Artificial Neural Network application for predicting in-flight particle characteristics of an atmospheric plasma spray process

https://doi.org/10.1016/j.surfcoat.2011.04.099Get rights and content

Abstract

Thermal spray consists of a group of coating processes that are used to apply metal or non-metallic coatings to protect a functional surface or to improve its performance. There are some 40 processing parameters that define the overall coating quality and these must be selected in an optimized fashion to manufacture a coating that exhibits desirable properties. The proper combination of processing variables is critical since these influence the cost as well as the coating characteristics.

Because of this high number of processing parameters, a major challenge is to have full control over the system and to understand parameter interdependencies, correlations and their individual effects on the in-flight particle characteristics, which have significant influence on the in service coating properties. This paper proposes an approach, based on the Artificial Neural Network (ANN) method, to play this role and illustrates the model's design, network optimization procedures, the database handling and expansion steps, and analysis of the predicted values, with respect to the experimental ones, in order to evaluate the network's performance.

Research highlights

► Artificial Neural Network (ANN) successfully predicted particle characteristics. ► Predicted values in overall show minimum scatter with the experimental values. ► Expansion of database improved trained ANN's performance. ► Results show possibility of ANN to set up a control system for the spray process.

Introduction

Atmospheric Plasma Spray (APS) is a versatile thermal spray process used for the application of metal or non metallic coatings on a variety of candidate materials; e.g., metals, ceramics, composites and polymers [1], [2]. In APS, a plasma gas mixture, which is generally a mixture of argon (primary plasma forming gas) and hydrogen (secondary forming gas), is injected inside an anode. A high intensity direct current arc (between 300 A and 700 A) is produced between the tip of a cathode and the cylindric anode [3]. A high enthalpy zone of partially dissociated and ionized gases is considered as the process zone for feedstock. The feedstock material, generally a powder that is transported with the carrier gas, is injected into the process zone of the plasma jet where it is heated above its melting point. The outcome is that the powder particles are simultaneously heated and accelerated towards the substrate.

The inertia of the incoming powder distribution defines their path in the jet. On striking the substrate, the particles flatten and solidify in a few microseconds to form thin lamellae, often called splats, and subsequent stacking of these splats into from 20 to 100 layers allows a coating to form. The coating exhibits a layered structure, whose properties such as size and distribution of porosity, oxide content, residual stress, macro cracks and micro cracks are strongly affected by the in-flight particle characteristics, which are affected by the spray parameters. Accurate control and appropriate combination of the spray parameters are important since these influence the performance and durability of the coatings [2].

Plasma jets are largely heterogeneous systems incorporating substantial radial and longitudinal variations of temperature and velocity. Over a radial distance of 30 mm (at atmospheric pressure in air), the temperature may drop sharply from 15,000 K to almost room temperature and the velocity may drop from 1500 ms 1 to several decades lower [4]. The feedstock particles are passed through the core of the plasma jet, which is the hottest portion, to provide maximum exposure for complete melting and acceleration of the particles.

It is difficult to set up the process control due to the involvement of a many process parameters in APS. There is an associated cost to optimize the thermal spray parameters for new coating materials. Therefore, there is a need to reduce the variables to manageable numbers. The in-flight particle characteristics (in-flight particle velocity, surface temperature and diameter) are important parameters that are extensively studied and are known to influence coating properties [5], [6], [7] and, thus, are considered to be indicators for process control [8], [9]. Although it is the particle surface temperature that is actually measured at all times, for simplicity in both this work and others [6], [10], [11], [12], it is referred to as ‘particle temperature’; i.e., it is implied that the surface temperature is being measured. The in-flight particle characteristics are sensitive to the processing parameters [13], [14], especially to the following power and injection parameters: arc current intensity, argon gas flow rate, hydrogen flow rate, argon carrier gas flow rate, injector stand-off distance and the injector diameter.

In-flight particle optical sensors are used for real time monitoring of the coating manufacturing process [15], [16]. These sensors are, however, unable to tune the parameters to the proper and optimum operating values when the jet reveals any fluctuations, which makes the process control incomplete. It would be desirable to have a feedback system that can predict the in-flight particle characteristics, involving average particle velocity, temperature and diameter, with respect to the variations of each input processing parameter. The input parameters could thus be adjusted beforehand to achieve the desired particle characteristics. However, this task becomes difficult due to the non-linearity and many permutations of thermal spray processes [11].

The Artificial Neural Network (ANN) method is proposed in this work to overcome the above-mentioned technical difficulties. This method is used to establish process controls by predicting the in-flight particle characteristics of the APS process from variations of the power and injection parameters. The correlations between particle characteristics and coating properties are of similar complexity as mentioned above, however it is not covered in this work.

The initial idea of the neural network implementation for the thermal spray process was presented by Einerson et al. [17]. In another study [12], a non-linear dynamic system based on ANN was used to complete the APS process control. The authors in [18] studied the use of ANN in the complex APS process. In the second part of their work [19] the authors described an example linking processing parameters with the in-flight particle characteristics.

Section 2 of this paper describes the proposed mathematical model, which includes a formal description of ANN; the need for database collection; expansion of the data base if required; pre- and post-processing steps; and the network training and optimization steps. Section 3analyzes the results in terms of the optimized and trained ANN that is used to simulate the experimental database. The conclusions are presented in Section 4.

Section snippets

Mathematical model

The following sections describe the development of a model for predicting the in-flight particle characteristics for various combinations of input parameters.

Results and discussions

The network, trained with an expanded database and with nine and eight neurons in the first and second hidden layers respectively, is chosen in this section to predict the in-flight particle characteristics from the input processing parameters of a plasma spray process.

The predicted values obtained from the neural network were compared with their respective experimental values and the relative error percentages (with respect to the experimental values) were calculated. All the relative error

Conclusions

The Artificial Neural Network (ANN) is used here to study and design a model in order to predict the output in-flight particle characteristics of an Atmospheric Plasma Spray (APS) process from the input power and injection parameters. It also establishes and helps in understanding the correlations between the output and input parameters. To perform this task we borrowed the database built from the experimental measurements (using DPV-2000 from TECNAR Automation Limited, St-Bruno, QC, Canada J3V

Acknowledgements

The authors would like to thank Swinburne University of Technology for providing Swinburne University Post Graduate Research Award (SUPRA) to facilitate the research.

The authors also appreciate the support of the Power and Energy Group and Thermal Spray Group (Industrial Research Institute Swinburne—IRIS) of Swinburne in assisting with this research.

T. A. Choudhury—is doing his PhD at Swinburne University of Technology, Australia, under the Faculty of Engineering and Industrial Science.

References (40)

  • E. Pfender

    Surf. Coat. Technol.

    (1988)
  • E. Pfender

    Thin Solid Films

    (1994)
  • S. Guessasma et al.

    Mater. Des.

    (2003)
  • S. Guessasma et al.

    Mater. Sci. Eng., B: Solid-State Mater. Adv. Technol.

    (2004)
  • A.J. Adeloye et al.

    J. Hydrol.

    (2006)
  • P. Fauchais et al.

    Pure Appl. Chem.

    (1994)
  • S. Guessasma et al.
  • C. Moreau et al.
  • J. Guilemany et al.

    On-line-monitoring control of stainless steel coatings obtained by APS processes

    (2002)
  • M. Friis et al.
  • L. Leblanc et al.

    J. Therm. Spray Technol.

    (2002)
  • M. Vardelle et al.

    Pure Appl. Chem.

    (1999)
  • J.F. Bisson et al.

    J. Therm. Spray Technol.

    (2003)
  • S. Guessasma et al.

    J. Therm. Spray Technol.

    (2004)
  • J.-E. Döring et al.
  • E. Lugscheider et al.
  • A. Refke et al.
  • C. Moreau
  • C.J. Einerson et al.
  • S. Guessasma et al.
  • Cited by (49)

    • Digital transformation of thermal and cold spray processes with emphasis on machine learning

      2022, Surface and Coatings Technology
      Citation Excerpt :

      This study highlighted the influence of in-flight particle characteristics on bond strength and the results were quantitatively demonstrated and predicted with the use of ANNs. In several other studies [112–121] optimized ANNs were developed and trained from experimental data for process optimization, control and coating property prediction purposes. More specifically, in [122] the wear data are used to create predictive ANN models for the material loss.

    • Prediction of control parameters corresponding to in-flight particles in atmospheric plasma spray employing convolutional neural networks

      2020, Surface and Coatings Technology
      Citation Excerpt :

      S. Guessasma et al. used artificial neural network to model the plasma spraying process, they analyzed and predicted the microstructure and property of coatings, such as porosity, the content of unmelted particles, and different components [27–31]. T. A. Choudhury et al. modeled plasma spraying process implementing different artificial neural networks and carried out modular analysis, then they predicted the status of in-flight particles according to different control process parameters by the trained ANN model [32,33]. A. F. Kanta et al. studied the relationship between control parameters and in-flight particle characteristics by analyzing the plasma spraying process of preparing the specific coating with the artificial neural network model [34–36].

    • Process modeling and parameter optimization of surface coatings using artificial neural networks (ANNs): State-of-the-art review

      2020, Materials Today: Proceedings
      Citation Excerpt :

      They concluded that the increase in particle temperature induced dense coatings and performance of gas-tightness is improved. T.A. Choudhary et al. [105] predicted the inflight particles characteristics of APS using ANN model. A multi-layered feed forward back propagation was employed to predict the in-flight particle characteristics such as particle temperature, particle velocity and particle diameter.

    View all citing articles on Scopus

    T. A. Choudhury—is doing his PhD at Swinburne University of Technology, Australia, under the Faculty of Engineering and Industrial Science.

    N. Hosseinzadeh—is the Senior Lecturer in the Faculty of Engineering and Industrial Science at Swinburne University of Technology, Australia.

    C. C. Berndt—is a Professor in the Faculty of Engineering and Industrial Science at Swinburne University of Technology, Australia, and is currently the Director of Industrial Research Institute Swinburne (IRIS).

    View full text