Topology optimization of multimaterial aircraft pylons using generalized shape function approach

arXiv:2609.27685 · cs.CE · Submitted 2026-09-23 · Read on arXiv

Listen

Radio episode about this paper

Transcript

Introduction to the show: ident: Computer Science. The day in computer science: software, systems and theory on arXiv.

Tom: I'm Tom.

Jane: And I'm Jane, here with Lu, Meng and our in-house large language model, Lalam.

Tom: We're looking at the paper: "Topology optimization of multimaterial aircraft pylons using generalized shape function approach".

Jane: Here's the gist of the paper. Topology optimization of multimaterial aircraft pylons using generalized shape function approach provides an optimized structural efficiency solution for aircraft pylons by employing a generalized shape function (gSF) method to handle up to 2n distinct material phases.

Tom: The Generalized Shape Function Methodology The gSF method extends traditional shape functions to a generalized n-linear form, mapping a compact set of natural coordinates onto the multimaterial simplex as design variables.

Jane: This construction guarantees physically valid material fractions and provides 2n material phases using only n design variables.

Tom: The density of any material m of a finite element is calculated using the formula involving the coordinates xim and design variables ζi.

Title and authors: Tom: So we're looking at this paper called "Topology optimization of multimaterial aircraft pylons using generalized shape function approach". It’s about optimizing these things for aircraft pylons, which are those main supports connecting an engine to the wing or fuselage.

Jane: Exactly. The title immediately tells you it’s tackling two big problems: optimizing the structure itself and handling multiple materials at once. It seems like they're proposing a new way to do this using something called a generalized shape function approach, or gSF.

Tom: Right, so instead of just one material per element, they're using this gSF method where you get up to 2n distinct material phases from only n design variables per element <ref:2609.27685#pg1,up to 2n distinct material phases>. That sounds like they’re trying to keep the design space manageable while allowing for more complexity in the final structure.

Lu: What's interesting about that is how they map those natural coordinates onto a design domain, which is a compact set of coordinates defining those materials as design variables. It guarantees physically valid material fractions and gets you 2n material phases using only n design variables, which is efficient for the number of materials you want to test <ref:2609.27685#pg1>.

Meng: Efficient for the number of materials is good, but how do they actually get those discrete results that matter? When you're doing topology optimization, you don't want fuzzy transitions between materials; you want clear material assignments.

Jane: That’s where they introduce a modified projection-based filter to achieve those crisp, discrete optimized solutions. They also calculate a non-discreteness measure called Mnd using a formula involving the total physical density and candidate materials to show how near binary the distribution is.

Tom: So they're not just throwing materials around; they’re actively filtering them out based on physical density to ensure the layout is sharp, and that Mnd value tells you exactly how close it is to being perfectly split between two materials. It sounds like they’ve put a lot of work into making sure the final geometry makes sense structurally.

Lu: And for the material properties themselves, they interpolate Young’s modulus using a formula where the exponent is set to three, which is what allows them to calculate the Young's modulus of each element correctly during optimization. That’s crucial because you can't just plug in a material property; it has to be calculated based on how much of that material is present.

Meng: Calculating the modulus this way sounds computationally heavy at scale. What about the actual optimization part? How are they minimizing compliance subject to all those volume constraints? That’s usually where things get tricky when you add these complex material interpolation steps.

Title and authors: Tom: They're using a modified SIMP method here, reconstructing it to incorporate this generalized shape function approach for the material interpolation, where Young’s modulus is interpolated based on the physical density and the material properties. And they are minimizing structural compliance, which means they’re trying to make the pylon as stiff as possible while staying within the given volume limits.

Jane: The objective function is written as minimizing F transpose U subject to various constraints, including all that loading, those boundary conditions, and those material volume limits we talked about earlier. They use a gradient-based optimization algorithm called the Method of Moving Asymptotes to update all those design variables.

Lu: And they exploit the geometric features of a typical pylon structure by selecting the midplane for optimization, which means they're performing multimaterial evolution on a 2D design domain representing that midplane, and then extruding that converged 2D solution to get the full three-dimensional pylon <ref:2609.27685#pg1,the geometric features of a typical pylon structure>.

Meng: It makes sense to simplify to a 2D domain for the initial optimization because it’s computationally cheaper, but how do you ensure this 2D result actually translates into a stable three-dimensional structure without introducing errors <ref:2609.27685#pg1>?

Jane: They handle that by having the pylon design exhibit geometric and loading symmetry along the z direction, which allows them to get that three-dimensional result by simply extruding the converged two dimensional optimized solution. It’s a standard trick to keep things stable in this context.

Tom: So they’ve basically taken a complex three dee problem and simplified it down to a 2D optimization on the midplane, then used symmetry to get the full shape back, which is smart for keeping the math tractable <ref:2609.27685#pg3>. But what's the real impact of this gSF method compared to older methods like standard SIMP?

Lu: The main improvement is that this gSF approach lets them handle up to 2n distinct material phases using only n design variables, which means they don't have to massively expand their design variable set just because they want more materials <ref:2609.27685#pg1,up to 2n distinct material phases>. This is a key scaling advantage for complex designs.

Meng: But I still worry about the computational cost of that density calculation and the projection filter. If you have many elements, running that filtering on every single element could slow down production runs significantly at scale.

Title and authors: Jane: That’s a fair concern, but they’ve shown it works for up to fourteen candidate materials in their testing. And their results show smooth and stable objective minimization during the convergence histories, which suggests the filter isn't causing wild oscillations.

Tom: Stability is important, but let's talk about the final result. The conclusion of this paper on "Topology optimization of multimaterial aircraft pylons using generalized shape function approach" is that this formulation not only satisfies those volume constraints but also yields sharply defined, optimized design topologies.

Lu: They achieve this because the algorithm strategically places each material where it contributes most to stiffness while keeping the volume fraction of each material at its maximum prescribed limit. It’s a very targeted placement strategy based on the optimization process itself.

Meng: Targeted placement is good, but how sure are we that this isn't just finding local optima in this complex design space? That’s always the risk with topology optimization when you start adding these custom filters and interpolation schemes.

Jane: They show that the Mnd values indicate the degree of discreteness of the optimized designs, which are very close to zero in all cases, indicating a near-binary solution, so they're not stuck in a messy compromise. This points toward reliably crisp material states.

Tom: So, we have a method that manages complexity by using generalized shape functions to handle many materials efficiently while maintaining sharp material transitions and good stability through their filtering technique. It seems like they’ve successfully bridged the gap between complex multimaterial modeling and practical structural design for aerospace applications.

Lu: And looking ahead, the implication here is that this gSF approach could become a standard way to tackle multimaterial problems in structural optimization across different industries where material interfaces are critical, not just aerospace.

Meng: For production, I'm still focused on the runtime. If we take this framework and try to run it on a very large pylon model with hundreds of elements, I need to know how much faster or slower this gSF method is compared to simpler methods like standard SIMP in a real-time simulation environment.

Jane: That’s the next thing we need to see. And for now, what this paper does is prove that you can get sharp material states while managing many materials without having an explosion in your design variable count, which is a big step forward for complex structures.

Tom: It's a solid piece of work because they deliver on the promise of handling complexity with efficiency and producing physically sensible results. That’s what we’re hearing about this topology optimization of multimaterial aircraft pylons using generalized shape function approach.

The paper's summary: Tom: So we're looking at this paper called "Topology optimization of multimaterial aircraft pylons using generalized shape function approach". This one shows how they used a generalized shape function method to optimize a pylon for multiple materials, handling up to twenty distinct material phases with just n design variables.

Jane: Right, so the main point is that they made the math more efficient by using this gSF thing. It’s about mapping those natural coordinates onto the design variables so you get those material fractions right and you can handle many materials without a huge explosion in the number of things you have to optimize.

Tom: Yeah, it’s about keeping the design space small while allowing for that complexity, which is what I like to see when we’re training models against complex physical constraints. They use this modified projection-based filter to get those sharp, discrete solutions they need.

Jane: And they have a specific formula there for calculating the non-discreteness measure, Mnd, which shows how close the material distribution is to being perfectly split between two materials. It’s a way to prove their solution is crisp and not fuzzy.

Tom: That filtering step is key because in real engineering, you don't want a design that looks smooth but has no clear material boundary where it matters structurally. They also have this formula for interpolating the Young’s modulus, using an exponent of three, so the optimization can actually calculate the element properties correctly.

Jane: That sounds like they are tackling a few different problems at once: efficient material representation, ensuring sharp physical results through filtering, and accurate property calculation during design. What this means for someone listening is that we’re getting a way to model complex structural materials in topology optimization without having to manually define every single material transition point.

Tom: It means they can optimize these things—like aircraft pylons—by focusing on the most important structural stiffness points, and the method naturally places the materials there while respecting those volume limits. They use a gradient-based algorithm called the Method of Moving Asymptotes to do all that math.

Jane: And when you look at their results, they show that even with four to seven materials, they use three dimensional shape functions, and for eight to fourteen materials, they switch to four dimensional shape functions. It scales up nicely depending on how many materials you need.

Tom: The conclusion is pretty clean—they manage the volume constraints and get sharply defined topologies where each material is placed where it adds the most stiffness while staying within its assigned fraction limit. That targeted placement strategy seems like it works well against those compliance minimization goals.

Jane: So, to sum it up, this paper provides a method that handles many materials efficiently, ensures sharp material assignments through filtering and density checks, and calculates element properties accurately for optimization. It’s a solid framework for tackling multimaterial design problems in structures like aircraft pylons. Now we’re going to see if those results hold up when we look at the actual structural validation tests next.

The paper's improvements: Tom: So we’re looking at what they suggest as improvements for this generalized shape function approach to optimizing multimaterial aircraft pylons. Basically, they’re saying their setup is solid but there are ways to refine the process further for real-world use.

Jane: Right, so the paper points out that while the method works well with many materials, there's always room to tweak how those material interpolations happen and how the filtering itself behaves. It’s about making it more robust for different kinds of pylon designs.

Tom: They are suggesting that they look at how they handle those near-binary solutions—the Mnd value—and try to tighten up the projection filter even more to ensure that discrete material states are even sharper across all the design variables.

Jane: That makes sense because if the filtering is a bit loose, you still end up with some kind of fuzzy transition where you want a hard switch. They’re exploring ways to make that transition cleaner, moving closer to zero for every element.

Tom: And they also suggest exploring different ways to set up those Young’s modulus interpolations; while the exponent three is what works for this test case, they are looking into whether other power values might give better results on more complex geometries or different material pairings.

Jane: That points to a future where you might have a library of these interpolation schemes and pick the best one based on what you’re optimizing—like if stiffness is more important than volume conservation in a certain design phase. It gives us more flexibility in the AI modeling side.

Tom: From an engineering standpoint, this means they are not just looking at one perfect setup but developing a family of approaches that might work better for specific failure modes on different aircraft. It’s about generalizing the method itself rather than just applying it to one specific pylon shape.

Jane: And what this changes for us is that instead of getting stuck with one fixed way to model material mixing, we get a toolkit where we can tailor the approach based on the complexity of the material system and how sharp those boundaries need to be.

Tom: So they’re pushing toward a more adaptable framework that handles both the efficiency gain from fewer design variables and the precision gain from better filtering and interpolation choices. It’s moving beyond just a successful test case to developing a general method.

Jane: That adaptability is what I find most important for future research because it shows how we can evolve these AI tools to handle more varied physical realities in structural design. We need those flexible methods if we want AI to truly do complex engineering synthesis.

Conclusion: Tom: So we’re wrapping up on "Topology optimization of multimaterial aircraft pylons using generalized shape function approach". This paper shows they achieved sharp, optimized designs for multi-material structures by using a generalized shape function method that handles many material phases efficiently.

Jane: Exactly, so the big takeaway is that you can get crisp material states while managing many materials without having to dramatically expand your design variable set. It’s a nice way to keep the math manageable while still getting high-quality structural results.

Tom: The implication here is that we can start using these gSF techniques for complex engineering problems where material interfaces are critical, not just in aerospace, but anywhere you need precise material distribution.

Jane: Right, and they proved that their method works with up to fourteen candidate materials and still gives very close to zero non-discreteness values across the board. That means the solutions aren't fuzzy compromises; they’re actually near-binary material distributions.

Tom: I'm glad they showed that smooth objective minimization during convergence histories, because that tells me the underlying math is stable and not just finding a local pit in the optimization landscape.

Jane: And for someone listening who doesn't read, it means we’re getting a tool that lets engineers design structures where every material gets exactly what it needs based on its physical contribution to stiffness.

Tom: It’s about strategic placement, right? The algorithm places each material where it contributes most to the overall stiffness while sticking strictly to those volume limits they set. That targeted strategy is what makes this formulation work so well.

Jane: It’s a very smart approach for structural optimization because it means the design isn't just random; it's driven by a clear physical goal, which is exactly what we want in engineering synthesis.

Tom: So, that’s the efficiency of the generalized shape function approach in action for aircraft pylons. We’ve seen how they use this method to handle complexity with reasonable design variables and sharp material states.

Jane: It really shows how advanced AI techniques are moving toward solutions that are not just mathematically interesting but genuinely useful for creating physical objects. That's a huge step forward in structural modeling.

Tom: Speaking of which, we’ve got some other interesting work on arXiv today that looks at real-time search ads and how they use hierarchical text representations.

Jane: Oh yeah, that’s a completely different field but it shows how these new representation learning techniques are being applied across so many domains. We'll take a quick look at those next.

Tom: Definitely, we'll see if those text representations can actually deliver real-time speed or if they just add more complexity to the system.

Swagatam Islam Sarkar, Prabhat Kumar

Indian Institute of Technology Hyderabad

cs.CE

Submitted: 2026-09-23

Updated: 2026-10-31

Comments: NCMDAO2026 conference paper

License: http://arxiv.org/licenses/nonexclusive-distrib/1.0/

Importance score: 70/100

The gist: Topology optimization of multimaterial aircraft pylons using generalized shape function approach provides an optimized structural efficiency solution for aircraft pylons by employing a generalized…

Key concepts

Generalized Shape Function (gSF)
This method extends traditional shape functions into a generalized n-linear form. It maps a compact set of natural coordinates onto the multimaterial simplex as design variables, allowing the optimization to handle up to 2n distinct material phases using only n variables.
Modified Projection-Based Filter
This filter is used to ensure the optimized solutions are near discrete (crisp) rather than smooth. It achieves this by calculating a non-discreteness measure (Mnd) based on the total physical density and candidate materials, indicating a near-binary material distribution.
Modified SIMP Method
The standard SIMP method is reconstructed to incorporate the gSF approach for material interpolation. This allows the Young's modulus of an element to be interpolated accurately based on its calculated physical density and the properties of the candidate materials.
Method of Moving Asymptotes
This is a gradient-based optimization algorithm used to update the design variables during the optimization process. It is employed within the framework to minimize structural compliance while respecting all defined loading, boundary conditions, and material volume limits.

Terminology

Summary

Topology optimization of multimaterial aircraft pylons using generalized shape function approach provides an optimized structural efficiency solution for aircraft pylons by employing a generalized shape function (gSF) method to handle up to 2n distinct material phases. p. 0

The Generalized Shape Function Methodology

The gSF method extends traditional shape functions to a generalized n-linear form, mapping a compact set of natural coordinates onto the multimaterial simplex as design variables. This construction guarantees physically valid material fractions and provides 2n material phases using only n design variables. p. 1 The density of any material m of a finite element is calculated using the formula involving the coordinates xim and design variables ζi. p. 2 To ensure near discrete 0-1 optimized solutions, a modified projection-based filter is adopted to achieve crisp, discrete, optimized solutions. p. 2 The non-discreteness (Mnd) is determined by the formula involving total physical density and candidate materials to indicate near-binary material distribution. p. 3

The Optimization Formulation

The modified SIMP method is reconstructed to incorporate the gSF approach for material interpolation, where Young’s modulus of an element is interpolated based on physical density and material properties. The objective function minimized in this work is structural compliance subject to volume constraints. p. 3 The optimization problem is mathematically written as minimizing the compliance F T U subject to various constraints including loading, boundary conditions, and material volume limits. p. 3 A gradient-based optimization algorithm, the Method of Moving Asymptotes, is employed to update the design variables. p. 4

Structural Design and Dimensionality

The study exploits the geometric features of a typical pylon structure by selecting the midplane for optimization. The design domain is discretized using finite elements, with the dimensionality of shape functions determined by the number of desired materials. p. 7 The pylon design exhibits geometric and loading symmetry along the z-direction, allowing the three-dimensional result to be obtained by extruding a converged two-dimensional optimized solution. p. 7

Performance Metrics and Results

The results demonstrate the capability of the framework to effectively optimize aircraft pylons with multiple candidate materials without requiring a considerable expansion of the design variable set. The optimized designs are presented using 2D shape functions for two and three materials, 3D SFs for four to seven materials, and 4D SFs for eight to fourteen materials. p. 7 The Mnd values indicate the degree of discreteness of the optimized designs, which are very close to zero in all cases, indicating a near-binary solution. p. 8 The pylon design exhibits smooth and stable objective minimization during convergence histories. p. 0

Conclusion of Findings

The paper concludes that the proposed formulation not only satisfies the prescribed volume constraints but also yields sharply defined, optimized design topologies. The formulated algorithm strategically places each material where it contributes most to stiffness while keeping the volume fraction of each material at its maximum prescribed limit. p. 8 This demonstrates that the proposed method minimizes pylon compliance under loading conditions while achieving crisp material states. p. 8

References

  1. M. Stefanovic and E. p. 8 Livne, “Structural design synthesis of aircraft engine pylons at certification level of detail,” Journal of Aircraft, vol. p. 8 58, no. 4, pp. 935–949, 2021 <ref:2609. p. 827685#pg2>

  2. H.-g. Kim, S. Kim, S.-h. Hong, H.-k. Choi, and S. p. 8-h. Kim, “Structural static test for validation of the structural integrity of an aircraft pylon,” Aerospace Science and Technology, vol. p. 8 122, p. 107402, 2022 <ref:2609. p. 827685#pg2>

  3. S. F. Zettel, M. B¨oswald, R. p. 9 D. Dewald, and R. p. 9 Winter, “Jet engine vibration model for estimating pylon–wing interface loads for realistic structural excitation,” CEAS Aeronautical Journal, pp. p. 9 1–14, 2025 <ref:2609. p. 027685#pg4>

  4. P. Kumar, “HoneyTop90: A 90-line MATLAB code for topology optimization using honeycomb tessellation,” Optimization and Engineering, vol. p. 9 24, no. 2, pp. 1433–1460, 2023 <ref:2609. p. 927685#pg5>

  5. M. P. Bendsøe and N. p. 1 Kikuchi, “Generating optimal topologies in structural design using a homogenization method,” Computer Methods in Applied Mechanics and Engineering, vol. p. 9 71, no. 2, pp. 197–224, 1988 <ref:2609. p. 927685#pg6>

  6. M. P. Bendsøe, “Optimal shape design as a material distribution problem,” Structural Optimization, vol. p. 9 1, no. 4, pp. 193–202, 1989 <ref:2609. p. 927685#pg6>

  7. J.-H. Zhu, W.-H. Zhang, and L. p. 9-Xia, “Topology optimization in aircraft and aerospace structures design,” Archives of Computational Methods in Engineering, vol. p. 9 23, no. 4, pp. 595–622, 2016 <ref:2609. p. 927685#pg8>

  8. A. Remouchamps, M. p. 1 Bruyneel, C. p. 9 Fleury, and S. p. 9-Grihon, “Application of a bilevel scheme including topology optimization to the design of an aircraft pylon,” Structural and Multidisciplinary Optimization, vol. p. 9 44, no. 6, pp. 739–750, 2011 <ref:2609. p. 927685#pg9>

  9. S. Coniglio, C. p. 1 Gogu, R. Amargier, and J. p. 9-Morlier, “Engine pylon topology optimization framework based on performance and stress criteria,” AIAA Journal, vol. p. 9 57, no. 12, pp. 5514–5526, 2019 <ref:2609. p. 927685#pg9>

  10. H.- Khalid and A. p. 9 A. Gomez-Gallegos, “Substituting ti-64 with aa2099 as material of a commercial aircraft pylon,” Advances in Materials Science, vol. p. 9 21, no. 2, pp. 77–92, 2021 <ref:2609. p. 027685#pg9>

  11. P. Kumar, “Topology optimization of pressure-loaded multi-material structures,” in Structural Integrity Conference and Exhibition, pp. p. 9 339–351, Springer, 2022 <ref:2609. p. 927685#pg9>

  12. S. I. Sarkar, N. Singh, A Saxena, and P Kumar, “A generalized shape function approach for multimaterial topology optimization,” International Journal for Numerical Methods in Engineering, vol. p. 9 127, no. 16, p e70401, 2026 <ref:2609. p. 927685#pg9>

  13. T Gao and W Zhang, “A mass constraint formulation for structural topology optimization with multiphase materials,” International Journal for Numerical Methods in Engineering, vol. p. 9 88, no. 8, pp. 774–796, 2011 <ref:2609. p. 927685#pg4>

  14. J Stegmann and E Lund, “Discrete material optimization of general composite shell structures,” International Journal for Numerical Methods in Engineering, vol. p. 9 62, no. 14, pp. 2009–2027, 2005 <ref:2609. p. 927685#pg5>

  15. R Tavakoli and S M Mohseni, “Alternating active-phase algorithm for multimaterial topology optimization problems: a 115-line matlab implementation,” Structural and Multidisciplinary Optimization, vol. p. 9 49, no. 4, pp. 621–642, 2014 <ref:2609. p. 927685#pg6>

  16. A Chandrasekhar and K Suresh, “Multi-material topology optimization using neural networks,” Computer-Aided Design, vol. p. 9 136, p e103017, 2021 <ref:2609. p. 027685#pg9>

  17. L Yin and G Ananthasuresh, “Topology optimization of compliant mechanisms with multiple materials using a peak function material interpolation scheme,” Structural and Multidisciplinary Optimization, vol. p. 9 23, no. 1, pp. 49–62, 2001 <ref:2609. p. 027685#pg6>

  18. W Zuo and K Saitou, “Multi-material topology optimization using ordered simp interpolation: multi-material topology optimization using ordered simp,” Structural and Multidisciplinary Optimization, vol. p. 9 55, no. 2, pp. 477–491, 2017 <ref:2609. p. 927685#pg6>

  19. X Yang and M Li, “Discrete multi-material topology optimization under total mass constraint,” Computer-Aided Design, vol. p. 9 102, pp. 182–192, 2018 <ref:2609. p. 927685#pg4>

  20. K Svanberg, “The method of moving asymptotes—a new method for structural optimization,” International journal for numerical methods in engineering, vol. p. 9 24, no. 2, pp. 359–373, 1987 <ref:2609. p. 9

Green — the paper says it, and the page is printed. Red — no passage carries it, so treat that sentence as unverified. The same check runs on your own text: check a summary against its source, $5.

Improvements for AI systems

  1. Improved structural design for aircraft pylons by enabling topology optimization of multimaterial structures using a generalized shape function approach that uses n natural-coordinate design variables per element to provide optimized designs up to 2n distinct material phases.

  2. The improved AI system can perform automated structural design synthesis for aircraft components, such as engine pylons, optimizing them for minimum compliance subject to volume constraints across multiple candidate materials without requiring a considerable expansion of the design variable set.

  3. The system can achieve near discrete optimized solutions by employing a modified projection-based filter where the projected density of material is determined as ¯ζi = 2 tanh βγ + tanh β((˜ζi + 1)/2 − γ) tanh βγ + tanh β(1 − γ).

  4. The AI system can ensure crisp, discrete optimized solutions in multimaterial topology optimization by calculating the non-discreteness Mnd using the formula Mnd = X / N sum j=1 4¯ρMj / (1 - ρ¯Mj) X 100%.

  5. The system can accurately interpolate Young's modulus for an element as Ee = Emin + 2 sum m=1 (¯ρme) p (Em − Emin), where p is set to 3, allowing the optimization formulation to correctly calculate the Young's modulus of each element.

  6. The system can optimize complex multimaterial structures by minimizing structural compliance f0 = F T U subject to volume constraints gm = V t m / V a m - 1 <= 0, using a gradient-based optimization algorithm like the Method of Moving Asymptotes.

Related papers