A novel methodology for data analysis of dynamic angle of repose tests and powder flow classification

The dynamic angle of repose test serves as an important assessment in powder rheology, offering valuable insights into the flow behaviour of granular materials [1], [2], [3], [4]. In pharmaceutical manufacturing, this test is one of the main tools to assess the flow properties of unconsolidated powders in motion [5], [6], [7]. The device is simple and consists of a horizontal cylinder partially filled with powder rotating around its axis [8]. The powder usually fills the device between 30% and 50% of its volume. Upon rotation, the granular material in the drum is carried upwards by the device’s walls. The pile formed is then dragged downwards by gravity, making the powder move downslope in a continuous flowing regime.

The powder physical properties determine the characteristics of the flow: it can be regular with a uniform profile, or intermittent and characterised by periodic clumps of material avalanching downwards. The rotation velocity, and consequently the powder flow speed, can be controlled and tuned to obtain various flow behaviours. The parallel flat sides of the device are made of a transparent material, allowing the flowing material to be observed and rheological data to be gathered.

The flow rheology is usually measured by collecting information on the average slope of the powder free surface and on its variation in time. An image analysis software is used to extract and process data from multiple snapshots of the powder flowing inside the device [9], [10], [11]. Despite its conceptual simplicity, the results from dynamic angle of repose test are difficult to interpret for cohesive powders, which are often used in the pharmaceutical industry [12]. Unlike their free flowing counterparts, cohesive powders exhibit irregular flow patterns characterised by fragmented surface profiles as well as adhesion to the test apparatus [13], [14], [15]. Both these issues make the identification of the free surface problematic [11], and can lead to unreliable results when used to measure the standard flow descriptors based on a precise identification of the powder profile. While describing the flow characteristics of easily flowing powders is straightforward [16], [17], deducing the properties of cohesive powders through this test can be considerably harder. Recognising the importance of cohesive powders in many fields [18], [19], such as pharmaceutical and food industries, we have developed a comprehensive data extraction and analysis framework based on image data acquired during dynamic angle of repose testing.

The aims of this work are multiple. Firstly, we want to maximise the rheological data that can be extracted from dynamic angle of repose tests [20]. Larger datasets translate into more information, thus enriching the understanding of flow properties upon testing of a material [21], [22]. Secondly, we want to define an optimal number of parameters to be extracted to describe the flow behaviour comprehensively and to understand the relations between these quantities. This will allow a better description of powder rheology in case some routinely measured quantities cannot be identified reliably or without consistent uncertainty [23], [24]. Thirdly, more descriptors will translate into a more nuanced classification of powder flow behaviour [25], allowing the proper categorisation of these granular materials in terms of similarities between their descriptors. This can simplify the identification of powders with similar flow dynamics [20], [26], [27] and, thus, surrogates of a target material [23], [28], [29], [30], [31], [32]. Finally, this post-processing framework is also compatible with Discrete Elements Method data, thus enabling us to compare the parameters measured experimentally with their numerical counterparts across various DEM calibration models [33], [34], [35]. The expanded pool of physical quantities enhances our ability to replicate real-world behaviours accurately, resulting in finely tuned models [36], [37].

The structure of our study is briefly outlined as follows. In the first part of our manuscript, we describe the data acquisition and analysis steps. Then, we explain all the powder flow descriptors implemented, providing visual examples. A Principal Component Analysis is used as a dimensionality reduction tool to handle the large amount of data gathered and provide a deeper insight into the most relevant flow parameters. All materials tested are then compared to identify similarity criteria between them. These are then compared to other criteria using fewer descriptors, highlighting the importance of more flow parameters when identifying powders with similar flow dynamics and surrogates.

Read more

Luca Orefice, Johan Remmelgas, Aurélien Neveu, Filip Francqui, Johannes G. Khinast, A novel methodology for data analysis of dynamic angle of repose tests and powder flow classification, Powder Technology, Volume 435, 2024, 119425, ISSN 0032-5910,