2.2.1. Numerical Solution Procedure
In the next step, we numerically evaluated the target variables (dimensionless volume flow rate
and dimensionless dissipation
) of our model for all 9,231 physically independent setups, using the shooting method. To this end, we calculated the velocity field in the flight clearance by solving the governing equations of our dimensionless model. A detailed description of the numerical solution procedure was given in [
26]. Transforming the boundary value into an initial-value problem, we derived explicit forms of the dimensionless momentum equations:
where the dimensionless viscosity was rewritten as:
The initial estimates of the integration constants
and
were taken from the Newtonian solution. Applying the Simpsons rule yielded the following equations for the velocity profiles:
To iteratively solve the unknowns, we used a Newton-Raphson scheme:
where
is the vector of unknowns,
the Jacobian matrix, and
is the vector of boundary conditions:
The velocity boundary conditions at the barrel surface will not be met unless the initial values are perfect. The converged solutions for the velocity profiles were then used to determine the dimensionless target variables and .
For all calculations, the dimensionless channel height was divided into 1000 equidistant segments. A solution was considered converged if the difference in dimensionless volume flow rate
between two iterations was smaller than 10
-8. Previous analyses have shown that these settings are sufficient to obtain mesh-independent results for our target variables [
26].
2.2.2. Numerical Results
Our parametric design study encompassing 9,231 independent setups provided numerical solutions for the dimensionless volume flow rate and the dissipation in the flight clearance as functions of the dimensionless input parameters , , and .
Figure 5 shows the dimensionless volume flow rate as a function of the dimensionless pressure gradient across the flight clearance for various power-law indices. For all setups, the curves are symmetrical about the point of pure drag flow (
); that is, an equidistant increase in the pressure gradient (positive or negative) affects the magnitude of the dimensionless volume flow rate equally.
The power-law index is a measure of the shear-thinning behavior of the polymer melt: the lower the power-law index, the more shear-thinning is the fluid. Assuming a Newtonian fluid, the widely-known linear behavior is evident: For (pure drag flow), the curve satisfies , while for , the zero-throughput condition is fulfilled .
Generally, the volume flow rates become negative if a critical pressure gradient is reached. From a mathematical viewpoint, this means that the direction of flow changes to the negative -direction. Physically, it implies that the pressure flow caused by the pressure build-up across the clearance exceeds the drag flow, which yields a negative net throughput. According to our model definition, this behavior is subject to strongly overridden melt-conveying zones. Positive flow rates, in contrast, are found in pressure-generating metering zones, in which the leakage flow reduces the net throughput. With decreasing power-law index, the critical pressure gradient shifts to lower values.
The curves become increasingly nonlinear and pressure-sensitive with decreasing power-law index. Two effects are evident: (i) For a given dimensionless volume flow rate, the more the dimensionless pressure gradient (positive and negative) increases, the less shear-thinning the fluid. (ii) For highly shear-thinning polymer melts, small variations in the pressure gradient can lead to pronounced variations in the volume flow rate.
The influence of the screw-pitch ratio on the dimensionless volume flow rate is less pronounced (
Figure 6). For positive pressure gradients, the target variable increases with increasing screw-pitch ratio, while the opposite behavior is evident for negative pressure gradients. This effect is caused by the influence of the flow along the flight clearance (in the
-direction) on the deformation rates, which becomes more pronounced the lower the screw-pitch ratio.
For markedly positive or negative dimensionless pressure gradients (
Figure 6b), the effect of the screw-pitch ratio decreases significantly since the flow is governed mainly by the pressure gradient across the flight clearance.
Figure 7 illustrates the influence of the power-law index on the dimensionless dissipation for a square-pitched screw with
. Viscous dissipation is mainly responsible for the temperature development in the channel. Due to inner friction, mechanical energy is converted into heat, causing a rise in melt temperature.
The dimensionless dissipation can be plotted as a function of either the dimensionless pressure gradient (
Figure 7a) or the dimensionless volume flow rate (
Figure 7b). For both representations, the curves are again symmetrical about the point of pure drag flow (
or
), where the target variable reaches a minimum. In general, the dimensionless dissipation increases if the pressure flow contributes to the flow characteristics; that is, the higher the dimensionless pressure gradient, the more pronounced is the frictional heat generation. Similarly, dissipation becomes highly dependent on the power-law index for strongly pressure-generating or pressure-consuming flows.
Three effects are observed: (i) For moderate pressure gradients, the more dimensionless dissipation decreases, the more shear-thinning the polymer melt. (ii) For higher magnitudes, in contrast, the more frictional heat generation decreases, the more Newtonian the fluid. (iii) For constant dimensionless volume flow rates, viscous heating increases with the increasing power-law index.
Figure 8 shows the influence of the screw-pitch ratio on the dimensionless dissipa-tion for a polymer melt with power-law index n = 0.2. For both constant dimensionless pressure gradients (
Figure 8a) and constant dimensionless flow rates (
Figure 8b), viscous dissipation increases with decreasing screw-pitch ratio. This result is again caused by the effect of transverse flow in the leakage gap. The influence of the screw-pitch ratio on dimensionless dissipation vanishes almost completely in strongly pressure-generating and pressure-consuming flows.
To represent the diverse characteristics of the leakage flow, we considered a wide range of dimensionless pressure gradients. Especially for highly shear-thinning polymer melts with low power-law indices, our extended dataset caused a significant nonlinear increase in the target variables, yielding values higher than
and
, as illustrated in
Figure 9. Recently, we have shown that the parameters are limited to
and
when analyzing the flow in metering channels [
26]. This comparison illustrates the increased complexity of the following symbolic regression analysis.