Annals of Occupational Hygiene Advance Access originally published online on October 27, 2004
Annals of Occupational Hygiene 2004 48(8):707-714; doi:10.1093/annhyg/meh068
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
© British Occupational Hygiene Society Published by Oxford University Press;
Development of Prototypes of Half-Mask Facepieces for Koreans Using the 3D Digitizing Design Method: A Pilot Study
1 Department of Occupational Health and Safety Engineering, Inje University Gimhae, Gyeongnam, 621-749, South Korea; 2 College of Design, Product Interaction Design Program, Inje University, Gimhae, Gyeongnam, 621-749, South Korea; 3 Department of Environmental Education, Gongju National University, Gongju, Chungnam, 314-701, South Korea
* Author to whom correspondence should be addressed. Tel: +82-55-320-3285; fax: +82-55-325-2471; e-mail: dhan{at}inje.ac.kr
Received 14 August 2003; in final form 11 June 2004
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
In the Korean market there are several respirators for industrial purposes. Some of them are imported from global manufacturers such as 3M, and the others are developed by domestic companies. However, some of the locally made respirators have a face-seal leakage problem because they were not designed taking into account Korean facial characteristics. This pilot study was conducted to develop three face models for a half-mask based on the Korean fit-test panel and also to design three well-fitting silicon prototypes (large, medium and small) for Korean faces. For a test panel, 50 subjects were selected on the basis of lip length and face length. Shapes of faces from the test panel were scanned by a 3D scanner (Vivid 900, Minolta). The facial dimension scales for three size groups were established through statistical analysis. To ensure that the shapes fit the mean facial dimensions of each size group, similarly shaped sample faces were selected from the test panel. Mean faces with representative facial dimensions as well as face models (mannequins) were made by the Rapid Prototyper (RP; Z400, Z Corps, USA). These were reshaped to fit the mean faces by clay-modeling. The face models were digitized to recheck the fit of the facial dimensions. On the basis of the face models, three types of masks were developed through digital modeling, and tangible models were prototyped by RP. Three differently sized prototype masks were made with silicon and examined by simple naked eye test. This procedure established the mask design process based on digital technology; it can be applied for similar design projects in the future. Further study is needed to verify whether these prototypes would be a good fit for Korean faces.
Keywords: design • facial dimension • half-mask facepiece • respirator • 3D digitizing
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
To find a respirator that fits, the wearer must be fit- tested in a country that has fit-test regulations such as the United States (CFR, 2002
). But in a country that has no fit-testing regulations, such as Korea, it is not easy for a worker to select a respirator that fits well.
In the Korean market there are several respirators for industrial purposes. Some of them are imported from global companies such as 3M, while others are developed by local manufacturers. A large number of locally made respirators have a face-seal leakage problem. The investigation of face-seal leakages reveals that global manufacturers' respirators perform better than those that are locally made (Han, 1999, 2000). The authors of the present study assert that there has been no scientific approach toward respirator design based on the study of Korean facial characteristics. In additional, locally made respirators may not fit Korean workers. Most local companies manufacture only medium sizes. It is, therefore, necessary in Korea to design respirators that are a good fit for Korean workers with small, medium and large faces.
Today, most manufacturers supply half- and full-face masks in multiple sizes that cover these facial groupings (Myers, 2000). However, many researchers have indicated that there is little or no correlation between these facial dimensions and the fit of half-mask respirators (Liau et al., 1982; Oestenstad et al., 1990; Oestenstad and Perkins, 1992; Brazile et al., 1998). Oestenstad and Perkins (1992) concluded that menton–subnasal (lower face) length alone was consistently indicated as being correlated or associated with the fit of each brand of half-mask respirator rather than face length and lip length. Brazile et al. (1998) concluded that respirator fit was not associated with facial dimensions based on race/ethnicity or gender and seemed to be associated with individual facial characteristics. Some research into the specific facial dimensions that affect face-seal leakage of a Korean mask were previously conducted (Han and Choi, 2003). In this research, no common facial dimensions were found to be significant for all nine brand/gender subgroups. Therefore, subjects for face models may be selected on the basis of the Korean fit-test panel developed previously by Han (1999).
In general, digital design methods include digital simulation and communication such as creating a model or transferring digital design data for concurrent engineering. These focus only on presenting the design concept or on sharing the product design data with the engineering side (Hsiao and Chuang, 2003). The human factor, in broad terms usability, is very important in the discipline of design. Digital technology may help to collect more detailed human data and to make a design suitable for the user's potential requirement.
To create a new design process based on this emerging technology, industrial respirators could be considered a good research case. In this project, as in other design processes, product structure analysis and usability tests were performed to derive important design factors. Furthermore, three differently sized (small, medium and large) digital face models were developed. For this process, a 3D scanner, a rapid prototyping machine and related software were used.
|
METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Test panel and face dimensions measured
To make face models (mannequins) for mask design a so-called ‘respirator test panel’, that is persons meeting specific facial dimension criteria to meet various facial sizes, needs to be selected.
A previous study indicated that three facial dimensions (face width, bitragion–menton arc and nose protrusion) should be preferentially considered when designing a half-mask respirator for Korean workers, but the study also implied that it would not be easy to consider certain facial dimension(s) when designing half-masks for Koreans because of no common facial dimensions to affect mask fit (Han and Choi, 2003). As stated earlier, many researchers have also indicated that there is little or no correlation between common facial dimensions and the fit of half-mask respirators. For this reason, the Korean test panel developed by the author for an earlier study (Han, 1999) was adopted to avoid controversy between facial dimensions and mask fit. The male and female test panels for half-masks consisted of 25 members (13 males, 12 females). But the test panel (subjects) used in this study consisted of 50 members (male 26, female 24) to enhance reliability as shown in Fig. 1. The facial dimensions illustrated in Fig. 2 were measured by an investigator who had been trained at Anthrotech (formerly Anthropology Research Project, Inc., Yellow Springs, OH 45387, USA) to ensure measuring accuracy. All measurements were made in centimeters to one decimal point using sliding and spreading anthropometric calipers and a steel measurement tape.
|
|
Scanning and measuring the face
3D scanning captures body dimensions in a fast and reproducible way. The number of anthropometrical variables that can be derived from a scanned human body is almost without limits (Rioux, 1997
; Yu et al., 2003). To scan the 50 subjects' faces fluently, scanner type, scanning place and procedure were arranged. Vivid 900 (Minolta, Japan), which is a very well-known laser slit-beam type 3D scanner, was used for face scanning. It captures more than 300 000 points at a scan. Its color scan function, which is possible exclusively in a scanner with such a resolution, is needed to obtain the color of the human face. The scanning process consisted of three steps: (i) preparation—hair adjustment, instruction for the scanning process and pose, (ii) scanning and (iii) approving the scanned data. The disadvantages of 3D scanning, compared to traditional anthropometry, include missing data caused by shading and occurrence of errors owing to movement of the human face. To minimize the problem, scanning was performed from three angles (front, 45° of left-hand side and right-hand side) for each person, using a rotating chair to ensure that the subject's eye direction does not move (Fig. 3). The size of the room was decided mainly by the scanning range (0.6–1.2 m) of the scanner and it was kept half-dark. After the scanning was completed, three scanned data for each subject were merged and smoothed for reducing irregularity on the surface caused by the scanning itself. For the merging and smoothing, Rapidform2000 (Inus Technology, Korea), a scan-data processing application was used.
|
The dimensions of the digitized face were measured using the Rapidform2000 software. Rapidform2000 is recognized as a comprehensive tool to convert physical data from a 3D scanner into digital data for scan-data processing, deformation, measurement and format transformation. Digital measuring differs from traditional measuring in this important aspect. The physical characteristics of traditional measuring tools, which can interact with the nature of human skin, are not present in digital tools. To solve the problem, a pre-marking phase should be included in the digital measuring process. A circular removable sticker with a diameter of 5 mm was good for this case.
The measuring differences that possibly occur in the two methods are as follows. Measuring dimension A (bizygomatic breadth) requires some pressure on the tragion with traditional calipers, which might reduce the length of A. Furthermore, the physical characteristics of a measurement tape usually produce the line as simply defined and not as a irregular line on the skin. Anthrotech defined dimensions H (bitragion–subnasal arc) and I (bitragion–menton arc) as ‘the surface distance between the right and left tragion across subnasal or menton’ (Anthrotech, 1997). Due to the tape's elasticity, measuring distance on a 3D human face results in a parabola, which is not on a flat plane. Because the plane, on which the line is located, is not flat, the surface distance is not the shortest one between two points, which is the general definition of distance in mathematics. Therefore, to measure the surface distance, a complete flat section of the face, which has all three points on it, has to be made and the section line distance between both tragions will have to be measured. Digital measuring is more proper and is based on the definitions for some aspects. With the newly made processes and methods, the face data of each of the 50 subjects were measured. To ensure reliability, the traditional physical measuring methods, using sliding calipers, spreading calipers and measuring tape, were employed simultaneously. When compared, the results obtained had a similar permissible error (95%).
Making three differently sized face models (mannequins)
Statistical analysis for each facial dimension and determination of mean digits of each size
To customize the masks, the authors decided that three different sizes must be made. Therefore the mean faces should be representative of three different size groups (large, medium and small). Dividing the 10 facial dimensions measured into three different sizes could be a reasonable way to make three groups. With the standard deviation of a quartile or an average, facial dimensions are divided into three groups based on scale (large, medium and small). The average of 10 facial dimensions for each group was computed, and the standard deviation for each was also computed. Then the Mahalanobis distance of each average value and observed value for the 10 parameters was calculated, and the standard deviation for each group was used as a weight to control for the influence of large-scale variables in this computing. The observed value of facial dimensions with the narrowest Mahalanobis distance was selected as a representative digit in a group.
Making representative mean face models
To make the mean facial shape, the subject whose face had the most closed mean digits in each group was selected through statistical methods. Table 1 shows the means of facial dimensions (mean digits) of each size group and facial dimension values of the most similarly shaped faces for mean digits of a size group in computer digitizing. Figure 4 shows an example of the most similarly shaped face in the large-size group through computer graphics. These subjects' facial shapes would be altered to fulfill the mean digits to obtain the natural surface of the mean face for each size group. This is done because exactly matching facial shape with each mean facial dimension is very difficult using the computer graphics method. Therefore, the authors used a traditional craft technique that put a handy-coat on the physical model reproduced by the Rapid Prototyper (RP; Z400, Z corps, USA). To make the selected shape most natural and to match exactly the mean digits, measuring templates were used during the craft process. After finishing the process, a 3D scanner was used to check whether the facial dimensions of each facial shape satisfied the mean digits. For the scanning, another 3D scanner (Opto-top, Breukmann, Germany), which has more accuracy than the Minolta Vivid 900, was used. Finally, three representative mean face models in three sizes were completed as shown in Fig. 5.
|
|
|
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Mask design and prototyping
The representative facial shape (mean face) for each group allows some part of the mask surface to fit the face; however, it is not enough to design a mask because several other factors need to be considered. To identify these factors, the shape characteristics of ordinary industrial masks were examined (Fig. 6). Three brands of masks were examined: 3M 6000 series (3M, Minnesota, USA), SG5121 (Samkong, Seoul, Korea) and YS1050ds (Youngseung, Seoul, Korea), which had been used in an earlier study on the relationship between fit predictors and facial dimensions for half-masks in Koreans (Han and Choi, 2003
). The examination included short, simple usability tests, structure analysis and sample product evaluation for the three brands. A half-mask respirator consists of an elastomeric body, strap and filter or cartridge. During the structure analysis, the strap seemed to affect the fitting performance as much as the elastomeric body shape. The authors were able to authenticate the function of the strap during the simple usability test as expected. For the test, three adult examiners were requested to put on and take off the mask and were asked to give their opinions about the fit. The entire procedure was recorded to aid analysis. It was revealed that the strap and wearing gear affect the fit of the respirator. In particular, the locally made mask's strap and chin support structure functioned poorly, and it seemed to be the cause of the face-seal leakage problem because it could not be tightened. An elastomeric body is made of one piece only, but its functions vary depending on its components. The inner part of the mask is to fit the mask to the face, and it should have sufficient surface area to prevent leakages. However, because of the structure of the human face, there are some restricted areas, such as the eyes, the nose and the mouth. The outer part constitutes the structure of the mask and it should not touch any of the surfaces on a face. Between the inside and outside parts, there is a linking part, which has a rounded section. It links the inside and outside parts to shape the mask, and gives a tension to the mask fit. The rounded section is comfortable and gives the mask some flexibility. However, the linking part of a sample mask has no rounded section and the outer part had an undefined shape. From these examinations, the following design factors were established. Figure 7 shows the three design components of half-mask facepieces arrived at in the study.
- Inner fitting part
- The surface area, which avoids covering the mouth, nostrils and eyes, should be as large as possible.
- From the front, the inside and outside line of the inner part should be offset from the outline of the outer part.
- The surface stream should be adopted from the surface of mean face models (mannequins), but should be defined as simply as possible.
- The chin, the lower part of the face, should be enclosed, but the surface should not be too close to the neck.
- It should be designed such that too much pressure on the nose is avoided.
- The surface area, which avoids covering the mouth, nostrils and eyes, should be as large as possible.
- Outer forming part
- The outline of the outer part should be wider than that in existing imported masks as Koreans have comparatively wider faces.
- From the front, the outside line of the outer part should be as simple as possible.
- There should be no contact between any part of the face, such as the nose, and this part of the mask.
- It should be designed such that the entire mask has a regular form.
- The outline of the outer part should be wider than that in existing imported masks as Koreans have comparatively wider faces.
- Linking part
- Make a regular round section through the part.
- No edge on the side to human face.
- Make a regular round section through the part.
|
|
Based on these factors digital modeling was done with Rhino3D modeler (Robert McNeel & Associates, USA). After making a simple outline of the outer part, the inner part was abstracted from the mean face models. The three parts were connected to complete the mask shape, as shown in Fig. 8. Three physical models were made by RP based on the digital models. These were transformed to silicon models via silicon moulding, which is similar to the material with which real masks are made, as shown in Fig. 9. A simple wearing test for fit without any accessories such as straps was performed on 10 members from the test panel and they reported feeling more comfortable than before. A simple naked eye test for fit was also conducted with face models, as shown in Fig. 10.
|
|
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
The main purpose of the project was to develop a mask using digital technology based on Korean facial characteristics to avoid face-seal leakages. Prototypes in three different sizes were developed and were evaluated favorably following a simple wearing test at the laboratory level. It was reported that the wearing fit was better than that of other existing masks and also that there was a noticeable difference between the former and the newly designed masks. These results should be established through a quantitative fit test in the near future. For the test, the elastomeric body should be made of real materials, and accessories, such as filter/cartridge and strap, should be attached to the elastomeric body as in existing masks.
Compared with traditional anthropometric measuring tools, digital face measuring methods were successful on account of the accuracy. If facial dimensions had been acquired through the traditional method, the mean face could not be made without improvising on the parameters. Digital scanning is good not only to obtain data from the face, but also to enable a real face surface to be utilized to make face models. However, as this was a pioneering effort there were some problems, which should be considered in future work. One of them is that the human body is not just the surface alone. In other words there is a skeleton under the surface. Traditional anthropometric methods constitute not just using measuring tapes and so forth, it has a human touch and feeling as well. To measure the human face digitally, we still need the help from a human assistant for activities such as marking, which could be found by palpation. Depending on only digitizing might rule out long-standing, valuable human-based methods altogether. As a long-term goal, it is suggested that the anthropometric measuring system for digital application including tools, dimension definitions and process must be studied as an independent research.
As mentioned earlier, the study was based on the concept that the main cause of face-seal leakage is the poor fit of the inner part of the elastomeric body facepiece. However, the authors found that inadequate fit might be just one of several causes of leakage, and other parts such as the outer part, linking part and strap could also be responsible for it. In fact, mask design is too complex for an ordinary design team. As in other design problem-solving processes, problem definition should be made at the initial stage of design even though it is not a conventional project for designers. One of the problems arose from choosing similar sample faces for making face models, which fulfil the mean facial dimensions. The selection was done by a statistics team. However, that was one sample face (large size) the statistics team suggested was not appropriate for making a mean face model: the sample face showed similar dimensions, but all the dimensions were slightly bigger than the parameters. This implies that correction should be done on every dimension, which means it's not a correction but a creation, which makes the correction plan useless. Therefore the large-size face was reselected on the basis of C (MNRL), D (MSNL) similarity, which is very difficult to be corrected. It was found that statistical similarity was different from design similarity. The authors concluded that to make a co-design process successful, continuous discussion and communication between different disciplines is vital.
|
CONCLUSION |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Through the co-design process with many disciplines, a pilot study on making three differently sized silicon prototypes (large, medium and small) for half-mask facepieces to fit Korean facial characteristics was performed. Before designing with computer graphics, three representative mean face models were proposed following the statistical analysis of facial dimensions. The prototypes were examined for fit by a simple naked eye test. After attaching accessories to these prototypes with real material, further study on fitting performance needs to verify whether they are a good fit for Korean faces. Due to the comparatively small number of subjects in the test panel, it might be difficult to generalize the result as a ready-to-market product. However, the process is considered as an almost complete one and it suggests that further study should aim to develop a product that can be launched in the market.
|
ACKNOWLEDGEMENTS |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
This work was supported by the Korea Science and Engineering Foundation (KOSEF) Grant (R05-2001-000-00684-0 (2002)). The authors would like to thank Professor K.-L. Choi, Department of Data Sciences, Inje University, for his statistical analysis. They also thank Ms S.Y. Chang, Design Department, Graduate School, Inje University, for her technical assistance.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMETHODSRESULTSDISCUSSIONCONCLUSIONACKNOWLEDGEMENTSREFERENCES |
|---|
Anthrotech (formerly Anthropology Research Project, Inc.). (1997) A short course in anthropometry. Yellow Springs, OH 45387, USA.
Brazile WJ, Buchan RM, Sandfort DR et al. (1998) Respirator fit and facial dimensions of two minority groups. Appl Occup Environ Hyg; 13(4): 233–7.
CFR (USA). (2002) Respiratory protection. Code of Regulations Title 29, Part 1910.134.
Han D-H. (1999). Fit testing for respirators and development of fit test panel for Koreans (in Korean). Korean Ind Hyg Assoc J; 9(1): 1–13.
Han D-H. (2000) Fit factors for quarter masks and facial size categories. Ann Occup Hyg; 44: 227–34.
Han D-H, Choi K-L. (2003) Facial dimensions and predictors of fit for half-mask respirators in Koreans. Am Ind Hyg Assoc J; 64(6): 815–22.
Hsiao SW, Chuang JC. (2003) A reverse engineering based approach for product form design. Design Studies; 24: 155–71.[CrossRef]
Liau YH, Bhattacharya A, Ayer H et al. (1982) Determinations of critical anthropometric parameters for design of respirators. Am Ind Hyg Assoc J; 43(12): 897–9.[Medline]
Myers WR. (2000) Respiratory protective equipment. In Harris RL, editor. Patty's industrial hygiene, 5th edn. New York: John Wiley & Sons, Inc, pp. 1501–5.
Oestenstad RK, Dillion HK, Perkins LL. (1990) Distribution of faceseal leak sites in a half-mask respirator and their association with facial dimensions. Am Ind Hyg Assoc J; 51(5): 285–90.[Web of Science][Medline]
Oestenstad RK, Perkins LL. (1992) An assessment of critical anthropometric dimensions for predicting the fit of a half mask respirator. Am Ind Hyg Assoc J; 53(10): 639–44.[Medline]
Rioux M. (1997) Color 3-D electronic imaging of the surface of the human body. Optic Lasers Eng; 28: 119–35.
Yu CY, Lo YH, Chiou WK. (2003) The 3D scanner for measuring body surface area: A simplified calculation in the Chinese adult. Appl Ergonom; 34: 273–8.[Medline]
CiteULike 
Connotea 
Del.icio.us What's this?
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||