The cool factor: Additive manufacturing is redefining cold plate design
Have you come across ECAM technology before? Short for Electrochemical Additive Manufacturing, ECAM is a room-temperature 3D metal printing method that produces complex, high-purity parts without the need for post-processing. By enabling exceptional precision and cost-efficiency, it opens the door to innovative design possibilities in thermal management systems.
Before diving into our project, it is worth highlighting a real-world example that shows why an effective cold plate design matters. In 2018, the GOES-17 weather satellite experienced critical overheating due to a blockage in its thermal system. This led to a loss of infrared imaging, which is a major issue for weather forecasting. It is a reminder that a reliable and optimized design is important to prevent mission-critical failures. In this blog, we’ll explore how we applied ECAM technology by Fabric8Labs and Simcenter CFD simulation software technology to push the boundaries of conventional cold plate design.
Competition: ASME K-16 and IEEE EPS Student Cold Plate Design Challenge
This year, we, Siemens Digital Industries Software Nottingham interns for 2024/25, took part in the ASME K-16 and IEEE EPS Student Cold Plate Design Challenge. Our goal was to explore how additive manufacturing could push the boundaries of traditional cold plate design. Using Simcenter CFD solutions, we modeled and simulated six CAD concepts, evaluating them based on three key performance factors:
- thermal resistance
- pressure drop
- mass
Four teams, including ours, were selected as finalists. We had the opportunity to manufacture our pure copper designs in collaboration with Fabric8Labs and present our results in front of academic and industry leaders at the IEEE ITherm 2025 conference in Dallas, Texas. This experience was both technically challenging and incredibly rewarding, allowing us to contribute meaningfully to real-world thermal engineering research.
Cold plate concept designs
We began by generating six concept models in CAD, which were then simulated using CFD to assess their performance. This helped us to identify the most promising design to refine and optimize for the final submission.
CFD simulation of cold plate design options
Early CFD simulations of cold plate Design 1 and Design 2 showed significantly lower Figure of Merit (FoM). This was due to high pressure losses across the cold plate, which were traced back to abrupt changes in geometry near the inlet. These sharp transitions caused flow recirculation, where the fluid swirls or reverses direction, leading to the formation of vortices and disrupting smooth downstream flow. Such flow instability reduced thermal performance and increased energy loss, making these designs less viable for further development.
Designs 4 and 5 achieved lower temperatures and reduced mass, but these advantages were offset by a slight increase in pressure loss. This highlighted the critical role of maintaining smooth fluid flow throughout the cold plate. While both designs outperformed the reference model in terms of FoM, the performance gain was small. Given the complexity and cost of additive manufacturing, the marginal improvement did not justify the additive manufacturing effort.
After comparing the FoM results from all initial simulations, the pyramidal Kagome lattice model appeared as the most balanced design in terms of thermal resistance, pressure drop and mass. When benchmarked against a conventional plane fin model, the Kagome design achieved an FoM that was 11.39% higher at inlet mass flow rate 1.85 L/min. This significant performance improvement made it the obvious choice for further development and optimization.
Why this type of design?
The pyramidal Kagome lattice model was not just chosen for its high FoM, it also highlights the potential of additive manufacturing. It was inspired by the Japanese basket-weaving patterns and tree-like biomimicry structures often used in architecture. It offers a highly efficient cooling solution by minimizing pressure loss, thermal resistance and mass, three critical factors for an optimal cold plate. Its intricate geometry would be difficult, if not impossible, to manufacture using conventional methods.
To meet the 3-mm thickness constraint, our final design included 768 individual pyramidal Kagome units, each with slanted rods positioned at a 55° from the base plane. The rods have a radius of just 0.2 mm, staying well within the 0.1 mm minimum feature size requirement. This cooling design is a perfect fit for ECAM technology by Fabric8Labs, which can manage the intricate geometry that conventional manufacturing methods struggle to produce. Let us take a closer look in the animation below.
Unlike conventional plane fin designs constrained by standard geometric patterns, the Kagome lattice draws inspiration from nature to optimize fluid flow and increase heat transfer within a more compact structure. Imagine it as a lightweight skeleton that acts like a thermal superhighway!
After deciding to proceed with this idea, we initially arranged the lattice layers at different angles, but this showed no significant improvement. Therefore, the lattice was aligned parallel to the fluid flow, resulting in a more uniform velocity profile and a further reduction in pressure drop across the cold plate. Additionally, we added a vertical rod at the center of each Kagome structure. The intersection point generates small-scale turbulence and disrupts boundary layer growth, which enhances heat convection. Finally, adding a pyramidal base beneath the structure increases the surface area in contact with the fluid and improves overall heat conductivity. If you are wondering whether this increases the mass, you are absolutely right. However, the reduction in thermal resistance and pressure loss compensates for the tiny mass added.
Final cold plate design analysis using CFD simulation
CFD analysis on the pyramidal Kagome lattice model. Our cold plate design was mated to the provided housing model, assuming the additive manufactured cold plate is made of pure copper with a thermal conductivity of 380 W/m·K, while all other components were considered insulated. A uniform heat flux of 350 W was applied at the bottom of the plate. The water flow rate at the inlet was set to 1.85 L/min at 20°C. A mesh refinement study was performed to ensure a level of high accuracy in results, that were be compared to experimental data later. Two of Simcenter’s CFD tools, Simcenter FLOEFD and Simcenter Flotherm XT*, both use SmartCells technology and Octree-based Cartesian automatic meshing, so in this study a finer mesh was automatically applied to the cold plate geometry and fluid regions that are more complex improving accuracy where needed, while coarser mesh was used in the surrounding domain for an efficient solve time.
Fluid velocity across the cold plate can be seen above. Higher fluid velocity prevents the thermal boundary layer from growing and increases the temperature gradient, which further improves the heat transfer rate. However, this also causes higher pressure loss due to eddy currents, swirling, and recirculation. To find the optimal balance, a parametric study was conducted to identify the mass flow rate that yields the highest FoM. By analyzing the maximum case temperature at the corners, the pressure drops across the cold plate, and its mass, the final simulation was performed to calculate the FoM.
The maximum temperature on the baseplate was 37.308°C. This is because the cold plate was positioned only at the center of the baseplate for heat conduction. However, a large section of the baseplate not in direct contact with the cold plate was observed to have a lower temperature due to effective heat convection. Thanks to our open framework design, we achieved a significant reduction in mass without compromising the strength needed to withstand fluid flow. From the design of experiments, the highest FoM from the simulation was 0.74489 at a mass flow rate of 2 L/min. Let us observe the fluid flow simulation with solid temperature and pressure visualization in the following video:
We decided to use the entire outlet section, promoting uniform fluid distribution and minimizing potential impingement effects, which enhances thermal spreading across the baseplate.
This approach ensures a balance between low thermal resistance, low pressure drop, and low mass, achieving the highest FoM. Conventional designs often require sacrificing one variable for another. But with the pyramidal Kagome lattice model, we optimized all three, maximizing efficiency without significant trade-offs. This optimal balance is what gives the final design its highest FoM, 0.7449, among all our simulated designs, making it both practical and powerful.
Does this cold plate actually work? Yes!
Before submitting the final design, we recognized that additive manufacturing at this scale requires precision. So, we created a proof of concept by printing prototypes using different conventional additive manufacturing methods. Initially, we used fused deposition modelling (FDM) with PLA at a 2:1 scale, due to the minimum feature size available here was 0.4 mm. A quarter of the model was printed, limited by the printer size, in 4 hours and 40 minutes.
For the Digital Light Processing (DLP) method, resin was used to print a full model. However, because of the minimum feature limitations, the model was scaled to 1.5:1 and printed in about an hour, followed by post-processing. These prototypes not only confirmed the manufacturability of our final design before submission but also highlighted the advantages of using ECAM technology provided by Fabric8Labs, which allowed us to print our intricate geometry without the traditional method limitations.
Why couldn’t we simply mill or cast this cold plate like other parts? The answer lies in its intricacy. Traditional methods struggle with tiny features such as 0.2 mm radius rods, internal cavities, and overhangs that require support structures. ECAM technology enabled us to realize our design as intended, practical, lightweight, and structurally efficient, without compromise. Fabric8Labs also printed the copper cold plate used for actual experimental testing by Intel Corporation and S-Pack Labs.
Results at IEEE ITherm 2025 – 2nd Place
Out of fifteen designs submitted from around the world, our team was selected as one of the finalists to present at the ITherm 2025 conference in Dallas, Texas. We were proud to secure 2nd place in the among the high quality teams competing in this year’s competition, and we highly value what we learned throughout the journey, using Siemens Digital Industries Simcenter software. Engineering is all about innovative problem-solving, and competitions like this give us the opportunity to stretch our skills, collaborate across disciplines, and explore industry-standard tools.
We recommend other student groups to take part in next year’s competition when announced, and we appreciate the hard work of the organizing committee and sponsors to provide this opportunity to student teams.
Have you come across other innovative designs that could only be realized through this ECAM technology? or applied CFD simulation in for design of thermal management solutions suited to additive manufacturing methods? Feel free to share your ideas or projects in the comments below, we’re always keen to connect with fellow engineers and enthusiasts.
Further resources reading
Fundamentals on-demand webinar: Heatsink thermal design – Key considerations for electronics cooling
Blog: How to design a heat sink for additive manufacturing
Blog: Using water cooling in electronics thermal management
Related liquid cooling topic relevant to cold plate design: Upcoming Siemens webinar on July 10th: Multiphysics design optimization of a power module cooling solution
*the CFD simulations in this study were performed using Simcenter Flotherm XT software, the meshing technology and SmartCells approach it use are based on Simcenter FLOEFD CAD-embedded CFD software technology.
