J. Phys. D: Appl. Phys., Vol. 12, 1979. Printed in Great Britain

Measurement of the coefficient of friction of floors R Brough, F Malkin and R Harrison British Ceramic Research Association, Queen's Road, Penkhull, Stoke-on-Trent ST4 7LQ Received 2 November 1977, in final form 16 August 1978 Abstract. An improved method of measuring the coefficient of sliding friction of floor products is described, whereby similar measurements are made on laboratory samples and floors. An average value over the surface and a trace showing the local variations in friction across the surface are obtained. The latter facilitates studies of the efficiency of geometric projections and other surface effects that are intended to provide resistance to slipping. Measurements on floors show that both organic and inorganic deposits contribute to a reduction of the coefficient of friction.

1. Introduction

Although some ceramic floor tiles have been in fixed positions where they have been walked on for hundreds of years, it is only comparatively recently that there has been a demand to relate frictional properties (the term most often used is slip-resistance) to particular areas of use. This has principally been required for industrial and recreational buildings. Ceramic tiles and other floor products with surface projections which are designed to provide resistance to slipping must also provide for ease of cleaning, drainage and even foot comfort. Both the geometry and the spacing of the surface projections are relevant to the required properties. The textures of the plane surfaces of unglazed ceramic floor products exhibit insignificant variations within each of the two main types, when examined either by means of a Talysurf or an electron microscope. Quarries are produced by extruding plastic clay bodies through steel orifices. Vitrified tiles are shaped from fine grains in steel dies at high pressure. They are fired at temperatures over 1100"C,and most vitrified tiles and a high proportion of the quarries have pore volumes of about 4%. An overriding consideration in choosing the method of measurement has been the need to agree with the known characteristics of surfaces in service. Whilst the true coefficient of friction can be measured over any convenient length of plane surface, the effective coefficient of friction has to include the resistance to motion of the surface projections. The disadvantage of many methods of measurement that have previously been adopted for floor coverings is that the sliders only make contact with the peaks of the geometrical projections in cases where penetration by heels of shoes to the lower plane surface has been shown to occur in practice. A system favoured by some workers utilises a whole shoe (Pangels 1962). One method of employing this system involves a person standing on an inclined plane which is tilted gradually until the angle is reached at which sliding starts. Improvements require the person to walk up and down the incline. 0022-3727/79/040517+ 12 $01.00

01979 The Institute of Physics

517

518

R Brouglz, F Mallcin and R Harrison

Another system employs the measurement of the frictional resistance to a slider mounted on the end of a pendulum arm. This is of particular interest, as a machine developed for comparison of wet road surfaces (Road Research Laboratory 1960) has been used by some workers to measure the coefficients of friction of floor tiles, both wet and dry. Useful reviews of techniques of measuring factors concerned with slipping are available (Levandowsky and Jacqmarcq 1970, Brungraber 1976). The techniques for the measurement of the coefficient of friction in these generally fail to make the proper distinctions between different types of relief surface.

2. Observations of walking action

Flat foot contact, typified by a person standing 011 an inclined plane, is clearly a state from which people rarely slip in practice. There are considered to be three actions which are principally involved when people slip during walking. The heel may slip forward on making contact with the ground, the toe may slip back in the action of pushing the body forward or, during a turning movement, slipping may be initiated from the ball of the outer foot. It was judged that heel slip was our primary concern. In order to determine parameters on which the design of the instrument is based we had therefore to examine the manner in which feet are presented to the floor during walking and running. Of particular concern was the horizontal speed of the heel at contact. Films were made at 64 fps of 30 different children and adults silhouetted against an illuminated screen graduated vertically and horizontally. The subjects were filmed walking and running, shod and barefoot, and in dry and wet conditions. The paths of heels and toes were plotted and these were used to determine the horizontal speeds and positions of feet through the whole process of walking. Typical examples of the plots of heel and toe movements are illustrated in figure 1. Each person was found to have a characteristic walk. Evidence of both heel and toe-slip was observed in the films. The

lal

L e f t foot

-

1b1

/

/

//

0 '

h

250

200

-

150 100 Direction of walk lcml

50

0

Figure 1. (a) Toe and (b) heel movements during walking. Subject, height 6 ft 1 in, stride 2.94 ft, walking speed 4.2 mph, maximum speed of toe 16.4 mph, of heel 14.9 mph.

Coejicient of.friction offloors

519

general pattern of behaviour enabled a judgment to be made of the geometry of the first contact of the heel. Analysis of the films showed that the average angle of contact of the heel was 21 O with a range of 11 to 32 '. Large variations were evident in the speeds of foot movement; even two persons travelling at the same approximate forward speed might exhibit different speeds and paths with heels and toes. Some attempts were made to simulate slipping and it was evident that slip did not occur from a static position and that, although initiation was at a slow speed, there was rapid acceleration. Two speeds were chosen as representative, 0.3 and 7.4 cm s-1. Further information was available on the changing area of contact (Building Research Station 1961). The surprising feature is how small is the area of contact at any instant. The area of contact and the loading were chosen to reproduce the approximate average pressure of the heel between first contact and removal of the rear foot from the ground.

3. Apparatus and procedure 3.1. General

The essential features of shoe-floor contact having been determined, adaptation to the measurement of sliding friction demanded an apparatus providing the intimate contact of a small slider at slow speed. Dr D Tabor and his colleagues had suggested that the Eldredge apparatus (Childs and Tabor 1967-8, Tabor 1956) might form a suitable basis, and it was established that this could be modified to have the characteristics that were necessary for the study of slip between shoe and flooring materials.

Figure 2. Diagram of the balanced arm of the apparatus. 1, direction of motion; 2, slider; 3, load; 4, leaf springs; 5 , transducer.

520

R Brough, F Malkin and R Harrison

The apparatus was developed in two forms: a laboratory model in which the surface under test moves on a trolley underneath the slider, and a floor model in which the slider moves over the surface. The balanced arm (figure 2) which holds the slider has a low-friction pivot allowing vertical movement. The direction of motion (1) of the slider in relation to the surface is indicated. The end holding the slider ( 2 ) under a load (3) has metal leaf springs 19 mm deep and 0.5 mm thick (4) which are deflected as the slider moves against surfaces. The lateral displacement, related to the tangential force on the slider, is determined by a linear transducer (5) connected to an X- Y recorder. The other coordinate is provided by the distance traversed across the surface. This is measured by means of a contact moving along a wound resistance. In addition to the trace showing the variation in coefficient of friction across the test surface, the signals are relayed to an integrator which gives the average coefficient. Each form of the apparatus can be used to measure over three lengths, 10, 20, or 40cm, determined by four suitably spaced microswitches. Wherever possible, the longest distance is used and this is equivalent to the distance across three 6 x 6 in vitrified floor tiles or quarries, and joints. The shortest distance is useful for examining small samples removed from site or specimens treated for wear, chemical attack or contamination in the laboratory. The normal load on the slider produces a pressure of 320 g cm-2. Although measurements are usually made at 0.3 cm s-1 and 7.4 cm s-1, other speeds in the range up to 18.45 cm s-1 may be obtained in the apparatus. 3.2. Sliders A small dimension (9 mm diameter) was originally adopted because this relates to the contact area of heels in practice, and a small slider moves easily on to fresh surface. Initially, hemispherical sliders providing an approximate contact angle of 21 O were used. When sliders of these dimensions were taken by hand over ceramic tiles and other materials having geometric projections, it was established that penetration to the lower plane surface was occurring in just those cases where observations in service and the film studies showed similar contact by heels. Certain rough-textured surfaces and some with geometric projections having narrow spacing could not be penetrated by the small sliders. This again corresponded with practical experience. Table 1. Effect of slider angle on coefficient of friction. ~~

Tile

Slider

~~~~

~~

Coefficient of friction TRRL rubber

Leather

Dry

Dry

_-

Wet

Slow

-_ A

B C

D

Flat N2 1.50 Angled 1.85 Flat 1.35 Angled Flat 1.80 1.40 Angled 1.75 Flat 1.45 Angled

Wet

___

___~_

Slow Fast

Slow

Slow Fast

0.66 0.53 0.97 0.82 0.94 0.86 0.96 0.88

0.73 0.61 0.70 0.62 0.82 0.71 0.79 0.72

0.55 0.53 0.71 0.67 0.78 0.73 0.77 0.75

__

0.46 0.38 1.07 0.83 1.10 0.93 1.12 0.95

0.73 0.58 0.72 0.58 0.77 0.67 0.76 0.69

Coeficient of friction offloors

52 1

The hemispherical sliders were satisfactory for measuring new floor products but increasingly the studies involved measurements of floors in service and tiles with surfaces that had undergone various treatments. In particular, measurements with the floor model usually showed some surface deposits, occasionally applied deliberately but more often occurring accidentally. It was then found to be necessary to use flat sliders that could be cleaned efficiently. If angled sliders are produced by using a wedge-shaped piece of slider material, the result is that this distorts and the measurements are unsatisfactory. It is essential to shape the end of the brass holder and to cover this with a uniform thickness of the slider material under examination. A comparison of the results on typical tile surfaces for horizontal-faced and angledfaced sliders is given in table 1. The ratio of the coefficients of friction of the angled slider to those of the flat slider range from 73 to 97 % but the majority are reasonably close to an average of 85 %. The rubber already adopted as a standard material for the Transport and Road Research Laboratory skid tester was used as a reference material. Bovine leather was the other material used as a standard. The choice of leather was obvious in the study, but sampling consistency posed a potential problem. It had been suggested that it would be sufficient to obtain samples from one position off the beasts and the small area of a slider could then be an undoubted advantage, since only relatively small samples of leather would be needed. It was ultimately found to be necessary to select specimens by measuring the coefficient of friction of a standard type of unglazed vitrified floor tile in dry conditions. This test was used to eliminate the occasional bad specimen and also to effect a regular check of each slider during its life. Pigskin was used as a slider material to represent the behaviour of bare human feet; this is particularly important in studies related to the use of ceramics in such areas as showers and swimming pools. 3 . 3 . Cleaning procedure

Freshly-fired ceramic floor products are clean because of the vitrification at high temperatures, and the main contamination is grease from handling after cooling from the kiln. For new tiles, therefore, the cleaning procedure that was found to be most convenient was to brush the surfaces under AnalaR acetone and then allow them to dry in the laboratory atmosphere. Subsequent work with tiles from service, laboratory chemical treatments and direct measurements on floors were much more complicated. Often the same technique could be used that had been adopted for new tiles but the deposits, although usually including grease, were found to be varied. Fortunately the prime consideration has been to measure the friction characteristics of surfaces as they are, and measurements have continued without cleaning the surface but by cleaning the slider after each traverse. Different paths are chosen for a series of measurements on one surface, the slider surface being renewed after each traverse by light abrasion with emery cloth (grade FF). 3 . 4 . Reproducibility

Repeated traverses with a slider of TRRL rubber over the surface of a new tile in dry conditions produce average coefficients of friction to within 0.05. In wet conditions sliders made of similar rubber give results reproducible to 0.02, and this value also applies to the selected leather sliders in both wet and dry conditions. These are the two materials

522

Coejicient of friction offloors

most widely used for sliders. Some synthetic shoe materials would be expected to be sensitive to temperature but all laboratory measurements were made within the range 20 i~3 "C, and therefore temperature was not a factor causing variations. The main difficulty on-site has proved to be that dry conditions cannot always be achieved because some working environments include permanent wet spillage or condensation.

4. Results

One significant feature for established ceramic products is that the average coefficients of friction tend to be similar in different conditions. Some plastics floorings exhibit wide variations. A typical trace across plane vitrified floor tile surfaces is shown in figure 3. This trace was obtained with leather in dry conditions at the slow speed of measurement. There is evidence of stick-slip behaviour.

J

Distance l c m l

Figure 3. Typical trace across three plane-surfaced vitrified tiles obtained with a leather slider on the dry surface at 0.3 cm s-1. J, tile joint.

The rubber used to obtain the results of table 2 is the standard material used for the TRRL skid tester. There was advantage in the availability of this established material but subsequent studies have shown that it is not typical of shoe solings and top-pieces (heels) because it is too soft. As expected, the average coefficients obtained from the integrator readings are lower for the smoother surfaces, but they show only relatively small differences between plane and relief 'non-slip' ceramic surfaces. The mean of such averages given in table 2 indicates a slight increase for relief surfaces with leather as slider but a decrease with rubber. Discrimination between the performance of plane and relief surfaces requires a study of the traces across the surfaces as well as the average friction values. When the trace is seen for a leather slider over vitrified tiles with a special relief surface (figure 4) the effect of the relief projections is to produce values as large as 6. The maximum height of the projections above the plane surface of the tile was 2-3 mm. The average coefficient of friction from this trace is 0.80 compared with 0.60 for a similar tile with a plane surface. The usefulness of the relief surface is that the projections furnish a substantial resistance to motion but the possibility of increasing the value still further by increasing the height of projection is limited by factors such as the comfort of persons moving over the floor and other properties such as impact resistance.

523

Coeficirnt of friction offloovs

n ~ ~ . ~ i _ m . i ~ l ~ i - ~ - le ~ P r o f l Joint

Joipt

Figure 4. Trace across three tiles with geometric projections on the surface, obtained with a leather slider on the dry surface at 0.3 cm s-l.

Table 2. (a) Mean values for ceramic flooring products. (b) Typical values for nonceramic flooring materials.

(4 Coefficient of friction at:

Description

0.3 cm s-I

7.4 cm s-l

Dry Rubber Leather

-~

_________

~

Wet

Wet

Rubber Leather

Rubber Leather

0.91 0.94 0.96 0.91 0.58

0.87 0.89 0.92 0.77 0.50

~________~_

‘Non-slip’ floor tiles Vitrified floor tiles Quarries Engobe finish Glazed finish

1.40 1.45 1.25 1.35 1.25

Description

Coeficient of friction at :

0.77 0.62 0.64 0.57 0.42

0.84 0.79 0.81

0.77 0.51

0.78 0.72 0.71 0.66 0.54

_____~_________

7.4 cm 5-1

0.3 cm s-l __

Dry

Wet

Wet -

Rubber Leather Concrete tile Parquet tile Epoxide PVC

0.85

0.70 0.55 1.20

0.62 0.29 0.34 0.46

___

Rubber Leather

Rubber Leather

0.76 1.64 1.36 0.93

1.01 0.90 0.71 0.95

0.81 0.67 0.57 0.65

0.75 0.55 0.57 0.56

524

R Brougk, I;Malkin and R Harrison

C

0

c

.0_ c L c

Profile

Figure 5. Trace across one tile with areas of rough texture, obtained with a leather slider on the dry surface at 0.3 cm s-1.

The traces of the coefficients of friction also show the effect of the joints. The increase in the coefficients at these positions can be seen for both plane and relief-surfaced tiles. Another type of relief tile surface had 18 mm square areas of roughened texture with 18 mm spaces of normally-textured surface between. The maximum height above the plane surface of the peaks in the areas of rough texture was 0.3 mm. The trace is shown in figure 5 and for comparison this was again obtained with a leather slider in the same conditions. The average coefficient of friction is 0-82, which is similar to the result for the product having the trace of figure 4. The distances between the projections on the surface of tiles are important in relation to the success of the products in service. One type of floor tile was available with a relief surface consisting of square blocks, but these were so close together that a spherical slider or a flat slider angled at 21 could not penetrate into the hollows between the blocks. The average coefficient was thereby exactly the same as for a similar but plane-surfaced tile. The tile behaved as though it had a plane surface. The method of measurement and in particular the provision of traces across the surface provides information that is of value in designing the best types of relief surfaces. Naturally this is limited by production and practicability factors. O

0

L

I

1 Speed of slider

L

10

I c m s-')

100

Figure 6. Variation of the coefficient of friction with speed on a plane-surfaced vitrified tile in wet conditions: a rubber, 0 leather, 0 pigskin.

Coeficient of friction offroors

525

Although other methods of measuring coefficients of friction such as the inclined plane and the skid tester are not able to identify the effects of the relief patterns, comparisons between the different methods for plane surfaces give more reasonable agreement although some anomalies have been noted. Over the range of speeds, the resistance to slipping of the relief projections remained significant. Speed is important in its effect on the behaviour of the slider material since distortion on these surfaces will occur more readily at higher speeds. Examples of the variation of coefficients with speed of slider for three types of slider materials are given in figure 6 . These results are all on one plane-surfaced vitrified floor tile in wet conditions. The rubber was the skid-tester type and therefore softer than soling rubbers. The pigskin is the slider material used to simulate human bare feet. The coefficients of friction for pigskin showed the smallest change and for all practical purposes are the same over the wide range of speeds. The coefficients of friction for a number of slider materials on one type of vitrified floor tile are given in table 3. Most of these are soling or top-piece materials. Table 3. Comparison of slider materials on one type of tile. ~

Shore A hardness

Coefficient of friction at: 0.3 cm s-1

Best grade leather TRRL skid tester rubber Crepe rubber Solid gristle rubber PVC

Composition rubber microcellular rubber Reaction moulded polyurethane EVA micromolecular Thermoplastic rubber Hard rubber composition Resin rubber SBR

95 55 38 67 78 84 51 65 54 36 90 95

7.4 cm s-1

Dry

Wet

Wet

0.67 1.67 1e06 0.93 0.78 0.82 1.07 0-90 0.60 1-07 0.89 0.72

0.76 0.95 0.76 0.79 0.79 0.88 0.92 0.92 0.61 0.81 0.86 0.77

0.72 1.12 0.90 1.00

0.84 0.91 1.09 1 *07 0.66 1.01 0.91 0.83

Leather is unique among the soling and top-piece materials that have been used as sliders. It has a fibrous nature not reproduced in any of the man-made materials and absorbs water, so that great care is necessary if a series of measurements is being performed wet and dry. In addition to the variation with speed, there are therefore effects of minor differences in wetness to be determined that may be of practical significance. One condition that is known to be hazardous in practice is that of walking from dry conditions on to wet floors and there is a reversed situation of wet footwear on some dry surfaces. The apparatus has the useful attribute of having measurements over lengths up to 40 cm, enabling these sorts of conditions to be reproduced. An example is given in figure 7, which shows the fall in the coefficient of friction for a rubber slider, initially dry, moving on to a wet patch of floor tiling. The portable apparatus has been used to measure floors with service lives of many years. Measurement is always first made of the floor surface after brushing only, so that the effect of any contamination can be known. Initial cleaning then consists of removal of grease with a solvent, although this can be extremely difficult in situ because of the

526

R Brough, F Malkirz and R Harrison

+

1-

O

c

c 2

0

c v

0

0

0

1 0

,

Distance lcm)

150

Figure 7. Trace over a surface changing from dry to wet conditions obtained with a rubber slider at 0.3 cm s-l.

-

0 C I

0 c

Clean area

Finger- m ar k ed area

1

C

f? 0

t l 0L-

(5

Figure 8. Reduction of the coefficient of friction because of grease on the surface obtained with a rubber slider at 0.3 cm s-l.

difficulty of isolating relatively small areas. The slightest trace of grease will considerably reduce the coefficient of friction. For instance, one touch with a finger may reduce the value by half, and this is illustrated in figure 8 which shows the effect of finger grease on terrazzo. Organic compounds are comparatively easy to remove, but after these have gone, persistent inorganic deposits remain. Analysis of tile surfaces by EDA shows that the principal elements which are added or are in additional proportions are silicon, calcium and phosphorus. The latter may be deposited from cleaning agents, but in some cases initial cleaning with phosphoric acid could be a contributory cause. Calcium may derive from water, especially in hard water areas but, together with silicon, there is also a possibility of redistribution over the surface after attack on the ceramic material by highly alkaline detergents. Another possible source of silicon is the sodium silicate used in some detergents. Inorganic deposits on vitrified ceramic tiles taken from site have had surfaces with the appearance of being glazed, and even when refired to 700 "C there was no change in the gloss value. The coefficients of friction may be reduced well below half the values for unaffected tiles, measured with a range of sliders both wet and dry. An example is shown in figure 9. The traces were obtained on one type of plane-surfaced vitrified tile with a leather slider at 0.3 cm s-1 in dry conditions. The changes in the coefficient of friction across a new tile are shown by the upper trace and the average value is 0.72. The lower trace was over a tile removed from a floor and the surface had a different appearance. The apparently smoother texture was primarily a consequence of the surface deposit and the average coefficient of friction had been reduced to 0.31. This measurement was made after cleaning to remove organic contaminants that might have occurred in handling or were present on the floor.

Coe$cient of friction osfloors

527

average 0.72

C

B

+

0

With surface deposit : average 0.31

0

U

Figure 9. Traces over a dry, plane-surfaced, vitrified tile with a leather slider at 0.3 cm s-1 showing the effect of deposits occurring because of inefficient cleaning.

The friction apparatus, and in particular the portable version, is an important tool in the investigation of the problem. It is significant that efficient rinsing of floors after cleaning either reduces, or completely prevents, the detrimental surface changes resulting from deposition.

5. Conclusions An apparatus has been developed which is suitable for the determination of coefficients of friction both on laboratory samples and on floors. Different slider materials may be used in various shapes for measurements over a range of speeds in dry or wet conditions. Coefficients of friction are obtained as an average and also in the form of a continuous trace over surfaces. The latter can be used in design processes to understand the relationship of surface profiles to friction characteristics. The results are consistent with known practical behaviour and, in particular, the advantages of certain relief surfaces, known to be efficacious, can be demonstrated. For particular types of slider in defined conditions the results could be used to classify flooring materials for slip resistance. The measurement technique is being applied to a problem of contamination of floor surfaces.

Acknowledgments

The authors thank Mr ADinsdale OBE, Director of Research, British Ceramic Research Association, for permission to publish this paper and express their appreciation of the expertise with designing and constructing instruments of Mr C J W Baker and his colleagues. Many of the measurements have been made by Mr C J Williamson. The energydispersive analyses were made by Dr SN Ruddlesden. Advice on soling and top-piece materials has been given by Mr D J Southwell of the Shoe and Allied Trades Research Association.

528

R Brough, F Malkin and R Harrison

References Brungraber R J 1976 NBS Techn. Note No. 895 Building Research Station 1961 BRS National Building Studies Research Paper No. 32 Childs T H C and Tabor D 1967-8 Proc. Instn Mech. Engrs 182 (3G) 7 Levandowsky L and Jacqmarcq M 1970 Bull. Soc. Franc. Ciram. 87 59-85 Pangels R 1962 Boden Wand Decke 8 (3) 136-43 Road Research Laboratory 1960 Road Note 27 (London: HMSO) Tabor D 1956 Proc. 11th Ann. Mtg ASLE, Pittsburgh in Lubrication Engineering (Nov-Dec.)