Mesure acoustique de protecteurs auditifs
Development of an acoustic bench to assess the sound pass through commercial earmuffs
1 Project Context
From January 2010 to May 2015, I have carried out my PhD as part of a collaborative research project on hearing protection between the École de technologie supérieure (ETS) and the Institut de recherche Robert-Sauvé en santé et en sécurité du travail (IRSST). The research aimed to develop new tools to: (1) better understand the Physics involved in the hearing protection devices (HPDs), (2) support the development of field measurement methods, (3) have a parametric and reliable tool that could be used to improve the HPDs designs. The research was fully funded by the IRSST (Montreal, Quebec).
My research focused on the sound attenuation of passive commercial earmuffs namely 3MTM EAR-MODEL-1000 and 3MTM PELTOR-OPTIME-98, using finite element modelling. For this last ear defender, I am grateful for 3MTM, for providing geometrical data and material parameters as part of a non-disclosure agreement. For the two studied earmuffs, I developed Finite Element (FE) models which were validated by experiments. Outcomes of the research were presented either in conferences, or in published articles or in my thesis.
The whole research project was also under the supervision a scientific advisor committee composed by: Elliott H. Berger, Scientific division 3M, USA; Karl Buck, Institut franco-allemand de recherches de Saint-Louis (ISL), France; William J. Murphy, National Institute of Occupational Safety and Health (NIOSH), USA; Nicolas Trompette, Institut national de recherche et de sécurité (INRS), France, to who I provided an overview of my research progression on a yearly basis. It is worth noting that at the beginning of the project, this panel of advisors were reluctant on the use of FE model to assess the true attenuation of ear defenders. Most of their works relied on pure measurements and analytical modeling.
The research was also carried out in a collaborative way with two other students, Guilhem Viallet and Martin Brummund, working on earplugs attenuation and the occlusion effect by an earplug.
2 Scientific objectives
The use of Finite Element modelling to assess the sound attenuation of ear defenders was (back in 2010) relatively new, and provided multiple advantages compared to laboratory measurements. The main advantage is the possibility to carry out parametric studies on geometries and material parameters without the need of manufacturing testing components. The second advantage is the possibility to run hypothetic studies, where, for instance, the sound field is neglected on specific component of the ear defender. These studies are generally used to understand the sound path transmission through each element of the model. At the time the project started, the industry relied on the Real-Ear-Attenuation-Threshold (REAT) measurement as a gold standard to evaluate the performance of the ear defenders. However, this subjective measurement method has been criticised for overestimating the sound attenuation of the tested ear defenders, widely influenced by the work environment, and user. The use of FEM to model hearing protector devices has emerged late 2000’s as a new way to develop better protections and support the development of Measurement-In-Real-Ear (MIRE) methods, to provide an objective measurement of the sound protection.
My research was driven by the need of getting better tools to better assess the earing protection of industrial workers exposed dangerous noises (85 dB(A) and more). At that time, very few works on numerical modelling of hearing protectors, were publicly available, generally providing a poor correlation between simulation results and validation measurement. During my research, I refined both aspects to increase the correlation and demonstrate that FEM could be used to predict the sound attenuation of earmuffs. I have improved acoustic measurement by developing a specific test bench with a specific methodology, I have improved the material characterisation of the comfort cushion, seen as the most difficult part to model, I have discussed the relevance of well-known parametric models.
3 Methodology and findings
My research was divided in two different activities: sound experiments to quantify the sound transfer path of the earmuff, the development of the numerical FE models for the low frequency band (below 400 Hz), including the characterisation of the earmuff components (geometry and material parameters). The model was then computed to higher frequencies up to 6.3 kHz and compared to the experimental results.
3.1 Sound experiments
3.1.1 Test bench development
For the first activity, I designed and manufactured an acoustic test bench to assess the insertion loss of the earmuffs and their uncoupled components, to (1) target the right level of modelling for each component and (2) have validation curves for the later developed model. It was designed in collaboration with the two other PhD students, so that it could be adapted for the needs of each specific research. This collaboration involved sharing designed ideas to incorporate each of our requirements, while considering the laboratory and space available. This acoustic test bench consisted of a rigid baffle fitted into a wall partition linking a semi-anechoic chamber to a reverberation chamber. In my case, the measurements were conducted in the semi-anechoic chambers, for which I used a loudspeaker to produce planar sound waves at normal incidence to that wall partition. I clamped the earmuff to the wall using a half headband and I measured thoroughly the clamping force. For my PhD purpose, I designed the baffle to integrate either a quarter-inch microphone or an ear simulator (coupler IEC-711) to study the dual protection (earplug and earmuff combined). The bench could integrate a silicone flesh replica (whose property complied with standard ANSI S12.42 -2010). To manufacture the flesh, I designed an aluminium mould using SolidWorks. The mould was manufactured by a local company (However, the final adjustment, taps and threads, and leaks control were done by me).
The experimental measurement system, as well as the methodology required some iterations. For instance, the first measurement campaign demonstrated a lack of stiffness of the wall support where the acoustic baffle was clamped. I also had to try various acoustic sealant to control acoustic leaks in my system. As an example, I tried petroleum jelly, which caused sliding motion over the baffle. I designed a specific methodology to assess individual sound path through the earmuff components. I designed dummy cushions made of lead, to assess the sound transfer path through the plastic cup, I designed a steel plate to study the sound transmission through the cushion, with or without its mechanical motion. I also assessed different type of foam liners and modified a comfort cushion to incorporate artificial acoustic leaks of different sizes.
3.1.2 Acoustic Measurement results
The measurement system with the methodology I developed demonstrated its capability in providing much better results compared to previous publications, but also a great reproducibility of the experiments. The measurements provided new findings, and somehow demystify some myths.
The first finding relates to the high insertion loss (attenuation) of the cushion material. Thanks to the experiments, I could measure an attenuation from 40 dB to 50 dB on the from 20Hz to 5 kHz. I could demonstrate that the mechanical behaviour (pumping motion) of the cushion is responsible of the degradation of the insertion loss at low frequency, and therefore it might not be necessary to model the sound excitation on the cushion flanks.
Another finding is the negative effect of the foam liner in mid-frequency. While this was presented in a one-third octave spectrum by others (but not mentioned), I could observe this phenomenon distinctively and point out the lack of sound attenuation was caused by the lack of absorption of the material in this frequency band.
I have demonstrated the earmuff attenuation is uniquely influenced at low frequency by the mass-spring system like made by the cup over the cushion and is influenced at high frequency by the cavity resonances under the cup.
3.2 Low frequency FE modelling
Based on the experimental result, I proposed two different ways to model the earmuffs and especially the cushion which is seen as the most complex part. The cushion was either model as a “spring foundation” (purely mechanical behaviour) or as an “equivalent viscoelastic solid” (allowing acoustic transmission through its flank).
For both models, I reversed engineer the equivalent geometrical and mechanical parameters: I reproduced the geometry of the compressed cushion (to mimic its real use) using cast moulds which were laser scanned using a coordinate measurement machine. I post processed the images using both Polyworks and Solidworks. To determine the equivalent mechanical parameters, I used a combination of two vibration methods (resonant measurement and a quasistatic measurement analysis) combined with a reverse finite element model.
Using the geometry and the material parameters, I implemented the FE models. Each model was developed in the low frequency band, up to 400 Hz using COMSOL Multiphysics. I compared the low frequency model (using both the equivalent solid model and spring foundation model for the comfort cushion) to the experimental curves and to a well-known analytical (lumped element) model. I could point this lumped parametric model is inaccurately described and relies on empirical parameters. Using results from both FE model and experiments, I could clarify the definition of these parameters. I could also compare the two approaches to model the cushion. In the low frequency band, both approaches to model the cushion were found relevant. I published the whole methodology and discussion in the Journal of the Acoustical Society of America, as well as in my thesis.
3.3 Computation to higher frequency range
I investigated the relevance of the model at higher frequency (up to 6.3 kHz) and compared with experimental results. I could observe some biases between the simulation and the experiments, but similarities were observed. Comparatively with previous research paper, a better correlation model/validation was found, but only when the sound transmission through the comfort cushion was neglected.
To better understand the physics, I adapted the energic approach proposed in Basten 1998 (Spatial matching of structural and acoustic modes in an airtight box) which helped to quantify which one of the structural domains (cushion, backplate and plastic cup) or cavity domain was controlling the vibroacoustic behaviour of the system. I carried out a modal analysis. The energetic approach seems similar to the SEA and consists in extracting both modal kinetic energies and modal potential energies from the structure and cavity domains. The identification of the coupled modes allowed me to determine, for each frequency band, if an improvement of the structure (e.g., Modify geometry, use of damping material), or of the cavity (add an insert) was needed.
4 Outcomes of the academic research
My academic research provided several findings that were presented in international conference, or in peer-reviewed articles I have authored and co-authored, accumulating more than 50 citations.
I have presented the description of the acoustic test bench, the methodology, the results and discussions at the International Congress of Acoustics in 2013 (Montreal) and later, I published in the journal Applied Acoustics in 2014. I presented additional results in my thesis, such as the effect of the silicone flesh and artificial leaks, which could not be published earlier. I had the opportunity to present an early development of the FE model in 2011, during the Acoustics Weeks in Canada 2011, Quebec City. The final development of the model was presented in the Journal of the Acoustical Society of America in 2014. During my cursus I also co-authors articles and conference proceedings on the overall project.
During our last project board meeting, the Scientific Advisor Committee recognized the progression made during these five years of research and found an interest in the numerical modelling activity.
The experimental result has seen some interest from the community, as it aimed to better understand the vibroacoustic behaviour of the earmuff. The article relating the results has been cited by other laboratory in the world working on hearing protection either developing new field measurement methods or experimenting new earmuff design.
5 Personal outcomes and other achievements
5.1 Project organisation
My PhD has been paced by various activities of different natures. I have been supervised on a weekly basis by my supervisors during five years. During this meeting I discussed the progression of my work, assess the target of my research, and make decision based on the progression.
During my first year, I worked on determining the context, the objectives, and the methodologies for the research, as part of a one-year course. I also followed two specific courses on acoustics at the École de Technologie Supérieure and the University of Sherbrooke.
Various events were also influencing my day-to-day research: the annual open house of the university, in January, followed by the annual poster contest of the Health and Safety Research Team, the scientific advisor committee preparation. There was also time when I focused on conference proceeding and article writing.
5.2 Technical skills
During my first year, I familiarised myself with sound scattering phenomenon that I had not studied before. For instance, I started to look at developing some mathematical models of the sound attenuation of hemispherical multi-layered shells. This model would look like a simple earmuff lying on a rigid surface.
I implemented the equation using FORTRAN and had to learnt quickly on how to code in this language.
I learn how to use COMSOL Multiphysics and improved my knowledge in MATLAB coding.
Through my research, I obviously learnt more on acoustic theory and modelling, but also increased my technical skills in engineering. I have developed expertise in acoustics and vibration, in reverse engineering to capture both geometries and material parameters.
I gained in analytical skills especially when it comes to analysing and assessing a complex system. I have applied these skills and techniques in my current role.
I have also experienced various challenges during my research, such as lack of computing and physical infrastructures. These challenges were overcome through the time, thanks to new equipment being delivered to our laboratory, or by new software release which provided bug corrections and fastest computation time.
5.3 Communication skills
As mentioned above, I have communicated my research through different media at destination to the scientific community. I have learned how to write a scientific paper, the submission and peer reviewed process. I also ensured the integrity of my research, by following the best practice and policies from the university on plagiarism. For this purpose, I used a referencing module (Zotero) to compile the literature review and track my notes. I maintained the database by adding new research papers or books I have used to carry out my research. While writing my thesis I thoroughly double-checked references, image copyrights, and data from other publications. I also ensured the integrity of the 3M intellectual property regarding the data they provided. I requested the permission to Elliott Berger (Scientific division 3M) to publish the mechanical parameters of the plastic elements of the Peltor-Optime-98. I also enquired about displaying the geometry of this earmuff.
The authoring task has been a long learning process with multiple iterations to account on comments from my different co-authors, account on the maximum pages allowed by the journal and account of the reviewers’ comments.
I have learned how to submit an article to a peer-reviewed journal, to select or reject potential reviewers. For instance, reviewers must be linked to the field, and able to understand the technical content of the paper, they also must be impartial and not present a conflict of interest. As an anecdote, there has been some confusion from a journal editor, who designated my supervisor as reviewer for one of my papers. My supervisor indeed declined the invitation to review, as he was supervising me, but was also a co-author.
As mentioned earlier, I have taken part to the annual open house at the university and local conferences organised by the Institut national de santé publique du Québec (equivalent of HSE), presenting my research on posters, using popular science. Each year, I have made a short selection of what could be presented to public. I designed the posters in a way they present the context, objectives, methodology and results with conclusion, while ensuring a certain graphical quality and readability. The poster was then submitted as part of a contest in January and was also presented during poster presentations sessions. These diverse oral experiences allowed me to become more confident in presenting to an audience. I acknowledge the quality of my poster and conference presentation have matured through my PhD.
In 2015, I was also pleased to attend a conference from Jean-Luc Doumont on scientific content presenting. I am still getting inspired from his guidance, such as minimising the text on presentation slides.
5.4 Securing additional funds
The Institut de recherche Robert-Sauvé en santé et en sécurité du travail (IRSST) sponsored my PhD program. As part of the administrative procedure proper to the university, I had to request the funds on a four monthly basis.
During my PhD, I have applied to various additional grants, and won money prizes as part of academic contests. Each year I entered the poster contest from Health and Safety Research Team of my university. In 2012, I successfully got 3rd cycle student scholarship, based on my academic progression.
I also engaged actively, with the university sport centre to know more about financial helps and could secure sports scholarships for years 2012, 2013, 2014.
5.5 Other activities
5.5.1 Test subject for hearing protectors
In parallel of my research, I have been enrolled as a test subject for other PhD research, hearing protector and earphones products. This helped me to get familiarised with the REAT method, and how subjective measurement methods are carried out.
5.5.2 Unexplored engineering activities
The biggest challenge of my PhD topic was the ambitious initial goal to get a FE model of the earmuff coupled to an artificial head and torso.
This would have opened the opportunity to a university exchange to compute the external sound field applied on the earmuff, using Boundary Element Methods (BEM). The plan was to solve the external sound field during a one-to-two-month trip in January 2013 at the Université Technologique de Compiègnee (France), as part of a co-supervision. I initiated the first contact in 2010, with Professor Mohamed-Ali Hamdi from UTC who attended the presentation of my PhD problematic plan by video conference.
For this purpose, I dedicated four months to 3D-model the Acoustic Test Fixture (ATF) GRAS 45CB-V1, using a laser scan (handy scan from Creaform), Polyworks (to clear the recorded point cloud), CATIA V5 (to recreate complex surfaces), and Solidworks (to form volumes). However, the development of the FE model at a lower level of complexity (earmuff coupled to a rigid baffle) was needed with a higher priority, and thus I have decided to abort that part of the project. I later released the geometry to another student, Marc-Andre Gaudreau, for a similar purpose, who used it to model noise attenuation of an earmuff couple to a manikin excited by a planar sound wave, using Finite Element Method.
5.6 Academic profile
I completed my PhD in 2015 with two published papers as main author in peer-reviewed journals and two international conference presentations. I am delighted by this accomplishment as it has helped me to start my professional career. My publications, especially the one on the sound transfer pat through the earmuff components has found interest from other researchers. My numerical model and the digitalisation of the acoustic test fixture (manikin) have been used by other students to keep exploring the sound attenuation prediction of earmuffs.
After I left the university, the Institut de recherche Robert-Sauvé en santé et en sécurité du travail (IRSST) created new PhD programs to refine the model of the earmuff. In 2017 Kevin Carillo provided an update with a refinement of the cushion comfort model, using anisotropy and a modified boundary condition. More recently, a new PhD candidate, Yu Luan, published a refined model of the sound attenuation using dual hearing protection (earmuff and earplugs).