Flow-priority optimization of additively manufactured variable-TPMS lattice heat exchanger based on macroscopic analysis

This study proposes a macroscopic modeling and optimization framework based on Darcy–Forchheimer theory to design variable-TPMS lattice heat exchangers with non-uniform channel widths, which experimental validation confirms achieve a 28.7% performance improvement over uniform lattice configurations.

The Gist

Imagine you have a very busy highway where two types of traffic are trying to pass each other without crashing: a stream of hot cars and a stream of cold cars. Their goal is to swap "heat" (like energy) as they pass. In a traditional heat exchanger, this highway is built with a uniform, repetitive pattern of walls (like a standard honeycomb or a grid). This works okay, but it's not perfect. Sometimes the hot cars get stuck in a jam, or the cold cars take a shortcut that doesn't let them swap enough heat.

This paper is about redesigning that highway using a special, mathematically perfect 3D pattern called a TPMS lattice (think of it as a complex, sponge-like structure that repeats in three dimensions). The researchers wanted to know: What if we didn't keep the walls uniform? What if we could stretch the hot lane wider in some spots and the cold lane wider in others, just where it helps the most?

Here is the breakdown of their journey, using simple analogies:

1. The Problem: The "One-Size-Fits-All" Trap

Usually, engineers build these heat exchangers with a uniform sponge-like structure. It's like building a city where every street is exactly the same width.

• The Issue: In a complex shape (like a U-turn or an L-shape), a uniform street width isn't always best. Sometimes the hot traffic needs a wider road to move faster, while the cold traffic needs a narrower, more winding path to slow down and swap heat better. A uniform design forces both to follow the same rules, which isn't efficient.

2. The Solution: The "Smart Sponge"

The researchers used Additive Manufacturing (3D printing with metal) to build a heat exchanger where the "sponge" isn't uniform. They wanted to change the thickness of the walls inside the sponge to control how much space the hot fluid gets versus the cold fluid.

• The Challenge: If you try to design this by looking at every single tiny pore in the sponge (like looking at every single brick in a wall), the computer takes forever to calculate it. It's like trying to design a city by counting every single pebble on the sidewalk.
• The Trick: They created a "Macroscopic Model." Instead of looking at every tiny pore, they treated the whole sponge as a "magic material" with average properties. They used a theory called Darcy-Forchheimer (which is like a rulebook for how water flows through a wet sponge) to predict how the fluid would move without needing to see every single hole.

3. The Optimization: Tuning the "Knob"

They treated the position of the sponge walls as a "knob" they could turn.

• The Knob: Imagine a dial labeled from -1 to +1.
  • Turn it left: The hot lane gets wider, and the cold lane gets squeezed.
  • Turn it right: The cold lane gets wider, and the hot lane gets squeezed.
• The Goal: The computer ran thousands of simulations, turning this knob in different spots across the heat exchanger, trying to find the perfect mix that would make the heat swap as fast as possible.

4. The Result: A Diagonal Dance

When the computer found the "perfect" design, it didn't look like a straight highway anymore.

• The Discovery: The optimal design made the hot and cold fluids cross each other diagonally, like two dancers weaving around each other, rather than just flowing straight past.
• Why it worked: This diagonal path forced the fluids to stay in contact with each other for a longer distance. It was like making the cars drive in a long, winding loop instead of a straight line, giving them more time to swap heat.
• The Score: This "smart" design improved the heat exchange performance by about 24% compared to the standard, uniform design.

5. The Reality Check: 3D Printing it

The researchers didn't just stop at the computer. They printed the design using metal powder and a laser (a process called Laser Powder Bed Fusion).

• The Test: They ran hot and cold water through the printed metal blocks.
• The Outcome: The real-world test matched the computer predictions very closely. The "smart" design really did work better than the uniform one.
• The Catch: The computer model was slightly too optimistic about how much pressure the water would lose (how hard the pump had to work). In the real world, the tiny channels in the "smart" design were so narrow that the 3D printer made tiny imperfections (like a slightly rough edge), which made the water struggle a bit more than the computer thought. However, the heat transfer benefit was still huge.

Summary

Think of this paper as a recipe for a better radiator. Instead of using a standard, uniform grid of tubes, the researchers used a computer to "bend" the internal walls of a 3D-printed metal sponge. They found that by making the lanes for hot and cold water uneven and diagonal, they could make the heat exchange much more efficient. They proved this works in real life, showing that 3D printing can create "smart" internal structures that are far superior to traditional, uniform designs.

Technical Summary

Technical Summary: Flow-Priority Optimization of Additively Manufactured Variable-TPMS Lattice Heat Exchanger

Problem Statement
While Triply Periodic Minimal Surface (TPMS) lattice structures are widely recognized for enhancing convective heat transfer through uniform flow distribution and boundary layer disruption, a uniform lattice configuration is not necessarily optimal for all heat exchanger geometries. Specifically, in complex flow paths where fluids follow U-shaped or L-shaped trajectories, a uniform lattice may lead to asymmetric flow resistance and suboptimal macroscopic flow patterns. Previous studies have highlighted the need for flow-path optimization and the potential of functionally graded lattices, yet rigorous numerical optimization of the relative flow priority between hot and cold channels in TPMS heat exchangers remains an open challenge. Furthermore, explicit topology optimization of intricate TPMS geometries is computationally prohibitive due to the high cost of resolving detailed micro-geometries during iterative design processes.

Methodology
This study proposes a macroscopic modeling approach to optimize the flow priority between hot and cold fluids in a two-fluid heat exchanger equipped with a TPMS Primitive lattice. The core methodology involves:

1. Macroscopic Porous-Medium Modeling: Instead of resolving the detailed TPMS geometry, the lattice is treated as a porous medium. The flow is governed by the Brinkman–Forchheimer equations, and heat transfer is modeled using an advection-diffusion approach with a volumetric heat-transfer coefficient ($h$). This coefficient, along with permeability ($\kappa$), drag coefficient ($\beta$), and effective thermal conductivity ($\lambda$), serves as an artificial property characterizing the unit-volume heat transfer capability.
2. Effective Parameter Derivation: These macroscopic parameters are derived using Representative Volume Element (RVE) based homogenization. The effective properties are computed by solving microscopic Navier-Stokes and energy equations on a single unit cell under specific boundary conditions.
3. Design Variable Definition: The isosurface threshold ($C$) of the TPMS Primitive function is utilized as the design variable. By varying $C$, the position of the solid walls is shifted, effectively altering the channel widths and the relative flow resistance between the hot and cold domains. A normalized design variable $d$ ($0 \le d \le 1$) controls this threshold.
4. Optimization Framework: An optimization algorithm (Method of Moving Asymptotes, MMA) is employed to maximize the heat transfer rate. The objective function is defined as the enthalpy difference between the inlet and outlet of the hot fluid. The optimization is performed on a planar heat exchanger with U-shaped counterflow trajectories, subject to fixed pressure drop constraints (500 Pa) at the inlets and outlets.
5. Validation: The optimal solution is verified through detailed-geometry simulations (using Star-CCM+) and experimental validation. Specimens were fabricated using metal Laser Powder Bed Fusion (LPBF) with 316L stainless steel, followed by leakage testing and thermal performance evaluation under steady-state conditions.

Key Contributions

• Macroscopic Optimization of Flow Priority: The study introduces a framework to optimize the relative dominance of hot and cold flows in TPMS heat exchangers by treating the isosurface threshold as a spatially varying design variable. This allows for the creation of non-uniform lattice distributions that guide macroscopic flow patterns.
• Computational Efficiency: By utilizing a macroscopic porous-flow approximation, the study achieves a 92.8% reduction in computational time compared to detailed-geometry analysis, making iterative optimization of complex lattice structures practically feasible.
• Experimental Verification of Optimized Design: The research provides experimental evidence that a macroscopically optimized, non-uniform lattice outperforms a uniform lattice in a U-shaped counterflow configuration, bridging the gap between theoretical optimization and additively manufactured reality.

Results

• Performance Improvement: The optimized lattice demonstrated a clear performance improvement over a uniform lattice. Detailed-geometry simulations showed an average enhancement of 24.2% in heat transfer rate, while experimental results confirmed an improvement of 23.3%.
• Flow Mechanism: The optimal solution induced a diagonal flow trajectory, increasing the effective interaction length between the hot and cold fluids compared to the predominantly vertical flow in uniform lattices. This resulted in a more uniform temperature distribution across the heat exchanger, reducing thermal imbalances near the inlet/outlet regions.
• Pressure Loss: While the macroscopic model tended to overestimate pressure losses compared to detailed simulations and experiments, the pressure loss differences between the uniform and optimal lattices were generally small. However, experimental results for the optimal lattice's hot-fluid channel showed higher pressure loss than predicted, attributed to geometric imperfections (throat narrowing) during the LPBF process when channel diameters approached the manufacturing resolution limit (~0.5 mm).
• Model Accuracy: The macroscopic model accurately predicted overall trends in flow and temperature distribution. Discrepancies in pressure loss were attributed to the assumption of perfect periodicity in the RVE derivation, which is violated in the optimized non-uniform structure, and simplifications in the volumetric heat-transfer coefficient derivation (e.g., assuming constant solid temperature within a cell).

Significance and Claims
The paper claims that optimizing the flow priority via spatial variation of the TPMS isosurface threshold is a viable strategy for enhancing heat exchange efficiency in complex flow configurations where uniform lattices are suboptimal. The study demonstrates that macroscopic analysis, despite inherent approximations, can reliably guide the design of high-performance heat exchangers that are manufacturable via metal AM.

The authors modestly acknowledge limitations, noting that the predictive accuracy of the macroscopic model depends on the pressure-gradient conditions used for parameter derivation and may degrade at higher flow rates or in regions of extreme geometric non-uniformity. They also highlight that geometric imperfections in narrow throats can significantly impact pressure loss, suggesting that design constraints on minimum feature size are necessary for practical applications. The work establishes a foundation for future research into the physical mechanisms of such optimizations and their application to other TPMS types and flow configurations.