Tribology Letters

, Volume 51, Issue 3, pp 377–383

Preliminary Friction Force Measurements on Small Bowel Lumen When Eliminating Sled Edge Effects

Authors

    • Department of Mechanical EngineeringUniversity of Colorado
  • Benjamin S. Terry
    • Department of Mechanical EngineeringUniversity of Colorado
  • Jonathan A. Schoen
    • Department of SurgeryUniversity of Colorado
  • Mark E. Rentschler
    • Department of Mechanical EngineeringUniversity of Colorado
Original Paper

DOI: 10.1007/s11249-013-0167-1

Cite this article as:
Lyle, A.B., Terry, B.S., Schoen, J.A. et al. Tribol Lett (2013) 51: 377. doi:10.1007/s11249-013-0167-1

Abstract

This study aims to produce experimental results for the coefficient of friction (COF) between the small bowel lumen and an edgeless, translating sled. Friction was measured as a function of sled speed under in situ and in vitro conditions. The results indicate that by eliminating edge effects, the COF between a stainless steel sled and the inner surface of the small bowel lumen is decreased. The average COF for in situ testing was found to be slightly lower than in vitro tests. Friction increased with increasing velocity. The friction forces ranged from 0.013 to 0.08 N, and COF values ranged from 0.007 to 0.054 under these conditions.

Keywords

Small bowelTribologyFrictionCapsule endoscopy

1 Introduction

A robotic capsule endoscope (RCE) inserted into the gastrointestinal (GI) tract via natural orifice translumenal endoscopic surgery (NOTES) may have the ability to perform tasks such as imaging, biopsies and targeted drug delivery that are not possible with the procedures and equipment available today [14]. For an RCE to be effective, it must be able to independently maneuver within the lumen of the GI tract, which includes controlling its position, speed and direction. For the capsule to perform desired tasks, it must be able to hold or reverse its position as the peristaltic contractions of the small intestine naturally force the device aborally. The capsule must therefore overcome two primary hindrances: (1) the lack of adequate reaction forces from the intestinal wall, and (2) low friction of the mucosal surface. Mobility will also be affected by changes in the chemical and mechanical structure of the mucosal surface, which are known to vary significantly in this region [5]. To properly design an RCE with the ability to traverse these surfaces, it is essential to quantify the physical and mechanical properties along the length of the small bowel. To optimize mobility, it is important to understand the parameters which can be manipulated to maximize or minimize surface friction interactions as desired.

In the experiments described in this paper, the coefficient of friction (COF) between the inner surface of porcine small bowel lumen and a polished stainless steel sled was measured. Friction was evaluated as a function of sled velocity, and in situ versus in vitro conditions were compared.

2 Background

Because of its complex nature, friction models specific to the small intestine surface are sparse. General friction models of varying complexity can be considered, ranging from basic Coulomb friction laws to friction models which account for velocity, temperature, contact area and mechanical properties of interacting surfaces [612]. However, general models are challenged to account for both the viscoelastic, heterogeneous structure of the small bowel tissue and contributions from the mucus layer which coats the lumen surface. Flexible columns of villi and microvilli extend from the intestinal surface and are protected and lubricated with two mucus layers, one firmly adherent and one loosely adherent [13]. Lai et al. define the bulk mucus as a non-Newtonian viscoelastic gel which possesses behavior similar to an elastic solid at low shear rates and reverts to viscous fluid behavior at high shear rates [14]. Thus, to properly characterize the surface friction of the small bowel, it may be helpful to independently consider the distinct contributions from the mucus layer and tissue surface.

There are a number of research groups which have produced experimental data reporting friction forces and coefficients between a variety of test coupons and the small bowel lumen surface [1523]. The range of reported friction forces and COF values ranges from 5 to 200 mN and 0.016 to 0.9, respectively, and is all obtained in vitro. The challenge in interpreting and comparing existing results is the multi-variable dependence of the friction behavior which changes with experimental design. Specimen preparation, environmental conditions and test coupon weight, contact area, geometry, material and velocity all vary in the published results. The individual contribution of each of these variables to the tissue friction response is not well known. As a consequence, some research groups have come to different conclusions as to the effects each of these variables has on the friction force.

Of those who have evaluated the friction–velocity relationship on open small bowel specimens, multiple groups have reported an increase in friction force with increasing velocity, while Kwon et al. found the effects to be negligible and an early study on the colon found no significant increase in friction with velocity [16, 17, 21, 23, 24]. The linearity of the friction–velocity relationship is not clear over the values which have been tested.

In previous work, we pulled cylindrical and flat polycarbonate sleds across the open specimen, both with curved leading edges [22]. During testing, we noticed mucus collecting at the leading edge of the sled which caused an increase in pull force as a function of displacement. Wang et al. observed similar results for a blunt rectangular block on an open specimen [21].

Closed specimen studies have evaluated the friction force as a function of velocity, surface area and geometry, with groups reporting that the friction forces increase with increases in velocity and capsule diameter, more than so for length increases [1517, 21]. This is attributed primarily to the increase in intestinal wall stress; however, the contribution to resistance force due to the buildup of mucus at the leading edge of the capsule has not been quantified. In open specimen testing, Kim et al. found no friction force dependence on contact area [17].

To develop a model to predict tissue response to a capsule with varying surface characteristics, it is useful to first obtain experimental data which are independent of capsule-specific geometries. In this work, we present preliminary results and methods that are used to evaluate the COF of the small intestine in contact with a test coupon with negligible sled edge effects. Using this baseline data, future studies will look at specific correction factors to account for geometry, weight and additional surface effects. With this in mind, this preliminary study was designed to evaluate the COF on the small bowel as a function of sled velocity as well as to compare in situ and in vitro results. To the authors’ knowledge, there are no existing studies which attempt to measure the friction behavior of the small bowel in situ.

3 Experimental Methods

To obtain experimental friction measurements, a novel tribometer was designed, as shown in Fig. 1, comprising a linear actuator (Haydon-Kerk, 25844-05-001ENG) which propels a load cell (Loadcell Central, ESP4-1 KG) along a linear slide, with the load cell pulling a flat, polished stainless steel sled upon the mucosal surface of an open porcine intestinal specimen. A motor driver (Sparkfun, A3967), data acquisition system (National Instruments, myDAQ) and bridge strain measurement module (National Instruments, USB-9237, USB-9162) control the motion of the sled and record force overtime.
https://static-content.springer.com/image/art%3A10.1007%2Fs11249-013-0167-1/MediaObjects/11249_2013_167_Fig1_HTML.gif
Fig. 1

Top view schematic of tribometer setup with pulleys, where the overhanging sled translates across the surface of a fixed, splayed small bowel tissue specimen

The sled in translation was designed to extend beyond the raised area of the splayed intestinal specimen on all sides, with the intent of eliminating effects from sled edge–tissue interactions. The overhanging sled was 20.3 cm in length, twice that of the tissue specimen. Sliding force measurements were collected as the sled traveled upon the middle 2.9 cm of the 10 cm long tissue specimen. Two pulleys were mounted fore and aft of the overhanging sled and served to align the central axis of the sled over the tissue specimen. Polymer string connected the front of the sled to the load cell and passed through the front and rear pulleys before re-attaching to the trailing end of the sled. A spring (with known constant) was positioned in series with the cable line to provide tension. As the load cell was driven forward by the linear actuator, the weighted sled was pulled over the fixed intestinal specimen while the pulley and cable system maintained alignment with the line of action of the load cell. Polished stainless steel was selected for the sled material so as to reduce surface roughness to the extent that the primary mechanisms of friction could be reduced to a combination of adhesion and fluid shear, rather than to deformations of asperities on the material surface.

3.1 Testing

3.1.1 In Situ Testing

In situ tests were performed on an anesthetized porcine model (IACUC protocol 87909(05)1D, Ref#100587). In order to reduce the amount of food debris in the gastrointestinal tract, the animal was placed on a Jell-O diet starting 48 h prior to surgery and replaced by a water-only diet for the final 24 h preceding surgery. The tests were performed by isolating a section of small bowel from the abdomen and cutting open along the longitudinal axis near the mesentery. With the mesentery—and therefore a constant blood supply—intact the bowel was brought to the tribometer and splayed flat upon a raised specimen tray with contact area 38.7 cm2. The tissue was fastened via metal barbs which protruded orthogonally from the vertical side of the specimen tray. Precisely controlled drops of saline were added to the lumen surface to support tissue hydration. The sled was weighted with 3.21 N, which was concentrated at the center of the sled. The tray contact area and sled weight were selected so as to simulate the body’s intra-abdominal pressure, in this case approximately 0.83 kPa [26]. In situ testing was conducted on two tissue samples from each of three regions (proximal, middle and distal) of the small bowel. Multiple runs were performed on each specimen sample at a velocity of 1 mm/s. Due to constraints on surgical time, only one velocity was tested in situ. The in situ testing setup is shown in Fig. 2.
https://static-content.springer.com/image/art%3A10.1007%2Fs11249-013-0167-1/MediaObjects/11249_2013_167_Fig2_HTML.jpg
Fig. 2

In situ testing using tribometer with pulleys

3.1.2 In Vitro Testing

Following in situ testing, the unused intestinal tissue from the test was harvested and stored in phosphate buffered saline (PBS) solution at 3 °C for approximately 2 h and was then brought to room temperature (20 °C) for testing. The tests conducted in situ were repeated on adjacent tissue specimens in vitro within 5 h of tissue excision. The mounting process, sled weight, tissue contact area and velocity were held constant for both tests. Similar to the in situ tests, the mucosa was hydrated by applying controlled allotments of PBS to the tissue surface using a syringe dropper.

3.1.3 Velocity Testing

A velocity study using the pulley tribometer setup was conducted in vitro on a separate porcine model which had not been subjected to in situ testing, but was put on a similar Jell-O and water diet for 48 h prior to tissue excision. A 0.94 N weighted sled was pulled across a tissue surface with approximate contact area of 38.7 cm2 at speeds of 0.5, 1, 2, 4 and 6 mm/s. This pressure was chosen to more closely match pressures indicated from other studies [17, 21, 22]. Each speed was evaluated on three tissue samples from three regions of the small bowel (proximal, middle and distal regions).

3.2 Data Analysis

For each test within this study, we obtained friction force measurements by collecting 1,000 data samples per second. The friction forces for each section of bowel were averaged over 2.9 cm of travel to form one data point for each run on the tissue sample. A friction coefficient was extracted using the Coulomb dry friction law given in Eq. (1), where Ff is the measured friction force, μ is the Coefficient of Friction and Fw is the applied load.
$$F_{f} = \mu F_{w} $$
(1)
95 % confidence intervals were then calculated to estimate a population mean with the sample means from each run. We then compared the sample means using an analysis of variance (ANOVA) test, assuming equal but unknown variances. The analysis was performed allowing for a 5 % probability of making a Type I statistical error.

4 Results

4.1 In Situ Versus In Vitro Results

The COF values from the in situ tests and in vitro tests are shown in Fig. 3. These results represent the mean COF along with the 95 % confidence interval for the referenced bowel location and testing conditions, and hence do not include the outliers. A two-factor ANOVA showed strong evidence that the COF values of the in situ and in vitro tissue samples were different in the proximal and distal sections (p = 0.003, p = 0.01), while the null hypothesis could not be rejected in the middle section (p = 0.053, F < Fcrit).
https://static-content.springer.com/image/art%3A10.1007%2Fs11249-013-0167-1/MediaObjects/11249_2013_167_Fig3_HTML.gif
Fig. 3

Coefficient of Friction for 3.21 N edgeless, polished stainless steel sled on small bowel mucosa; in situ versus in vitro results. Error bars represent a 95 % confidence interval around the mean

The temperature of the specimen surface was recorded throughout the course of testing and ranged from 24.2 to 28.5 °C for in situ tests and 19.1 to 21.4 °C for in vitro tests.

4.2 Velocity Results

The velocity study indicates that the COF increases with increasing velocity. A single factor ANOVA was applied to compare differences in COF versus velocity, and the difference was found to be significant (p < 0.05). There were no significant differences between region for each velocity. The data from measurements taken from the velocity test are shown in Fig. 4, again, representing the mean COF and 95 % confidence intervals, with outliers excluded.
https://static-content.springer.com/image/art%3A10.1007%2Fs11249-013-0167-1/MediaObjects/11249_2013_167_Fig4_HTML.gif
Fig. 4

Coefficient of Friction for 0.94 N edgeless, polished stainless steel sled on small bowel mucosa: velocity results using pulley tribometer. Error bars represent a 95 % confidence interval around the mean

5 Discussion

5.1 Edge Effects

The elimination of edge effects appears to reduce the overall contribution of geometry-dependent drag force to the total friction force. There is a challenge in making a direct comparison between the results of this study to existing studies, due to the multi-variable dependence of the friction response between the intestine tissue surface and a surface with which it is in contact. Kim et al. found that with an increase in weight, the measured friction force increased while the friction coefficient decreased [17]. The weights used for this study were one to two orders of magnitude larger than those used in other studies; however, the measured friction forces in this study were of the same magnitude.

This reduction in friction is likely explained by the lack of fluid drag which would be induced by mucus collecting at the leading cross-sectional area for a capsule or block, as well as limited mechanical engagement of the sled edge with tissue. At this point, the exact reduction in the friction coefficient is not known, due to differences in sled weight, velocity, contact area, and tissue sample variations among the studies and the lack of an appropriate model to normalize these parameters. For this comparison, the contact area between the overhanging sled and the tissue was nearly six times larger than that of the sled in the previous study [22]. The use of stainless steel for sled material likely further reduced the friction resistance between the sled and the tissue surface when compared to aluminum, ABS plastic, polycarbonate or ceramic used by other researchers [15, 17, 2123].

5.2 In Situ Versus In Vitro

In considering in situ versus in vitro conditions for each region, there does appear to be a difference in the COF in the proximal and distal sections of the small bowel at the 95 % confidence level. While the ANOVA analysis for the middle section failed to reject a null hypothesis at the 95 % level, indicating that the means were the same in situ and in vitro, a comparison at the 90 % confidence level showed that the means could be considered different. This may provide sufficient evidence to indicate that in vitro tests may slightly overestimate the tissue’s friction response. There were noticeable contractile responses within the bowel wall during testing. We believe any differences in friction to be as a result of the elevated tissue surface temperature, muscle wall response to the load, continuous mucus replenishment and maintained blood flow to the in situ test specimen. Additionally, the distal portion of the bowel trended toward slightly larger COF values than those measured in the proximal and middle sections. Despite 48 h of fasting, food debris was occasionally noted within the folds of the villi of the distal bowel specimens, which may have increased the friction resistance.

It should be noted that significant convective cooling decreased the surface temperature of the in situ specimen from its physiological state by approximately 10 °C. This was in close range to the surface temperatures recorded for the in vitro testing, which may explain why we did not observe the magnitude of difference we expected.

5.3 Velocity

This velocity study confirmed the work of others which showed that increasing the sled velocity increases the friction resistance [17, 21, 23]. The behavior may be attributed to initial adhesive bonding between the mucus and the sliding tray and viscoelastic properties of the intestinal tissue [25, 27]. As noted by Kim et al. at low velocities, stress relaxation occurs within the tissue and mucus-sled re-bonding can be achieved [17]. However, beyond a certain velocity increase, the tissue responds more as an elastic solid, and the mucosal bonds are not allowed sufficient time to develop as the mucus is sheared across the surface of the tissue. This behavior corresponds to that described by Olsson et al. and Gong et al. where surface bond strength between the mucus and sled serves as the primary friction mechanism at low velocities and as velocity increases, shear forces within the mucus layer become dominant [6, 12]. To what extent this relationship is linear is unknown. A preliminary evaluation indicates that the relationship is nonlinear, but requires additional study to verify. The three regions of bowel responded similarly to changes in sled velocity. There were no statistically significant differences between bowel region at any of the test velocities. This may be explained by more complete fasting for this animal.

These findings suggest the need to model the intestine as a system consisting of a viscoelastic tissue with a mucus lubrication layer, the thickness, hydration and rheological properties of which influence the friction forces it produces. The fluid flow of the mucus and the resistance it produces against an object moving in the opposing direction will be of utmost importance when quantifying the in vivo conditions the robot must overcome. We also know that various pathologies can affect the thickness and rheological properties of the mucus layer which may play a role in an endoscopic robot’s ability to actuate as desired [14]. The behavior of the friction–velocity relationship ultimately depends on which surface component, bulk mucus layer or the underlying viscoelastic tissue, dominates the friction interaction.

5.4 Additional Observations

In the majority of cases, we observed that the first run on each tissue specimen yielded a higher friction force than subsequent runs. The mucus layer and morphology of the tissue surface did not appear to change significantly, but some flattening was observed. This indicates that the friction resistance that an endoscopic capsule encounters may vary depending on whether it is moving across a section of tissue for the first time, or reversing over a previously traveled segment which has not had its mucus layer replenished.

There appears to be large variability in the friction response of the small bowel within the pig population. There was nearly one order of magnitude difference between the COF measured in the animal used for the in situ and in vitro testing as compared to the animal used for the velocity studies. Age, health, diet and digestion history are all contributors to this variability. A similar range should therefore be expected within the human population.

The ultimate goal was to collect raw in situ friction measurements with only the native mucus as a lubricant layer. However, without knowing the true impact of the addition of saline to the tissue surface, it was more important to hold all possible variables constant between the in situ and in vitro tests. Therefore, saline was carefully and sparingly administered via a syringe dropper for both in situ and in vitro testing. Yoshida et al. considered the tissue’s hydration condition to be the primary factor affecting the coefficient of friction [23]. When under a load, the water content of the surface layer decreased as it was forced out through the sides of the sled-intestine interface, and they concluded that friction depends essentially on water content of the surface layer. While saline is intended to simulate a replenished, hydrated mucus layer, it has viscosity properties lower than mucus and may serve to dilute the natural mucus. Its presence may contribute to a falsely low friction value and we recommend that it be used conservatively.

6 Conclusions and Future Work

The findings from this research indicate that the friction coefficient between the small bowel surface and a stainless steel sled is reduced when edge effects are eliminated. With contributions from capsule-specific geometry removed, this may allow certain relationships (i.e., friction-velocity, friction-contact area) to be more evident. In this study, we found the range of friction forces and friction coefficients to be 0.013–0.08 N and 0.007–0.054, respectively, both depending on velocity, tissue condition (in situ vs. in vitro) and whether it was the first or subsequent pass over the tissue. The aforementioned range is the full range of measured values, to include outliers. We believe these COF measurements may be of use to researchers who wish to model middle body segments of the RCE which do not present edge effects and to researchers looking at fundamental numerical or analytical models that require a normalized COF which they can further supplement with geometric correction factors.

The friction behavior of the intestinal surface in contact with stainless steel does not conform to Coulomb’s law. There is a dependence on velocity and future work will include an investigation of the function of this relationship. The rheological properties of the mucus layer should be further quantified experimentally and compared for in vivo and in vitro conditions. Additionally, it may be useful to evaluate the friction forces on a “stripped” mucosal surface in order to determine whether the tissue surface or mucus layer dominates when considering friction interactions with an object sliding upon it.

Future work will include controlling both the temperature and humidity of the ambient air surrounding the tissue specimen to more closely simulate the biological environment of the tissue, thus eliminating the need to apply saline directly to the tissue surface. This will preserve the condition of the tissue and mucus layer allowing the viscoelastic properties of the specimen to be properly observed. Additionally, the sled weight and contact area will be adjusted to simulate the pressure exerted by the contractile forces from the muscular wall in response to a solid bolus, which have been measured in vivo [28, 29]. A subset of this work has been conducted and preliminary results have been published [30].

Copyright information

© Springer Science+Business Media New York 2013