,â International journal of pharmaceutics, vol. 518, pp. 138â154, 2017.
 N. Bostijn, J. Dhondt, A. Ryckaert, E. SzabÃ³, W. Dhondt, B. Van Snick, V. Vanhoorne, Ch. Vervaet, and Th. De Beer, âA multivariate approach to predict the volumetric and gravimetric feeding behavior of a low feed rate feeder based on raw material properties,â International Journal of Pharmaceutics, vol. 557, pp. 342â353, 2019.
 B. J. Johnson, S. GarcÃa-MuÃ±oz, M. Sen, J. Hanson, D. Slade, and N. V. Sahinidis, â(366d) stochastic analysis and modeling of pharmaceutical screw feeder mass flow rates,â in Proceedings of the 2020 Virtual AIChE Annual Meeting, 2020.
 B. Van Snick, J. Dhondt, K. Pandelaere, J. Bertels, R. Mertens, D. Klingeleers, G. Di Pretoro, J. P. Remon, Ch. Vervaet, Th. De Beer, V. Vanhoorne, âA multivariate raw material property database to facilitate drug product development and enable in-silico design of pharmaceutical dry powder processes,â International journal of pharmaceutics, vol. 549, pp. 415â435, 2018.