First Principles Design vs Generative Design: Which Delivers Better Results in AM Heat Exchangers?

By Michael Fuller, Founder, Conflux Technology

Conflux’s very existence was built from a deep desire to pioneer heat exchange and use additive manufacturing. Publicly, we often discuss the importance of First Principles Design and how we use it in our approach to creating our cutting-edge heat exchangers.

In conversations with our customers, the discussion often surfaces around whether to use First Principles Design, or lean into Generative Design Tools. It’s a fair question but it’s not an either-or proposition. These two approaches aren’t adversaries; when combined strategically by a thermal design expert they can significantly accelerate innovation and industrialisation in thermal systems.

Conflux’s focus is squarely on serial production of high-performance, certifiable AM heat exchangers. We are constantly driving innovations that progress this, at scale. Achieving series production efficiently and reliably requires both rigorous physics-based design and the smart application of advanced computational tools.

But let me state the obvious: tools alone aren’t magic. They’re only as good as the thinking that guides them. Knowing how and when to apply each method, and how to integrate them, is fundamental to our continued success. We’re building heat exchangers for serial production, not for niche prototypes.

So, with that in mind, let’s dig in and break it down.

First principles design provides an essential and solid foundation for developing Conflux heat exchangers.

The Discipline of First Principles Design

What is First Principles Design?

First Principles Design, or “reasoning from first principles,” is an engineering approach that breaks problems down to fundamental physical truths and rebuilds solutions from the ground up. Rather than relying on conventional assumptions or iterative tweaks of past designs, engineers using first principles define precise relationships between performance targets and the physical mechanisms that achieve them.

Using First Principles Design, Conflux developed this Water Charge Air Cooler for in an annular configuration for a motorsport application.

How First Principles Design Drives Insights

Designing heat exchangers from first principles starts with core physics: thermodynamics, fluid mechanics, and materials science. You understand how heat transfer happens, what drives pressure losses, and how geometries can manipulate those effects.  In practice, this means specifying critical boundaries for a design such as required heat transfer rates, allowable pressure drops, spatial constraints, mechanical integrity under operating loads, and manufacturability parameters specific to AM. This foundation becomes the “language” with which any subsequent tools, including generative design, must work.

Our geometries are specific to the AM process we’ve refined over the years. Our “design library”, built over the past 10 years, contains knowledge about how each shape behaves under real manufacturing and real fluid conditions. For each new application brief, we start with a framework built from first principles, tied directly to our manufacturing capability.

Importantly, designing AM heat exchangers through first principles doesn’t mean ignoring computational tools. Rather, it’s about defining the correct problem and constraints for those tools to solve.

The Role of Generative Design Tools

What is Generative Design?

Generative Design tools are advanced software applications that use algorithms and artificial intelligence to explore a wide range of design options based on user-defined parameters and constraints. These tools can rapidly produce novel geometries, exploring broad parametric sweeps of design variables, often beyond what a human might conceive unaided, and can help reduce design time.

The Limitations of Generative Design Tools

However, generative design doesn’t work in isolation. The algorithm requires an informed “seed,” the fundamental constraints, the boundary conditions, parameters, heat transfer efficiency, low pressure drop, flow-resistance and any other physical principles that define whether a solution is viable. These things are not known by the ‘tool’ unless you tell it. Without this foundation, it can easily generate shapes that are mathematically elegant but physically useless or impossible to manufacture.

For instance, generative algorithms might optimise for surface area but be unaware that increasing surface area could impose unacceptable pressure drops in certain applications. Similarly, it won’t intuitively know if wall thicknesses are manufacturable in the chosen AM process without explicit inputs.

Contrary to the notion that generative tools empower non-specialists, deep thermal engineering expertise is essential to assess whether proposed geometries truly improve performance. Without this expertise, non-specialists risk being misled by results that appear promising but fail to deliver viable thermal performance or manufacturability.

What about Gyroids for Heat Exchange?

In most instances, the conversation around gyroids has become more cut-and-dry with clearer knowledge and a better understanding in key markets as to which applications benefit from these geometries, and which applications do not.

What is a Gyroid?

Gyroid geometries, with their continuous, triply periodic minimal surfaces, are considered by some to be an emblem of what additive manufacturing can achieve. Derived through mathematical equations and often generated using computational design tools, they offer a striking visual representation of complex, highly efficient structures. Their high surface-area-to-volume ratio and self-supporting nature make them particularly attractive for certain heat exchanger applications. However, their performance is highly application-dependent.

Does Conflux make Gyroids?

Conflux’s performance-led approach evaluates each geometry against the specific thermal, mechanical and flow requirements of the target environment. In many of our high-performance applications, such as those in motorsport or aerospace, the need to tightly control pressure drop is paramount. These systems often demand compact, lightweight, effective heat transfer with minimal flow resistance. Additionally, gyroids impose inherent symmetry in surface area distribution between fluid domains, which can be problematic when fluids have significantly different thermal conductivities and require asymmetrical heat transfer surfaces, for instance allocating more surface area to the fluid with lower conductivity to improve heat transfer efficiency. Therefore, in this context, gyroids often present too much pressure loss to be a fit-for-purpose option. That said, we remain geometry-agnostic.

Gyroid structure engineered by Conflux

Conflux’s Approach: Combining the Best of Both Worlds

Conflux’s methodology is not about choosing one approach over the other. It’s about integration.

We start from first principles because it provides a solid foundation. Our team defines precise performance targets and physical limits, drawing on our proprietary knowledge base, our ever-expanding design library and performance database, and decades of thermal-fluid expertise.

This knowledge informs:

• The required heat transfer per unit area of our Conflux core geometries
• The manufacturing constraints tied to our specific AM processes
• The specific nuance of process optimisation for serial additive manufacturing including automated post-processing to enable the cost-efficiency enablers required to accelerate AM adoption through scale
• The certification pathways demanded by industries like aerospace

Generative design then becomes a powerful accelerator. Once the problem space is correctly defined, we can deploy computational tools to explore optimised configurations rapidly, reducing design cycles and speeding time-to-market for serial production. But, to reiterate, it is only because of our deep domain knowledge that we can leverage these tools to optimise performance.

Furthermore, our Conflux Production System (CPS) underpins this integrated approach. CPS combines advanced Design for Serial Additive Manufacturing (or what we call ‘DfSAM’) with proprietary manufacturing know-how and process automation. It ensures that even highly complex geometries are manufacturable at scale, consistently, and with the quality levels required for certified industrial applications.

Conflux’s cartridge heat exchanger for the Pagani Utopia combines advanced Design for Serial Additive Manufacture (DfSAM) with proprietary manufacturing know-how and process automation.

When it comes to engineering high-performance, certifiable heat exchangers, first principles design remains essential. It gives precise control over designs, ensures manufacturability from the outset and allows us to meet stringent performance and regulatory requirements.

While automated tools can suggest promising geometries, it’s only the user of these tools who holds the nuanced understanding needed to deliver the best, application-specific solutions for motorsport, aerospace and industrial clients.

Generative design tools offer powerful capabilities; however, they are of very limited value in AM heat exchanger design without deep domain knowledge. You must be able to critically evaluate what’s being proposed or risk being led astray by designs that look impressive but fail in practice. Without the ability to critically evaluate the outputs, these tools can be risky for early concepts in advanced applications.

The real opportunity lies in bringing these two approaches together. Our ever-expanding design library and performance database underpin our first principles methods, while generative design tools help us explore and refine optimised solutions in shorter timeframes. Used together, they’re not simply complementary; they’re mutually dependent. It’s the combination of first principles insights and generative design exploration that enables us to deliver innovations redefining what’s possible in high-performance heat exchangers and serial additive manufacturing.