|Abstract:||Periodic flow reversal which is associated with boiling instabilities is a very commonly seen phenomenon in small scale heat removal applications, for instance, microchannel heat sinks for cooling of electronics. Due to the industry’s strict requirement for thermal stability, previous research on flow instabilities in microchannels has been largely focused on heat sinks for electronic cooling. There is a lack of understanding of the effect of flow reversal in microchannel heat exchangers, which are used in air conditioning and refrigeration applications.
This thesis first presents the visualization of flow reversal in a microchannel as part of a realistic automotive evaporator. The flow regimes inside of the microchannel are recorded under both two phase feeding and liquid only feeding modes. In the two-phase feeding mode, churn, bubbly/slug, and annular flows occur alternatively, but the period of each flow regime is not constant. Flow reversal is only witnessed occasionally in the bubbly/slug flow regime. In the liquid feeding mode, only liquid and bubbly/slug flows are observed. Reverse flow occurs periodically. The duration of flow reversal is much shorter in the two-phase feeding mode than in the liquid feeding mode. Most likely it is due to higher upstream resistance caused by two-phase refrigerant feeding.
After confirming the existence of flow reversal in microchannel heat exchangers which are used in air conditioning applications, this thesis presents a newly developed mechanistic model of bubble dynamics in a single microchannel, which demonstrates how flow reversal is generated. The comparison between high-speed visualization of in-channel flow regimes and simulation results shows that this model is capable of capturing the transient flow regime and slug velocity inside of a single microchannel and predicting flow reversal. The model quantitatively demonstrates the mechanism of flow reversal. Within one periodic cycle, the evolution of flow regime, pressure distribution along channel length and mass flux at tube inlet are well correlated with each other. The local pressure peak caused by the build-up of downstream flow resistance can cause a positive pressure gradient, which induces flow reversal.
In the third part of the thesis, the effects of channel geometries on flow reversal are presented. Multiple evaporators with different geometries are tested using R134a as the refrigerant. In each evaporator, the reversed vapor flow is vented out of the inlet header and the flow rate is measured. It has been found experimentally that under the same heat flux, superheat and channel length, the microchannel evaporator with smaller diameter generates more reversed vapor flow (per unit mass flow rate of supplied liquid refrigerant) at a higher frequency. A simulation generated by the aforementioned mechanistic model under the same condition demonstrates that smaller diameter creates more rapid growth of vapor slugs and allocates more flow resistance to the downstream section. As a result, the incoming flow is quickly decelerated and the positive pressure gradient ends up covering larger upstream areas, all of which leads to more vapor flow reversal at a higher frequency. Experiments also show that the microchannel evaporator with longer tubes (when diameter is the same) produces less reversed vapor flow (per unit mass flow rate of supplied liquid refrigerant) at a lower frequency, but flow reversal is less sensitive to channel length than to channel diameter. Simulation results reveals a similar pattern of flow regime and pressure development within one periodic cycle for channels with the same diameter but different lengths, confirming that flow reversal is less sensitive to channel length. Consistent with the experimental results, simulation also predicts less vapor flow reversal at a slower frequency in the longer channel. This is due to relatively more buildup of upstream resistance which is caused by a higher refrigerant velocity in the longer channel, especially at the beginning of a periodic cycle when large dryout area exists.
In the fourth part of the thesis, the effects of refrigerant thermophysical properties on flow reversal are presented. Four refrigerants (R134a, R1234yf, R245fa and R32) are tested in the same system. Heat flux and superheat are maintained the same. It has been found experimentally that R245fa which has similar heat of vaporization with R134a but much larger specific volume difference between vapor and liquid phase generates more reversed vapor flow volumetrically (per unit mass flow rate of supplied liquid refrigerant) than R134a. A simulation generated by the aforementioned mechanistic model under the same conditions demonstrates that a larger specific volume difference generates more drastic growth of vapor slugs and concentrates more flow resistance to the downstream. As a result, more vapor flow reversal is generated at a higher frequency. Although the simulated frequency results are not validated by the experimental results, the predictions of the reversed vapor flow rates for R134a and R245fa are well confirmed by the experiments. Experiments using R134a, R1234yf and R32 show that the volumetric flow rate of reversed vapor (per unit mass flow rate of supplied liquid refrigerant) is not sensitive to heat of vaporization, although increasing heat of vaporization reduces the absolute volumetric flow rate of the reversed vapor. Frequency of flow reversal increases as the heat of vaporization of the refrigerant decreases. Simulation predictions are consistent with the experimental results. It is shown that the effects of heat of vaporization have similar magnitude on reverse vapor flow and supplied liquid flow, but smaller heat of vaporization creates more drastic vapor slug expansion which increases the frequency.
In the last part of this thesis, the effects of boiling instabilities on heat transfer performance of a microchannel heat exchanger are investigated. In the same facility as introduced in the first part of the thesis, two heat exchangers with identical heat transfer areas are employed. One of them is equipped with an artificial inlet restriction. The two heat exchangers are operated under identical conditions. The heat exchanger without artificial flow resistance is subject to more severe boiling instability and consequently generates four times more reverse vapor flow than the other one. The comparison of capacities under identical operating conditions reveals that more reverse flow helps to improve cooling capacity by up to 13.3%. Meanwhile, numerical simulations of bubble dynamics coupled with heat transfer are carried out for both heat exchangers. Results show that in the heat exchanger with more reverse flow, the refrigerant side heat transfer coefficients are enhanced, especially in the upstream part of a channel where the flow velocity is relatively low.