(198h) Application of Population Balance Models for Pharmaceutical Screening Process Development

Screening is a fundamental process in solid oral dosage form manufacturing. The screening process controls the particle size distribution of the coarse materials such as granules, ribbons, and slugs through impacts created by the rotating impellers and screens. Control of the particle size distribution is important in implementing Quality by Design in drug product manufacturing since the particle size distribution influences the flowability and compatibility in the subsequent processes that affect the critical quality attributes such as uniformity of dosage units, assay, and dissolution finally. The population balance equation is ideal for correlating the incoming material properties and process parameters with the screened granules’ particle size distribution. In the pieces of literature, the mechanistic modeling of the screening process has been researched [1–3]. The process models demonstrated the prediction capability for both dry and wet granulation operations [4-6]. However, the practical implementation of the first-principle models for the screening process development, such as the risk assessment through understanding the potential size reduction mechanism, has not been well studied yet. In the current work, a population balance modeling approach was applied for the screening process using a set of roller compaction experiments. The hyperparameters in the population balance equation were fitted to elucidate the potential size reduction mechanisms, including the effect of the incoming material properties and process parameters on the screened granules’ particle size distribution. Through the estimated cause and effect relationship, a target region of the process was visualized utilizing the parameters considered in the process model to identify the unit process design space.

The current study utilized a population balance equation that traces the trajectories of the number and mass of particles over time in the given discretized particle size fractions. A set of ordinary differential equations for the number and mass of particles in each size fraction was obtained by the cell average technique proposed by Kumar et al. [3]. With the cell average technique, the continuous distribution of the generating particles is allocated to the discrete particle size fractions to provide the prediction capability with maintaining computational cost. The population balance comprises the number density function, breakage function, breakage rate kernel, and discretized particle size classes. The breakage function is defined as the volume-weighted combination of the localized disintegration and the multiple fragmentations to be a suitable general-purpose model. The breakage rate kernel is defined as the empirical model proposed by Austin and Luckie [7] and Gotsis [8]. In this study, the breakage rate kernel is coupled with a Heaviside step function to allow the particles below the critical size to pass through the screen immediately. A dry granulation process train composed of roller compaction and a subsequent screening mill was used for experiments. The dry granulation used a roller compactor FP90 (Freund turbo corporation, Japan) that has counter-rotating rolls with a diameter of 90 mm and a roll width of 30 mm. A formulation mainly composed of mannitol and microcrystalline cellulose was used to conduct thirteen roller compaction runs at a constant impeller speed with a screen size of 1.5 mm.

The experiment dataset showed a positive correlation between ribbon thickness and ribbon density. The population balance model showed good prediction performance for the particle size distribution in the dry granulation process. The model suggested that the incoming material properties in roller compaction, e.g., ribbon thickness and ribbon density, affected the fragmentation mode. The multiple fragmentations became dominant as the ribbon density and thickness increased, spontaneously leading to more frequent breakage events during screening. The greater number of breakage events resulted in a broader particle size distribution since the coarser particles close to the critical particle size below which the particles can pass through the screen remained in the screened granules. The fine particle fraction, such as cumulative particle size x10 to x30, remained unchanged even when ribbon density and ribbon thickness increased because the more significant number of breakage events caused accumulating generation of fines. Given the observed correlation among ribbon density, ribbon thickness, size distribution, the span of screened granules, and the volume fraction of the localized disintegration in breakage function z, the target region of z was confirmed. Target incoming material properties were predicted using the acceptable region of z. As demonstrated in the case study, the expected potential cause and effect relationship and the visualization via model parameter provided a better process understanding. Note that the results suggested a smaller number of breakage events of the ribbons thick and dense enough to reduce the volume fraction of localized disintegration in the breakage function will provide a narrower size distribution of screened granules. Installing in-line two-stage screening mills that aid the gentle and fast crushing of coarse ribbons and subsequent sizing of the daughter particles might produce a tighter size control according to the mechanistic model-based interpretation.

Another case study is ongoing to capture the effect of incoming material attributes and process parameters on the screening process model. The case study will use the data from a continuous wet granulation process, including twin-screw granulation with water addition, fluid bed drying, and screening mill. The wet granulation experiment was already completed using ConsiGma 25 (GEA engineering, Wommelgem, Belgium) with a formulation composed of mefenamic acid, lactose monohydrate, corn starch, crospovidone, and povidone. The prediction capability, size reduction mechanism, potential process risk factors, and the discussion on the control strategy will be presented at AIChE Annual Meeting 2023 in November, together with the roller compaction case study.

[1] Diemer Jr., R.B., et al., 2005. Interpretation of size reduction data via moment models. Powder Technology 156, 83–94.

[2] Hounslow, M.J., et al., 2001. Tracer Studies of High-Shear Granulation: II. Population Balance Modeling. AIChE J. 47, 9, 1984–1999

[3] Kumar, J., et al., 2008. An efficient numerical technique for solving population balance equation involving aggregation, breakage, growth and nucleation. Powder Technol. 182, 81–104.

[4] Barrasso, D., et al., 2013. Population Balance Model Validation and Prediction of CQAs for Continuous Milling Processes: toward QbD in Pharmaceutical Drug Product Manufacturing. J. Pharm. Innov. 8, 147–162

[5] Olaleye, B., et al., 2019. Population balance modelling of ribbon milling with a new mass-based breakage function. Int. J. Pharm. 571, 118765

[6] Reynolds, G., et al., 2010. Modelling of pharmaceutical granule size reduction in a conical screen mill. Chemical Engineering Journal 164, 383–392.

[7] Austin, L.G., Luckie, P.T., 1972. Methods for determination of breakage distribution parameters. Powder Technol. 5, 215–222.

[8] Gotsis, C., Austin, L.G., Luckie, P.T., Shoji, K., 1985. Modeling of a grinding circuit with a swing-hammer mill and a twin-cone classifier. Powder Technol. 42, 209–216.