mae 493n_503t_lec 2
TRANSCRIPT
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 1/29
“Tribology in Mechanical Engineering”
MAE 493N/593T
Dr. Konstantinos
A.
Sierros
West Virginia University
Mechanical & Aerospace Engineering
ESB Annex
263
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 2/29
Course material
You can read and download course material from the following blog;
http://wvumechanicsonline.blogspot.com
e‐handouts will be also uploaded in the blog
Please print responsibly!
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 3/29
Surfaces
• No real surface can possess perfect geometry
• Surface roughness and errors in the shape (waviness)
Roughness
Waviness
Total profile
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 4/29
Surfaces
Before polishing After polishing
http://www.leicester‐ils.co.uk/electropolishing.html
• Stainless steel
surface
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 5/29
Roughness and waviness
• Waviness
(b):
Undulations
with
relatively
long
wavelengths
(mm
scale)• Roughness (c): Variations with much shorter wavelengths
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 6/29
Contact between surfaces
•
When two surfaces contact each other, the
surface asperities (tips of the
surface roughness) must first carry the load
http://www.theshortspan.com/features/friction.htm
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 7/29
Contact between surfaces
• Geometric texture
of
surfaces
depend
on;
a.Production route
b.Nature of underlying substrate
Atomic lattice resolution of a mica
surface. Height image. Image size 6.6 nm x
6.6 nm,
z‐range
0.4
nm
http://www.azonano.com/details.asp?ArticleID=2076
• Atomically smooth surfaces
• If 2 mica (clean) surfaces touch each other;
apparent contact area ≈
true contact area
• Squeezing 2 small
rubber
surfaces
together;
apparent contact area ≈
true contact area
(elastic deformation of asperities)
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 8/29
How do we examine surfaces?• Optical microscopy (up to 1000x mag)
•
Scanning Electron Microscopy ‐
SEM (high
magnification up
to
300000x
and
better field of depth)
•
Transmission Electron Microscopy – TEM (even
higher magnification, 750000x)
•
Atomic
Force
Microscopy
– 3D
surface
topography and true atomic resolution
in high vacuum
•
Profilometry – 3D surface topography, longer
scans and
5 nm
vertical
resolution
diamond stylus
surface
Stylus
movement
MEMS –
Sandia Labs
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 9/29
Statistical nature of surfaces
• No
single
numerical
parameter
can
adequately
describe
surface
geometry• Centre line average roughness (Ra
) and Root mean square roughness (Rq
)
are the simplest parameters
dx y
L
R
L
∫ =0
1α
Length
Coordinate in the surface
Height of surface measured above
mean level
( At mean level: area of material above = area of the voids below)
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 10/29
Making smooth surfaces and using Ra
Decreasing roughness
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 11/29
Issues with average roughness (Ra
)
• Different
topographies
may
have
the
same
Ra value!
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 12/29
How do we overcome this problem?
• Using
the
root
mean
square
roughness
value
(Rq
)
⎟⎟ ⎠
⎞
⎜⎜⎝
⎛
= ∫
L
q dx L R y0
21
In this
case;
y becomes
y2
as
compared
to
Ra
equation
and this squared term gives greater significance to surface variation
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 13/29
Abbott bearing area curve
• We
should
be
able
to
estimate
the
bearing
area•
Bearing area
is the proportion of the nominal contact area between two
opposing surfaces that are in true contact
It
is
reallybearing
length!
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 14/29
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 15/29
Statistical treatment of surfaces
•
We first define Φ(z) which is the probability distribution function* of
the z axis
•
For Φ(z) ‐
Plot the number of times a particular value of z occurs in the
data vs. the value of z and scale the most close data fitting curve so the
area under
it
is
unity;
*A mathematical function that is used to model the frequencies and probabilities of
events over time.
∫
+∞
∞− =Φ 1)( dz z
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 16/29
Statistical treatment of surfaces
•
For real surfaces, a ‘bell‐shaped’
curve is observed and Φ(z) is described by a
Gaussian distribution
(Large
number
of
randomly
occurring
events)
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 17/29
Statistical treatment of surfaces•
In general, it is not always easy to judge if any given surface
deviates
from
Gaussian
distribution•
To further investigate we can plot the number of times a particular
value of z occurs in the data vs. the value of z, on normal probability axes
SD of
data
A truly random
(Gaussian) distribution
falls
on
a
straight
line
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 18/29
Engineering surfaces
• In general, random surfaces in engineering are the exception and not
the rule
•
Most surfaces are tailored to our needs and surface patterns are
apparent in most cases (depending on magnification of observation)
Si nanostructure, 200 nm pitch
Fabricated by lithography and DRIE
Chang‐Hwan Choi, UCLA
Turning on a lathe
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 19/29
Statistical treatment of surfaces• We can attempt to quantify the shape of the distribution curve
• We can
use
the
moments
of
the
distribution*
*A set of numbers that describe various properties of a distribution are called
moments of the distribution
∫
+∞
∞− Φ= dz z zm
n
n )(nth
moment
• For n=1, we generally choose the level from which z is measured
to give us
m1
=0
• For n=2, and taking into account that m1
=0 we arrive at the following;
q
Rm == σ 2
SD of the distribution
root mean square
roughness
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 20/29
Skewness, Sk
dz z zSm
k )(1 3
33
3
Φ== ∫
+∞
∞−σ σ
3d
moment
• If Sk
=0, the distribution is symmetrical
∫ +∞
∞−
Φ= dz z zmn
n )(
Skewness Sk
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 21/29
Skewness, Sk
• If Sk
=0 the distribution is symmetrical (a)
• If a process starts removing the peaks (b)
Sk
<0
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 22/29
Kurtosis, K
dz z zK m
)(1 4
44
4 Φ== ∫ +∞
∞−σ σ
4th
moment
• A
Gaussian
distribution
has
K=3• Distributions with sharper peaks than Gaussian have K<3
∫ +∞
∞−
Φ= dz z zmn
n )(
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 23/29
Considerations
(dealing only with height distributions)
• These statistical parameters are based on random data and they themselves are
subject to
random
statistical
variations
• Therefore, we may not be able to tell if a given value of Sk
or K represents a true
property of the surface or it has been influenced by the sampling process itself
• There is no information on the horizontal dimension and spatial
distribution!
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 24/29
Surface height distribution
•
Based only on height distribution the two surfaces below give identical
roughness parameters
•
Their tribological response may be different since (a) exhibits
an ‘open’
texture
and (b) exhibits a ‘closed’
texture
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 25/29
Surface ‘horizontal descriptors’
•
In order to take into account the horizontal dimension of the surface one should
consider possible parameters such as;
‐ Number of peaks per unit length of profile
‐ Number of crossing points per unit length with the mean level
• None
of
the
above
parameters
are
intrinsic properties
of
the
surface•
It is widely accepted that surfaces exhibit a form of self similarity with increasing
magnification
•
Therefore; Peak density, zero‐crossing density etc are all highly dependent on
the
sampling
interval
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 26/29
Self ‐similarity
Autocorrelation functions (ACF)
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 27/29
Autocorrelation functions (ACF)
• Approach to incorporate information on spatial variations
• The ACF is
defined
as;
⎟
⎟⎟
⎠
⎞
⎜
⎜⎜
⎝
⎛
+×= ∫ ∞→
L
L
dx x z x z
L
t 0
2)()(
1lim
1)( τ
σ
ρ
Sampling lengthSD of z
Autocorrelation functions (ACF)
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 28/29
• When τ=0, ρ(τ)=1 and the curve
must decay
from
a value
of
unity
• As τ
goes to infinity, ρ(τ) becomes
asymptotic to zero
• The form of the curve decay provides
info on
the
spatial
distribution
of
roughness
Summary
8/9/2019 MAE 493N_503T_Lec 2
http://slidepdf.com/reader/full/mae-493n503tlec-2 29/29
Summary
• No real engineering surface can possess perfect geometry
• There are
several
techniques
of
surface
examination
• Engineering surfaces are different compared to ‘real’
surfaces
• Surface roughness parameters
• Attempts to incorporate information on spatial variations
Out of class activity
You can attempt to solve the following problem;
Find the values of Ra, Rq, Sk, and K for a surface whose profile is given by
z(x)=αcos(πx/α)