PSI - Issue 54

Available online at www.sciencedirect.com Structural Integrity Procedia 00 (2023) 000 – 000 Available online at www.sciencedirect.com ^ĐŝĞŶĐĞ ŝƌĞĐƚ

www.elsevier.com/locate/procedia

ScienceDirect

Procedia Structural Integrity 54 (2024) 1–2

International Conference on Structural Integrity 2023 (ICSI 2023) Editorial – ICSI2023 Pedro Moreira*, Paulo Tavares INEGI – Institute of Science and Innovation in Mechanical and Industrial Engineering, Porto, Portugal

In the last two years, since the virtual ICSI2021, research activity in Structural Integrity has seen an exponential increase, spilling over several exciting areas in materials, methods, and applications. For the most part, this has been fueled by both the necessity of diversifying energy sources and the societal pressure to cope with climate change and the issues brought about by the required technological development. Research into metal (mis)behavior in the presence of Hydrogen, to cite an example, which has long been an important topic in Structural Integrity, gradually came under the focus of a growing number of scientists due to its importance in Hydrogen storage. In testimony of the importance of this topic, nearly one-fourth of all ICSI2023 accepted abstracts focusing on Hydrogen Embrittlement-related topics. Concomitantly, the experimental activity related to experimental validation in new simulation concepts and applications, and novel applications for validated simulation models which, for instance, enable sensor virtualization, together with disruptive sensing technologies from the realm of science fiction a mere couple of years ago, are paving the way for a revamped R&D arena, with implications in most Engineering fields. From the very start, five editions ago, ICSI has focused on all aspects and scales of structural integrity, ranging from basics to future trends, with special emphasis on multi-scale and multi-physics approaches, and applications to new materials and challenging environments. Current research topics in the realm of Structural Integrity targeted by ICSI2023 include, but are not limited to Fracture and Fatigue, Stress Analysis, Damage Tolerance, Durability, Crack Closure, Joining Technologies, Nanomechanics and Nanomaterials, Ageing, Coatings Technology, Environmental Effects, Structural Health Monitoring, New materials, Surface Engineering, Integrity of biomechanics structures in and many other exciting research topics. In 2023, the ICSI organization focused on inviting lecturers related to topics that dominate the contemporary status in Structural Integrity, such as Prof. Frank Cheng from the University of Calgary, working in the field of corrosion engineering and pipeli ne reliability, Prof. Su Taylor from Queen’s University in Belfast, devoted to the development of structural health monitoring of sustainable infrastructure; Prof. José Correia from Porto University, fully engaged in Structural Integrity of energy infrastructure; Prof. César Azevedo well known for his contribution in structural failure and Prof. Zagorac, from Belgrade University, working in the field of material behavior under extreme

* Corresponding author. Tel.: +351 229578710 E-mail address: pmoreira@inegi.up.pt

2452-3216 © 2023 The Authors, Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers

2452-3216 © 2023 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers 10.1016/j.prostr.2024.01.048

Pedro Moreira et al. / Procedia Structural Integrity 54 (2024) 1–2 Author name / Structural Integrity Procedia 00 (2019) 000 – 000

2 2

conditions. Like previous editions, ICSI2023 has been organized into a general track and a number of thematic symposia. Apart from Procedia Structural Integrity, a special issue from Engineering Failure Analysis will cover ICSI2023. The response to the organization’s efforts has been outstanding: Thirteen symposia were proposed; the number of abstract submissions was kept at a similar level to previous editions, with around 200 approved for oral communication. These are very challenging times for the organization of meaningful conferences, but the organizers strived to make this 5th Edition a memorable one, which will stimulate both young and well-known researchers in the field to contribute further to ICSI.

Available online at www.sciencedirect.com Available online at www.sciencedirect.com Available online at www.sciencedirect.com

ScienceDirect

Procedia Structural Integrity 54 (2024) 225–232 Structural Integrity Procedia 00 (2023) 000–000 Structural Integrity Procedia 00 (2023) 000–000

www.elsevier.com / locate / procedia www.elsevier.com / locate / procedia

© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0 ) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http: // creativecommons.org / licenses / by-nc-nd / 4.0 / ) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers. Keywords: Acoustic Emission; Akaike Information Criterion; Time of Arrival; Syntactic Foams; Cenosphere Abstract Innovation in material science and the introduction of new structural materials into the commercial and industrial market are forcing researchers to look for modern, sophisticated and advanced structural monitoring tools. The Acoustic Emission (AE) technique is one of the most widely used Structural Health Monitoring (SHM) tools. Despite its remarkable advantages over all other passive Non-Destructive Evaluation (NDE) tools, its applicability is constantly being questioned. This is due to the complex time-frequency characteristics of acoustic waves, especially when used in highly noisy environments or highly non-homogeneous structures. In this study, an advanced signal processing technique is used to analyse the acoustic signals generated by di ff erent failure modes of composite specimens. A novel procedure is proposed to extract the Time of Arrival (ToA) of the AE signals. This approach uses a modified version of the Akaike Information Criterion (AIC), which is applicable also to signals with a low signal-to-noise ratio. The proposed method is tested on the AE signals generated from the static tensile test of the syntactic foam, cenosphere-reinforced unsaturated polyester composites. The ToA of the AE signals successfully identifies the damage evolution stages in the syntactic foam. © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http: // creativecommons.org / licenses / by-nc-nd / 4.0 / ) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers. Keywords: Acoustic Emission; Akaike Information Criterion; Time of Arrival; Syntactic Foams; Cenosphere International Conference on Structural Integrity 2023 (ICSI 2023) Advanced Acoustic Emission Signal Processing Techniques for Structural Health Monitoring Claudia Barile a , Vimalathithan Paramsamy Kannan a, ∗ , Giovanni Pappalettera a , Caterina Casavola a a Dipartimento di Meccanica, Matematica e Management, Politecnico di Bari, Via E.Orabona 4, 70125 - Bari, Italy Abstract Innovation in material science and the introduction of new structural materials into the commercial and industrial market are forcing researchers to look for modern, sophisticated and advanced structural monitoring tools. The Acoustic Emission (AE) technique is one of the most widely used Structural Health Monitoring (SHM) tools. Despite its remarkable advantages over all other passive Non-Destructive Evaluation (NDE) tools, its applicability is constantly being questioned. This is due to the complex time-frequency characteristics of acoustic waves, especially when used in highly noisy environments or highly non-homogeneous structures. In this study, an advanced signal processing technique is used to analyse the acoustic signals generated by di ff erent failure modes of composite specimens. A novel procedure is proposed to extract the Time of Arrival (ToA) of the AE signals. This approach uses a modified version of the Akaike Information Criterion (AIC), which is applicable also to signals with a low signal-to-noise ratio. The proposed method is tested on the AE signals generated from the static tensile test of the syntactic foam, cenosphere-reinforced unsaturated polyester composites. The ToA of the AE signals successfully identifies the damage evolution stages in the syntactic foam. International Conference on Structural Integrity 2023 (ICSI 2023) Advanced Acoustic Emission Signal Processing Techniques for Structural Health Monitoring Claudia Barile a , Vimalathithan Paramsamy Kannan a, ∗ , Giovanni Pappalettera a , Caterina Casavola a a Dipartimento di Meccanica, Matematica e Management, Politecnico di Bari, Via E.Orabona 4, 70125 - Bari, Italy

1. Introduction 1. Introduction

The Acoustic Emission (AE) technique is considered as one of the most formidable Nondestructive Evaluation (NDE) for structural health monitoring. Since its introduction in the 1960s, its definitions, use and applications have been documented by several researchers [Sause et al. (2012), Barile et al. (2020)]. In short, in this technique, the stress waves generated by a material / structure undergoing deformation due to external forces are used to analyse the characteristics of the deforming material / structure. The stress waves are analysed in two ways: parameter-based anal ysis and signal-based analysis. The parameter-based analysis is useful for in-situ analysis while serving as a passive The Acoustic Emission (AE) technique is considered as one of the most formidable Nondestructive Evaluation (NDE) for structural health monitoring. Since its introduction in the 1960s, its definitions, use and applications have been documented by several researchers [Sause et al. (2012), Barile et al. (2020)]. In short, in this technique, the stress waves generated by a material / structure undergoing deformation due to external forces are used to analyse the characteristics of the deforming material / structure. The stress waves are analysed in two ways: parameter-based anal ysis and signal-based analysis. The parameter-based analysis is useful for in-situ analysis while serving as a passive

∗ Corresponding author. Vimalathithan Paramsamy Kannan E-mail address: pk.vimalathithan@poliba.it ∗ Corresponding author. Vimalathithan Paramsamy Kannan E-mail address: pk.vimalathithan@poliba.it

2452-3216 © 2023 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers 10.1016/j.prostr.2024.01.077 2210-7843 © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http: // creativecommons.org / licenses / by-nc-nd / 4.0 / ) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers. 2210-7843 © 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http: // creativecommons.org / licenses / by-nc-nd / 4.0 / ) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers.

Claudia Barile et al. / Procedia Structural Integrity 54 (2024) 225–232 C. Barile et al. / Structural Integrity Procedia 00 (2023) 000–000

226

2

NDE tool, however, it is not as e ffi cient as the signal-based analysis. The signal-based analysis is time-consuming and cost-intensive in monitoring large structures. Researchers tried to bridge the gap between these two types of analysis by introducing some new parameters such as entropy, complexity etc [Burud and Kishen (2021), Barile et al. (2023)]. Time of Arrival (ToA) is one of the most e ffi cient and robust parameters in the analysis of stress waves / acoustic waves. Several researchers, over the years, have introduced di ff erent techniques for estimating the ToA of the AE signals. While most of them function e ffi ciently, they all share a common limitation: their e ffi ciency depends on the Signal-to-Noise ratio (SNR) of the acquired signal. None of the available techniques are applicable for signals with lowSNR. In this research work, a novel approach, which is based on the popular Akaike Information Criterion (AIC) is introduced to estimate the ToA of the AE signals. Its e ffi ciency is tested by characterizing the damage evolution stages of syntactic foam test specimens under tensile loading. The ToA of the AE signals generated from the static tensile test are estimated and used for analysis.

2. Methods

2.1. Estimation of Time of Arrival

The novel procedure proposed to estimate the ToA of the AE signal is schematically presented in Figure 1. First, the signal must be converted to its characteristic form in order to overcome the limitations regarding the SNR of the signals. Several researchers have proposed di ff erent methods such as the envelope of the signal, absolute level or RMS of the signal as their characteristic form [Siracusano et al. (2016), Sedlak et al. (2008)]. In this work, a characteristic function based on Allen’s function is proposed. Consider an AE signal X of length N as shown in Equation (1).

Fig. 1. Schematic Representation of the Procedure for Estimating ToA

X = x 1 , x 2 , x 3 , ..., x N

(1)

The characteristic function of signal X is calculated using Equation (2). X ( n ) = | x ( n ) | + R | x ( n ) − x ( n − 1) | , R = 4 (2) x ( n ) is the signal datapoint at n and R = 4 is calculated based on trail-and-error method. Figure 2 shows the original signal and its characteristic form compared to the envelope and absolute level of the signal.

Claudia Barile et al. / Procedia Structural Integrity 54 (2024) 225–232 C. Barile et al. / Structural Integrity Procedia 00 (2023) 000–000

227

3

Fig. 2. An Example Signal, its Envelope, Absolute Level and Characteristic Form

After this, the signal is segmented into k segments, each having equal length. The Yule-Walker autoregressive (AR) model of order M is fit to each segment. The coe ffi cients of the AR model are given by a i m , where m = 1 , 2 , 3 , .., M . The AR model is given in Equation (3) [Carpinteri et al. (2012)].

M  m = 1

a i

m x t − m + e i t

x n =

(3)

e i t is the Gaussian noise with zero mean and a standard deviation of σ i , which is used to create the AR model. Now the goal is to divide the signal into a deterministic part of intervals 1 , ..., k and a non-deterministic part of intervals k + 1 , ..., K (where K is the total number of segments) and to identify the segment where the ToA may appear. The approximate likelihood function that divides the deterministic and non-deterministic parts is given by

exp    −

2   

n i  j = p i    x j −

m x j − m   

2  i = 1    1 σ 2

i 2 π    2

M  m = 1

1 2 σ 2 i

a i

G ( x ) =

(4)

The maximum likelihood function of Equation (4) is given by

n i  j = p i    x j −

m x j − m    2

M  m = 1

1 n i

σ 2

a i

(5)

i , max =

For splitting the AR models of the segments k into non-deterministic (before ToA) and deterministic (post ToA) parts, i = 1 , 2. For i = 1 , 2, Equation (5) can be solved by assigning p 1 = 1, p 2 = k + 1, n 1 = k and n 2 = n − k . The entropy of σ 2 i , max , k is calculated for k = 1 , 2 , 3 .., K by considering that the maximum likelihood solution is a series of data for k . The entropy used in this study is Renyi’s entropy, which is given by H ( k ) = − log    K  k = 1 P ( k ) 2    , P ( k ) = σ 2 i , max , k (6) By minimizing H ( k ), the segment where ToA may appear can be identified. ToA appears in the segment where H ( k ) is minimum. Now, in order to find the precise ToA, Akaike Information Criterion (AIC) is used [Kitagawa and Akaike (1978)]. AIC ( w ) = wlog [ var ( x { 1 , w } )] + ( N − w − 1) log [ var ( x { w + 1 , N } )] (7)

Claudia Barile et al. / Procedia Structural Integrity 54 (2024) 225–232 C. Barile et al. / Structural Integrity Procedia 00 (2023) 000–000

228

4

where, w is the range of the signal data points in the identified segment k and N is the total length of the segment. The point at which AIC returns the minimum value is the ToA of the signal. Few sample signals are taken for analysis and their ToA are estimated and presented in Figure 3.

Fig. 3. Estimation of ToA by the Proposed Approach for Sample Signals

3. Experimental Methods

In order to experimentally validate the application of ToA, the following experiment is carried out. Cenosphere reinforced unsaturated polyester resin composites (syntactic foams) are prepared using the resin cast moulding pro cess. In the author’s previous works, syntactic foams are prepared from vinyl ester resin systems and cenosphere composites [Vimalathithan and Vijayakumar (2018)]. The same procedure is followed in this work to prepare the un saturated polyester syntactic foams while keeping the catalyst and accelerator for the polymerization process the same as in the previous work. Tensile specimens with rectangular cross-sections are prepared according to ASTM D3039 - “Standard Test Method for Tensile Properties of Polymer Matrix Composite Materials”. To record the acoustic waves generated during the tensile tests, a wide-band piezoelectric sensor is attached to the surface of the test specimen in the gauge length. The piezoelectric sensor has a minimum acquisition frequency of 200 kHz and resonant frequencies at 250 kHz and 550 kHz. The signals which cross the detection threshold of 35 dB are acquired and amplified by 40 dB before recording them at a sampling frequency of 2 MHz. The length of the signals acquired are 1024 samples and the ToA is estimated by using the procedure explained in Section 2 by setting k = 32 segments with each segment having an equal length of 32 samples.

4. Results and Discussions

Three cenosphere-reinforced unsaturated polyester syntactic foams are tested in this study. Before going to the AE results based on the ToA, the mechanical test results are explained in this section. Understanding the mechanical behaviour of these test specimens may help in comparing them with the AE results.

Claudia Barile et al. / Procedia Structural Integrity 54 (2024) 225–232 C. Barile et al. / Structural Integrity Procedia 00 (2023) 000–000

229 5

Fig. 4. Stress-Strain Behaviour of Cenosphere-reinforced Unsaturated Polyester Syntactic Foams

4.1. Mechanical Test Results

The stress-strain behaviours of the syntactic foam test specimens are presented in Figure 4. For the sake of brevity, longitudinal strain behaviour is discussed here. All three specimens show linear stress-strain behaviour with apparently no yielding, which is a characteristic feature of thermoset polymer systems.

Fig. 5. SEM Micrographs of the Fractured Surface of Cenosphere / Unsaturated Polyester Syntactic Foams

Claudia Barile et al. / Procedia Structural Integrity 54 (2024) 225–232 C. Barile et al. / Structural Integrity Procedia 00 (2023) 000–000

230

6

It has been reported in the literature that the addition of cenospheres in thermoset polymer resins results in im proving their elastic modulus, while negatively influencing the tensile strength [Deepthi et al. (2010), Kulkarni and Mhanwar (2012)]. In the author’s previous work, the influence of the weight percentage of cenospheres in the improve ment of structural properties of the thermoset polymer systems is reported [Vimalathithan and Vijayakumar (2018)]. The average elastic modulus of the syntactic foams is 3.33 GPa. Specimens 1 and 2 show very similar stress-strain behaviour, while the strain deformation is comparatively high in Specimen 3. Since these specimens are prepared in a mould box, there is a possibility of agglomeration of cenosphere particles, which may have negatively influenced the tensile strengths of Specimens 1 and 2. SEM micrographs of the fracture surfaces are presented in Figure 5. Progressive fracture in the form of beach marks can be found in Figures 5a-5d, which confirms the brittle failure in the specimens. In most cases, the cracks initiate from the boundary edge of the test specimen and propagate through the surface (Figure 5(b)) and they culminate around the cenosphere particles, which resulted in their removal from the resin surface (Figure 5(c)). This mechanism possibly has improved the elastic modulus and fracture toughness of this composite system. Only in very few cases, the symmetrical cleaving of the cenosphere can be observed (Figure 5(d)).

4.2. Acoustic Emission Test Results

Fig. 6. ToA and Peak Frequency of the AE signals from a) Specimen 1, b) Specimen 2, and c) Specimen 3 classified using Density-based Unsuper vised Clustering

In thermoset polymer systems, the failure modes that generate AE signals are generally matrix cracking / friction or rubbing between the damaged portions of the matrix and the crack growth [Sause et al. (2012), Barile et al. (2020)]. However, in these syntactic foam test specimens, other failure modes apart from those above two can be observed, for instance, removal of cenospheres and fracture in cenosphere. These may also generate AE signals which vary in their

Claudia Barile et al. / Procedia Structural Integrity 54 (2024) 225–232 C. Barile et al. / Structural Integrity Procedia 00 (2023) 000–000

231

7

time, frequency or time-frequency characteristics. To understand the failure characteristics, the ToA of the AE signals are classified with respect to their peak frequencies using a density-based unsupervised clustering technique. First, the neighbourhood clustering density threshold is calculated using the k-nearest neighbour (k-NN) search method. Then using this density threshold the ToA of the AE signals are classified. The results are presented in Figure 6. More details about the density-based unsupervised clustering and the estimation of the clustering threshold are well-documented in the literature [Ester et al. (1996)]. Based on the density of ToA distribution, the AE signals are classified into four clusters (excluding the outliers) with respect to their peak frequencies. Clusters 1 and 2 have very few signals, while most of the signals are classified into Cluster 3, which has between ToA 325 µ s and 420 µ s and peak frequencies above 850 kHz (note that their ToA is greater than 420 µ s in Specimen 3). Typically, brittle fracture in an unsaturated polyester system is associated with crack growth through the surface of the material. The AE signals associated with the crack growth events often have larger peak frequencies. Considering the number of signals present in high-frequency clusters, Cluster 3 and Cluster 4, it is safe to assume that these signals are responsible for the final failure of the test specimens.

Fig. 7. Comparison between AE results and Mechanical Results for a) Specimen 1, b) Specimen 2, and c) Specimen 3

In order to validate this, the classified AE signals are compared with the strain evolution during the tensile loading. The results are presented in Figure 7. In all three specimens, the AE signals in Clusters 1 and 2, which have ToA between 370 µ s and390 µ s and peak frequencies less than 500 kHz can be found throughout the strain evolution stages. These signals could be generated from the local dislocations such as matrix cracking in the vicinity of voids, friction between the matrices, etc. AE signals in Cluster 3, however, begin to appear when the strain reaches around 3000 µϵ in all three specimens. Considering the peak frequency characteristics of these signals and their wide distribution of ToA, it can be assumed that the nonuniform and non-localized stress distribution in the syntactic foams begins around 3000 µϵ , which initiated several local microcracks. The non-localized damage localization might be the reason why the ToA has a larger distribution in these clusters.

Claudia Barile et al. / Procedia Structural Integrity 54 (2024) 225–232 C. Barile et al. / Structural Integrity Procedia 00 (2023) 000–000

232

8

The AE signals in Cluster 4 begin to appear abundantly just before the failure: 5000 µϵ in Specimen 1, which failed at 5125 µϵ ; 4300 µϵ in Specimen 2 and 6000 µϵ in Specimen 3. Possibly, these signals could represent the damage mode responsible for the final failure of the test specimens. Theoretically, these signals could be generated from the cracks which culminated around the cenosphere failed by the removal of these spherical particles from the resin surface. Perhaps, additional analysis using some other in-situ techniques might validate this hypothesis. Nonetheless, the AE signals, which are classified based on their ToA and peak frequencies can identify the di ff erent damage evolution stages in the syntactic foams under tensile loading. Perhaps, an additional in-situ technique may be useful in associating specific failure modes to each cluster of AE signals. It must be noted that the main objective of this research work is to validate whether the ToA estimated using the proposed method can be used for damage characterization and structural health monitoring. The presented results show that the estimated ToA of the AE signals can potentially be used in structural health monitoring applications. A novel method for estimating the Time of Arrival (ToA) of the Acoustic Emission (AE) signals is proposed in this research work. The proposed method is also applicable to signals with a low signal-to-noise ratio and signals with multiple amplitude peaks. The applicability of the ToA estimated by the proposed method in structural health monitoring applications is validated in this research work. The ToA of the AE signals generated from the tensile tests of cenosphere-reinforced unsaturated polyester syntactic foams are estimated. The estimated ToA, when classified along with the peak frequencies of the AE signals can potentially identify the di ff erent failure modes in test specimens. However, the applicability of this proposed method must be controlled for the AE signals generated from the larger structures. Barile, C., Casavola, C., Pappalettera, G. and Kannan, V.P., 2020. Application of di ff erent acoustic emission descriptors in damage assessment of fiber reinforced plastics: A comprehensive review. Engineering Fracture Mechanics, 235, p.107083. Barile, C., Casavola, C., Pappalettera, G. and Kannan, V.P., 2023. Interpreting the Lempel–Ziv complexity of acoustic emission signals for identi fying damage modes in composite materials. Structural Health Monitoring, 22(3), pp.1708-1720. Burud, N. and Kishen, J.C., 2021. Damage detection using wavelet entropy of acoustic emission waveforms in concrete under flexure. Structural Health Monitoring, 20(5), pp.2461-2475. Carpinteri, A., Xu, J., Lacidogna, G. and Manuello, A., 2012. Reliable onset time determination and source location of acoustic emissions in concrete structures. Cement and concrete composites, 34(4), pp.529-537. Deepthi, M.V., Sharma, M., Sailaja, R.R.N., Anantha, P., Sampathkumaran, P. and Seetharamu, S., 2010. Mechanical and thermal characteristics of high density polyethylene–fly ash cenospheres composites. Materials & Design, 31(4), pp.2051-2060. Ester, M., Kriegel, H.P., Sander, J. and Xu, X., 1996, August. A density-based algorithm for discovering clusters in large spatial databases with noise. In kdd (Vol. 96, No. 34, pp. 226-231). Kitagawa, G. and Akaike, H., 1978. A procedure for the modeling of non-stationary time series. Annals of the Institute of Statistical Mathematics, 30, pp.351-363. Kulkarni, M.B. and Mahanwar, P.A., 2012. E ff ect of methyl methacrylate–acrylonitrile-butadiene–styrene (MABS) on the mechanical and thermal properties of poly (Methyl Methacrylate)(PMMA)-fly ash cenospheres (FAC) filled composites. Journal of Minerals and Materials Characteri zation and Engineering, 11(04), p.365. Sause, M.G., Mu¨ller, T., Horoschenko ff , A. and Horn, S., 2012. Quantification of failure mechanisms in mode-I loading of fiber reinforced plastics utilizing acoustic emission analysis. Composites science and technology, 72(2), pp.167-174. Sedlak, P., Hirose, Y., Enoki, M. and Sikula, J., 2008. Arrival time detection in thin multilayer plates on the basis of Akaike information criterion. Journal of Acoustic emission, 26, pp.182-188. Siracusano, G., Lamonaca, F., Tomasello, R., Garesc`ı, F., La Corte, A., Carn`ı, D.L., Carpentieri, M., Grimaldi, D. and Finocchio, G., 2016. A frame work for the damage evaluation of acoustic emission signals through Hilbert–Huang transform. Mechanical Systems and Signal Processing, 75, pp.109-122. Vimalathithan, P.K. and Vijayakumar, C.T., 2018. Characterization of cenosphere-reinforced vinyl ester composites. Journal of Elastomers & Plastics, 50(2), pp.95-106. 5. Conclusion References

Available online at www.sciencedirect.com Structural Integrity Procedia 00 (2023) 000 – 000 ^ĐŝĞŶĐĞ ŝƌĞĐƚ Available online at www.sciencedirect.com ^ĐŝĞŶĐĞ ŝƌĞĐƚ Structural Integrity Procedia 00 (2023) 000 – 000

www.elsevier.com/locate/procedia

www.elsevier.com/locate/procedia

ScienceDirect

Procedia Structural Integrity 54 (2024) 498–505

© 2023 The Authors. Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0 ) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers Abstract The article describes methods for creating parameterized geometry and presents the CAD models used in the study. Aerodynamic optimization methods in the context of shape optimization are described. The problem of optimization of the shape of the nacelle support suspended under the wing of the UAV is presented. The methodology of the study assuming two optimization processes is comprehensively described. The first assumes optimization of the shape of the profile, while the second assumes optimization of the height of the cantilever. Two optimization methods were used, namely design of experiments and gradient optimization method. The optimization results of the two methods were compared and conclusions were drawn. Translated with www.DeepL.com/Translator (free version) © 2023 The Authors, Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers 1 Department of Fundamentals of Machinery Design, Silesian University of Technology, Konarskiego 18A, 44-100 Gliwice, Poland, Abstract The article describes methods for creating parameterized geometry and presents the CAD models used in the study. Aerodynamic optimization methods in the context of shape optimization are described. The problem of optimization of the shape of the nacelle support suspended under the wing of the UAV is presented. The methodology of the study assuming two optimization processes is comprehensively described. The first assumes optimization of the shape of the profile, while the second assumes optimization of the height of the cantilever. Two optimization methods were used, namely design of experiments and gradient optimization method. The optimization results of the two methods were compared and conclusions were drawn. Translated with www.DeepL.com/Translator (free version) © 2023 The Authors, Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers Keywords: Optimization; aerodynamics; gradient; parameterization; DOE 1. Introduction Introduction In the era of growing needs for unmanned aerial vehicles, optimization of their aerodynamic structure is becoming a key research issue. In order to increase the efficiency and reliability of the operation of these devices, it is necessary to develop an effective methodology for optimizing their design. 1 This work will discuss the concept of a methodology for optimizing the aerodynamic structure of unmanned aerial vehicles. The use of this method allows for fine-tuning the shape of the surface to meet the needs of a specific application, leading to improved aerodynamic properties and increased energy efficiency. Keywords: Optimization; aerodynamics; gradient; parameterization; DOE 1. Introduction Introduction In the era of growing needs for unmanned aerial vehicles, optimization of their aerodynamic structure is becoming a key research issue. In order to increase the efficiency and reliability of the operation of these devices, it is necessary to develop an effective methodology for optimizing their design. 1 This work will discuss the concept of a methodology for optimizing the aerodynamic structure of unmanned aerial vehicles. The use of this method allows for fine-tuning the shape of the surface to meet the needs of a specific application, leading to improved aerodynamic properties and increased energy efficiency. International Conference on Structural Integrity 2023 (ICSI 2023) Aerodynamic optimization of an UAV Wojciech Skarka 1 * , , Bartosz Rodak 1 , International Conference on Structural Integrity 2023 (ICSI 2023) Aerodynamic optimization of an UAV Wojciech Skarka 1 * , , Bartosz Rodak 1 , 1 Department of Fundamentals of Machinery Design, Silesian University of Technology, Konarskiego 18A, 44-100 Gliwice, Poland,

* Corresponding author. Tel.: +48 32 237 14 91 E-mail address: wojciech.skarka@polsl.pl

2452-3216 © 2023 The Authors, Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers 2452-3216 © 2023 The Authors, Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers * Corresponding author. Tel.: +48 32 237 14 91 E-mail address: wojciech.skarka@polsl.pl

2452-3216 © 2023 The Authors. Published by ELSEVIER B.V. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0) Peer-review under responsibility of the scientific committee of the ICSI 2023 organizers 10.1016/j.prostr.2024.01.112

Wojciech Skarka et al. / Procedia Structural Integrity 54 (2024) 498–505 Bartosz Rodak/ Structural Integrity Procedia 00 (2019) 000 – 000

499

2

An unremarkable field when it comes to adapting surface shape is parametric geometry. Parametric geometry is a branch of mathematics that deals with describing and modeling shapes using mathematical functions. Unlike traditional geometry, in which figures are described by fixed coordinates, in parametric geometry figures are described by variable parameters. Parametric geometry is particularly useful in fields such as design, engineering and computer graphics, where accurate and precise modeling of shapes is necessary. In parametric geometry, figures are described by functions, which makes it easy to change their shape by changing parameter values. And consequently, thanks to parametrization, optimization algorithms can be easily combined with geometry editing. 2 This paper will present the theoretical basis of the optimization methodology and its application in practice, taking into account the various problems that can occur during the optimization process. 2. Parameterization and geometry Parametric geometry allows shapes to be described and modeled using mathematical functions. In optimization problems, these functions are used to describe many variables that can affect the final result. In this way, it is easy to change the value of the variables and determine the best solution for a given problem. Given this, generating a fully parameterized geometric model is crucial in an optimization problem. One of the main tools of parametric geometry is the equation of a curve. A curve is described by a parametric equation, which defines the position of a point on the curve depending on the value of a parameter. It is also possible to create solids using curve equations to model more complex shapes. A linear spline [1] will be used in solving the optimization problem using the design of experiments (DOE) method [2], which will be described in detail later in the paper. 3

2.1. CAD models

A CAD model of the connection between the wing and the nacelle was created. Separately - the model of the connecting element itself and the model of the whole with the wing and nacelle in order to simplify simulation and later verification of the whole structure. The connecting element is shown in Figure 1a, while the coordinates of the nodes of its sketch and the radius of the rounded leading

Fig. 1. (a) the connecting element; (b) coordinates of the nodes of its sketch and the radius of the rounded leading edge. edge are shown in Figure 1b.

The coordinates of the sketch nodes are input parameters in parametric geometry. The linear spline is based on these nodes. By parameterizing the sketch nodes, it is possible to modify the shape of the spline element at will. In a separate simulation for the optimal height of the connecting element, the input parameter will be the height of the connecting element. In this simulation, an analysis will be carried out on the entire structure to determine the optimal length of the connecting element between the wing and the nacelle.

Fig. 2. - CAD model of the entire wing structure, nacelle and connecting element Such a procedure will save time and computing power resources. Figure 2 shows a CAD model of the entire structure. It is assumed that the shape and size of the wing and nacelle are fixed. The nacelle is symmetrical in each axis.

Wojciech Skarka et al. / Procedia Structural Integrity 54 (2024) 498–505 Bartosz Rodak/ Structural Integrity Procedia 00 (2019) 000 – 000

500

3

3. Aerodynamic optimization methods Aerodynamic optimization is the process of finding the shape of an object that provides the minimum aerodynamic drag. Aerodynamic drag is the force that opposes the movement of an object through the air. The lower the aerodynamic drag, the faster and more efficiently an object can move. In this case, the finite element method was used in the ANSYS Fluent environment in combination with the optimization module. The methods they used were design of experiments [2] and gradient method [3], [4]. The idea of optimisation of the particular fragment is to solve the partial fragment of the shape decisive in the opinion of the designer or aerodynamicist about critical aerodynamic features such as drag or lift force. Examples of such important parts of the structure are, for example, the front wheel surrounding in the car [5], the exterior mirror in the car, or the mounting of the gondola under the wing or fuselage. Such fragments can be solved with certain restrictions independently of the rest of the structure. such verified computational methodologies can be used and developed in models developed in accordance with the MBD (Model Based Design) methodology [6] or for the development of generativemodels [7]. 4. Profile optimization by Design of Experiments method Full aerodynamic simulation of the element connecting the wing to the nacelle was carried out in the ANSYS Fluent environment. The dimensions of the individual points and their allowable values in the design of experiments [2], [3] module are shown in Table 1. Instead of the first point, a rounding was used on the leading edge with a radius that was an input parameter. The height of the connecting element in this test is fixed at 50mm

Table 1. Summary of node dimensions and their limits.

Direction

Nominal dimension

Maximum dimension

Minimum dimension

Point

0 0 2

X Y X Y X Y X Y

0 0

0 0

0 0

40mm

36mm

44mm

2 3 3 4 4 Edge round

13mm 130mm 4mm 200mm 0 1mm

11,7mm 117mm 3,6mm 180mm 0 0,5mm

14,3mm 143mm 4,4mm 220mm 0 8mm

Air velocity: 30m/s After simulation, the aerodynamic drag for the input geometry was calculated to be 0.0681N Once the simulation of the connecting element with the base geometry was done, the optimization process proceeded. The first step in optimization using the design of experiments method is to create an optimization plan based on the input parameters. The experiment plan was created in the design of experiments module in ANSYS software. The maximum and minimum values of the coordinates of each spline node, i.e. the input parameters, were taken into account. The experiment plan was to obtain the maximum entropy for the input parameters. For each experiment, an identical simulation was performed as for the base geometry. Table 2 below shows which input parameters gave the best result in terms of the smallest output parameter value.

Table 2. Coordinates of spline nodes, forming the optimal geometry.

Direction

Nominal dimension

After optimization dimension

Point

0

X

0

0

Wojciech Skarka et al. / Procedia Structural Integrity 54 (2024) 498–505 Bartosz Rodak/ Structural Integrity Procedia 00 (2019) 000 – 000

501

4

0 2

Y X Y X Y X Y

0

0

40mm

43,7mm 11,8mm 140,5mm 3,7mm 194,2mm 0 2,3mm

2 3 3 4 4 Edge round

13mm 130mm 4mm 200mm 0 1mm

The new geometry achieved an aerodynamic drag of 0.0617N. Which is a drop of 9.39% from the initial model. The following will show the differences between the input and output models. Comparing the input geometry to the output geometry, it can be seen in Figure 3 that the output geometry has a much slimmer profile and more rounding of the leading edge. Also in Table 2 above, it can be seen that node number two is positioned closer to the axis of symmetry and shifted more towards the trailing edge making the connecting element slimmer and therefore more aerodynamic. As for the pressure distribution in the symmetry axis, Figure 4a shows what effect the rounding of the leading edge has on the pressure generated in front of the connecting element. Analyzing the turbulent flows in Figure 4b, it can be seen that due to the more slender silhouette and the rounding of the leading edge, turbulence was significantly reduced at the nacelle-wing connecting surface.

Fig. 3. - Input to output geometry comparison (from top: input, output)

Fig. 4. (a) Comparison of pressure distribution on the element and on the plane of symmetry (from the top: initial, optimized); (b) comparison of visualization of turbulent flows on the coupling surface (from the top: initial, optimized)

Wojciech Skarka et al. / Procedia Structural Integrity 54 (2024) 498–505 Bartosz Rodak/ Structural Integrity Procedia 00 (2019) 000 – 000

502

5

5. Profile optimization by Gradient optimization method The CAD model, FEM mesh, boundary conditions, solver settings of the tested geometry is the same as in the case of optimization by the DoE method described in Chapter 4 [2]. The Adaptive Single-Objective method is a gradient-based algorithm [3], [4] that provides an improved global optimization result. It supports a single objective, multiple constraints and aims to find the global optimum. It is limited to continuous and producible input parameters. In practice, this means that the program generates twenty initial experiments, solves them, and then, based on the principles of gradient optimization, creates successive points according to the direction (ⅈ) and step ℎ (ⅈ) . A total of one hundred and forty experiments are set. An identical simulation was performed for each experiment. Table 3 below presents the dimensions giving the lowest aerodynamic drag found using gradient optimization.

Table 3. Coordinates of spline nodes, forming the optimal geometry.

Direction

Nominal dimension

After optimization dimension

Point

0 0 2

X Y X Y X Y X Y

0 0

0 0

40mm

43,6mm 11,8mm 125,5mm 4,4mm 186,7mm 0 1,2mm

2 3 3 4 4 Edge round

13mm 130mm 4mm 200mm 0 1mm

The new geometry achieved an aerodynamic drag of 0.06038N. That is, 2.88% less than the geometry obtained using DoE optimization. 6. Comparison of optimization methods Optimization using the DoE method reduced aerodynamic drag by 9.39%, while optimization using the gradient method reduced aerodynamic drag by 12.27%. The DoE method has the advantage of giving extensive information about what input parameter has what effect on the output parameter, so you can estimate how a change in the input parameter will affect the result. In Table 4 below, you can see what parameters the two methods have chosen. Figure 5 you can see the differences in how the two methods searched the range of input parameter values. The DoE method uniformly searched the entire range of parameters, while the gradient algorithm compacted the search closer to the optimum over time.

Table 4. Comparison of the coordinates of the spline nodes, created using the DoE and the gradient method.

Direction

After optimization dimension by DoE method

After optimization dimension by gradient method

Point

0 0 2 2 3 3

X Y X Y X Y

0 0

0 0

43,7mm 11,8mm 140,5mm 3,7mm

43,6mm 11,8mm 125,5mm 4,4mm

Wojciech Skarka et al. / Procedia Structural Integrity 54 (2024) 498–505 Bartosz Rodak/ Structural Integrity Procedia 00 (2019) 000 – 000

503

6

4 4 Edge round

X Y

194,2mm 0 2,4mm

186,7mm 0 1,2mm

7. Element height optimization A full aerodynamic simulation of the entire system containing the wing, the nacelle and the optimized connecting element was carried out. The input parameter was the distance between the wing chord and the geometric center of the nacelle. The output parameter is the aerodynamic drag generated by the entire structure. The wing model was created on a standardized NACA 633-618 profile. The wing span for the simulation is 1m. The model of the measuring nacelle was created on the basis of a symmetrical NACA EPPLER 863 STRUT 9 profile. And the model of the connecting element is an optimized version of the connecting element from the previous chapter. The entire structure is shown in Figure 6. solver settings are identical to those for the simulation when optimizing the shape of the connecting element. The optimization was performed using the design of experiments method due to the fact that this method gives more information about the course of the optimization process.

Fig. 5. - The way to search for the optimal solution (top: DoE, bottom: Gradient)

The optimization process proved that, the highest drag is provided by a design with the nacelle very close to the wing. With distances of 1mm to 7mm between the surfaces of the nacelle and the wing, the aerodynamic drag was about 3.7N. A clear decrease in the generated aerodynamic drag can be observed at a distance of 8mm between the surfaces of the nacelle and the wing, that is, when the length of the connecting element counting from the wing chord to the structural plane of the nacelle was 75mm. The lowest resistance was shown by

Fig. 6. - Model used in optimizing the height of the connecting element

the geometry with the length of the connecting element from about 90mm, while the minimum resistance of 2.96N was obtained with the length of the connecting element of 100mm. This corresponds to distances between the nacelle surface and the wing surface of 23mm and 33mm, respectively. Below is presented a graph (Figure 7) showing how the length of the connecting element affects the generated aerodynamic drag

Fig. 7. - Effect of the height of the connecting element on aerodynamic drag.

Wojciech Skarka et al. / Procedia Structural Integrity 54 (2024) 498–505 Bartosz Rodak/ Structural Integrity Procedia 00 (2019) 000 – 000

504

7

By analyzing the distribution of pressures on the surfaces of the elements and in the plane of symmetry as in Figure 8, the supposition of a low-pressure system generated behind the nacelle was confirmed. The low air pressure in the area behind the nacelle effectively increases the aerodynamic drag generated by the structure. Looking at the airflow comparison in Figure 9, it was noted that when the nacelle is close to the wing significant air turbulence and turbulence behind the nacelle is generated.

Fig. 8. - comparison of pressure distribution on the elements and in the plane of symmetry. (from top: optimal, suboptimal)

Fig. 9. - Comparison of visualization of air streams colored depending on their speed (from the top: optimal, suboptimal)

This leads to the conclusion that in order not to create air turbulence behind the object, the nacelle should be moved away from the wing in order to separate the air currents.

8. Summary

• Optimization by the DoE method is relatively simple to perform and provides an answer to the question of what input parameters have the greatest impact on the outcome of the experiment. • Optimization of the shape by the DoE method resulted in a reduction of aerodynamic drag by 9.39% relative to the input geometry.The geometry due to optimization was significantly slenderized, which reduced the generated aerodynamic drag.The biggest impact on the reduction of drag is the parameter describing the position of the second node in the Y axis. Thanks to this parameter, the element gains slenderness. Increasing the slenderness also had a positive effect on turbulent flows and air turbulence • Optimization of the height of the connecting element revealed that seating the nacelle as close to the wing as possible does not yield the best results. As the length of the connecting element increases, the drag value decreases to 2.96N at 100 mm of connecting element length, which corresponds to a distance of 33 mm between the surface of the nacelle and the surface of the wing. Further lengthening of the connecting element does not bring a decrease in aerodynamic drag. This is due to the generated turbulence, turbulence and backflow behind the nacelle at small distances between the wing and the nacelle. By the above-mentioned factors, at small distances between the nacelle and the wing, there is a low pressure area behind the nacelle. • The gradient optimization method resulted in a 12.27% reduction in generated aerodynamic drag relative to the input geometry. the gradient algorithm slimmed the profile of the connecting element to an even greater extent than the DoE method. Which resulted in an additional 2.88% decrease in aerodynamic drag relative to DoE.