A re-examination of joint roughness

coefficient (JRC)

WONG Ngai Yuen, Louis (王毅遠)

PhD(MIT), BSc(HKU)

Associate Professor
Department of Earth Sciences
The University of Hong Kong

*formerly Nanyang Technological University, Singapore

Roughness ?

Sandpaper Silk cloth

http://www.my-walls.net/silk-cloth-material-texture/
http://en.wikipedia.org/wiki/Sandpaper

2
Shear strength between rock surfaces

Common shear strength models

 Mohr-Coulomb model (cohension and friction angle)

 Bilinear model
 JRC-JCS model (Barton criterion)

Bilinear model (Patton, 1966)

The irregularity of discontinuity surfaces could be approximated by
asperity angle i + basic friction angle b

At low normal stresses, shear loading causes the discontinuity

surfaces to dilate as shear displacement occurs

4
 Bilinear model

Intersection line

JRC-JCS model (Barton criterion)

 shear surfaces become continuously damaged as asperities are sheared

 failure locus stabilizes at an angle b

Barton (1973, 1976) 6

 Roughness
profiles and
corresponding
JRC values

Barton and Choubey (1977)

Field estimates of JRC

• The length of the surface of interest may be several

metres or even tens of metres

• How to determine JRC value for the full scale surface?

8
 e.g. Length = 2 m
Amplitude = 20 mm

JRC = 4.5

Barton (1982)
9

Comments
• quick and general judgments of joint roughness


• subjective assessment 

• not entirely adequate for quantifying the rock

joint roughness profile 

10
 There are solutions…
• Plenty correlations of JRC with both statistical and
fractal parameters (Tse and Cruden 1979; Reeves 1985; Maerz et al. 1990
Yu and Vayssade 1991; Xie and Pariseau 1994; Aydan et al. 1996; Yang and Chen
1999; Yang et al. 2001; Grasselli and Egger 2003; Tatone and Grasselli 2010)

• Tse and Cruden (1979):

JRC = 32.2 + 32.47 log Z2
where Z2 ：the root mean square (slope-based roughness parameter)

• Different fitting coefficients


11

Slope based parameter

Slope 1
Amplitude = A

Slope 2
Amplitude = A

Slope 2 > Slope 1, so …… 12

 Amplitude based parameter
Slope = m

Amplitude 1

Slope = m

Amplitude 2

Amplitude 2 > Amplitude 1, so …… 13

Outline
1. Background (√)
2. Motivation (√)
3. Research objectives
4. Methodology
5. Results
6. Summary and conclusions
14
 Research Objectives
• Revisiting the correlation between roughness
parameter Z2 and JRC

• Understanding reasons of discrepencies?

• More representative correlations?

15

Outline
1. Background (√)
2. Motivation (√)
3. Research objectives (√)
4. Methodology
5. Results
6. Summary and conclusions
16
 Methodology

Digitalization of profiles
• Download paper containing the
original profiles (Barton and
Choubey 1977) from the the website
of the Rock Mechanics and Rock
Engineering

• No printing and scanning!

• Check horizontality of the profiles

• Check cleanliness of the profiles

Barton & Choubey (1977)
17

 The coordinates of the points lying on the individual profiles

retrieved for those RGB values < 255 are identified and
stored (max RGB = 255 = white).

 About 360-370 pixels in the horizontal direction (x axis) can

be obtained for each profile and the 10 cm scale bar

 Due to line thickness, an array of points at one particular x.

Inaccurate
representation!!
18
 Line has a thickness!!

Individual
pixels 16

 Obtain central line of each profile by averaging the values of

y coordinates at each x coordinate (not manually).

 The interval of the x coordinate is about 0.27 mm, which is

obtained by

profile length / no. of pixels along the horizontal direction

 Ready for calculating Z2

2 1
1 xL
1
Z 2   ( ) dx  [
dy
2
(yi1  yi ]
) 2 2

L x0 dx M (x)
L = length of the profile
x = sampling interval
M = number of the sampling points 20
 Recall - slope based parameter

Slope 1
Amplitude = A

Slope 2
Amplitude = A

Slope 2 > Slope 1, so slope 2 is rougher 21

Z2 values are calculated at three different sampling intervals

 0.27 mm (small)
 0.54 mm (medium)
 1.08 mm (large)

22
 How about previous results (small sampling intervals)?

Note: reversal
of axes

23

Different Z2 values?

Two potential sources of error:

1. Human error - operator’s trace may deviate away from the profile

2. Line thickness - operator has to consistently trace the central line

24
 How about previous results (small sampling intervals)?

25

To illustrate the significance of the potential errors, profile thickness

is computed at the 0.27 mm sampling interval

……

……
 “max” and “min” = maximum thickness and minimum thickness of each profile
 “std.” = standard deviation
 “rta” = ratio of mean profile thickness to profile amplitude
 “profile amplitude” = distance between the highest point (ymax) and the lowest
point (ymin) along the profile

Results: average “mean thickness” of the ten profiles is 0.492 mm, and the
mode of thickness of all profiles is 0.54 mm.
26
 27

Segmentation of JRC:6-8 (profile 4) and JRC:8-10 (profile 5)

Local higher plateau

 Z2 values of three segments are calculated based on the same 0.27 mm

sampling interval

 profile 4 should be rougher than profile 5?

28
 Outline
1. Background (√)
2. Motivation (√)
3. Research objectives (√)
4. Methodology (√)
5. Results
6. Summary and conclusions
29

Back to the earlier research objectives?

1. Revisiting the correlation between roughness
parameter Z2 and JRC (√)

2. Understanding reasons of discrepencies (√)

3. More representative correlations?

30
Proposal: include a new parameter called normalized amplitude Anor

For a particular joint profile, Anor is defined as the ratio of the

respective profile amplitude to the maximum profile amplitude among
the 10 JRC profiles.

Profile 8 (JRC 14-16) has the maximum profile amplitude of 6.62 mm

JRC  k1  log(Z 2 )  k2  Anor

k3
 k4
where k1, k2, k3 and k4 are coefficients to be solved
31

JRC  41.17log(Z 2 )  4.93Anor

1.53
 26.72

32
 33

Outline
1. Background (√)
2. Motivation (√)
3. Research objectives (√)
4. Methodology (√)
5. Results(√)
6. Summary and conclusions
34
Summary and Conclusions
Methodology
data cursor and document laser scanner (previous)
vs
MATLAB digitization (present)

Interesting finding
Z2 values not always increasing with JRC values

35

Summary and Conclusions (continued)

A new proposed correlation

Slope-based parameter

JRC = 32.2 + 32.47 log Z2 (Tse and Cruden (1979)

vs

Slope-based + amplitude-based parameter

JRC  k1  log(Z 2 )  k2  Anor

k3
 k4
36
Summary and Conclusions (continued)
Recommendation – automatic measurement of surface
roughness by photogrammetry or laser scanning

37

Research

Old problem  New approach?

New problem  Old approach?

New problem  New approach?

Gao, Y. and Wong, L.N.Y.* (2015) "A modified correlation between roughness
parameter Z₂ and JRC", Rock Mechanics and Rock Engineering, 48(1), pp 387-396.
38
 Thank you

Louis N.Y. Wong PhD(MIT), BSc(HKU)

The University of Hong Kong

([email protected])