Artificial Bees Crack the Rubber Blend Puzzle for Cheaper, Tougher Car Parts

Every car on the road depends on rubber components that must survive heat, vibration, oil, and years of mechanical abuse. Chloroprene rubber, better known by its trade name neoprene, has long been a workhorse of the automotive industry because of its excellent resistance to weathering, ozone, and chemicals. The problem is cost: chloroprene rubber is relatively expensive, and manufacturers are under constant pressure to cut material expenses without sacrificing performance. A new study published in Polymer Bulletin offers a data-driven answer to that dilemma, showing how a swarm-intelligence algorithm borrowed from the behavior of honeybees can find the sweet spot where cost savings and mechanical performance meet.

The research, carried out by Derya Deliktaş of Kütahya Dumlupınar University, Ezgi Aktar Demirtas of Eskişehir Osmangazi University, and Mert Göksüzoğlu of the SAMPA Automotive R&D Center in Samsun, Türkiye, tackles a deceptively simple question: what happens when you blend chloroprene rubber with cheaper natural rubber, and how do you find the exact formulation that delivers the properties an engineer actually needs? Blending the two elastomers is an obvious strategy for balancing cost and performance, but the chemistry is far from trivial. The proportions of natural rubber, along with the amounts of accelerators, retarders, and vulcanization agents added to the mix, all interact in complicated ways that determine how the final material behaves.

To map this complicated landscape, the team first turned to Response Surface Methodology, a statistical framework that has been a staple of industrial optimization since the mid-twentieth century. Rather than testing every possible combination of ingredients, which would require an impractical number of experiments, the researchers used a Central Composite Design to select a structured set of experimental runs. This design allows a quadratic response surface to be fitted over the formulation space, meaning that the relationship between each ingredient and the resulting material property can be captured with a relatively small number of trials. From these experiments, the team built regression models describing three key response variables: hardness, tensile strength, and the loss factor known as Tan δ.

Each of these three properties matters for a different reason. Hardness, measured in Shore A units, reflects how resistant the rubber is to indentation and deformation, which is critical for seals, mounts, and bushings. Tensile strength, expressed in megapascals, measures how much stretching force the material can withstand before breaking, a fundamental durability metric. Tan δ, the ratio of the loss modulus to the storage modulus, captures the damping behavior of the rubber, essentially how efficiently it converts vibration energy into heat. For automotive components such as engine mounts and suspension bushings, damping is what keeps noise and harshness out of the cabin, making Tan δ a crucial dynamic property.

Here lies the central conflict that makes rubber formulation so difficult: the objectives pull against each other. A formulation that maximizes tensile strength may deliver mediocre damping. One that minimizes Tan δ for superior vibration isolation may compromise strength. Improving hardness can come at the expense of flexibility. Because these trade-offs cannot be resolved by optimizing one property at a time, the researchers framed the problem as multi-objective optimization, and this is where the artificial bees enter the picture.

The Artificial Bee Colony algorithm, first developed by Turkish computer scientist Derviş Karaboga and collaborators, is a metaheuristic inspired by the foraging behavior of honeybee swarms. In the algorithm, artificial bees explore the search space in three roles: employed bees exploit known food sources, which in this context correspond to promising formulation candidates; onlooker bees choose among those sources with a probability proportional to their quality; and scout bees abandon exhausted sources and search randomly for new ones. This division of labor allows the algorithm to balance exploitation of good solutions with exploration of uncharted regions, avoiding the premature convergence that plagues simpler optimization methods. The ABC algorithm has proven effective in fields ranging from structural truss design to vehicle routing, and the Turkish research team brought it to bear on rubber chemistry.

To combine the three competing objectives into a form the algorithm could optimize, the researchers applied scalarization methods, which convert multiple objectives into a single weighted function. They compared three approaches: the classical weighted sum method, the Tchebycheff method, and the Conic Scalarization Method, a technique developed by mathematician Refail Kasimbeyli that can characterize efficient solutions even in nonconvex problems. The comparison revealed a clear winner. The Conic Scalarization Method delivered the most robust optimization performance, offering flexible trade-offs between mechanical strength and dynamic properties that the other methods could not match. This finding is significant for practitioners because the choice of scalarization technique can determine whether an optimization finds genuinely useful compromise solutions or gets stuck in inferior corners of the design space.

The practical payoff of the approach is best illustrated by the numbers. When the team prioritized minimization of Tan δ by assigning it a weight of 0.8, the optimizer produced a formulation yielding a Tan δ of just 0.076, with a hardness of 49.53 Shore A and a tensile strength of 16.23 megapascals. That is a formulation tailored for components where vibration damping is paramount. When hardness was prioritized instead, the resulting blend achieved 57.15 Shore A, a tensile strength of 19.38 megapascals, and a Tan δ of 0.198, suited to parts that must resist deformation. Prioritizing tensile strength with the same weight produced the most impressive strength figure of all: 24.86 megapascals, accompanied by 53.60 Shore A hardness and a Tan δ of 0.238. In other words, a single optimization framework can hand engineers three very different rubber recipes, each tuned to a different design requirement.

What makes this work notable beyond the rubber lab is the way it stitches together two traditions that rarely collaborate so closely. Response Surface Methodology provides statistically validated models grounded in real experimental data, while the swarm-based optimizer searches those models efficiently across thousands of candidate formulations without a single additional laboratory test. The study builds on the team’s earlier 2024 work in Polymer Bulletin, which applied response surface methodology alone to the same chloroprene and natural rubber blend system, and it reflects a broader trend in materials science of pairing designed experiments with artificial intelligence and metaheuristic algorithms. Similar hybrid approaches have recently been applied to polymer nanocomposites, machining of composites, and biodiesel production, suggesting that the RSM-plus-metaheuristic pipeline is becoming a standard tool for modern materials engineering.

For the automotive industry, the implications are concrete. Rubber component manufacturers working with chloroprene blends can now adjust their formulations to hit targeted hardness, strength, or damping values with confidence, rather than relying on trial-and-error compounding. Because natural rubber is cheaper than chloroprene, every formulation that shifts the balance toward more natural rubber while preserving performance translates directly into cost savings at industrial scale. The researchers, who acknowledge the facilities provided by the SAMPA Automotive R&D Center, describe their CSM-based multi-objective ABC approach as a versatile tool for developing cost-effective, high-performance automotive rubber components. As material costs and performance demands continue to rise together, it seems fitting that the answer to designing better rubber may have come from mimicking the collective intelligence of bees.

Subject of Research: Multi-objective optimization of chloroprene/natural rubber blend mechanical and rheological properties using response surface methodology and the artificial bee colony algorithm

Article Title: Response surface-guided multi-objective optimization of CR/NR blend mechanical and rheological properties using artificial bee colony algorithm

Article References: Deliktaş, D., Demirtas, E. A., & Göksüzoğlu, M. (2026). Response surface-guided multi-objective optimization of CR/NR blend mechanical and rheological properties using artificial bee colony algorithm. Polymer Bulletin, 83(11), Article 618. https://doi.org/10.1007/s00289-026-06653-6

Image Credits: AI Generated

DOI: 10.1007/s00289-026-06653-6

Keywords: chloroprene rubber, natural rubber, rubber blends, vulcanization, response surface methodology, artificial bee colony algorithm, multi-objective optimization, conic scalarization method, tensile strength, hardness, Tan delta, automotive rubber components