Skip to main content
SpringerLink
Log in

Data-driven design exploration method using conditional variational autoencoder for airfoil design

  • Research Paper
  • Published:
Structural and Multidisciplinary Optimization Aims and scope Submit manuscript

Abstract

An objective of mechanical design is to obtain a shape that satisfies specific requirements. In the present work, we achieve this goal using a conditional variational autoencoder (CVAE). The method enables us to analyze the relationship between aerodynamic performance and the shape of aerodynamic parts, and to explore new designs for the parts. In the CVAE model, a shape is fed as an input and the corresponding aerodynamic performance index is fed as a continuous label. Then, shapes are generated by specifying the continuous label and latent vector. When CVAE is applied to mechanical design, it is desired to draw shapes that reproduce the specified aerodynamic performance. In ordinal CVAE, the model is trained to minimize reconstruction loss and latent loss, and it is usually optimized considering the sum of these losses. However, the present study shows that the optimal network is not always optimal in terms of reproducing the aerodynamic performance. The proposed method is verified using two numerical examples: a two-dimensional (2D) airfoil and a turbine blade. In the airfoil example, we demonstrate the effects of latent dimension, and in the turbine design example, we demonstrate that the proposed method can be applied to a real turbine design problem and reduce the design time.

This is a preview of subscription content, log in via an institution to check access.

Access this article

Log in via an institution

Price excludes VAT (USA)
Tax calculation will be finalised during checkout.

Instant access to the full article PDF.

Institutional subscriptions

Fig. 1
Fig. 2
Fig. 3
Fig. 4
Fig. 5
Fig. 6
Fig. 7
Fig. 8
Fig. 9
Fig. 10
Fig. 11
Fig. 12
Fig. 13
Fig. 14
Fig. 15
Fig. 16
Fig. 17
Fig. 18

Similar content being viewed by others

References

  • Abadi M, Agarwal A, Barham P, Brevdo E, Chen Z, Citro C, Corrado GS, Davis A, Dean J, Devin M, Ghemawat S, Goodfellow I, Harp A, Irving G, Isard M, Jia Y, Jozefowicz R, Kaiser L, Kudlur M, Levenberg J, Mané D, Monga R, Moore S, Murray D, Olah C, Schuster M, Shlens J, Steiner B, Sutskever I, Talwar K, Tucker P, Vanhoucke V, Vasudevan V, Viégas F, Vinyals O, Warden P, Wattenberg M, Wicke M, Yu Y, Zheng X (2015) TensorFlow: large-scale machine learning on heterogeneous systems. Software available from tensorflow.org

  • Abbot IH, von Doenhoff AE, Stivers L Jr (1945) Summary of airfoil data. NACA-TR-824

  • Akiba T, Sano S, Yanase T, Ohta T, Koyama M (2019) Optuna: a next-generation hyperparameter optimization framework. In: Proceedings of the 25th ACM SIGKDD international conference on knowledge discovery and data mining, pp 2623–2631

  • Bendsøe MP, Sigmund O (2003) Topology optimization: theory, methods and applications, 2nd edn. Springer, Berlin

    MATH  Google Scholar 

  • Bonaiuti D, Zangeneh M, Aartojarvi R, Eriksson J (2010) Parametric design of a waterjet pump by means of inverse design, CFD calculations and experimental analyses. J Fluids Eng 132(3):031104

    Article  Google Scholar 

  • Brown NC, Mueller CT (2019) Design variable analysis and generation for performance-based parametric modeling in architecture. Int J Archit Comput 17(1):36–52

    Google Scholar 

  • Cao S, Peng G, Yu Z (2005) Hydrodynamic design of rotodynamic pump impeller for multiphase pumping by combined approach of inverse design and CFD analysis. J Fluids Eng 172(2):330–338

    Article  Google Scholar 

  • Davidson TR, Falorsi L, de Cao N, Kipf T, Tomczak JM (2018) Hyperspherical variational auto-encoders. In: 34th conference on uncertainty in artificial Intelligence (UAI-18)

  • Dorney DJ, Lake JP, King P, Ashpis D (2000) Experimental and numerical investigation of losses in low-pressure turbine blade rows. Int J Turbo Jet Eng 17(4):241–253

    Google Scholar 

  • Drela M (1989) Xfoil: An analysis and design system for low Reynolds number airfoils. In: Mueller TJ (ed) Low Reynolds number aerodynamics, lecture notes in engineering, vol 54. Springer, Berlin, pp 1–12

  • Gaggero S, Vernengo G, Villa D, Bonfiglio L (2020) A reduced order approach for optimal design of efficient marine propellers. Ships Offshore Struc 15(2):319–328

    Article  Google Scholar 

  • Goto A, Zangeneh M (2002) Hydrodynamic design of pump diffuser using inverse design method and CFD. J Fluids Eng 124(2):319– 328

    Article  Google Scholar 

  • Hinton GE, Salakhutdinov RR (2006) Reducing the dimensionality of data with neural networks. Science 313(5786):504–507

    Article  MathSciNet  Google Scholar 

  • Imaizumi M, Fukumizu K (2019) Deep neural networks learn non-smooth functions effectively. J Mach Learn Res 89:869–878

    Google Scholar 

  • Kennon SR, Dulikravich GS (1985) Inverse design of coolant flow passage shapes with partially fixed internal geometries. Int J Turbo Jet-Engines 3:13–20

    Google Scholar 

  • Kingma DP, Welling M (2013) Auto-encoding variational bayes. In: The International Conference on Learning Representation (ICLR)

  • Lee I, Choi KK, Gorsick D (2010) System reliability-based design optimization using the MPP-based dimension reduction method. Struct Multidiscipl Optim 41(6):823–839

    Article  Google Scholar 

  • Li X, Ning S, Liu Z, Yan Z, Luo C, Zhuang Z (2020) Designing phononic crystal with anticipated band gap through a deep learning based data-driven method. Comput Methods Appl Mech Eng 361:112737

    Article  MathSciNet  Google Scholar 

  • Liu MY, Breuel T, Kautz J (2017) Unsupervised image-to-image translation networks. In: Advances in neural information processing systems 30. pp 700–708

  • Narducci R, Grossman B, Haftka RT (1995) Sensitivity algorithms for an inverse design problem involving a shock wave. Inverse Probl Eng 2(1):49–83

    Article  Google Scholar 

  • Nita K, Okita Y, Nakamata C, Kubo S, Yonekura K, Watanabe O (2014) Film cooling hole shape optimization using proper orthogonal decomposition. In: ASME Turbo expo 2014: turbine technical conference and exposition, pp GT2014–27239

  • Nita K, Okita Y, Nakamata C, Kubo S, Yonekura K, Watanabe O (2017) Turbine blade. US Patent 9,759,069

  • van den Oord A, Vinyals O, Kavukcuoglu K (2017) Neural discrete representation learning. In: Advances in neural information processing systems, vol. 30, pp 6306–6315

  • Pedregosa F, Varoquaux G, Gramfort A, Michel V, Thirion B, Grisel O, Blondel M, Prettenhofer P, Weiss R, Dubourg V, Vanderplas J, Passos A, Cournapeau D, Brucher M, Perrot M, Duchesnay E (2011) Scikit-learn: machine learning in Python. J Mach Learn Res 12:2825–2830

    MathSciNet  MATH  Google Scholar 

  • Petrucci DR, Filho NM (2007) A fast algorithm for inverse airfoil design using a transpiration model and an improved vortex panel method. J Braz Soc Mech Sci Eng 29(4):354–365

    Article  Google Scholar 

  • Pu Y, Gan Z, Henao R, Yuan X, Li C, Stevens A, Carin L (2016) Variational autoencoder for deep learning of images, labels and captions. In: Advances in neural information processing systems 29, pp 2352–2360

  • Raifai S, Vincent P, Muller X, Glorot X, Bengio Y (2011) Contractive auto-encoders: explicit invariance during feature extraction. In: Proceedings of the 28th international conference on machine learning (ICML)

  • Selig MS, Maughmer MD (1992) Generalized multipont inverse airfoil design. AIAA J 30 (11):2618–2625

    Article  Google Scholar 

  • Sohn K, Lee H, Yan X (2015) Learning structured output representation using deep conditional generative models. In: Advances in neural information processing systems 28, pp 3483–3491

  • Sokolowski J, Zolesio JP (1992) Introduction to shape optimization. Springer, Berlin

    Book  Google Scholar 

  • Sun L, Gao H, Pan S, Wang JX (2020) Surrogate modeling for fluid flows based on physics-constrained deep learning without simulation data. Comput Methods Appl Mech Eng 361:112732

    Article  MathSciNet  Google Scholar 

  • Bui-Thanh T, Damodaran M, Willcox K (2004) Aerodynamic data reconstruction and inverse design using proper orthogonal decomposition. AIAA J 42(8):1505–1516

    Article  Google Scholar 

  • Tan RK, Zhang NL, Ye W (2020) A deep learning-based method for the design of microstructural materials. Struct Multidiscip Optim 61(4):1417–1438

    Article  MathSciNet  Google Scholar 

  • Tolstikhin I, Bousquet O, Gelly S (2018) Wasserstein auto-encoders. In: The International Conference on Learning Representations (ICLR)

  • Tortorelli DA, Michaleris P (1994) Design sensitivity analysis: overview and review. Inverse Probl Eng 1(1):71–105

    Article  Google Scholar 

  • Wang C, Yan Xu L, Sheng Fan J (2020) A general deep learning framework for history-dependent response prediction based on ua-seq2seq model. Comput Methods Appl Mech Eng 372:113357

    Article  MathSciNet  Google Scholar 

  • Xu H, Chen W, Zhao N, Li Z, Bu J, Li Z, Liu Y, Zhao Y, Pei D, Feng Y, Chen J, Wang Z, Qiao H (2018) Unsupervised anomaly detection via variational auto-encoder for seasonal KPIs in web applications. In: Proceedings of the 2018 world wide web conference, pp 187–196

  • Yonekura K, Hattori H (2019) Framework for design optimization using deep reinforcement learning. Struct Multidiscip Optim 60(4):1709–1713

    Article  Google Scholar 

  • Yonekura K, Kanno Y (2015) A flow topology optimization method for steady state flow using transient information of flow field solved by lattice Boltzmann method. Struct Multidiscip Optim 51(1):159– 172

    Article  MathSciNet  Google Scholar 

  • Yonekura K, Kanno Y (2016) Erratum to: A flow topology optimization method for steady state flow using transient information of flow field solved by lattice Boltzmann method. Struct Multidiscip Optim 54 (1):193–195

    Article  MathSciNet  Google Scholar 

  • Yonekura K, Watanabe O (2014) A shape parameterization method using principal component analysis in application to shape optimization. J Mech Des 136(12):121401

    Article  Google Scholar 

  • Yu Y, Hur T, Jung J, Jang IG (2019) Deep learning for determining a near-optimal topological design without any iteration. Struct Multidiscip Optim 59(3):787–799

    Article  Google Scholar 

  • Zhang Y, Ye W (2019) Deep learning-based inverse method for layout design. Struct Multidiscip Optim 60(2):527–536

    Article  MathSciNet  Google Scholar 

  • Zhengming W (1985) Inverse design calculations for transonic cascades. In Turbo Expo: Power for Land, Sea, and Air, pp 85-GT-6

Download references

Author information

Authors and Affiliations

  1. The University of Tokyo, 7-3-1 Hongo, Bunkyo-ku, Tokyo, 113-8656, Japan

    Kazuo Yonekura & Katsuyuki Suzuki

  2. IHI Corporation, 1 Shin-nakahara-cho, Isogo-ku, Yokohama, 235-8501, Japan

    Kazuo Yonekura

Authors
  1. Kazuo Yonekura
    View author publications

    You can also search for this author in PubMed Google Scholar

  2. Katsuyuki Suzuki
    View author publications

    You can also search for this author in PubMed Google Scholar

Corresponding author

Correspondence to Kazuo Yonekura.

Ethics declarations

Conflict of interest

The authors declare that they have no conflict of interest.

Additional information

Responsible Editor: Ming Zhou

Replication of result

The source code of the neural networks used in the numerical experiments section is implemented using TensorFlow on Python. Even the NACA airfoil data are generated using Python. The turbine data are generated by an in-house code of IHI Corporation. We are willing to provide the data and source code upon a reasonable and responsible request. However, the copyrights of the in-house code for turbine data generation are reserved with IHI corporation and cannot be disclosed without permission from IHI Corporation.

Publisher’s note

Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.

Rights and permissions

Reprints and permissions

About this article

Check for updates. Verify currency and authenticity via CrossMark

Cite this article

Yonekura, K., Suzuki, K. Data-driven design exploration method using conditional variational autoencoder for airfoil design. Struct Multidisc Optim 64, 613–624 (2021). https://doi.org/10.1007/s00158-021-02851-0

Download citation

  • Received:

  • Revised:

  • Accepted:

  • Published:

  • Issue Date:

  • DOI: https://doi.org/10.1007/s00158-021-02851-0

Keywords

Search

Navigation