Portfolio / Publications & research

Journal article / 2026

Design of Herringbone Microchannels for Enhanced Mixing in Augmented Cooling Slurries

Anagh Dave, Vivek V. Manepalli, Roshith Mittakolu, Clayton Pullins, Michael T. Barako, Samuel Graham, and Damena Agonafer

IEEE Transactions on Components, Packaging and Manufacturing Technology ·

How can microchannel geometry help a microencapsulated phase-change material (MEPCM) slurry use more of its latent heat? This numerical study explores herringbone ribs that generate secondary vortices, improve transverse mixing, and promote particle melting. Regression-based heat-transfer and friction correlations support multi-objective optimization with NSGA-II, balancing cooling enhancement against pressure drop.

  • Microchannel cooling
  • MEPCM slurries
  • NSGA-II optimization
View publication on IEEE Read the research explained
Herringbone microchannel research figure showing the channel array and rib geometry, simulated flow trajectories, and a cross-sectional enthalpy map with flow vectors under sidewall and bottom heating
Geometry and simulated flow. (a) Channel array and herringbone-rib geometry; (b) flow trajectories through the ribbed channel; (c) cross-sectional enthalpy distribution and flow vectors with heat entering through the sidewalls and bottom. Numerical visualization, not experimental imaging.
Open full-size figure

The mechanism / Illustrative animation

Forward flow. Cross-channel mixing.

Flow along a ribbed microchannel Top-view schematic. Cyan traces travel left to right over repeated herringbone ribs, with sideways deflections representing transverse transport. Paths are illustrative, not computed streamlines. 01 / Along the channel Bulk flow Top view / rib pattern simplified Secondary circulation across the channel Cross-sectional schematic of two counter-rotating circulation loops between heated sidewalls and above a heated bottom wall. Orange arrows indicate heat entering from both sides and below. Actual vortex structure varies with geometry and operating conditions. 02 / Across the channel Secondary circulation / schematic Heat through sidewalls and bottom
How to read this schematic. Cyan moving traces represent downstream transport and cross-channel circulation; orange arrows indicate heat input. Secondary motion can bring slurry between the warmer wall region and the channel interior, promoting heat exchange and particle melting. These are hand-drawn explanatory paths, not CFD results or tracked particles. Geometry, vortex structure, speed, and color are not quantitative; this animation does not show enthalpy or melting fraction. The original research figure above remains unchanged.

Static view. Use Play schematic to start the illustrative motion.

Behind the results

Why mixing matters for phase-change cooling.

Adding a heat-absorbing material to a coolant is only part of the solution. The channel must also help that material reach the heat. This study connects coolant behavior, rib geometry, and pumping requirements rather than treating them as separate design choices.

How mixing spreads the wall's heat into the coolant

Cross-sectional views looking along the flow. In both views, coolant also travels downstream, out of the page. Heat enters through both sidewalls and the bottom wall; orange edges and inward arrows mark these heated surfaces.

Limited cross-channel transport

Heat concentrated near the sidewalls and bottom with limited transverse transport Orange regions near both sidewalls and the bottom fade into a blue channel interior. Inward arrows outside these three walls show heat input, and short arrows inside represent heat diffusion. Capsules near the interior have less thermal exposure. This is a qualitative concept, not a computed temperature field. Less-exposed interior Heat through sidewalls and bottom

Heat still diffuses into the fluid, but weak transverse motion limits exchange with the interior. Near-wall fluid warms while some particles farther away receive less thermal exposure.

Herringbone-induced mixing

Secondary circulation redistributes energy across the channel Heat enters through both sidewalls and the bottom wall. Two circulation loops exchange fluid between these heated surfaces and the channel interior. Relatively cooler fluid warms as it returns along the heated sidewalls and bottom. Capsules are carried between regions. Mixing redistributes energy but does not make temperature perfectly uniform or remove the absorbed heat. Exchange with the interior Heat through sidewalls and bottom

Secondary circulation exchanges fluid between the heated sidewalls, bottom, and channel interior. Relatively cooler return fluid absorbs heat as it passes these surfaces. This exchange spreads energy through more of the flowing slurry and renews the fluid next to the heated walls.

Conceptual comparison, not CFD contours. Background colors suggest warmer wall regions and cooler interior regions; orange arrows show heat input or transport of warmer fluid, and cyan arrows show the cooler return flow. Outlined circles represent capsules, not their melting fraction. Geometry, colors, and paths are qualitative, not quantitative temperature or enthalpy data. Ribs are omitted from these cross-sections for clarity; real vortex patterns vary along the channel.

From fluid motion to better heat transfer

1. Renew the near-wall fluid. Cooler fluid arriving near the wall can sustain a steeper local temperature gradient, increasing heat transfer for a given wall-to-bulk temperature difference. Equivalently, a higher effective heat transfer coefficient can reduce the temperature difference needed to remove a fixed heat load.

2. Give more particles access to heat. Circulation changes particle thermal histories. If the PCM reaches its melting range and has enough time to absorb heat, some of the added energy melts the core rather than only raising its temperature. More stored enthalpy therefore does not necessarily mean a proportionally higher temperature.

3. Carry the absorbed energy downstream. Mixing redistributes heat; it does not destroy it. At steady state, with the same heat input to the coolant and the same mass flow, the bulk specific-enthalpy rise remains Q̇ / ṁ, neglecting other energy terms. The wall temperature and cross-sectional distribution can change even when that bulk energy rise is unchanged. A downstream heat exchanger must still reject the heat.

Mixing does not guarantee complete melting or a perfectly uniform temperature. The benefit depends on inlet state, residence time, geometry, and the additional pressure drop.

1. The problem: available latent heat is not always fully used

A microencapsulated phase-change material (MEPCM) slurry carries tiny capsules in a liquid. The material inside each capsule absorbs latent heat as it melts, in addition to the sensible heat absorbed as the coolant warms. The shell keeps the phase-change material contained.

That extra heat capacity is useful only when the particles experience suitable temperatures for long enough. With limited cross-channel transport, particles near a heated wall and those in the cooler channel interior can have different thermal histories. Simply adding more particles does not ensure that their melting capacity will be used effectively.

2. The mechanism: ribs move coolant across the channel

The angled herringbone ribs redirect part of the downstream flow into secondary, cross-channel motion. This circulation exchanges fluid between the wall region and the channel interior, improving transverse mixing and promoting particle melting.

The ribs are passive features: they have no moving parts, but they are not energetically free. They also resist flow. The design question is therefore not just how to strengthen mixing, but how much additional pressure drop is justified by the thermal benefit.

3. The method: numerical study, correlations, and optimization

The numerical study examines rib height and thickness, Reynolds number, and particle concentration. Reynolds number characterizes the balance of inertial and viscous effects in the flow; concentration changes how much phase-change material is carried by the slurry.

Regression correlations summarize the computed heat-transfer and friction behavior. NSGA-II, a multi-objective genetic algorithm, uses those relationships to search for trade-offs between heat-transfer enhancement and flow resistance. A Pareto trade-off means improving one objective requires sacrificing the other; it is not a single universally best design.

4. The findings: stronger heat transfer, with a hydraulic cost

The selected balanced numerical design has a heat transfer coefficient of 7.3 × 104 W/(m²·K) and a pressure drop of 37.9 kPa. A larger heat transfer coefficient indicates stronger heat exchange for a given area and wall-to-fluid temperature difference, using the study's definitions.

The reported enhancement is approximately 3.8× relative to straight microchannels with MEPCM slurry and 1.6× relative to herringbone microchannels with single-phase water. The first comparison changes the channel geometry; the second changes the coolant. These are distinct baselines, not interchangeable performance claims.

Neither ratio means the device runs that many times cooler or uses that much less pump power. Hydraulic power depends on both pressure drop and volumetric flow rate; electrical input also depends on pump efficiency. The comparisons should not be interpreted as equal-pumping-power results.

5. The significance: design the channel and coolant together

The engineering takeaway is that latent-heat capacity and access to the heated region must work together. Herringbone geometry can help a slurry make better use of phase change, but its value depends on the operating conditions and the acceptable pressure-drop budget.

This is numerical research, not experimental validation of a production cooler. The correlations should stay within their fitted ranges. Hardware adoption would additionally require checking slurry stability, capsule durability, material compatibility, manufacturing constraints, and performance in a complete cooling loop; those are qualification needs, not results established by this visualization.

This explanation summarizes the study and provides engineering context. For model assumptions, parameter ranges, and comparison conditions, see the full IEEE publication.