Khalifa, Ali Ahmad

In this study, a recently developed data-driven model for the collision-induced agglomerate breakup (CHERD 195, 2023) is evaluated. It is especially intended for Euler–Lagrange simulations of flows with high mass loadings, where coupled CFD–DEM predictions are too expensive. Therefore, a surrogate model relying on the hard-sphere approach in which agglomerates are represented by effective spheres was developed. Based on a variety of DEM simulations, artificial neural networks were trained to predict the post-collision number of arising fragments, their size distribution and their velocities. In the present contribution, the agglomerate collision model is assessed using the particle-laden flow through a T-junction. Since two fluid streams with agglomerates are injected at both opposite ends, the setup is particularly suitable for investigating breakage caused by collisions. Two flow configurations (laminar flow at Re = 130 and turbulent flow at Re = 8000) and two different powders (primary particle diameter of 0.97 and 5.08 micrometers) are taken into account. The latter allows to study the influence of the strength of the agglomerates on the collision-induced breakage. The laminar case offers the possibility to evaluate the effect of the collision angle in detail. The collision-induced breakage proves to be the most dominant deagglomeration mechanism in both the laminar and turbulent flow scenario. Nevertheless, the role of the fluid stresses and especially the drag stress becomes more prominent in the turbulent case, while in the laminar flow their effects are negligible.

The present contribution is a follow-up of a recently conducted study to derive a data-driven model for the breakage of agglomerates by wall impacts. This time the collision-induced breakage of agglomerates and concurrently occurring particle agglomeration processes are considered in order to derive a model for Euler--Lagrange methods, in which agglomerates are represented by effective spheres. Although the physical problem is more challenging due to an increased number of influencing parameters, the strategy followed is very similar. In a first step extensive discrete element simulations are carried out to study a variety of binary inter-agglomerate collision scenarios. That includes different collision angles, collision velocities, agglomerate sizes and powders. The resulting extensive database accounts for back-bouncing, agglomeration and breakage events. Subsequently, the collision database is used for training artificial neural networks to predict the post-collision number of arising entities, their size distributions and their velocities. Finally, it is shown how the arising data-driven model can be incorporated into the Euler--Lagrange framework to be used in future studies for efficient computations of flows with high mass loadings.

The present study proposes a novel modeling approach for predicting the wall-impact breakage of agglomerates in wall-bounded particle-laden turbulent flows based on artificial neutral networks (ANN). The suggested model is especially useful for efficient Euler-Lagrange simulation methods relying on the hard-sphere approach and the equivalent-sphere model for the agglomerate structure, allowing LES predictions of high mass loadings. Based on the impact conditions, i.e., the impact velocity, the impact angle, the number of included primary particles and the diameter of the primary particles, the outcomes of the breakage events are forecasted using two pre-trained artificial neural networks. The first network is concerned with the prediction of the possibility of breakage and the resulting fragment size distribution, whereas the second network predicts the post-breakage velocities of the fragments. The supervised training of the employed networks relies on a database obtained by extensive DEM simulations of agglomerate wall-impacts covering wide ranges of impact conditions, which were partially reported in Khalifa and Breuer (2020, 2021) for developing a breakage model based on a dimensional analysis and regression techniques. In the present contribution, the database mainly comprising the normal or oblique impact case is extended by adding results for the practically relevant shear impact case of wall-bounded particle-laden flows at extremely small impact angles, i.e., 3◦ and practically flat (0.2◦ ). In addition, the breakage behavior of agglomerates containing very small numbers of particles are investigated under different impact angles and primary particle sizes. The ANN model is employed in Euler-Lagrange LES predictions of duct flows taking three Reynolds numbers and agglomerates of two powders distinguished by the size of the primary particles into account. The results obtained are compared with those based on a previous regression model (Khalifa and Breuer, 2021). In general, a good agreement between the results is found. However, the new ANN model is more widely applicable since the shear impact case is taken into account, which leads to subtle differences enhancing the reliability of the predictions.

The present thesis is concerned with the development of predictive models for the deagglomeration of cohesive particle entrained in turbulent wall-bounded flows. The models are derived for an existing efficient Euler-Lagrange simulation strategy, in which agglomerates are represented by single spheres possessing effective diameters. In this context, serious meaningful advancements to the state of the art are achieved concerning the following issues: (i) the description of the structural features of agglomerates, (ii) the breakup by fluid-induced stresses, and (iii) the wall-impact breakage. The model of the structural features endows the spheres used for representing agglomerates with properties allowing to reasonably determine the effective diameter used in the Lagrangian tracking and the strength of the agglomerate needed for evaluating the possibility of breakage. In the model for the fluid-induced breakup ideas from the literature are revisited and extended to derive breakup criteria. The approach relies on a comparison between the fluid stresses exerted on an agglomerate along its Lagrangian trajectory with a critical stress threshold defined by the strength of the agglomerate. Three types of stresses are considered, which are the turbulent, the drag, and the rotary stress. Furthermore, a special emphasis is put on the post-breakup treatment, i.e., the arising velocities of the disintegrated fragments. For the derivation of the wall-impact breakage model, an extensive number of detailed discrete element simulations of single agglomerates is carried out. Wide ranges of different impact conditions such as the impact velocity, the impact angle, the number of primary particles and the size of the particles are taken into account. The results are analyzed based on useful parameters allowing to quantify the number of arising fragments, their size distribution, and the distribution of their post-breakage velocities. To establish relationships between the impact conditions and the breakage parameters two approaches are explored. The first relies on conventional dimensionality reduction and regression techniques, whereas the second employs feed-forward artificial neural networks. Lastly, the described LES-based Euler-Lagrange methodology is applied to assess the performance of the new models based on three investigations: 1- funnel-duct flows inspired by an experimental investigation to study the breakup of agglomerates in a lab-scale disperser, 2- duct flows, and 3- the flow in pipe bends inspired by experimental and numerical studies on the effect of the bend design on the deagglomeration performance of dry powder inhalers. The results obtained demonstrate the clear advantage of the simulation strategy in offering reasonable predictions at affordable computational costs. Furthermore, important insight into the breakage behavior are gained. These include the understanding of the relative importance of the different breakage mechanisms and the identification of the critical fluid and particle properties for breakup. In addition, the interaction between breakage and other competing phenomena such as agglomeration is explored.

The breakage of agglomerates due to wall impact within a turbulent two-phase flow is studied based on a recently developed model which relies on two artificial neural networks (ANNs). The breakup model is intended for the application within an Euler-Lagrange approach using the point-particle assumption. The ANNs were trained based on comprehensive DEM simulations. In the present study the entire simulation methodology is applied to the flow through two sharp pipe bends considering two different Reynolds numbers. In a first step, the flow structures of the continuous flow arising in both bend configurations are analyzed in detail. In a second step, the breakage behavior of agglomerates consisting of spherical, dry and cohesive silica particles is predicted based on the newly established simulation methodology taking agglomeration, fluid-induced breakage and breakage due to wall impact into account. The latter is found to be the dominant mechanism determining the resulting size distribution at the bend outlet. Since the setups are generic geometries found in dry powder inhalers, important knowledge concerning the effect of the Reynolds number as well as the design type (one-step vs. two-step deflection) can be gained.

The present study completes the development of a model for predicting the effect of wall impacts on agglomerates in turbulent flows. Relying on an Euler-Lagrange hard-sphere approach this physical phenomenon is described in an efficient manner allowing practically relevant multiphase flow simulations at high mass loadings. In a recent study \citep{khalifa2020data} conditions for the onset of breakage and the resulting \added[id=2]{fragment} size distribution were derived. In the present investigation a data-driven description of the post-breakage kinetics of the fragments is developed based on extensive DEM simulations taking a variety of impact conditions (impact velocity, impact angle, agglomerate size) into account. The description relates the velocity vectors of the fragments after breakage to three parameters: The reflection angle, the spreading angle and a velocity ratio of the magnitude of the fragment velocity to the impact velocity of the agglomerate. Relying on the DEM results Weibull distribution functions are used to describe the parameters of the wall-impact model. The shape and scale parameters of the Weibull distributions are found to mainly depend on the impact angle of the agglomerate. Consequently, relationships between the shape and the scale parameters and the impact angle are established for each of the three parameters based on a fourth-order regression. This allows to determine the velocity vectors of the fragments randomly based on the corresponding Weibull distributions of the reflection angle, the spreading angle and the fragment velocity ratio. The devised model is evaluated in a turbulent duct flow at five Reynolds numbers and three agglomerate strengths given by powders consisting of primary particles of different size. The analysis first concentrates on the pure wall-impact breakage but then also includes agglomerate breakup due to turbulence, drag forces and rotation allowing to determine the shares of the different physical phenomena. It is found that with increasing Stokes number the wall-impact breakage occurs less effectively due to the reduced responsiveness of the agglomerates to the secondary flow motions in the duct. However, in the very high range of St$^+$ other mechanisms such as the turbophoresis and the lift force augment the breakage at walls. Comparing the contributions of the different breakage mechanism reveals that the wall impact is dominant at the lowest Reynolds numbers, whereas the drag stress prevails at high Re.