Journal of Ergonomics
Open Access

An Analysis of Walkway Surface Micro Roughness and its Impact on Human Slip Potential

Anthony Marletta^{* *}Correspondence: Anthony Marletta, Department of Ergonomics, Indiana University of Pennsylvania, New York, USA, Email:

Author info »

Introduction

From an occupational perspective, according to the 2023 Liberty Mutual Workplace Safety Index, slips, trips and falls on the same level are the second most common cause of occupational injuries in the workplace in 2022. These injuries accounted for direct costs of $8.98 billion and 15.3% of workplace injuries and illnesses (Liberty Mutual Research Institute for Safety, 2022), estimated to represent only a fraction of the total cost of these injuries [1-2]. Furthermore, the Bureau of Labor Statistics (BLS) has analyzed non-fatal occupational injuries that have led to unemployment, and their analysis shows trends with an increase in the proportion of same-level falls to overall occupational injuries on a year over year basis (BLS, 2019). According to the Centers for Disease Control (CDC), falls are the fourth leading cause of unintentional deaths in the United States based on annual, aggregate data from 1981 to 2019 (CDC, 2021). The Centers for Disease Control also reported that between 2001 and 2018, “unintentional falls” were the leading overall cause of nonfatal injuries treated in hospital emergency departments in the U.S. especially in patients over the age of 65 (CDC NEISS, 2019). Also, according to the National Electronic Injury Surveillance System (NEISS, 2019), which collects injury statistics from hospital emergency rooms across the U.S., injuries on stairs, ramps, landings, and floors are the leading cause of injury for both genders in every age group except for age group 5 through 14 years. Specifically, slips related to bathtubs or showers accounted for 11,391 of the 40,939 total cases (27%) of slip and fall injuries reported between 2013 and 2017 [3-6].

One of the most prevalent causes of same-level fall events among the general population is the loss of traction or slip resistance between shoe soles and floor surfaces [7]. However, when looking at slips, trips and falls on the same level, it is important to consider that both slips and trips represent two significantly different precursors to the outcome of a fall [8,9]. Various floor surface roughness parameters have been highlighted as having correlation to slip resistance and overall slip risk [10]. (Chang found that various topographic characteristics (e.g., Ra, Rpm, Da, WDq and Wtm) of a given floor surface can significantly impact traction performance between the shoe sole and the floor surface [11-13]. However, to date, there has been no adoption of any single tribometer in the United States, due to the variability related, in part, to the operator, device, test condition and surface preparation [5]. Further, the utilization of a floor surface available friction value is dependent on the tribometer, and that specific available friction result signifies different levels of floor surface slip risk depending on each different tribometer in varying environments [10]. Further factors that contribute to the uncertainty of tribometer measurement and their applicability to slip risk include differences in the test method (such as both the number and direction of tests), tribometer condition (such as wear and lubrication of moving parts), test foot type, environmental conditions, and test sample properties [14]. Blanchette, et al., highlights the use of a reference surface based on human subjects testing which can be utilized as means for identifying a relative safety threshold value when applying the specific tribometer, environment and user operating the device [2]. However, because surface friction measurements should not be used as a standalone indicator for assessing slip risk due to the variability in tribometer measurement, surface roughness testing could act as a complement to surface friction measurements when measuring floor surface slip risk. However, there is limited research-based guidance indicating specific criteria levels for floor surface roughness as it relates to the COF of a floor surface and its relationship to slip and fall events. Kim, et al., found a range where floor surface roughness increased overall slip resistance on various floor surfaces; however, they only measured in a wet, soapsuds environment with one tribometer and one floor surface, which limited the power and applicability of their results to the general industry [11]. Chang analyzed the effect of surface roughness on slip resistance measurements and was able to establish a relationship between two of the tribometers out of the five utilized in his study. Said found relationship was between friction index measured in Rpm (the average of the maximum height above the mean line) and Rpk (the reduced peak height) indices. While Chang was able to highlight a relationship between the two measures of surface microroughness and slip resistance measurements, as well as which surface roughness parameters correlated best with measured friction, he was not able to highlight a specific range nor criteria level for those roughness parameters that could be utilized in industry [15].

Research hypotheses

The research tested the following research hypotheses:

H0: Floor surface roughness, as measured by the parameter Ra and Rz, user, floor surface type and/or tribometer type are not significant predictors of floor surface slipperiness as defined by the tribometer.

HA: Floor surface roughness, as measured by the parameter Ra and/or Rz, user, floor surface type and/or tribometer type are significant predictors of floor surface slipperiness as defined by the tribometer.

Assumptions, limitations, and delimitations

The study had the following assumptions:

• Tribometers used in this study were used in an accurate manner, following the manufacturer’s operational guidelines for each machine.

• The research instruments were assumed to be operated in the same manner with minimum variance.

• The intra-tile homogeneity of ceramic tiles which were tested were considered consistent.

• The data collection instruments used in this research were assumed to be reliable with minimum error and variance.

The study is limited to the following:

• The research instruments utilized in this study are subject to a certain amount of unknown source error in the measurements due, in part, to user error as well as inherent tribometer error.

• The intra-tile homogeneity of ceramic tile can be inconsistent and therefore consistency of ceramic tile results is subject to some error associated with its homogeneity within the same tile sample.

• The application of this study is limited to only the three floor surface types that have been measured in this study.

The study is delimited to the following:

• This study focused only on the three floor surfaces selected, as testing a greater variety of floor surfaces at varying degrees of surface roughness is not practical given the resources allocated for this study.

• The study used three tribometers to measure surface slipperiness values, the Slip-Test Mark IIIB, BOT 3000E and English XL.

Materials and Methods

Apparatus/Tools

The purpose of this study was to evaluate three common floor surfaces: Ceramic tile with a polished finish, ceramic tile with a matte finish, and a porcelain enamel bathing surface all with varying degrees of floor surface roughness, measured in both the Rz and Ra roughness parameters, to identify relative roughness thresholds (captured in the Ra and Rz parameters) for floor surface manufacturer and end-user utilization.

Sand blasting system: Ceramic tiles with both a polished finish and a matte finish were cut into 12’’ x 12’’ samples by the retailer. The porcelain enamel surface samples were cut into 9’’ x 10’’ or 9’’ x 12’’ samples based on the available size of the bathing surface utilized for measurement in this study. Once the samples were cut, they were sandblasted to systematically alter the surface roughness on the surface samples. The air pressure at the inlet and exposure time were the control variables in the sand blasting process used to vary the surface roughness. The exposure time of sand blasting were 2, 4, 6, and 7 minutes. The air pressure at the inlet for sand particles was 80 psi and the distance between the tile and the exit nozzle of the sand particles was 3 inches [3-5]. The nozzle was kept perpendicular to the sample surface. The particles that were used consisted of 99% to 100% coal slag. Four different processes, which consist of different combinations of exposure times and air pressures, have been selected for the sandblasting. 12 different surfaces were generated under the sand blasting process, with each sample type having four different levels of surface roughness in addition to its unaltered form creating 15 total surfaces for testing [16].

Surtronic duo profilometer: This instrument measures surface topography in numerous scales, including the two scales selected for this study, Ra (arithmetic mean deviation) and Rz (maximum height above the profile). These parameters were selected as they were indicated in Chang’s study to have correlations to transitional friction when compared to other surface parameters. Also, the Health and Safety Executive references the Rz parameter in their standard guidance on surface microroughness and friction (HSE, 2007). Surtronic Duo Profilometer is manufactured by Taylor Hobson manufacturers. The device model number is M112- 4952. The device serial number is 04916265C495. This device was inspected by the manufacturer on February 21st, 2020. The manufacturer certifies that the device measures ± 2%+0.004 μm. The users conducted a pre-calibration on a calibration plate designed to read 21.5 μm (Rz) and 5.81 μm (Ra) in order to ensure the accuracy in the device measurements. The device then was post-calibrated on the same calibration plate to ensure the device performed as expected throughout all of the test measurements [17].

Tribometers: The objective through using a tribometer is to simulate a slip on the floor surface to measure the maximum amount of available friction that may exist at the foot and floor surface interface. In this study the BOT 3000E, English XL VIT and the Slip-Test Mark IIIB tribometers were utilized. The three tribometers were used to compare the floor surface roughness and the surface’s slipperiness value as determined by the tribometer. All tribometer measurements were conducted based on the requirements outlined in ASTM F2508 and from the tribometer manufacturers.

English XL VIT: The English XL Variable Incidence Tribometer (VIT), is manufactured by Excel Tribometers. The test foot for the English XL was 32 mm in diameter and was made of Neolite. The resulting measurements for this machine were taken in “slip index,” or slip resistance index. For each floor surface test with the English XL VIT, the machine was operated until four slips were experienced. After each slip, the machine was rotated 90 degrees on the floor surface. The average reading when utilizing the Certified Test Foot Calibration Tile should be ± 0.03 accuracy according to the English XL VIT user guide. The XL measures surface traction on a scale having a range from 0.0 to 1.0, with values at the lower end indicating danger and values in the upper range showing increased degrees of safety. The threshold value for relative safety was highlighted for this study by the use of surface G as a threshold reference surface based on the human subject studies put forth (indicating that surface as a means for relative safety transition based on the frequency of slip events on that surface).

Slip-test mark IIIB: The Slip-Test Mark IIIB is a Variable Angle Tribometer manufactured by Slip-Test Inc. Slip-Test Inc. certifies that each tribometer model has undergone certification procedures in accordance with ASTM F2508. The precision results for the Slip-Test Mark IIIB are detailed based on the machine’s ability to rank order tiles A-D in accordance with ASTM F2508. The manufacturer of the Slip Test Mark IIIB describes the output as Transitional Coefficient of Friction (TCOF) as measured at the point of heel strike or test foot strike to full slip. The test foot for the Mark IIIB was 76 mm x 76 mm and consisted of Neolite material with 15 grooves running parallel to the slip direction mimicked by the machine. Measurement results obtained from the Slip-Test Mark IIIB are reported as Slip Resistance Values or SRV. The threshold value utilized by the Mark IIIB is the same as the threshold value utilized with the English XL, which was based on the Surface G test results for each tribometer.

BOT 3000E: The BOT-3000E is manufactured by Regan Scientific Instruments and is capable of testing static and dynamic coefficient of friction. The BOT-3000E is an automated tribometer which utilizes a physical printout as well as a digital file for the recording of verification tests, location, and friction measurements taken. For purposes of application in this study, the use of surface G as a threshold reference surface was utilized for this tribometer.

This BOT-3000E device was calibrated on November 30 , 2020, by TCNA (Tile Council of North America) an outside organization the Regan Scientific utilizes for tribometer validation. TCNA utilizes the test method for ASTM F2508-16 “Standard Practice for Validation, Calibration, and Certification of Walkway Tribometers Using Reference Surfaces”, and Section 16.2.2 Interlaboratory Study (ILS) according to ASTM E691.

Tribometer users: The study utilized four tribometer users to conduct all testing. A second individual was utilized to record the tribometer results for each floor surface, while a third individual monitored the testing protocol to ensure consistent technique and correct recording among all tribometer types utilized in this study. Floor surface order was randomized across users testing for available friction with the tribometers.

Floor surface specifications: One floor surface utilized for this study was the porcelain enamel surface. This surface was manufactured by American Standard and cut from the “Princeton 60 inch by 30 inch Integral Apron Bathtub,” Model Number: 2391202.020. In order to minimize tribometer error associated with dual concavity, the side apron portion of the bathtub was cut and utilized for testing. The other two floor surfaces utilized in the study were flat ceramic-based tiles utilized in [2]. Tile E and Tile G. Tile E was a polished ceramic-based surface while Tile G has a matte finish. These surfaces were selected for numerous reasons. Firstly, at the time of writing, these tiles are not yet officially included into ASTM F2508 but are likely to be the new reference surfaces utilized in that standard. Also, it is important that the surface maintains its characteristics over time in order to maintain the efficacy of any future application of these surfaces and the results from this study. Tile E and Tile G were selected specifically due to their likelihood for consistency in future research applications, as well as their slip risk indications from prior study Blanchette, et al., Tile G indicated no heel slips and one toe slip out of 37 human trials, and Tile E indicated 16 heel slips and 14 toe slips out of 37 human trails. Therefore, the tribometer test results in this study on these same floor surfaces were utilized as a reference for a slippery floor surface (Surface E) and a transitional floor surface (Surface G) [2], (Table 1).

Table 1: Floor surface specifications.

Experimental procedure

This study utilized three tribometers (BOT 3000E, English XL VIT, and Slip-Test Mark IIIB) to measure the floor surface available friction (expressed in different slipperiness values as determined by each individual manufacturer). The contaminant utilized on the floor surfaces in this study was distilled water for the English XL and Mark IIIB, a SLS solution for the BOT 3000E, based on the manufacturer’s guidelines. Four repeated friction measurements were taken under wet conditions for each surface by each of the four tribometer users testing with each tribometer. Four readings were taken with each tribometer, one per cardinal direction, based on the tribometer manufacturer’s guidelines as well as in accordance with ASTM F2508. All users were given a copy of the manufacturer’s operation instructions prior to the use of each machine, and were required to read those instructions. Between tests, each floor surface was wiped clean with a microfiber towel. The porcelain enamel surface was clamped down using a Pony 10-in Hand screw Clamp made by Jorgensen, and an 18-in bar clamp made by Jorgensen in order to validate the surface was flat. The value measured by each tribometer was assumed to represent the available friction measured by that tribometer unless the manufacturer specified the required use of a conversion factor [15]. Due to time constraints and user/ resource availability, the BOT 3000E was not utilized for testing on the bathing surface. A surface profilometer (Surtronic Duo) was used to make ten parallel surface roughness measurements, 10 mm apart in the direction of the friction measurements, at ten different locations of each floor surface sample. The vertical resolution of this profilometer is 16 nm. The tip of the stylus has a radius of 5 μm. The nominal force on the stylus is 100 mg. The cut-off length utilized for this study was 0.8 mm, with a traverse length of 5 mm. This measurement length was selected based on the study findings from Chang which highlighted 0.8 mm as a length with strong correlation with transition friction. The surfaces were measured in both Ra and Rz roughness scales. These were selected as they were indicated in Chang’s study to be relatively good indicators for slip resistance and roughness correlation. The device was calibrated both before and after the data is collected to ensure device accuracy and repeatability.

Data analysis: The data from the study was analyzed using the IBM SPSS Version 27 software package. The data was initially checked for accuracy of recording into the software as well as checked for missing data. To determine significance an alpha ( ) level of .05 was used.

Descriptive statistics: All variables were characterized(e.g., tribometer type, floor surface type, user, and surface roughness parameter (both Ra and Rz were measured)) using mean, median, range, minimum, maximum, and standard deviation on all of the floor surfaces. Descriptive statistics were used to assess at which roughness level (measured in Ra and/or Rz) do all three tribometers define the surface as non-slippery when wet for each floor surface type.

Linear regression: A linear regression was performed to test the hypothesis as shown in Table 2 to determine which independent variables in the hypothesis are significant predictors of slipperiness for each individual tribometer.

Table 2: Null and alternative hypothesis.

The following assumptions were tested for the linear regression:

Assumption #1: The dependent variable is measured on a continuous level.

Assumption #2: The independent variable is measured at the continuous level.

Assumption #3: There is independence of observations, and the dependent variable should have mutually exclusive and exhaustive categories.

Assumption #4: There needs to be a linear relationship between any independent variables and the dependent variable.

Assumption #5: The data needs to show homoscedasticity.

Assumption #6: The data must show normality in the residuals, or that the residuals of the regression line are approximately normally distributed. None of the linear regression model assumptions were violated (Table 2).

Results

To first determine the correlation between Ra and Rz, a Spearman’s rho correlation test was run. The results show that Ra and Rz are highly correlated (r=.976) indicating that they are likely to evaluate similar characteristics of the property being evaluated. As a result, Rz and Ra were tested separately to determine which one is the better measure of available friction as measured by each tribometer.

BOT 3000E linear regression

To test whether user, floor surface type, and roughness measurements (Ra or Rz) affect the available friction measurement from the BOT 3000E tribometer, a linear regression analysis was performed twice (one with Rz and the other with Ra). The assumptions of linear regression were tested and found to be met. The ANOVA results for the BOT 3000E Regression using, Roughness (Ra), User, and Tile Type as predictors reveals that the regression is statistically significant (p-value<.001) with an adjusted R2 of 37.1%. This means the three predictors together can explain 37.1% of the variability in available friction. The Coefficients results using the same variables indicate that tile type and user do not have a statistically significant effect on available friction (p-value ranged from .427 to 1.000 for user and was .461 for tile type); however, Ra roughness measurement is a statistically significant predictor (p-value<.001). As a result, the linear regression was run again with only Ra as a predictor. The ANOVA results for Roughness (Ra) revealed that the regression is statistically significant (p-value<.001) with an adjusted R2 of 39.2% (Table 3). This means the predictor (Ra) can explain 39.2% of the variability in available friction which is a higher value than the three predictors tested together. The Coefficients results indicate that Ra roughness measurement is a significant predictor of slipperiness (p-value<.001). When the regression was ran again using Rz instead of Ra, the ANOVA results for the BOT 3000E, Roughness (Rz), User, and Tile Type reveal that the regression is statistically significant (p-value<.001) with an adjusted R2 of 40.3%. This means the three predictors can explain 40.3% of the variability in available friction when using Rz. The Coefficients results indicate that tile type and user do not have a statistically significant effect on available friction (p-value ranged from .416 to 1.000 for user and was 0.585 for tile type); however, Rz roughness measurement has a statistically significant effect (p-value<.001). The linear regression in this case was run again with only Rz as a predictor. The ANOVA results for Roughness (Rz) reveal that the regression is statistically significant (p-value<.001) with an adjusted R2 of 42.5%. This means the predictor (Rz) can explain 42.5% of the variability in available friction which is a higher value than the three predictors tested together. When comparing Ra and Rz, while both measures of floor surface topography are significant predictors of available friction with the BOT 3000E, the adjusted R2 result for Ra (about 39%) was able to predict slightly less variability when compared to the Rz result (about 43%) for the BOT 3000E tribometer. By removing the other variables, it can be determined that Rz can predict about 43% of the variability in available friction readings by the BOT 3000E, which is a 4% improvement when compared to Ra.

Table 3: Roughness and available friction results (average by tribometer) by tile type.

English XL linear regression

To test whether user, floor surface type, and roughness measurements (Ra or Rz) affect the available friction measurement with the English XL tribometer, a linear regression analysis was performed twice (one with Rz and the other with Ra). The assumptions of linear regression were tested and found to be met.

The ANOVA results for the English XL Regression using, Roughness (Ra), User, and Tile Type as predictors reveals that the regression is statistically significant (p-value<.001) with an adjusted R2 of 83.0%. This means the three predictors can explain 83.0% of the variability in available friction. The Coefficients results indicate that tile type and user do not have a statistically significant effect on available friction (p-value ranged from 0.320 to 0.640 for user and was 0.390 for tile type); however, Ra roughness measurement is a statistically significant predictor (p-value<.001). As a result, the linear regression was run again with only Ra as a predictor. The ANOVA results for Roughness (Ra) reveal that the regression is statistically significant (p-value<.001) with an adjusted R2 of 82.1%. This means the predictor (Ra) can explain 82.1% of the variability in available friction which is about the same when compared to the three predictors tested above. The Coefficients results indicate that Ra roughness measurement is a significant predictor of slipperiness (p-value<.001). When the regression was ran again using Rz instead of Ra, the ANOVA results for the English XL, Roughness (Rz), User, and Tile Type reveal that the regression is statistically significant (p-value<.001) with an adjusted R2 of 79.4%. This means the three predictors can explain 79.4% of the variability in available friction.

The Coefficients results indicate that tile type and user do not have a statistically significant effect on available friction (p-value ranged from 0.366 to 0.670 for user and was .633 for tile type); however, Rz roughness measurement has a statistically significant effect (p-value<.001). The linear regression was run again with only Rz as a predictor. The ANOVA results for Roughness (Rz) reveal that the regression is statistically significant (p-value<.001) with an adjusted R2 of 79.5%. This means the predictor (Rz) can explain 79.5% of the variability in available friction which is a slightly lower value than the three predictors tested together. The Coefficients results indicate that Rz roughness measurement was found to have a statistically significant effect (p-value<.001). When comparing Ra and Rz, while both measures of floor surface topography are significant predictors of available friction with the English XL, the adjusted R2 result for Rz (about 80%) was able to predict slightly less variability when compared to the Ra result (about 82%). By removing the other variables, it can be determined that Ra can predict about 82% of the variability in available friction readings by the English XL.

Mark IIIB linear regression

To test whether user, floor surface type, and roughness measurements (Ra or Rz) affect the available friction measurement from the Mark IIB tribometer, a linear regression analysis was performed twice (one with Rz and the other with Ra). The assumptions of linear regression were tested and found to be met. The ANOVA results for the English XL Regression using, Roughness (Ra), User, and Tile Type as predictors reveal that the regression is statistically significant (p-value<.001) with an adjusted R2 of 54.6%. This means the three predictors can explain 54.6% of the variability in available friction. The Coefficients results indicate that tile type and user do not have a statistically significant effect on available friction (p-value ranged from .398 to .739 for user and was .665 for tile type); however, Ra roughness measurement is a statistically significant predictor (p-value<001). As a result, the linear regression was run again with only Ra as a predictor. The ANOVA results for Roughness (Ra) reveal that the regression is statically significant (p-value<.001) with an adjusted R2 of 55.7%. This means the predictor (Ra) alone can explain 55.7% of the variability in available friction which is about the same when compared to the three predictors tested together. The Coefficients results indicate that Ra roughness measurement is a significant predictor of slipperiness (p-value<.001).

When running the regression again using Rz instead of Ra, the ANOVA results shown in Table 2 reveal that the regression is statistically significant (p-value<.001) with an adjusted R2 of 56.3%. This means the three predictors can explain 56.3% of the variability in available friction. The Coefficients results indicate that tile type and user do not have a statistically significant effect on available friction (p-value ranged from .389 to .734 for user and was .590 for tile type); however, Rz roughness measurement has a statistically significant effect (p-value<.001). The linear regression in this case was run again with only Rz as a predictor. The ANOVA results for Roughness (Rz) reveal that the regression is statistically significant (p-value<.001) with an adjusted R2 of 57.8%. This means the predictor (Rz) can explain 57.8% of the variability in available friction, which is a higher value than the three predictors tested together. The Coefficients results indicate that Rz roughness measurement was found to have a statistically significant effect (p-value<.001).

In summary, when comparing Ra and Rz, while both measures of floor surface topography are significant predictors of available friction with the Mark IIIB, the adjusted R2 result for Ra (about 56%) was able to predict slightly less variability when compared to the Rz result (about 58%). By removing the other variables, it can be determined that Rz can predict about 58% of the variability in available friction readings by the Mark IIIB (Figure 1).

Figure 1: Tribometer standard deviation result by surface. Note: : BOT 3000E AVG; : English XL AVG; :Mark IIIB AVG.

Tribometer results

As shown in Table 3, the average standard deviation for the BOT 3000E and Mark IIIB was 0.03, while the English XL was 0.04. The greatest standard deviation on all tribometers collectively was on surface E1, which was also the surface with the lowest available friction measured by all tribometers in the study. These measures are indications of repeatability and reproducibility of the aggregated results on each of these floor surfaces. Based on the results highlighted in Table 4, the roughness levels which all tribometers agreed that a surface was non-slippery when wet varied based on each individual surface type. Using the results from surface G1 as a reference for a relative safety threshold for each tribometer, the following roughness thresholds were measured. For Tile G, the level where all tribometers found the surface transition from slippery to non-slippery was at surface G1 (the reference surface for the study), indicating a threshold value of 1.57 µm Ra, and 8.6 µm Rz or greater for relative safety. For Tile E, the range where all tribometers found the surface transition from slippery to non-slippery was between 1.14 µm- 5.95 µm Ra, and 8.00 µm-32.00 µm Rz. The lower end of the range is where one or more of the tribometers found the surface to be relatively non-slippery, while the higher end of the range is where all tribometers agreed the surface was non-slippery. This indicates that greater than 5.95 µm Ra and 32.00 µm Rz (higher results) would provide for a conservative threshold estimate for relative safety. In a similar manner, for the Porcelain Enamel (PE) surface, the range where all tribometers found the surface transition from slippery to non-slippery was 0.15 µm-3.78 µm Ra, and 0.8 µm-22.30 µm Rz. This indicates that greater than 3.780 µm Ra and 22.30 µm Rz would provide for a conservative threshold estimate for relative safety (Table 4). Tables 5 and 6 represent the results from the surface topography measurements (captured in Ra and Rz) which were repeated 10 times on each surface. Table 6 was developed to provide guidance for ranges of surface roughness and available friction relative slippery/nonslippery thresholds. The value highlighting the greatest available friction measurement from the range of all three tribometers was utilized as guidance for the table (Tables 5-6).

Table 4: Floor surface topography measured in Ra.

Table 5: Floor surface topography measured in Rz.

Table 6: Surface roughness threshold guidance by surface

Discussion

The purpose of this study was to analyze three different floor surfaces (ceramic tile with a polished finish, ceramic tile with a matte finish, and a porcelain enamel bathing surface) with varying degrees of floor surface micro roughness measured in both the Rz and Ra roughness parameters to highlight relative safety thresholds for floor surface manufacturer and end-user utilization. The study results identified that tile type and user were not significant predictors of available friction, while surface parameters Rz and Ra were the only significant predictors of available friction for all tribometers. These results agree with Chang, et al., that there is a strong correlation between Ra and Rz and available friction. These findings support the utilization of surface micro roughness parameters Ra and Rz as a relative means for predicting available friction [4]. This indicates that surface micro roughness measurement in Ra or Rz could be utilized as another data point/measurement to give a general indication of floor surface slip potential. Surface roughness can act as a complement to tribometer measurement to further analyze floor surface slip risk based on its correlation to available friction. However, a caveat to consider when looking at the implications of Ra and Rz as a measure of available friction is that both Ra and Rz are measures of surface topography that highlight specific parameters on a floor surface Chang, et al., For example, a floor surface that has sharp pointed peaks or sharp pointed valleys could have the same Ra or Rz values and therefore, the implications for the utilization of Ra or Rz as a standalone indicator of available friction may not be as accurate and effective as expected [4,5].

Chang, et al., highlights the use of one single surface parameter as a means for standalone friction prediction in a glycerol environment as limited and inadequate, even given the larger 8 mm cut-off length compared to the 0.8 mm cut-off length utilized in this study (longer cut-off length is correlated to higher prediction of friction, but is not accessible in most surface profilometers outside of a laboratory setting [6]. However, the use of multiple surface parameters which have each individually been shown to correlate to available friction may be sufficient to predict available friction Chang, et al Therefore, the use of both Ra and Rz measurement together, as a complement to tribometer measurement, may be optimal for available friction prediction/ slip risk indication, when limited to measurement outside of a laboratory environment [7]. Since the correlation between Ra and Rz is so high (r=.976), it can also be expected that there may be a limited improvement in the available friction prediction when using both of these parameters concurrently, compared to using a single parameter which was found to have the highest correlation with available friction based on the tribometer type identified in this study. The surface roughness parameters Ra or Rz were able to explain between 39% to 83% of variability in available friction measurements, depending on which tribometer was used. The predictor surface roughness Rz was a slightly greater (3% more) predictor for Mark IIIB and BOOT3000E while the Ra was a slightly greater (3% more) for the English XL based on the percentage of variability in available friction explained (Adjusted R2 value). This slight variability between Ra and Rz could be acceptable due to equipment and measurement errors [17].

The wide range of unexplained variability especially for the BOT3000E and Mark IIIB tribometers (since Ra or Rz were able to only explain around 50% of the variability) could be due to multiple variables. One variable that could have contributed to the unexplained variability is the lack of intra-tile homogeneity of the ceramic tiles tested in this study, supporting that the consistency of ceramic tile is subject to error associated with its homogeneity within the same tile sample. Siegmund, et al., highlighted that on surface G, variability between different tile samples of the same surface in the same lot accounted for between 6% and 77% of the variance in the Mark IIIB data, and the different samples of surface E accounted for between 1% and 31% of the variance in the English XL results [17].

Another factor which can increase the study error within the results is the variability within the tribometer test result accuracy. There are numerous known factors that contribute to the variability of tribometer accuracy including variability between specific tribometer measuring units, tribometer condition and function (such as wear and moving parts), operator performance, and test foot type (such as material, shape, and groove pattern) Siegmund, et al., In this current study, only one unit per tribometer was utilized by each user, which does not allow for an assessment of the user-unit interaction as the same user only was able to use one unit per tribometer type [17]. This was significant because test result implications for any specific tribometer was determined by only one unit of that tribometer, rather than a set of units. This restricts the ability for the variable user to be separated from the variable tribometer unit being utilized by that user. Siegmund, et al. showed that between 51% and 82% of the total variance in tribometer measurement was explained by the user-unit interaction, with 40%-82% of that variance being from the user, further highlighting a potential source of variability in this current study [17]. However, the linear regression results in this current study for all three tribometers revealed that user was not a significant source of variability in any of the tribometer test results, which is contradictory to the results highlighted by Siegmund, et al. One explanation for this finding could be the limited number of users as well as the aforementioned limitation related to the lack of units utilized in this study. Another explanation of the variability between different tribometer’s explanation of microroughness could be due to the variability in the test foot, especially for the English XL when compared to the Mark IIIB [17].

Localized hydroplaning could have been more prevalent in the English XL testing because there are no groves in the test foot, compared to the Mark IIIB which had a 15-groove test foot by Persson. Also, the size of the English XL (32 mm in diameter) test foot is significantly smaller when compared to the Mark IIIB test foot (76 mm x 76 mm) which could have contributed to the differences in the variability explained by the two tribometers [14]. From a repeatability and reproducibility standpoint, the average standard deviation for the BOT 3000E and Mark IIIB was 0.03, while the English XL was 0.04 with the greatest being on surface E1 for all tribometers collectively. One area that has been highlighted as a source of variability in both the ILS and this study was that the use of distilled water as a contaminant can create challenges to ensuring a continuous unbroken film of water under the test foot. This inconsistency of unbroken film under the test foot factors could be one explanation for the increased standard deviation for some surfaces compared to the ILS reported findings and therefore less consistency in the tribometer test method process Chimich, et al. Also, surface porosity and permeability could impact the amount of liquid contaminant on the floor surfaces measured and therefore relate to some degree of variability in the tribometer results [9].

The surface roughness threshold guidance varies depending on the type of surface. For Tile G, the level where all tribometers found the surface transition from slippery to non-slippery was at surface G1 (the reference surface for the study), indicating a threshold value of 1.57 µm Ra, and 8.60 µm Rz or greater for relative safety. For Tile E, the range where all tribometers found the surface transition from slippery to non-slippery was between 1.14 µm-5.95 µm Ra, and 8.00 µm-32.00 µm Rz. This indicates that greater than 5.95 µm Ra and 32.00 µm Rz would provide for a conservative threshold estimate for relative safety. For the porcelain enamel surface, the range where all tribometers found the surface transition from slippery to non-slippery was 0.15 µm-3.78 µm Ra, and 0.80 µm-22.3 µm Rz. This indicates that greater than 3.78 µm Ra and 22.3 µm Rz would provide for a conservative threshold estimate for relative safety for these surfaces. These findings are much lower than the results from Kim, et al. where an available friction threshold of 0.4 was utilized by a pendulum-type hydraulic dynamic friction tester [11]. Their findings identified a lower bound Ra value of 17 µm; however, these results cannot be directly compared to this current study results due to the use of soapsuds environment in their study while distilled water was utilized in this study for the English XL and Mark IIIB and a SLS solution for the BOT 3000E.

The wide range of variance in safety threshold values identified between floor surfaces in this study could be attributed to the lack of values within the ranges highlighted in surfaces E and PE as significant threshold ranges. For surface E, this study did not utilize any samples within the wide range of 1.14 µm-5.95 µm Ra, and 8.0 µm-32.0 µm Rz, which minimized/limited the specificity that could be highlighted by this study for application to a specific threshold value or a smaller threshold range. For surface PE, a similar issue existed due to the lack of surfaces tested within the roughness range that was found to be significant of 0.15 µm- 3.78 µm Ra, and 0.8 µm-22.3 µm Rz. However, since surface G was utilized as a reference surface for relative safety transition in this study, this created a value that was set for a relative safety threshold on that surface (1.57 µm Ra, and 8.6 µm Rz).

The HSE (2007) established a slip potential classification chart, that classifies a surface microroughness (Rz) that is less than 10 µm as a high slip risk, while between 10 µm-20 µm as a moderate slip risk, and greater than 20 µm as a low slip risk. However, this current study highlights relative slip risk as a function of the tribometers being used based on the threshold values highlighted by use of surface G1 as a reference surface with each tribometer. Therefore, the results from this study specify a roughness threshold range and conservative value for relative safety rather than slip potential classifications established by HSE (2007). This current study also highlights specific safety threshold guidance which varies for each of the three surface types assessed, rather than the general slip potential classification for all surfaces (one fits all approach) provided by the HSE. This finding is significant since it identified a range of about 4 μm for Ra and 23 μm for Rz of variance in safety threshold guidance between surfaces assessed in this study, compared to the general slip potential classifications highlighted by the HSE (2007). When utilizing surface G1 as a basis for a relative safety threshold, tribometer sensitivity implications for surface microroughness can be reviewed. With the manufacturer baseline measurement of .34 with the English XL, .59 with the Mark IIIB, and .47 with the BOT3000E as the basis for safety transition (based on G1 results compared to human subjects testing from Blanchette, et al., there was an increase in available friction measured by the English XL from .34 to .53 (56% increase), from .59 to .71 (20% increase) for the Mark IIIB, and from .47 to .66 (40% increase) for the BOT 3000E. Following this initial increase in G1, the BOT3000E was found to have inconsistent results between the remaining G2-G5 surfaces with varying levels of surface roughness [2]. For the Mark IIIB, following the initial increase of 20% above the threshold value for safety, none of the other G surfaces (G2-G5) produced a result that it differentiated by more than ± 0.02 TCOF. However, the English XL measured an increase from G2 to G3 by 8% increase (.53 to .57) and an increase from G3 to G4 by 33% increase (.57 to .76). When reviewing the results for surface E1, a surface that was assessed in Blanchette, et al., and found to have a 16 heel slips and 14 toe slips over 37 human trials, it can be determined that the results from the tribometers on E1 would highlight a surface that has a high human slip risk [2]. Then, as the surface microroughness on surface E1 (0.05 μm Ra and 0.5 μm Rz) increased to E2 (1.14 μm Ra and 8.0 μm Rz), a rise in available friction was measured by all tribometers, but the percent increase varied by tribometer (88% increase with the BOT3000E, 15% increase with the English XL, 62% increase with the Mark IIIB). However, when microroughness was increased from E2 (1.14 μm Ra and 8.0 μm Rz) to E3 (5.95 μm Ra and 32.0 μm Rz), tribometer available friction results varied by tribometer type (9% decrease with the BOT3000E, 95% increase with the English XL, 3% increase with the Mark IIIB). These values between E2 and E3 highlight the relative safety transition point for available friction based on the English XL and Mark IIIB results, with similar available friction results to G1 results. This could highlight a safety threshold indicator for microroughness on surface E to be just below the levels measured on surface E3.

The increase in the English XL available friction results was more consistent throughout the roughness ranges assessed in this study when compared to the Mark IIIB (which did not distinguish the roughness levels as consistently as the English XL did with measurements greater than the relative safety threshold). The BOT 3000E test results which were higher than the safety threshold, were inconsistent. These results suggest a potential increase in tribometer sensitivity for microroughness above the relative safety threshold for the English XL compared to the Mark IIIB and BOT3000E.

However, this study did not assess test foot size, groove design, force application, test foot angle, among other variables as a means to directly compare the tribometers, therefore this suggested increase in tribometer sensitivity would need further assessment for future application. This study also did not evaluate the influence of mechanical operating principles of each individual tribometer and their interaction with surface microroughness.

Conclusion

Surface micro roughness in both Ra and Rz parameters were found to be significant predictors of available friction on all surfaces measured by all three tribometers. The predictor surface roughness (Rz) was able to explain 56% of the variability in available friction measured by the Mark IIIB and 43% of the variability in available friction measured by the BOT3000E. The predictor surface roughness (Ra) was able to explain 82% of the variability in available friction measured by the English XL. This furthers the use of surface micro roughness parameters Ra and Rz to be used as a relative means for predicting available friction. The results from this study also indicate surface roughness relative safety threshold guidance which varies for each of the three surface types assessed. While there are limitations to tribometer measurement and surface microroughness measurement both of these methods have been correlated as a unique indicator for slip risk prediction. With each method having varying degrees of precision and error, the use of these techniques can be applied generally as complementary methods for a more comprehensive assessment of walkway surface slip risk.

Limitations

The study findings should be interpreted with the consideration of its limitations. First, the limited number of tribometer measurements on each surface, and thus the potential variability in tribometer test results must be noted. This study utilized four users on every surface with only one unit per tribometer type. The study was conducted using three common tribometers, however tribometers are known to have high levels of variability, especially given the single laboratory environment with only one unit per tribometer.

References

1. American society of testing and materials, F3132-Standard practice for selection of walkway surfaces when considering pedestrian safety. 2023.
2. Blanchette MG, Lee-Confer J, Brault JR, Rutledge B, Elkin BS, Siegmund GP. Human slip assessment of candidate reference surfaces for walkway tribometer validation: An update to standard ASTM F2508. Journal of Testing and Evaluation. 2021;50(2):20210240.

   [Crossref] [Google scholar]

3. Chang W. The effect of surface roughness on the measurement of slip resistance. Int. J. Ind. Ergon. 1999;24(3):299-313.

   [Cross ref] [Google scholar]

4. Chang W, Kim I, Manning D, Bunterngchit Y. The role of surface roughness in the measurement of slipperiness. Ergonomics. 2001;44(13):1200-1216.

   [Crossref] [Google scholar] [PubMed]

5. Chang W. The effects of surface roughness and contaminants on the dynamic friction between porcelain tile and vulcanized rubber. Safety Science. 2002;40(8):577-591.

   [Cross ref] [Google scholar] [PubMed]

6. Chang W, Hirvone, M, Grönqvist R, Aguilera W. The Effects of Cut-Off length on surface roughness parameters and their correlation with transition friction. Human Factors. 2004;46(19):1713-1717.

   [Crossref] [Google scholar]

7. Chang W, Matz S, Grönqvist R, Hirvonen M. Linear regression models of floor surface parameters on friction between Neolite and quarry tiles. Applied Ergonomics. 2010;41(1):27-33.

   [Crossref] [Google scholar] [PubMed]

8. Chang, Wen-Ruey, Sylvie Leclercq, Thurmon E, Lockhart , Roger Haslam. State of science: occupational slips, trips and falls on the same level. Ergonomics. 2016;59(7):861-883.

   [Cross ref] [Google scholar] [PubMed]

9. Chimich DD, Al-Salehi L, Elkin BS, Siegmund GP. Contaminant film thickness affects walkway friction measurements. Frontiers in public health. 2022;10(1);915140.

   [Cross ref] [Google scholar] [PubMed]

10. An examination of novel roughness parameters to be used in conjunction with the HSE Slips Assessment Tool (SAT). Health and Safety Executive. 2017;3(1):1-34.

    [Google scholar]

11. Kim, Hsiao, Simeonov. Functional levels of floor surface roughness for the prevention of slips and falls: Clean-and-dry and soapsuds-covered wet surfaces. Applied Ergonomics. 2013;44(1):58-64.

    [Crossref] [Google scholar] [PubMed]

12. Liberty Mutual Research Institute for Safety. Workplace Safety Index. 2023.

    [Google scholar]

13. National estimates of the 5 leading causes of Nonfatal injuries treated in hospital emergency departments, United States. NEISS. 2013.

    [Google scholar]

14. Persson BNJ. Sliding on lubricated surfaces. In: sliding friction. NanoScience and Technology. 2000.

    [Google scholar]

15. Powers, Christopher M, Mark G, Blanchette, John R. Brault. Validation of walkway tribometers: Establishing a reference standard. Journal of Forensic Sciences. 2010;:50(2):366-370.

    [Crossref] [Google scholar] [PubMed]

16. Redfern MS, Cham R, Gielo-Perczak K, Grönqvist R, Hirvonen M., Lanshammar H. Biomechanics of slips. Ergonomics. 2001;44(13):1138-1166.

    [Crossref] [Google scholar] [PubMed]

17. Siegmund G, Blanchette M, Brault J, Chimich D, Elkin B. Quantifying the uncertainty in tribometer measurements on walkway surfaces. Ergonomics. 2021;64(3):396-409.

    [Crossref] [Google scholar] [PubMed]