A numerical tool to predict powder behaviour for pharmaceutical handling and processing

Highlights

• DEM modelling allows the prediction of behavioural properties of powders during handling and processing.

• Problematics of segregation, mixing or cohesion can be modelled and predicted with the developed approach.

• Experimental results validate the numerical model.

• Behaviour of powder during a wet granulation, or compression process was predicted, especially the cohesion, the granule size evolution and the compressive force.

Abstract

In this article, we discuss the potential of a numerical model to predict the behaviour of powders during pharmaceutical handling and processing. The model is based on the discrete element method (DEM) where simple experiments are required to calibrate the model used such as angle of repose tests. In a first step, the numerical calibration provided the coefficient of static friction and surface energy of the powders tested. This model has been validated with experimental data in different configurations: powder and granule settling by vibration, powder flow within a polydisperse system and mixing study in a rotary drum. Each comparison with the experimental data leads to a good accuracy of the model. In a second step, the method was used to predict numerically the granule formation during the wet granulation process and the manufacture of tablets during the compression process. Therefore, mixing (segregation), compression (mass uniformity and homogeneity of the components in the tablet) or cohesion problems can be modelled and predicted in order to limit some of the manufacturing problems during the industrialisation of pharmaceutical products such as granules or tablets, providing a scale-up of the process.

Introduction

The powder flow phenomenology has been further processed. This parameter intervenes and qualifies the powder behaviour in different granular media, used, treated and handled in several fields, such as pharmaceutical, chemical, food processing, mineral, building … Granular media are qualified by several phenomena resulting from different behaviours and properties controlled by their physics, such as the flowability. The flow of the powders is governed by interparticle forces which have a fatal influence on their fluidity. The understanding of the granular media handled, and in particular their flow, is a necessity, especially in the pharmaceutical field. The treated powder for pharmaceutical use must have the required flow properties in order to guarantee a best handling in the manufacturing chain of different pharmaceutical forms, to guarantee both efficacy and security of the final product.

It is well known that in pharmaceutical fields, the rarity of free-flowing powders is quite noticeable. The use of non-free-flowing powders leads frequently to many problems during their processing, such as agglomeration of the particles due to high cohesion, which often causes rat-holing or formation of vaults in the machine hoppers for the manufacture of different pharmaceutical forms during discharge. Hence, a pre-treatment and transformation of powders into granules by the granulation process is required in order to confer them the desired properties.

Several methods for powder flow characterization are widely described in literature. Most of these methods are very easy to use and process [1]. They are classified into direct methods such as shear cells: translation [2], rotation, annular [3], uniaxial compression and tensile strength, and indirect methods such as angle of repose [4], Carr index, flow through an orifice and compressibility. These methods are more adapted to industrial constraints in terms of quality control and have an official status by certification on the European Pharmacopoeia [5,6].

The rheology and behavioural laws of granular flow have been the subject of many research studies, from both experimental and numerical points of view. Experimental investigations have clearly shown the effect of the particle size distribution of pharmaceutical excipient powders on their flow. This is mostly done with measurements of static angle of repose and compressibility. This was correlated to the coefficients of static friction of the powders [7]. Shear cell method was also used experimentally for the same purpose [8,9] with applications of the Jenike methodology, as well as dynamic analyses of the FT4 Powder Rheometer [10].

The main drawback of these experimental methods is the large amount of required powder, which could be expensive and sometimes difficult to produce. Moreover, numerous experimental studies have been carried out to characterize the powder's physical properties. For example [11], have shown relationship between the filling properties of the matrices in the compression cells and the static flow properties such as static angle of repose and compressibility. But the powder's flow problems are still relevant in other various pharmaceutical applications such as mixing [12], granulation [13], die filling [14], tableting …

For this reason, the interest towards numerical investigations has increased in order to better understand and find prompt solutions to the powder handling problems in the pharmaceutical field. Some authors used dimensionless approach to characterize the pharmaceutical processes like [15] Or [16] to predict the granulation behaviour, or [17] to characterize coating process. But this method is limited and cannot take into account some phenomena like hydrodynamic.

The Discrete Element Method (DEM), originally proposed by [18], is of much interest nowadays for the description, modelling and simulation of different granular media, where the positions, velocities and physical states of the particles are controlled. DEM has been used to investigate strategies to solve fine powder problems [19], to study inhalers [[20], [21], [22]], to monitor the flow behaviour of pharmaceutical powders as a function of some parameters, such as particle surface property [23] or Triboelectric charging [24]. A further study of the effect of adhesion on the powder flow behaviour was highlighted by the use of the cohesive JKR model [25] by introducing surface energy values to simulate simple cohesive systems behaviour [26,27]. The DEM enabled also to study the effect of particle shape on the general flow of a granular system in a rotary batch seed coater [28]. It allowed to obtain results close to the experimental ones, with a calibration of the coefficients of static friction. But most of these numerical studies which aim to investigate the link between the effects of different physical properties and powder flow behaviour have been done on rather simple systems with several assumptions. This gives results which are rather comparable to experiments but generally do not provide any validation.

These often concern monodisperse systems with large particles of rather spherical shape that are easy to simulate in order to reduce calculation times. Reality dictates the opposite, especially in the pharmaceutical field, where very fine powders in polydisperse systems are generally used in industrial drug manufacturing chains. The challenge of getting as close as possible numerically to the experimental results is significant: by using the particle size distributions the closest to the experimental attributes, the shape, the number of particles, as well as the different mechanical properties of the sample to ensure an optimum predictive numerical approach.

In this context, the aim of the present study is to show the potential of the DEM to predict the powder behaviour during process used in the pharmaceutical industry.

The present article exposes in a first part, the experimental material and the numerical model. The second part is divided into three parts: 1- the characterisation of the powders, 2- the calibration of the model, which consists in the adjustment of numerical input parameters to fit experimental values. 3- Simulation of different pharmaceutical processes.

The challenge is to numerically reproduce the experimental trials while respecting the polydispersity of the systems and other key parameters. Such an understanding could lead to a mapping of the behavioural trends for different powder samples, but also to assure the process scale-up.

Thus, this paper will focus on:

• Model calibration of two coefficients for non-granulated and granulated powders.

• Experimental versus numerical results for three powder handling: powder settlement, flow and mixing.

• Numerical prediction for the wet granulation process and a numerical compression test.