Characterization of Fabric-to-Fabric Friction: Application to Medical Compression Bandages

2.1 Compression bandages

Compression bandages are textile medical devices designed for the treatment of some venous or lymphatic pathologies. The treatment consists in wrapping a stretched bandage around the limb, hence applying an interface pressure on the limb thanks to the tension in the fabric. The interface pressure, which is the active principle of the therapeutic treatment, depends on several parameters such as the bandage tension, the leg geometry, the number of layers, and other parameters such as bandage-to-bandage contact interactions [12]. Moreover, bandage-to-bandage frictional properties prevent bandage slippage and are thought to have an impact on interface pressure variation between rest state and exercise.

First, the frictional properties of several medical compression bandages were investigated under conditions close to their use conditions. The selected bandages were the Biflex^® 16, the Biflex^® 17, the Flexideal^®, the Biflexideal^® and the Cotton bandage, all manufactured by Thuasne (Levallois Perret, France), but also the bandage Rosidal^® K, manufactured by L&R (Rengsdorf, Germany). Their characteristics are given in Table 1. These bandages consisted in a representative sample of commercially available bandages, with different textile structures. It is expected that the characterization of their frictional properties will provide a range variation of friction coefficient.

In a second phase, the impact of applied pressure and stretch on frictional properties was evaluated for only two bandages: the Biflex^® 16 (B16) and the Biflex^® 17 (B17) (Figure 1), which were selected because they have a very similar fabric structure but different mechanical properties (stiffness). They mainly differ in their core elastic yarn, whose force at stretch = 1.3 is higher for the B17 (12.3 cN) than for the B16 (5.7 cN). According to the manufacturer’s recommendations, these bandages should be applied on the lower limb with a 1.3 stretch (stretch = L/L₀, with L the post-application length of the bandage and L₀ its initial length).

Figure 1

Cross-sectional images of bandages B16 (top) and B17 (bottom) obtained in X-ray micro-tomography (Phoenix Nanotom^® S) in the weft (left) and warp (right) planes and at three different stretch levels (1.0, 1.2, and 1.4).

2.2 Kawabata Evaluation System (KES)

The Kawabata Evaluation System (KES) was designed for the standard evaluation of fabric hand [6]. It is composed of four instruments [13]. Six characterization tests can be performed thanks to these different devices: tensile and shear tests, pure bending test, compression test, and friction and roughness measurements. Only the last of these four systems was of interest in the present study: the KES-FB4. In this study, it was partly redesigned to characterize fabric-to-fabric dynamic friction coefficient.

2.2.1 Characterization of fabric frictional properties

2.2.1.1 KES-FB 4

The purpose of this device is to mimic a human finger touching the fabric. For this, a 5mm long friction contactor, composed of ten 0.5 mm diameter steel piano strings (mimicking the finger ridges), is in contact with the fabric (Figure 2). This contactor is linked to the frictional force sensor and a normal load (49.8 cN) is applied on the contactor (equivalent to a 50 g weight). Then, the fabric underwent a 3 cm translation motion, with return, under the friction contactor, at a constant speed equal to 1 mm/s. The instantaneous dynamic friction coefficient is computed thanks to the following equation (Amonton’s law):

(2) μ(displacement)=|frictional forcenormal force|

Figure 2

Illustration of the two friction contactors: A, the initial one from the KES and B, the one designed for fabric to fabric friction characterization.

It is then averaged over the fabric displacement, the first and last half centimeter of the translation motion (i.e., the acceleration and deceleration phases) not being taken into account for this computation (Figure 7). Consequently, the friction coefficient obtained and provided by the device is the dynamic friction coefficient.

The KES is only suitable for fabrics to metal friction characterization. Moreover, the contact area of the sensor could be quite small with respect to the fabric pattern. However, it is a reliable and simple measurement tool [14], hence the idea to use this module but to design a new friction contactor.

2.2.1.2 Fabric-to-fabric friction contactor

Within the aim of proposing a simple characterization method of fabric-to-fabric friction, a new friction contactor was designed (Figure 2B and Figure 3B). It was composed of a metal cube on which the fabric was fully taped with the required stretch (to prevent any slippage between the fabric and the friction contactor on which it was set on). Different weights could be set on the contactor to adapt the normal load within the experimentally observed range of pressure [12] (Table 2).

Figure 3

(A) Stretching frame: the fabric is set in the jaws with no stretch, then the translation of the jaw stretches in a homogeneous way the fabric. Once stretched, the fabric is fixed in the holding frame and removed from the jaws. (B) Holding frame and friction contactor used for the measurement of fabric to fabric friction coefficient.

Table 2

Different normal loads applied for the characterization of friction coefficient

Applied weight (g)	Normal force (cN)	Pressure (kPa)	Pressure (mm Hg)
50.8	49.8	2.22	16.6
65.4	64.2	2.85	21.4
80.5	79.0	3.51	26.3
95.2	93.4	4.15	31.1
106.1	104.1	4.63	34.7

The contactor was linked to the KES-FB4 frictional force transducer. However, the automatic data postprocessing was based on the assumption that the normal load was 48.9 cN (corresponding to a 50 g weight). Therefore, a correction of the data given by the KES-FB4 was required:

(3) μ=k*μKES−FB4 with k=regular normal load(48.9cN)applied normal load

2.2.1.3 Tensioning and holding frames

The pressure applied on the limb by compression bandages is the direct consequence of fabric stretch, and it is the active principle of the therapeutic treatment. Thus, it was primordial to perform the experiments on stretched bandages. For that purpose, a stretching frame was designed to set the fabric with a homogeneous stretch (Figure 3A). The fabric was set in a pair of jaws without any stretch. The translation of one of the jaws stretched the fabrics in a homogeneous way, until the expected stretch was reached. Then, the fabric was fixed (and fully taped) in the holding frame (Figure 3A and 3B), which maintained the fabric at the desired stretch. The fabric was then removed from the jaws and the frame was set on the KES moving platform.

2.3 Experimental protocols

2.3.2 Frictional properties of different medical compression bandages

The friction coefficient was measured for six different commercially available medical compression bandages with various structures (Table 1). The applied stretch was chosen to be in accordance with the actual bandage application:

1. stretch = 1.3 for the Biflex^® 16 and Biflex^® 17

2. stretch = 1.5 for the Flexideal^®, Biflexideal^® and Rosidal^® K

3. no stretch for the Cotton bandage.

The cotton bandage is not a compression bandage but a padding layer (applied under a compression bandage) and is usually applied with very little stretch. The applied load was set to 2.85 kPa (21.4 mm Hg, i.e., similar to use conditions) and friction was measured along the warp direction. Five measurements, with five different samples, were performed for each bandage.

2.4 Statistical analysis

All mean results are given with their 95% CI.

The significance of the difference between the friction coefficients at different stretch levels or pressure levels was assessed with the Kruskal–Wallis nonparametric test, as the samples were very small (n = 5). Then, the individual effects were tested with a Mann–Withney U-test corrected with the Bonferroni method ( α=0.05Number of tests ). Difference was considered as significant if p < 0.05.

3.1 Friction coefficient for different bandages

Bandage-to-bandage friction coefficients were measured for six commercially available bandages (Figure 4). The range of variation of the measured friction coefficient was between 0.5 and 0.7. For bandages with similar structures, Biflex^® 16 and Biflex^® 17 but also BiflexIdeal^® and FlexIdeal^®, increasing the fabric density tended to increase the friction coefficient. Indeed, the friction coefficient was higher for the Biflex^® 17 (0.57±0.03) than for the Biflex^® 16 (0.53±0.02) and the same trend was observed for the BiflexIdeal^® (0.51±0.01) and the Flexideal^® (0.63±0.03). The maximum friction coefficient was observed for the bandage Rosidal^® K (0.65±0.03).

Figure 4

Bandage-to-Bandage friction coefficient for different medical compression bandages (mean value ± 95% CI).

3.2 Influence of applied pressure

Dynamic friction coefficient μ was measured with varying applied pressure for two bandages with very similar structures, Biflex^® 16 (B16) and Biflex^® 17 (B17). The results are presented in Figure 5. For bandage Biflex^® 16, the friction coefficient tended to decrease at the initiation of pressure increase and then was stable. For bandage Biflex^® 17, a trend of decrease in the friction coefficient was also observed when pressure increased. Significant differences were observed in friction coefficient for varying pressure (p = 0.03 for the B16 and p = 0.01 for the B17), only between applied pressures equal to 2.22 kPa and 4.15 kPa (Figure 5).

Figure 5

Evolution of friction coefficient with regards to applied pressure (*significant difference).

3.3 Influence of bandage stretch

Frictional behavior of both bandages B16 and B17 was characterized at different levels of stretch. The results, presented in Figure 6, showed that stretch had no significant impact on friction coefficient of bandage B17 (p = 0.58), but it had a significant impact for the bandage B16 (p = 0.006) between stretches equal to 1.2 and 1.3.

Figure 6

Evolution of friction coefficient with regards to bandage stretch (*significant difference).

4.1 Bandage structure

Following the KES protocol, the friction coefficient was the mean of that measured for a 20 mm displacement forward and backward. This friction coefficient varied quite periodically within the displacement range (Figure 7). As it was previously observed by Ajayi et al. [16], it is possible to correlate the obtained pattern (Figure 7) with the fabric surface topography (weft yarn density).

Figure 7

Examples of curves for two bandages samples B16 and B17, along the warp direction.

These experiments were not designed to investigate the impact of fabric structure on frictional properties. However, the measurements performed for the six bandages tend to be in agreement with a previous study [17], which noticed that an increase in fabric construction (i.e., the number of weft yarn per centimeter) led to an increase in the friction coefficient. Further investigations would be needed to conclude about a real impact of fabric structure on friction coefficient.

4.2 Influence of applied pressure

In the range of the tested applied pressures, very little changes in the friction coefficient were observed. Even if most of the changes were nonsignificant, increasing normal load seemed to result in a decrease of friction coefficient. This has already been observed by Ajayi [16] and Das et al. [4], where the relationship between the normal load and the frictional was found to be logarithmic. Here, the variability in the friction coefficient makes this logarithmic model impossible to verify and friction coefficient must be considered as constant.

4.3 Influence of fabric stretch

Contrary to applied pressure, stretch had an impact on bandage thickness. According to the tomographic images (Figure 1), the structure of the fabrics was subject to important changes along the warp direction but did not vary along the weft direction. Indeed, these bandages showed no Poisson effect once stretched (i.e., the width of the bandage remains stable whatever its stretch) as shown on Figure 1. However, stretching the fabric tended to decrease its density. It could have been expected that the friction coefficient decreased when stretch increased, as the number of weft yarn per cm decreased, but it was not the case. Figure 1 also clearly shows that stretching the fabric changed the period of the structure, but did not change the pattern. A simple geometrical model based on moiré description was built to quantify the average interaction area between the fabric with a 1.3 stretch (on the movable pad) and the one with variable stretch (from 1.1 to 1.5). This model shows that, because of the small pad dimension, the interaction area is almost constant independently of the stretch.

4.4 Limitations

In the present study, the frictional behavior of fabric was approximated with Amonton’s law. This represents a strong hypothesis as it is known that fabric frictional behavior is more complex. Nonetheless, this law is very commonly used in numerical simulation.

To compute the friction coefficient, the KES removed the first and last 5 mm of the translation (which corresponds to the initiation of the translation motion). Thus, the computed friction coefficient was the dynamic friction coefficient. This system did not permit to characterize the static fabric-to-fabric friction coefficient. Another limitation of the KES was the translation speed. Even though previous studies did not observe any significant effect of the sliding speed on the dynamic friction coefficient [16, 18], it was not possible to characterize the frictional properties at varying speeds with the KES.

The area of the friction contactor should be designed with regards to the weaving pattern of the fabric. If the pattern is too large, the friction contactor area has to be enlarged. This may be the cause of difficulties depending on the normal load, which would be required for the characterization. Indeed, increasing the contact area also means increasing the applied weight, which might disturb the translation motion of the KES.

Finally, the fabrics tested in the present study were mostly woven fabrics. But it was proven that the fabric structure and its fibers components impact on the friction coefficient [8, 19].