Experiment based evaluation of the pore network model
6.2.4 Impact of local temperature variation
If drying of the PN is assumed as a quasi‐steady invasion process in the capillary controlled regime (and if furthermore the temperature dependency of vapor diffusion is disregarded), the order of pore and throat invasion is dictated by the entry pressure threshold of the competitive liquid elements, as already discussed above. As the invasion pressure of a liquid filled element is not only associated with its radius but also with its temperature, temperature variations affect the order of emptying if the pore or throat size distribution is narrow. More clearly, Fig. 5‐12 reveals a very small distribution of rm in the primary PN
(induced by the small height of the PN). The impact of temperature can be revealed by PN simulations using the temperature field specified in Fig. 3‐8a and else identical parameter setting as before. Simulations with PNM level 1 and temperature dependency of surface tension (thermal impact on vapor transport is neglected here) are compared to simulations with PNM level 2‐4 (taking the dependence of vapor transport on temperature into account) in the following.
Figure 6‐11 shows that independent of the model option a similar invasion behavior as shown before (Figs. 6‐1b, 6‐5, 6‐7, 6‐9) can be expected if temperature is assumed to vary locally as in the drying experiment. The good agreement of the phase patterns in Fig. 6‐11 is explained with the negligible ratio of condensation, which is vc 0.06 using PNM level 2, c v 0.08 using PNM level 3 and vc 0.06 using PNM level 4 and CLO3 (i.e. the PNM in which the labeling is adjusted so as to prevent fluctuation of cluster merging). In general, the drying front appears more stabilized than in the isothermal drying simulation with the benchmark model (Fig. 6‐1b), which is associated with the impact of the temperature variation on the order of invasion. The breakthrough occurs at S 0.47 (in PNM level 1‐3) and S 0.45 (in PNM level 4, CLO3). It is furthermore evident that the effect of the temperature variation on the stabilization of the drying front is very similar to the effect of viscosity and single pore volumes. The drying rate curves presented in Fig. 6‐12 reveal, independently of the PNM option, a quasi CRP at the start of drying (similarly as in Fig. 6‐8). This period is on a higher level than found in Fig. 6‐2b, but also much shorter (the drop
PNM level 1 and
T PNM level 2 PNM level 3 PNM level 4 and CLO3 Figure 6‐11 Phase distributions from drying with temperature variation from experiment and separate invasion of liquid pores. S = [0.79, 0.4, 0.2].This shows that the impact of the slight spatial variation of temperature between
constant temperature and ii) different initial drying behavior. Despite this, very similar drying times are predicted with the different versions of the model (Fig. 6‐12). The deviations in Fig. 6‐12a are explained as follows. With PNM level 1 the temperature dependency of the vapor pressure Pv*(T) was neglected with the purpose to illustrate the impact of temperature on capillary invasion. This can affect vapor diffusion within the partially saturated zone and lead to an increase in drying time because of the slightly lower drying rates (indicated by the
different slopes of the saturation curve for S 0.6). Deceleration of drying using
PNM level 3 and PNM level 4 and CLO3 is explained with the condensation and refilling of the void space. More clearly, the condensed liquid volume must again be evaporated which slightly increases drying time. a) b) Figure 6‐12 a) Drying curves and b) drying rate curves corresponding to Fig. 6‐11. Note that the drying rate curves of the four different simulations are almost overlapping. Experimental drying curves in gray.
Interestingly, comparison of the temperature affected drying of the PN with separate invasion of the pores (PNM level 1 with temperature dependent interface curvature) with the PN simulation in which the pore volumes are assigned to their largest throat neighbors (benchmark model with temperature dependent interface curvature of throats) reveals very similar drying behavior (upper rows in Figs. 6‐7 and 6‐11). The drying curves are almost overlapping in Fig. 6‐13. This shows that in the PN under study the capillary pressure variation due to the existence of single pores plays a minor role for the order of invasion if
T is high enough. This is because the capillary pressure distribution related to the pore radius variation is smaller than compared to the temperature variation. A different behavior is expected if either T is decreased or if the effective pore radii have greater variance. From the above discussion it can be concluded that the separate invasion of liquid pores can alter the order of invasion if the impact of temperature is disregarded. If however, the temperature distribution is taken into account, the separate invasion of the pores has a minor impact for the given PN structure and temperature range (a similar finding was obtained for the comparison of viscous and non‐viscous drying under the impact of temperature variation). This finding can be helpful to reduce the computational effort, namely if the separate invasion of pores is disregarded (because this reduces the invasion events from 7351 to 4851). It was also found, that the cluster growth mechanism can be
neglected in the further study of this temperature range as the amount of condensed liquid volume is very small. Apart from that, it must be concluded that none of the options of the PNM studied so far can reliably predict the drying time. This is explained with an underestimation of the duration of the CRP shortly after the start of drying as well as an underestimation of the drying rate during the period of drying in which the drying front recedes from the PN surface (RFP). This might be explained with the existence of liquid films in the crevices of pore throats in the microfluidic network, which is discussed in the following sections.
a) b)
Figure 6‐13 Comparison of drying simulations with different assignment of pore volumes:
a) drying curves and b) drying rate curves corresponding to the images in upper rows in Figs. 6‐7 (denoted as benchmark) and 6‐11 (denoted as PNM level 1). Experimental drying curves in gray.