Toward identifying efficient sound absorbers, we have formulated and analyzed the micro-perforated panels (MPPs) comprising cylindrical pores of arbitrary axial profiles for out-of-plane wave propagation. First, the forward problem was solved wherein an exact analytical expression for the absorption coefficient α was developed for these MPPs. The dependence of α was analyzed for various profiles comprising convex and concave-shaped sections including divergent and convergent linear and sinusoidal profiles, conic sections, Kilroy, and chirp shapes. The MPPs with pores having a diverging shape at the entrance were found to have higher sound absorption. The non-symmetric fluctuations in the profile led to fluctuations in the spectrum of α. Aiming to address the long-standing challenge of low-frequency sound absorption, we solved the inverse problem to identify the axial profile of the cylindrical pore for maximum sound absorption for frequencies up to 2500 Hz. Given the advances in additive manufacturing, the results of this comprehensive work help in designing MPPs comprising complex-shaped pores suitable for a particular spectral regime.

The absorption of sound has been of interest to the acoustic (sonic) and ultrasonic communities for attenuation of sound waves in fluids and solids.1 The traditional way to absorb sound is by the use of porous viscoelastic materials2 and the recent focus emphasizes biodegradability in these materials.3 Over the past three decades, structured materials were developed to control the propagation of sound and elastic waves in solids;4,5 these materials were referred to as acoustic metamaterials (AMs). Phononic crystals of metals6 and polymers,7 micro-perforated panels,8 topological insulators,9 etc., at various length scales are some examples of acoustic metamaterials, which have had applications as waveguides,10 isolators,11 sensors,12 and controlling the flow of sound.13 The characteristic size of AMs often dictates the spectral regime of where the “meta” effect is observed. For example, the central wavelength of the bandgap of a phononic crystal is approximately the size of the period of the crystal. The spectral regime of special characteristics is often a narrow band. Micro-perforated panels (MPPs) are the only type of AMs, similar to the traditional acoustic materials,14 where (a) the attenuation of the sound is over a broad spectral regime15 and (b) the mechanism is by sound absorption. Furthermore, the micrometer-sized perforations lead to absorption in the audible regime. The effectiveness of MPPs is measured in terms of the sound absorption coefficient α, which represents the fraction of the sound absorbed.

MPPs are thin structured composite panels comprising one panel perforated with a large number of sub-millimeter-sized pores, followed by an air cavity, and then a rigid panel.16 MPPs are used for sound absorption in the sonic regime. The development of the understanding of MPPs began in the year 1975 with the work by Maa17 and subsequent developments by him.18,19 The full thickness of an MPP lies between 50 and 100 mm, with the thickness of the perforated panel typically ranging from 0.5 to 3 mm, the pore diameter typically in the range 50–500  μm, and the air cavity of at least 50 mm for better performance. The shapes of the pores are often upright cylinders,19 and in rare cases, slanted20 or tapered cylindrical.20,21

Additive manufacturing technologies have aided in the rapid prototyping of AMs,22 including MPPs comprising pores of tapered21 and petal-shaped23 for experimental verification. Furthermore, MPPs comprising multiple walls,24 honeycomb backing,25,26 and coiled-up channels backing27,28 have also been developed due to the advancements in manufacturing as well as the high efficiency in sound absorption over a broadband spectral regime.

Wave propagation through MPPs is often analyzed in terms of the equivalent electrical circuit rather than solving the boundary value problem.16 However, these circuit-based analyses cannot be extended when the profile of the pore is complex. Furthermore, in a large part of the literature on MPPs, the analysis is made for normally impinged waves. Efforts for waves with wavevector lying in the plane of the panel have only begun recently.29 In this work, we focused our efforts on the propagation of waves with wavevectors perpendicular to the plane of the MPP. We refer to this as out-of-plane wave propagation.

In a project dedicated to the development of sound-absorbing materials, we explored two types of acoustic metamaterials: locally resonant materials30 and MPPs, in addition to biodegradable sound-absorbing materials.31 In the work focused on MPPs, we explored the study of waves propagating in-plane and out-of-plane. The study was also extended for MPPs comprising multiple shells in each pore for both kinds of wave propagation. In this work, we present the analysis for wavevectors perpendicular to the plane of MPP, i.e., out-of-plane wave propagation. In existing MPPs, the shape of the pore is upright cylindrical. In a more advanced MPP, the pore is replaced by a multi-shelled-pore.24 As additive manufacturing techniques can rapidly prototype complex shapes,32,33 first, we chose to investigate MPPs comprising symmetric pores, but with the radius of the profile varying arbitrarily with depth. Starting from the first principles, exact analytical expressions were determined for the generic profile of radius, and various easily fabricated cases were analyzed. Numerical cases of standard concave and convex geometries of pores including conic sections, Kilroy, and chirp shapes were analyzed to establish the effect of shape on the absorption spectrum.

The need to optimize MPPs arose due to the long-standing challenge in attenuating the sound for frequencies between 20 and 2000 Hz of the audible regime.34,35 This noise control or minimization is crucial not only for indoor environments such as offices and houses but also for engineering applications such as ground and aerial vehicles, and industrial machinery.36 Hence, in tandem with the developments of the forward analysis, where the architecture of the MPP or AM is altered, efforts were also made to maximize the desired outcome. Some examples are maximization of the bandgap in phononic crystals by varying the geometry by topology optimization37 or genetic algorithms,38 increasing the sound absorption of Helmholtz resonator-based absorbers using deep neural network efficiency,39 or more recently by the use of machine learning models for AMs.40 While inverse problems are solved by a variety of numerical methods,41 the traditional method involves the use of heuristic,42 non-heuristic optimization43 algorithms, regularization methods,44 and the machine and deep learning models.40,45 Among the many methods, traditional optimization methods fail to achieve the desired output due to the complex nature of the equations. Genetic algorithms (GAs) operate by simulating the process of natural selection, where candidate solutions undergo successive generations of reproduction, mutation, and selection to evolve toward optimal or near-optimal solutions to a given problem. In essence, GAs are particularly well-suited for exploring large solution spaces, especially in cases where the objective function is complex or difficult to evaluate analytically. A survey of literature also established the success of using GAs for various AMs such as phononic crystals,38 locally resonant materials,46 and MPPs.47 Hence, we attempted to maximize the sound absorption over a broad spectral regime by GA, rather than aiming for a maximum over a local spectral regime. In the next part of the work, we solved the inverse problem for out-of-plane propagation in an MPP and identified the axial profile of the symmetric pore to have maximum broadband soundabsorption for frequencies below 2500 Hz.

The plan for the remaining paper is as follows: Starting with the theoretical preliminaries, we formulate and determine the velocity fields for a pore of the arbitrary axial profile in Sec. II A. The impedance and absorption coefficient are determined in Sec. II B. The results of various numerical cases of the forward model are discussed in Sec. III. The inverse problem to identify the shape of the pore for maximum absorption is detailed in Sec. IV. We end with the conclusions in Sec. V.

The schematic of a typical MPP is shown in Fig. 1(a). The pore diameter is d, the thickness of the panel is τ, and the spacing between the panel and wall is D. These geometric parameters { d , τ , D } represented in the figure can be tuned to achieve the desired bandwidth of absorption. As the MPP is a parallel arrangement of a large number of small pores, the following assumptions are made in line with the literature: (a) The distance between the pores is greater than their diameter so as to neglect the effect of a pore on another and (b) the distance between pores must be small compared to the wavelength of sound, to neglect the reflection of sound by the solid part of the panel.

FIG. 1.

Schematic of (a) MPP; (b) pore of an MPP, with R ( x ) as the radius of cross section; (c) a ring element of radius r ( x ) < R ( x ) of a pore, with dr as the radial thickness and dx as the axial thickness; and (d) free body diagram of the ring element. (e) Schematic of air cavity of length D, with x = D representing the rigid back wall of MPP.

FIG. 1.

Schematic of (a) MPP; (b) pore of an MPP, with R ( x ) as the radius of cross section; (c) a ring element of radius r ( x ) < R ( x ) of a pore, with dr as the radial thickness and dx as the axial thickness; and (d) free body diagram of the ring element. (e) Schematic of air cavity of length D, with x = D representing the rigid back wall of MPP.

Close modal
The schematic of a single pore of varying cross section is shown in Fig. 1(b). The axis of symmetry is along x ^. As is the case for modeling MPPs, Ref. 48, p. 229, the air in the pore is considered a collection of thin shells, each sliding axially based on the viscosity of air and pressure gradient. Accordingly, we consider an annular ring-type element of radius r ( x ) and thickness d r, as shown in Fig. 1(c). The free body diagram of this element is shown in Fig. 1(d). The axial driving force due to negative pressure gradient is
d f driving = ( 2 π r d r ) ψ ( x ) d x ,
(1)
where ψ ( x ) is the pressure gradient. The opposing frictional force on the inner surface f r and outer surfaces f r + d r of the ring are
f r = 2 π r d x μ u r and f r + d r = f r + f r r d r ,
(2)
where μ is the coefficient of viscosity of air and u is the particle speed. The inertial force on the element is
d f intertia = i ω ρ 0 u ( 2 π r d r ) d x ,
(3)
where ρ 0 is the density of air and ω is the angular frequency. Per Newton’s second law of motion,
d f driving ( f r + d r f r ) = d f inertia .
(4)
Substituting Eqs. (1) to (3) in Eq. (4), we get17–19 
( 2 r 2 + 1 r r i ω ρ 0 μ ) u = ψ ( x ) μ .
(5)
The solution of this non-homogeneous partial differential equation has two parts. For the homogeneous part,
( 2 r 2 + 1 r r i ω ρ 0 μ ) u h = 0 ,
(6)
using the Ansatz u h = j = 0 a j r j + n and equating coefficients of the polynomial to zero gives a 1 = 0 and
j 2 a j i ω ρ 0 μ a j 2 = 0 a j = a j 2 i ω ρ 0 μ j 2 .
(7)
Therefore,
u h = j = 0 ( 1 ) j a 0 ( k x i ) 2 j 2 2 j ( j ! ) 2 = a 0 J 0 ( k x i ) ,
(8)
where k x = r ( x ) ω ρ 0 / μ can be termed as the depth-dependant perforation constant and J q ( x ) is the Bessel function of the first kind and qth order. The particular solution u p is obtained by setting all the partial derivatives in Eq. (5) to zero, i.e.,
( i ω ρ 0 μ ) u 2 = ψ ( x ) μ u 2 = ψ ( x ) i ω ρ 0 .
(9)
Therefore, the final solution is
u ( r , x ) = u h + u 2 = a 0 J 0 ( k x i ) + ψ ( x ) i ω ρ 0 .
(10)
The average speed u ¯ x at the circular cross section of radius r ( x ) is
u ¯ x = ψ ( x ) i ω ρ 0 ( 1 2 k x i J 1 ( k x i ) J 0 ( k x i ) ) .
(11)
The volume speed of the element is
U ¯ x = u ¯ x A x = u ¯ x π r 2 ( x ) ,
(12)
where A x is the cross-sectional area of the disk element located at depth x.
The acoustic impedance of the circular disk element of thickness d x and radius r ( x ) is
d Z x = ψ ( x ) d x U ¯ x .
(13)
Substituting Eqs. (11) and (12) in Eq. (13), and integrating gives the impedance for the entire pore as
Z pore = i ω ρ 0 0 τ 1 A x [ 1 2 k x i J 1 ( k x i ) J 0 ( k x i ) ] 1 d x .
(14)
Therefore, the specific acoustic impedance of the pore is
z pore = i ω ρ 0 0 τ ( r 1 r x ) 2 [ 1 2 k x i J 1 ( k x i ) J 0 ( k x i ) ] 1 d x ,
(15)
where r 1 is the radius of upper part of pore, i.e., at x = 0. As the pores of MPP are in a parallel connection for which the pressure difference (analogous to potential) is the same across each perforation, the combined acoustic impedance of the pores is Z pores = Z pore / N, where N is the number of pores. As Z = z / S, with S as the cross-sectional area, specific acoustic impedance is
z pores = z pore ν , with
(16)
ν = Sum of cross{-}section area of all pores {Area of the rectangular wall}
(17)
being the fraction of perforation.
Each cavity beneath the pore can be considered as a one-dimensional medium of air driven by a harmonic plane wave source at x = 0 and terminated by rigid boundary at x = D, as shown in Fig. 1(e). The pressure inside the cavity can be written as
p = A e i ( ω t + κ ( D x ) ) + B e i ( ω t κ ( D x ) ) ,
(18)
where κ = ω / c 0 is the wavenumber. Considering rigid boundary conditions at x = D, i.e., u = 0 or p / x = 0, we have
p = A e i ω t [ e i κ ( D x ) + e i κ ( D x ) ] = 2 A cos [ κ ( D x ) ] e i ω t .
Using Euler’s equation ρ 0 u / t = p / x,
u = 2 A κ i ρ 0 ω sin [ κ ( D x ) ] e i ω t .
(19)
Therefore, the acoustic impedance of cavity is
z cavity = ( p u ) x = 0 = [ i ρ 0 ω κ cot ( κ D κ x ) ] x = 0 .
On applying limits, acoustic impedance becomes
z cavity = i ρ 0 c 0 cot ( κ D ) .
The specific acoustic impedance of MPP including the pore and air cavity is
z MPP = z pores + z cavity .
(20)
The reflection coefficient r is
r = z MPP ρ 0 c 0 z MPP + ρ 0 c 0 .
(21)
As the back end of the MPP is considered rigid, the absorption coefficient α is conveniently evaluated as
α = 1 | r | 2 .
(22)
Combining all expressions, we have the absorption coefficient as
α = 1 | z MPP ρ 0 c 0 z MPP + ρ 0 c 0 | 2 = 4 Re ( Z MPP ) [ 1 + Re ( Z MPP ) ] 2 + [ Im ( Z MPP ) ] 2 ,
(23)
where Z MPP = z MPP / ρ 0 c 0 is the total relative acoustic impedance. The end corrections of MPPs of arbitrary shapes pores remain the same as that for the MPP comprising cylindrical pores. For completeness, we have provided them in the  Appendix.

We first validated the theory developed for MPPs of arbitrarily shaped profiles by reconstructing the absorption coefficient spectrums for two cases from the literature: (a) MPPs comprising upright cylindrical pores of different thicknesses and pore diameters20 and (b) MPPs comprising tapered pores with 0.8 mm diameter on the exposed side and 0.5 mm diameter on the other side.20 In all the profiles of the pores presented, a constraint 0.1537 < r < 0.3074 mm for the radius of the pore was enforced, based on the literature.48 All computations were implemented in MATLAB.49 

In exploring alternative geometries that may offer unique acoustic characteristics, we considered pore profiles of conic sections such as parabolas and ellipses along with circular ones. The profiles are shown in the inset of Fig. 2(a). The lower and upper limits of 0.15 and 0.3 mm radii for the profiles were set. The circular profile was constructed by pulling apart two identical semi-circular sections at a certain distance. The ellipse-based profiles are formed by carving out the middle portion of a full ellipse. Unlike circular profiles, joining the two sections of the ellipse-based profiles would not result in a complete ellipse as the middle portion of the full ellipse was carved out. Hence, we see a good overlap in the circular and elliptical pore profiles.

FIG. 2.

Sound absorption coefficients of MPPs comprising (a) conic sections and (b) circular sections as pore profiles. All of these profiles were within the limits of minimum and maximum allowed radius as 0.1537 and 0.3074 mm, respectively.

FIG. 2.

Sound absorption coefficients of MPPs comprising (a) conic sections and (b) circular sections as pore profiles. All of these profiles were within the limits of minimum and maximum allowed radius as 0.1537 and 0.3074 mm, respectively.

Close modal

A comparison of sound absorption for MPPs with pores of different geometric shapes circle, parabola, ellipse, and hyperbola is shown in Fig. 2(a). We note that for a pore with fixed minimum and maximum radial distance, the value of sound absorption decreases for pores of ellipse, circle, parabola, and then hyperbola, especially in the low-frequency regime. We further considered MPPs comprising circular pores, but the pore separation is more in one case. The schematics for these pore profiles along with the corresponding absorption coefficients are shown in Fig. 2(b). For comparison, the spectrum of α of an MPP comprising cylindrical pores is also overlaid. It can be seen that the absorption is better for circle 1 profile, then circle 2 and the least for the cylinder. As cavities enhance resonance and mixing, the absorption is higher for the pore with circle 1 profile than that for the pore with circle 2 profile, despite the latter having a part of the profile lower than the cylinder.

The sound absorption coefficients of MPPs comprising two units of convergent-divergent (CD) and divergent-convergent (DC) pores are shown in Fig. 3(a). The results of an MPP with cylindrical pores of equivalent radius are also superposed for comparison. Even with an increase in the number of cycles, we find that the α spectrum of MPPs with cylindrical pores lies between that of MPPs comprising DC or CD pore profiles. Further comparing with Fig. 4, we find that α is independent of the number of repetitions of the profile. The conclusions differ slightly for the MPPs comprising pores of CD- and DC-styled sinusoidal profiles, shown in Fig. 3(b). We find that neither the phase of the profile nor the number of repetitions of the CD- or DC-styled sinusoidal profiles altered the spectrum of the absorption coefficient. The difference between α’s of sinusoidal CD and DC pores from that of the corresponding triangular pores can be attributed to the smooth transition between the converging and diverging parts of the pores in the former.

FIG. 3.

Sound absorption graph of MPPs with (a) convergent–divergent (CD) and divergent–convergent (DC) pores of two cycles and (b) sinusoidal pores of two cycles with a mean radius of 0.225  mm. In (a) and (b), the result of an MPP with cylindrical pores of radius 0.225 mm is overlaid.

FIG. 3.

Sound absorption graph of MPPs with (a) convergent–divergent (CD) and divergent–convergent (DC) pores of two cycles and (b) sinusoidal pores of two cycles with a mean radius of 0.225  mm. In (a) and (b), the result of an MPP with cylindrical pores of radius 0.225 mm is overlaid.

Close modal
FIG. 4.

Sound absorption coefficients of MPP comprising one unit of (a) and (b) divergent–convergent (DC), and (c) and (d) convergent–divergent (CD) pores. In (a) and (d), the minimum radial distance is 0.1537 mm with varying maximum radii. In (c) and (d), the maximum radial distance is 0.3074  mm with varying minimum radii.

FIG. 4.

Sound absorption coefficients of MPP comprising one unit of (a) and (b) divergent–convergent (DC), and (c) and (d) convergent–divergent (CD) pores. In (a) and (d), the minimum radial distance is 0.1537 mm with varying maximum radii. In (c) and (d), the maximum radial distance is 0.3074  mm with varying minimum radii.

Close modal

Sound absorption coefficients of MPPs with DC of cycles n = 1 with either a fixed minimum or maximum radius are shown in Figs. 4(a) and 4(b), respectively. The same for n = 1 cycles MPPs of CD pores are shown in Figs. 4(c) and 4(d). We note that for DC pores with fixed minimum radius, shown in Fig. 4(a), an increase in maximum radius led to an increase in α and an increase in the peak-frequency of α. For DC pores with a fixed maximum radius, shown in Fig. 4(b), an increase in minimum radius led to a decrease in α as well as the frequency at which peak in α was observed. The increase in absorption can be attributed to the expansion of air due to the widening of the pore, leading to more relaxation times or equivalently more absorption. For CD pores with a fixed maximum radius, shown in Fig. 4(c), an increase in minimum radius led to increase in α and an increase in the peak-frequency of α. For CD pores with fixed minimum radius, shown in Fig. 4(d), an increase in maximum radius led to a decrease in α and decrease in the frequency at which peak in α was observed. As the mean radius increases for CD pores, the pore becomes more inclined, and hence a part of the incoming wave gets reflected and goes out, thereby decreasing the absorption characteristics. The values of α increase with the mean radius as the pore becomes widened and incoming waves have a larger part to enter. From these figures shown in Fig. 4, we see that the absorption characteristics of MPPs with cylindrical pores lie midway in between that of MPPs comprising DC or CD pore profiles. Furthermore, we conclude that an increase in the curvature of the pore either (concave or convex) enhances sound absorption.

Kilroy was a meme that became popular during World War II. We designed four types of curves based on the Kilroy profile. The first curve was designed using the equation y = log ( | sin ( k 1 x ) / k 2 x | ) , where k 1 and k 2 are constants. Considering k 1 = 10 4 and k 2 = 10 3, curve 1 was obtained and further scaled to fit the radius of the pore in the range of 0.15–0.3 mm. By flipping curve 1, curve 2 was obtained. By appending curve 1 with curve 2 and fitting into the same range, curve 3 was obtained. By appending curve 2 with curve 1, curve 4 was obtained. The schematics for these four variations of Kilroy-curve based pores are shown in Fig. 5(a) and Fig. 5(b) along with their sound absorption spectrums of the corresponding MPPs. As the pressure wave propagates within the pore of curve 1 or 3, shown in Fig. 5(a) or (b), despite a slightly converging entry, the increasing number of cavities helps in having larger values of α than that of the profile with curve 2 or 4, respectively. For all four variants of the Kilroy profiles, we find that the broad spectrum of spatial frequencies has induced a broad fluctuation in the spectrums of α. The spatial fluctuations being more for the profiles of curves 3 and 4 as compared to that from pores made of curves 1 and 2, the same was also reflected in the corresponding spectrums of α. The spatial frequencies of curves 1 and 2 are identical and hence, the corresponding frequencies at which fluctuations occur in the αs of these curves are also identical. The same conclusion holds for the cases discussed in Fig. 5(b). The spatial frequency of the sinusoidal pores did not induce any fluctuations in the spectrum of α due to the symmetry of the profile. Furthermore, we found that the spatial fluctuation in the pore profiles also locally altered the frequency at which a maximum exists in α.

FIG. 5.

Sound absorption coefficients of MPPs comprising pores of (a) and (b) Kilroy profiles and (c) and (d) chirp profiles.

FIG. 5.

Sound absorption coefficients of MPPs comprising pores of (a) and (b) Kilroy profiles and (c) and (d) chirp profiles.

Close modal

A cosine-based linear chirp profile was generated using the function y = chirp ( t , f 0 , t 1 , f 1 ) in MATLAB. The instantaneous frequency at time t = 0 is f 0 and the instantaneous frequency at time t 1 is f 1. We set f 0 = 0, t 1 = 0.005, and f 1 = 1000 and obtained four different variations of the chirp curve, similar to that of the Kilroy pore. The schematics are shown in Figs. 5(c) and 5(d) along with their sound absorption spectrums. The results and reasons remain the same as that of the Kilroy profiles. Despite the converging profile (chirp curve 1), shown in Fig. 5(c), the α of chirp curve 1 is marginally higher than that of chirp curve 2 due to the increase in the number of cavities. The same reason can be attributed to the α of chirp curve 3 as compared to that of chirp curve 4, Fig. 5(d). For all the spectrums shown in Figs. 5(c) and 5(d), as the fluctuations are more closely spaced along the length, very few fluctuations are seen in α of chirp curve 1 and chirp curve 3 as compared to that of chirp curve 2 and chirp curve 4.

We have numerically estimated the profile of the pore for out-of-plane propagation in MPPs so as to maximize the spectrum of the absorption coefficient. The discretized profile of the symmetric pore was obtained by maximization of the absorption coefficient. The following procedure was used: Initial spectrums of α ( ω ) were generated using profiles considered in the forward problem for J = 2000 discrete frequencies. The physical properties of air { ρ 0 , μ } and geometric parameters { τ , D } were considered constant and grouped together as V = { ρ 0 , μ , τ , D }. The discretized set of radii U = [ r 1 , r 2 , , r J ] along the length of the axisymmetric pore are to be determined for maximum absorption. The function E used to maximize absorption α ( ω ) is
E ( ω , U ) = | 1 α anly ( ω ; U ; V ) | 2 ,
(24)
where α anly is the absorption coefficient estimated from the forward model and | | denotes the absolute value of the argument. The constraint on radii was enforced as U [ 0.1537 , 0.3074 ] mm. The error functional
F = p = 1 J E ( ω j ; U )
(25)
was minimized to get optimum values of U. The gradient-based optimization methods were found to give unsuitable values for U. Hence, we used the genetic algorithm function “ga” in MATLAB. The hybrid genetic algorithm comprising the “ga” and the standard constraint optimization algorithm “fmincon” and “sequential quadratic programming” was implemented. The maximum number of iterations were set as 10 4, and the tolerances on the variables U and the functional F as 10 10. The spectrums of α generated from the forward model for different profiles of pores were fed as the initial population.

The obtained optimized pore profile (in red) constructed from U is shown in Fig. 6(a). Further, the reverse of the optimized profile (in black) along with the minimum and maximum radius are also compared. Since the symmetry of the pore was not imposed along x ^, we intend to compare the corresponding the spectrum of α for the optimized profile along with that of the reversed optimum profile. The reverse of the optimized profile (in black) along with the minimum and maximum radius are also compared. The corresponding sound absorption coefficients for these four pore profiles are shown in Fig. 6(b). It is clear that the obtained spectrum has maximum absorption not just around the a local peak, but all through the spectrum. The spectrum of the reversed profile has a spectrum intermediate to that of minimum and maximum profiles. This observation is in line with our reasoning from forward models, that a diverging entry to the pore leads to more absorption.

FIG. 6.

(a) Pores with optimized profile, minimum radius, maximum radius, and reverse of optimized profiles and, (b) the corresponding absorption coefficients.

FIG. 6.

(a) Pores with optimized profile, minimum radius, maximum radius, and reverse of optimized profiles and, (b) the corresponding absorption coefficients.

Close modal

While the initial calculations indeed indicated the effectiveness of circular and elliptical cross sections in achieving higher absorption coefficients, it is important to note that the inversion process considers the optimization of the entire pore profile, rather than individual sections in isolation. The inclusion of a linear variation in the profile likely serves to introduce additional acoustic impedance variation, thereby enhancing sound absorption properties within the desired frequency range. Furthermore, we note that the GA algorithm has not led to an arc-shaped divergent section for the following reasons: (a) the GA in pursuit of the global optimum may prioritize certain configurations that offer a more balanced trade-off between complexity and performance and (b) the optimization process may have converged toward a solution that exhibits greater efficacy in achieving the desired acoustic objective of maximum overall absorption coefficient across a broad spectral range.

The main advancement of this work is the establishment of analytical expressions for absorption coefficient α of microperforated panels (MPPs) comprising cylindrical pores of arbitrary axial profiles. After developing the generalized theory, we identified α spectrum of MPPs comprising pores of conic sections, divergent and convergent pores, sinusoidal pores, Kilroy pores, and chirp pores. Given the progress of additive manufacturing techniques, the second advancement from this work is on identifying the optimized axial profile of the pore for maximum sound absorption. The novelty of the work is in developing the theory from first principles. The immediate extension to this work is to determine the absorption coefficient of MPPs comprising pores of arbitrary non-symmetric three-dimensional pores. Such analysis is of use to understand MPPs comprising slanted and zig-zag-shaped cylindrical pores. The forward model could also be used to design MPPs with a desired peak in the spectrum of α.

This work was performed with partial funds through funds from Initiation Grant (No. IITK/ME/2017445) from the Indian Institute of Technology Kanpur and SERB Early Career Research Award (No. ECR/2018/02341) to C.C.

The authors have no conflicts to disclose.

O. V. Vigneswar: Data curation (equal); Formal analysis (equal); Investigation (equal); Methodology (supporting); Software (equal); Validation (lead); Visualization (equal); Writing – original draft (supporting); Writing – review & editing (supporting). C. Chandraprakash: Conceptualization (equal); Data curation (equal); Formal analysis (equal); Funding acquisition (lead); Investigation (equal); Methodology (lead); Project administration (lead); Resources (lead); Software (equal); Supervision (lead); Validation (supporting); Visualization (supporting); Writing – original draft (lead); Writing – review & editing (lead).

The data that support the findings of this study are available within the article.

For a cylindrical pore of radius r 0 = d / 2 the boundary condition is u = 0 at r = r 0 . Applying this condition to Eq. (10) results in
u = ψ i ω ρ 0 ( 1 J 0 ( k i ) J 0 ( k r 0 i ) ) ,
(A1)
where k r 0 = r 0 ω ρ 0 / μ is the perforation constant for the surface. The average speed u ¯ is
u ¯ = 2 r 0 2 0 r 0 u r d r .
(A2)
Using Eqs. (A1) in (A2) and the identity 0 x 0 J 0 ( x ) x d x = J 1 ( x 0 ) x 0 , we get
u ¯ = ψ i ω ρ 0 ( 1 2 k r 0 i J 1 ( k r 0 i ) J 0 ( k r 0 i ) ) .
(A3)
The specific acoustic impedance of pore z pore is
z pore = Δ p / u ¯ ,
(A4)
where Δ p is the sound pressure difference between the ends (positive). Substituting Eq. (A3) in Eq. (A4) and also using the relation ψ = Δ p / τ,
z pore = i ω ρ 0 τ [ 1 2 k r 0 i J 1 ( k r 0 i ) J 0 ( k r 0 i ) ] 1 .
(A5)
Depending upon the values of k r 0, the equation for impedance changes.17–19,
  • k r 0 < 1
    z pore = 4 3 i ω ρ 0 τ + 32 μ τ d 2 .
  • k r 0 > 10
    z pore = i ω ρ 0 τ + 4 μ τ d ω ρ 0 2 μ ( 1 + i ) .
  • 1 < k r 0 < 10 (in this work)
    z pore = 32 μ τ d 2 ( 1 + k r 0 2 32 ) 1 / 2 + i ω ρ 0 τ ( 1 + ( 9 + k r 0 2 2 ) 1 / 2 ) .
As the dimensions of the pores are of the same order as the thickness boundary layer, viscous and thermal effects were accounted for. As the air flow enters the perforation, due to the vibration of air molecules an additional viscous effect is also produced. This viscosity effect was considered in the real part of impedance.19,50 The apparent increase in panel thickness is caused by the radiation of a cylindrical mass of air via the opening. This increase in thickness is accounted by using a correction term in the imaginary part of impedance.50 Taking into account these end corrections, the final expression for impedance is
z pore = 32 μ τ d 2 ( ( 1 + k r 0 2 32 ) 1 / 2 + k r 0 d τ 2 32 ) + i ω ρ 0 τ ( 1 + ( 9 + k r 0 2 2 ) 1 / 2 + 0.85 d τ ) .
(A6)
1.
A. B.
Bhatia
,
Ultrasonic Absorption: an Introduction to the Theory of Sound Absorption and Dispersion in Gases, Liquids, and Solids
(
Courier Corporation
,
1985
).
2.
J.
Allard
and
N.
Atalla
,
Propagation of Sound in Porous Media: Modelling Sound Absorbing Materials
(
John Wiley & Sons
,
2009
).
3.
T.
Bhuvana
,
A.
Tiwari
, and
C.
Chandraprakash
, “
Green fabrication of cellulose-rich agricultural residues for scalable and biodegradable acoustic boards
,”
Indus. Crops Products
204
,
117404
(
2023
).
4.
J.
Li
and
C. T.
Chan
, “
Double-negative acoustic metamaterial
,”
Phys. Rev. E.
70
,
055602
(
2004
).
5.
C.
Chindam
,
A.
Lakhtakia
,
O. O.
Awadelkarim
, and
W.
Orfali
, “
Acoustic scattering from microfibers of parylene C
,”
J. Appl. Phys.
116
,
134905
(
2014
).
6.
V.
Sharma
and
C.
Chandraprakash
, “
Quasi-superhydrophobic microscale two-dimensional phononic crystals of stainless steel 304
,”
J. Appl. Phys.
131
,
184901
(
2022
).
7.
C.
Chindam
,
A.
Lakhtakia
, and
O. O.
Awadelkarim
, “
Parylene-c microfibrous thin films as phononic crystals
,”
J. Micromech. Microeng.
27
,
075012
(
2017
).
8.
N.
Gao
,
Z.
Dong
,
H. Y.
Mak
, and
P.
Sheng
, “
Manipulation of low-frequency sound with a tunable active metamaterial panel
,”
Phys. Rev. Appl.
17
,
044037
(
2022
).
9.
F.
Xi
,
Y.
Tang
, and
L.
Hu
, “
Phase-change in topological chiral phononic crystal for directional coupling switch
,”
J. Appl. Phys.
134
,
155106
(
2023
).
10.
V.
Sharma
and
C.
Chandraprakash
, “
Fabrication and bandgaps of microscale metallic phononic crystals
,”
Inter. J. Adv. Eng. Sci. Appl. Math.
15
,
159
166
(
2023
).
11.
J.-S.
Chen
,
T.-Y.
Chen
, and
Y.-C.
Chang
, “
Metastructures with double-spiral resonators for low-frequency flexural wave attenuation
,”
J. Appl. Phys.
130
,
014901
(
2021
).
12.
A.
Gueddida
,
Y.
Pennec
,
V.
Zhang
,
F.
Lucklum
,
M.
Vellekoop
,
N.
Mukhin
,
R.
Lucklum
,
B.
Bonello
, and
B.
Djafari Rouhani
, “
Tubular phononic crystal sensor
,”
J. Appl. Phys.
130
,
105103
(
2021
).
13.
M.
Nakayama
,
T.
Matsuoka
,
Y.
Saito
,
N.
Uchida
,
K.
Inoue
,
H.
Mitani
,
S.
Akasaka
, and
S.
Koga
, “
A practically designed acoustic metamaterial sheet with two-dimensional connection of local resonators for sound insulation applications
,”
J. Appl. Phys.
129
,
105106
(
2021
).
14.
S.
Kumar
,
K.
Jahan
,
A.
Verma
,
M.
Agarwal
, and
C.
Chandraprakash
, “
Agar-based composite films as effective biodegradable sound absorbers
,”
ACS Sustain. Chem. Eng.
10
,
8242
8253
(
2022
).
15.
L.
Shen
,
Y.
Zhu
,
F.
Mao
,
S.
Gao
,
Z.
Su
,
Z.
Luo
,
H.
Zhang
, and
B.
Assouar
, “
Broadband low-frequency acoustic metamuffler
,”
Phys. Rev. Appl.
16
,
064057
(
2021
).
16.
L. L.
Beranek
and
T.
Mellow
,
Acoustics: Sound Fields and Transducers
(
Academic Press
,
2012
).
17.
D. Y.
Maa
, Theory and design of microperforated panel sound-Absorbing Constructions,
Scientia Sinica
18
(1),
55
71
(
1975
).
18.
D. Y.
Maa
, “
Microperforated-panel wideband absorbers
,”
Noise Control Eng. J.
29
,
77
84
(
1987
).
19.
D.-Y.
Maa
, “
Potential of microperforated panel absorber
,”
J. Acoust. Soc. Am.
104
,
2861
2866
(
1998
).
20.
K.
Sakagami
,
M.
Morimoto
,
M.
Yairi
, and
A.
Minemura
, “
A pilot study on improving the absorptivity of a thick microperforated panel absorber
,”
Appl. Acoust.
69
,
179
182
(
2008
).
21.
C.
Jiang
,
X.
Li
,
W.
Xiao
, and
B.
Zhang
, “
Acoustic characteristics of microperforated plate with variable cross-sectional holes
,”
J. Acoust. Soc. Am.
150
,
1652
1662
(
2021
).
22.
G. Y.
Song
,
Q.
Cheng
,
B.
Huang
,
H. Y.
Dong
, and
T. J.
Cui
, “
Broadband fractal acoustic metamaterials for low-frequency sound attenuation
,”
Appl. Phys. Lett.
109
,
131901
(
2016
).
23.
M.
Duan
,
C.
Yu
,
W.
He
,
F.
Xin
, and
T. J.
Lu
, “
Perfect sound absorption of Helmholtz resonators with embedded channels in petal shape
,”
J. Appl. Phys.
130
,
135102
(
2021
).
24.
L.-j.
Li
,
B.
Zheng
,
L.-m.
Zhong
,
J.
Yang
,
B.
Liang
, and
J.-c.
Cheng
, “
Broadband compact acoustic absorber with high-efficiency ventilation performance
,”
Appl. Phys. Lett.
113
,
103501
(
2018
).
25.
S.
Xie
,
D.
Wang
,
Z.
Feng
, and
S.
Yang
, “
Sound absorption performance of microperforated honeycomb metasurface panels with a combination of multiple orifice diameters
,”
Appl. Acoust.
158
,
107046
(
2020
).
26.
Y.
Tang
,
S.
Ren
,
H.
Meng
,
F.
Xin
,
L.
Huang
,
T.
Chen
,
C.
Zhang
, and
T. J.
Lu
, “
Hybrid acoustic metamaterial as super absorber for broadband low-frequency sound
,”
Sci. Rep.
7
,
1
11
(
2017
).
27.
F.
Wu
,
Y.
Xiao
,
D.
Yu
,
H.
Zhao
,
Y.
Wang
, and
J.
Wen
, “
Low-frequency sound absorption of hybrid absorber based on micro-perforated panel and coiled-up channels
,”
Appl. Phys. Lett.
114
,
151901
(
2019
).
28.
Q.
Xu
,
J.
Qiao
,
G.
Zhang
, and
L.
Li
, “
Low-frequency sound-absorbing metasurface constructed by a membrane-covered and coiled Helmholtz resonator
,”
J. Appl. Phys.
133
,
075106
(
2023
).
29.
D.-Y.
Maa
, “
Theory of microslit absorbers
,”
Acta Acustica-Peking-
25
,
481
485
(
2000
).
30.
S.
Beniwal
and
C.
Chandraprakash
, “
Alumina–copper woodpile-kind locally resonant phononic crystal
,”
Appl. Phys. A.
129
,
844
(
2023
).
31.
K.
Bikumalla
,
T.
Bhuvana
,
A.
Tiwari
, and
C.
Chandraprakash
, “
Binder-free, non-carcinogenic surfactant-based cellulose-rich bagasse foams as acoustic panels
,”
J. Appl. Polym. Sci.
141
,
e55096
(
2024
).
32.
C.
Choi
,
S.
Bansal
,
N.
Münzenrieder
, and
S.
Subramanian
, “
Fabricating and assembling acoustic metamaterials and phononic crystals
,”
Adv. Eng. Mater.
23
,
2000988
(
2021
).
33.
Z.
Zhang
, “
A review on additive manufacturing of wave controlling metamaterial
,”
Inter. J. Adv. Manufact. Technol.
124
,
647
680
(
2023
).
34.
H.
Leventhall
, “
Low frequency noise and annoyance
,”
Noise Health
6
,
59
(
2004
).
35.
B.
Berglund
,
P.
Hassmen
, and
R.
Job
, “
Sources and effects of low-frequency noise
,”
J. Acoust. Soc. Am.
99
,
2985
3002
(
1996
).
36.
X.
Sagartzazu
,
L.
Hervella
, and
J.
Pagalday
, “
Review in sound absorbing materials
,”
Arch. Comput. Methods Eng.
15
,
311
342
(
2008
).
37.
G.
Yi
and
B. D.
Youn
, “
A comprehensive survey on topology optimization of phononic crystals
,”
Struct. Multidiscipl. Optim.
54
,
1315
1344
(
2016
).
38.
B. W.
Zong-fa Liu
and
C.
fu He
, “
Band-gap optimization of two-dimensional phononic crystals based on genetic algorithm and FPWE
,”
Waves Random Complex Media
24
,
286
305
(
2014
).
39.
K.
Mahesh
,
S.
Kumar Ranjith
, and
R. S.
Mini
, “
Inverse design of a Helmholtz resonator based low-frequency acoustic absorber using deep neural network
,”
J. Appl. Phys.
129
,
174901
(
2021
).
40.
Muhammad
,
J.
Kennedy
, and
C.
Lim
, “
Machine learning and deep learning in phononic crystals and metamaterials—A review
,”
Mater. Today. Commun.
33
,
104606
(
2022
).
41.
H. T.
Banks
,
K.
Holm
, and
F.
Kappel
, “
Comparison of optimal design methods in inverse problems
,”
Inver. Prob.
27
,
075002
(
2011
).
42.
C.
Chindam
,
A.
Lakhtakia
,
O. O.
Awadelkarim
, and
W.
Orfali
, “
Relative permittivity of bulk parylene-c polymer in the infrared regime
,”
J. Electromagnetic Waves Appl.
29
,
2139
2146
(
2015
).
43.
P.
Sun
,
Z.
Zhang
,
H.
Guo
,
N.
Liu
,
W.
Jin
,
T.
Yuan
, and
Y.
Wang
, “
Topological optimization of hierarchical honeycomb acoustic metamaterials for low-frequency extreme broad band gaps
,”
Appl. Acoust.
188
,
108579
(
2022
).
44.
Q.
Leclére
, “
Acoustic imaging using under-determined inverse approaches: Frequency limitations and optimal regularization
,”
J. Sound Vib.
321
,
605
619
(
2009
).
45.
R.
Zaheer
and
H.
Shaziya
, “A study of the optimization algorithms in deep learning,” in
2019 Third International Conference on Inventive Systems and Control (ICISC)
(IEEE, 2019), pp. 536–539.
46.
H.
Xiang
,
X.
Ma
, and
J.
Xiang
, “
Optimization for a locally resonant phononic crystal of square spiral with circle inside
,”
IEEE Access
7
,
145988
145995
(
2019
).
47.
Y.
Qian
,
K.
Cui
,
S.
Liu
,
Z.
Li
,
D.
Shao
,
D.
Kong
, and
S.
Sun
, “
Optimization of multi-size micro-perforated panel absorbers using multi-population genetic algorithm
,”
Noise Control Eng. J.
62
,
37
46
(
2014
).
48.
I. B.
Crandall
,
Theory of Vibrating Systems and Sound
(
D. Van Nostrand Company
,
1926
).
49.
The Mathworks Inc, ‘MATLAB; https://www.mathworks.com/products/matlab.html.
50.
U.
Ingard
, “
On the theory and design of acoustic resonators
,”
J. Acoust. Soc. Am.
25
,
1037
1061
(
1953
).