Professional Documents
Culture Documents
Impact of Frother on the Terminal Velocity of Small Bubbles
PII: S0301-7516(08)00210-X
DOI: doi: 10.1016/j.minpro.2008.12.004
Reference: MINPRO 2133
Please cite this article as: Navarra, A., Acuña, C., Finch, J.A., Impact of Frother on
the Terminal Velocity of Small Bubbles, International Journal of Mineral Processing (2009),
doi: 10.1016/j.minpro.2008.12.004
This is a PDF file of an unedited manuscript that has been accepted for publication.
As a service to our customers we are providing this early version of the manuscript.
The manuscript will undergo copyediting, typesetting, and review of the resulting proof
before it is published in its final form. Please note that during the production process
errors may be discovered which could affect the content, and all legal disclaimers that
apply to the journal pertain.
ACCEPTED MANUSCRIPT
T
A. Navarra,
P
Department of Applied Mathematics and Industrial Engineering,
École Polytechnique de Montréal,
RI
Montreal, QC, Canada
SC
C. Acuña,
Chemical Engineering Department,
Universidad Católica del Norte,
NU
Antofagasta, Chile
J.A. Finch
Department of Mining and Materials Engineering,
MA
McGill University
Montreal, QC, Canada
ED
Abstract
Using a circulating flow that balances buoyancy and drag, small bubbles (< 1mm) are held in a
PT
column that classifies them according to their terminal velocities. Without frother, the terminal
velocities fall between the values predicted by Hadamard-Rybczynski for fluid spheres, and
those predicted by Stokes for hard spheres. Although it is commonly believed that industrial
CE
surfactants have little to no impact on such small bubbles, this study demonstrates a trend
comparable to that of larger bubbles, namely that the addition of frother can retard the bubbles
even beyond the predictions of the hard sphere model. Hence the motion of small bubbles
AC
appears to be impeded by mechanisms similar to those acting on larger bubbles. The frothers
studied were MIBC and Dowfroth 250.
ACCEPTED MANUSCRIPT
1) Introduction
Understanding bubble swarm dynamics is the key to effective control of a variety of processes,
T
mineral flotation being one important example. An expanding knowledge of how bubbles
P
interact one-on-one, as well as globally within a swarm, is helping researchers and operators
RI
discover optimal surfactant chemistries and operating strategies. In terms of bubble properties,
SC
the most important surfactant in mineral flotation is the frother, added to control (reduce) bubble
size and promote froth formation. There is a continuing debate on the general impact that
NU
frothers have on elementary bubble properties (Harris, 1982; Pugh, 1996; Grau and Laskowski,
MA
2006; Acuña et al., 2007). A clear understanding of beneficial bubble characteristics is
complementary to the design of frothers, as well as the search for optimal operating conditions
ED
(Cappuccitti and Finch, 2007). The work here isolates the effect of frothers on bubble terminal
2) Theory
CE
water: (i) the motion of a fluid sphere balanced by buoyant and drag forces, and (ii) the motion
of a rigid sphere balanced by these same forces. The former seems to describe the rise of air
bubbles through water. However, the spherical regime dominates for small bubbles (d < 1 mm in
water), in which even a small amount of contamination induces a viscous layer at the air-water
interface, causing the surface to become ‘rigid’ (Clift et al., 2005a). The nature of the
contamination layer will be discussed in Section 2.2, but for now consider the cases depicted in
Fig 1.
Fig 1) An idealized spherical bubble, without any contamination at the air-water interface (a). A bubble with a
viscous layer separating the air from the water (b). An idealized rigid sphere (c).
ACCEPTED MANUSCRIPT
Given the presence of a viscoelastic layer (Fig 1b), the behaviour of a rising should be between
that of an idealised fluid sphere (Fig 1a), and that of an idealised hard sphere (Fig 1c), as the
layer obstructs the transmission of shear forces from the water to the air. Indeed, terminal
velocities can fill the entire spectrum from an ideal fluid sphere all the way to an ideal rigid
sphere (and beyond, if other mechanisms are considered, as discussed in Section 2.2).
P T
The terminal velocity of an ideal (Newtonian) fluid sphere rising through another ideal fluid was
RI
obtained by Rybczynski (1911) and independently by Hadamard (1911) for low Reynolds
SC
numbers (Re < 1) (Dukhin et al., 1998). Taking the viscosity and density of water to be much
NU
gd 2 ρ
U HR = (1)
MA 12η
where g is the gravitational constant, d is the bubble diameter, ρ is the density of water and η is
ED
the viscosity of water. The terminal velocity of a rigid sphere (Fig 1c) is the classical result of
PT
Stokes (1851),
CE
gd 2 ρ
U St = (2)
18η
AC
To derive these two equations, Re < 1 is assumed to simplify (linearize) the Navier-Stokes
equations. Bubbles ranging from 0.4 to 1.0 mm in diameter fall in an intermediate regime (1<
Re < 100), but Eq's 1 and 2 are still essential points of comparison because of their simplicity,
and because an analytical solution is inaccessible in this intermediate regime (Duhkin et al.,
1998). There are non-linear contributions in the intermediate region and departure from the
Almost universally added to mineral flotation pulps, frothers molecules help form the
thermodynamically difficult to displace1, and the bubble starts to become rigid. There are,
T
however other secondary mechanisms believed to diminish the energy available for upward
P
RI
motion. In fact, terminal velocities may fall below USt (Dukhin et al., 1998); evidently, the
obstruction of shear is one means to decrease the terminal velocity, but it cannot be the only
SC
means.
NU
The debate over secondary mechanisms has spanned nearly a century (Dukhin et al., 1998). As a
MA
bubble increases in speed, it is deformed into an oblate spheroid, storing potential energy
according to Hooke’s law. When the bubble diameters are below 1 mm, any deformation is
ED
linear and results in symmetric oblate spheroids, as confirmed by Duineveld (1995), who
Wobbling is a secondary kinetic phenomenon especially important for large bubbles (d > 1 mm).
CE
Vertical terminal velocities are further diminished if there is a helical component to the bubble
AC
spin along its own axis, the spin would take up kinetic energy, further decreasing the terminal
velocity. Linear and nonlinear deformation (storage of potential energy), as well as helical
motion and wobbling (secondary kinetic mechanisms), are common in bubbles but the organised
A spherical bubble experiences shear stress as it rises through the pulp, causing a tension
gradient in the longitudinal direction; the rear frother at the rear of the bubble tends to a
1
There is an entropy gradient pointing radially out of the bubble.
ACCEPTED MANUSCRIPT
compression limit, resulting in a stagnant cap. (The stagnant cap can be considered ultra-
viscous.) Therefore the tension gradient is coupled with a frother concentration gradient, in
which there is a frother shortage at the front, and a frother surplus at the rear. This results in an
T
ongoing adsorption of frother molecules in the front, with a simultaneous desorption of frother
P
RI
molecules from the rear. This action, due to the coupled tension and concentration gradients, is
the Marangoni Effect, and it induces surface perturbations, another secondary phenomenon.
SC
From this discussion, frothers induce two kinds of energy storage: in the viscous layer, shear is
NU
prevented from reaching the inner air and is trapped as potential energy; secondly, an additional
MA
amount of energy is expressed as surface perturbations and the motion of frother molecules,
hence a form of kinetic energy. The potential energy storage is comparable to the action of a
ED
mechanical spring or an electronic capacitor, while the kinetic energy storage is comparable to
Whenever a material, or more generally a physical space, stores energy in a conservative manner
CE
(be it kinetic or potential energy), the continued addition of energy favours dissipation, e.g.
AC
friction and the like. A steady terminal velocity is attained only when the rate of energy storage
is balanced by the dissipation. A priori, the bubble would have been subject to an unsteady
force, which is often described in two parts: the basset force and the added mass force (Brennen,
2005). The basset (a.k.a. the ‘history’) force addresses the viscous effects and the delay in the
The added mass force is due to the acceleration of the adjacent water (it has the same effect as an
increase in the bubble mass). According to Gélinas et al. (2005), frother molecules organize
water over some considerable distance (due to H-bonding). Perhaps the added water carried
ACCEPTED MANUSCRIPT
alongside the frother molecules is enough for a marked change in the acceleration as well as the
terminal velocity.
3) Experimental Setup
T
To study the effect of frother on the terminal velocities of bubble populations below 1mm, an
P
RI
experimental apparatus consisting of a trapezoidal column subtended by a surge tank, was placed
in cycle with a pump (Fig 2). The water is pulled from the bottom of the Plexiglas column and
SC
sent back up to the surge tank, so that the buoyancy forces pushing the bubbles upward are
NU
opposed by the downward drag from the water. When in balance, the net effect is that the
bubbles stay at a more-or-less fixed position in the column. Given the trapezoidal shape of the
MA
column, different heights have different cross-sectional areas, hence different downward water
velocities. Thus bubbles report to a height corresponding to their terminal velocity. Large
ED
bubbles have high terminal velocities, so they tend to report near the top (where the cross-
PT
sectional area is small); conversely, smaller bubbles have lower terminal velocities, hence they
Fig 2) Schematic of the experimental apparatus. The input and bypass valves are closed during operation, so that
water is circulated up directly to the surge tank, and down through the trapezoidal column back to the pump.
AC
Initially 40L of water (Antofagasta, Chile tap) enters through the input valve, passing through a
carbon filter and into the previously empty apparatus. At this point the input valve is closed and
the pump is turned on. The water exiting the pump is flow regulated with a valve and orifice
plate. Parallel to the orifice plate is a bypass used for cleaning purposes, containing a second
carbon filter. Between experiments, the bypass is opened and water is circulated to remove any
When the flow valve is completely open, the pull causes a vortex in the surge tank, through
which air is introduced into the column. This causes a turbulent mix of air throughout the entire
column, and under these conditions, the desired quantity of frother is added to the system via the
T
top of the surge tank. Conditioning with frother is allowed for ten minutes, and then the flow
P
RI
rate is lowered, upon which the largest bubbles escape through the surge tank. The remaining
bubbles are smaller than 1mm in diameter. With a combination of digital photography and
SC
image analysis (using Matlab), the diameter of these bubbles is measured as a function of their
NU
terminal velocities.
MA
The terminal velocities could be varied via two operating parameters: the volumetric flow rate of
water (Q), and the height at which the photographs are taken (y). For each of the frother
ED
chemistries, samples were taken under two flow rates and at three heights, giving six data points
per chemistry. The heights were at ⅓, ½ and ⅔ up the column, hence y values of 50 cm, 75 cm
PT
and 100 cm, respectively, and Q = 650 mL/s and 850 mL/s were the two flow rates. Five
CE
chemistries were examined: filtered water, 0.1 mmol/L MIBC, 0.5 mmol/L MIBC, 0.1 mmol/L
DF250, and 0.5 mmol/L DF250. This gives a total of thirty data points, each corresponding to
AC
ten pictures taken under a given combination of frother chemistry, column height and flow rate.
The frother chemistries are above the critical coalescence concentration (CCC), i.e. coalescence
has a negligible impact on bubble size. Depending on other operating conditions, the CCC is ca.
0.08 mmol/L for MIBC and 0.03 mmol/L for DF250. (For a more complete discussion on CCC,
the reader is referred to Nesset et al., 2007; Laskowski, 2003; Cho and Laskowski, 2002).
Q
U=
wl
ACCEPTED MANUSCRIPT
where w = 6 cm is the width of the column, and l is the length which is a linear function of the
height y above the base of the trapezoid. Using the dimensions given in Fig 5, the terminal
velocity becomes:
P T
25Q
U= (3)
RI
3000 − 14 y
SC
where Q is in mL/s and y is in cm; the resulting velocity is given in cm/s.
NU
The experimental setup described in this section has two attractive features: (i) it imposes an
innate classification of bubbles according to terminal velocities, and (ii) the desired bubbles are
MA
retained long enough to guarantee sufficient frother adsorption time to attain terminal velocity
(Zhang and Finch, 2001). (Holding a bubble in the trapezoid for the typical ten minutes of an
ED
In a typical trial, the temperature increased by ca. 4°C due to pumping. To determine the
CE
η = a ⋅ 10 b /(T −c ) (4)
AC
where a = 2.414⋅10-5 Pa⋅s, b = 247.8 °K, c = 140.0 °K, and the temperature T is taken in °K, on
the Kelvin scale. Eq 4 is in excellent agreement with data (Mott, 2006) pertaining to the
It is usual to consider Terminal Velocity vs. Bubble Size, (Clift et. al, 2005a). But the apparatus
described has made the terminal velocity the independent variable. Accordingly, the results of
12ηU
d HR = (5)
gρ
18ηU
T
d St = (6)
gρ
P
RI
Note that for a given velocity the Stokes result, dSt, is always larger than the HR result by a
SC
factor of (1.5)½ . This underscores the logic that the more energy is taken up by the interfacial
layer, the larger the diameters become for equivalent velocity. Therefore experimental diameters
NU
exceeding dSt would demonstrate the presence of secondary mechanisms that go beyond the
MA
action of an ultra-viscous layer. In the results section dHR and dSt are calculated.
For dimensionless analysis, the implied drag coefficient CD may be taken as,
ED
4 gd
CD = (7)
3U 2
PT
(By using Eq 2 to substitute USt as U , it can be found that CD,St = (24/Re) which is commonly
CE
referred to as Stokes’s Law.) Another relevant dimensionless quantity is the implied correction
AC
U St = kU
d2
k= 2
(8)
d St
Fig 3 shows two typical images, each capturing bubbles at y = 49 cm and in a 0.1 mmol/L
solution of DF250. The left picture has a flow rate of Q = 650 mL/s, giving a downward velocity
T
of 7.0 cm/s at the middle of the field of view; the right picture has Q = 850 mL/s, giving a
P
velocity of 9.2 cm/s. Since the bubbles are within the spheroid regime, they are appropriately
RI
characterized with the equivalent circular diameter; that is, the diameter giving the same circular
SC
area as the 2D projection of the bubble. The conditions in Fig 3a gave a Sauter mean bubble
diameter (d32 = Σdi3/Σdi2) of 0.45 mm with a standard error of 0.01 mm; the conditions of Fig 3b
NU
gave a mean of 0.59 mm, with the same standard error of 0.01 mm. (Note that the Sauter mean
MA
carries intrinsic data about the balance between buoyancy and drag.)
Fig 3) Sample photographs. In both pictures, the solution contains 0.1 mmol/L DF250. In (a) the water is
flowing downward at a velocity of 7.0 cm/s, and in (b) the velocity is 9.2 cm/s. The Matlab algorithm uses the
ED
rectangle on the left of the pictures (10 mm long) as a basis of comparison to measure the bubble diameters.
It was expected that for a given condition, a fixed velocity would imply a single corresponding
PT
bubble diameter. Instead, there was some spread in bubble diameters. Fig 3 is a snapshot of a
CE
set of bubbles that was in constant motion, leaving and re-entering the frame. A single bubble
would stay at a fixed height (velocity) in the column for an instant, and then accelerate upward
AC
halting at another height for several seconds, before returning to its initial height. Undoubtedly
there were non-uniform velocity fields, for example due to wall effects or bubble interactions.
(In the future, it would be desirable to measure bubbles near the center of the column, where the
wall-effects would be diminished. It would also be of interest to unleash single bubbles with d <
1 mm.)
The quantitative results are summarized in Table 1. Fig 4 displays bubble diameter as a function
of terminal velocity in the absence of frother (i.e. filtered water), along with the predictions made
Fig 4) Terminal Velocity vs Bubble Diameter, in Filtered Water. Diamonds mark the Sauter mean diameter
as a function of the terminal velocity. Thin, vertical black lines mark the bubble diameters covered by one
standard error above and below each of the means. Rectangles denote the predicted diameter ranges, from
Hadamard-Rybczynski (left side of the rectangles) to Stokes (right side of the rectangles), as a function of
terminal velocity.
T
These experimental results generally fall between dHR and dSt, hence agreeing with the
P
explanation in Section 2.1. It was found that the highest combination of height (y ~ 100 cm) and
RI
flow (Q ~ 850 mL/s or U = 12.8 cm/s) produced diameters 1.5 times larger than dSt, which will
SC
be discussed below.
NU
Virtually all of Fig 5 demonstrates that frothers decrease bubble rise velocities below the result
of Stokes, which confirms the importance of the secondary mechanisms. The comparison
MA
between Figs 4 and 5 suggest that the perturbations on the air-water interface bear a significant
impact on small spherical bubbles, resulting in diameters that are higher than dSt, for equivalent
ED
velocity.
Fig 5) Bubble diameters are displayed as a function of terminal velocities, under different frother chemistries (The
PT
markings follow the same convention as in Fig 4): a) 0.1 mmol/L MIBC, b) 0.5 mmol/L MIBC, c) 0.1 mmol/L
DF250 and d) 0.5 mmol/L DF250.
CE
Fig 6a demonstrates that the filtered water consistently yields higher drag coefficients for
AC
equivalent Reynolds numbers. This is the trend reported in Clift et al. 2005a accepted for larger
bubbles. A similar conclusion can be drawn from Fig 6b: the correction factor was below unity
for all of the points pertaining to filtered water, and above unity for nearly all of the frother
points (the only discrepancy is the lowest velocity point for 0.1 mmol/L MIBC). Note that the
filtered water points seem to follow a smooth trend, whereas the frother points are somewhat
erratic.
Fig 6) The implied drag coefficient CD plotted as a function of Re (a); the implied correction factor for
Stokes’s Equation k plotted as a function of Re (b).
ACCEPTED MANUSCRIPT
For the filtered water case, the rightmost point had been omitted from Fig 6 because it was the
only case in this chemistry where the Sauter mean bubble size (0.68 mm) was higher than the
P T
Stokes result (dSt= 0.45 mm), suggesting that it was the only point that was heavily influenced by
RI
the secondary mechanisms discussed in Section 2.2. It also had more than twice as many
SC
measurable bubbles than any other point for the filtered water chemistry. (198 bubbles were
observed at U = 12.8 cm/s in the filtered water, which more than doubles the 90 bubbles that
NU
were observed at U = 10.9 cm/s). Furthermore, the Reynolds number was roughly twice as big
MA
as the next highest value (Re = 99.77 for U = 12.8 cm/s, compared to Re = 50.22 cm/s for U =
10.9 cm/s). The anomalously high bubble diameter is tentatively attributed to swarm
ED
Table 1) Summary of Experimental Results. U = terminal velocity, Q = volumetric flow rate, y = distance
PT
between the camera and to the bottom of the column, T = temperature in the column, n = number of measurable
bubbles, d32 = Sauter mean diameter in the swarm, Re = Reynolds number, σ = standard deviation away from the
Sauter mean, SE = σn-½ = standard error of the mean.
CE
5) Conclusions
AC
terminal velocities of bubbles having diameters smaller than 1mm; using a trapezoidal column,
such bubbles were held in place at a given velocity and the corresponding bubble size (Sauter
mean) determined. In the case of filtered water, the bubble size was within the range given by
the Hadamard-Rybczynski and Stokes models. For two common industrial frothing agents, the
alcohol MIBC and the polyglycol Dowfroth 250, the velocity was below that of the Stokes
model. This indicates that secondary mechanisms are involved in the slowing of bubbles smaller
than 1 mm; this trend had already been accepted for bubbles larger than 1mm.
ACCEPTED MANUSCRIPT
Acknowledgements
This work was conducted in the Chemical Engineering Department of the Universidad Católica
del Norte in Antofagasta (Chile), which provided materials and technical staff to this project.
T
Financial assistance was provided by the Chair in Mineral Processing at McGill University,
P
funded by NSERC (Natural Sciences and Engineering Research Council of Canada) through the
RI
CRD (Collaborative Research and Development) program sponsored by (2007) CVRD Inco,
SC
Xstrata Process Support, Teck Cominco, Agnico-Eagle, Shell Canada Energy, Flottec, SGS
NU
References
MA
1. C. Acuña, J.E. Nesset and J.A. Finch, ‘Impact of Frother on Bubble Production and
Behaviour in the Pulp Zone,’ Proceedings Copper/Cobre 2007, Mineral Processing, Vol
II (Eds. R del Villar, J. E. Nesset, C. O. Gomez, and A. W. Stradling), CIM, pp 197-210.
ED
T
13. R.L. Mott, ‘Applied Fluid Mechanics,’ 6th Edition, Prentice Hall, Pearson Prentice Hall,
P
Upper Saddle River (NJ), 2006.
RI
14. J.E. Nesset, J.A. Finch and C.O. Gomez, ‘Operating Variables Affecting the Bubble Size
in Forced-Air Mechanical Flotation Machines,’ Proceedings 9th Mill Operators’
SC
Conference, Fremantle (WA), 2007, pp 55-65.
15. R. J. Pugh, ‘Surface Chemical Studies on Particle Stabilized Froths,’ Proceedings 23rd
International Mineral Processing Congress, 2006, pp 205-225.
NU
16. Y. Zhang and J.A. Finch, ‘A note on single bubble motion in surfactant solutions,’
Journal of Fluid Mechanics, vol. 249, 2001, pp 63-66.
MA
ED
PT
CE
AC
ACCEPTED MANUSCRIPT
T
Fig 1) An idealized spherical bubble, without any contamination at the air-water interface (a). A bubble with a
P
viscous layer separating the air from the water (b). An idealized rigid sphere (c).
RI
SC
NU
MA
ED
PT
CE
AC
Fig 2) Schematic of the experimental apparatus. The input and bypass valves are closed during operation, so that
water is circulated up directly to the surge tank, and down through the trapezoidal column back to the pump.
ACCEPTED MANUSCRIPT
P T
RI
SC
NU
MA
ED
Fig 3) Sample photographs. In both pictures, the solution contains 0.1 mmol/L DF250. In (a) the water is
flowing downward at a velocity of 7.0 cm/s, and in (b) the velocity is 9.2 cm/s. The Matlab algorithm uses the
PT
rectangle on the left of the pictures (10 mm long) as a basis of comparison to measure the bubble diameters.
CE
P T
RI
SC
NU
MA
ED
PT
CE
AC
Fig 5) Bubble diameters are displayed as a function of terminal velocities, under different frother chemistries (The
markings follow the same convention as in Fig 4): a) 0.1 mmol/L MIBC, b) 0.5 mmol/L MIBC, c) 0.1 mmol/L
DF250 and d) 0.5 mmol/L DF250.
ACCEPTED MANUSCRIPT
P T
RI
SC
NU
Fig 6) The implied drag coefficient CD plotted as a function of Re (a); the implied correction factor for
Stokes’s Equation k plotted as a function of Re (b).
MA
Filtered Water
y T
ED
y T
Trial U (cm/s) Q (mL/s) (cm) (ºC) n d32 (mm) Re σ (mm) SE(mm)
7 7.0 650 49 28 12 0.25 20.82 0.04 0.01
8 9.2 850 49 28 123 0.44 47.68 0.15 0.01
9 8.2 650 73 27 91 0.50 48.32 0.21 0.02
10 10.7 850 73 27 63 0.54 67.37 0.24 0.03
11 9.7 650 95 24 56 0.53 56.55 0.26 0.03
12 12.7 850 95 24 213 0.67 93.37 0.29 0.02
MIBC (0.5 mmol/L)
y T
Trial U (cm/s) Q (mL/s) (cm) (ºC) n d32 (mm) Re σ (mm) SE(mm)
13 7.1 650 49 22 250 0.57 41.75 0.21 0.01
14 9.2 850 49 22 350 0.58 55.92 0.21 0.01
15 8.0 650 69 25 146 0.43 38.38 0.16 0.01
16 10.4 850 69 25 661 0.57 67.11 0.24 0.01
ACCEPTED MANUSCRIPT
T
19 7.0 650 49 22 112 0.45 33.17 0.14 0.01
P
20 9.2 850 49 22 539 0.59 56.54 0.15 0.01
21 8.2 650 73 25 164 0.48 43.99 0.19 0.01
RI
22 10.7 850 73 25 1242 0.59 70.29 0.16 0.00
23 10.2 650 100 27 1383 0.45 53.39 0.12 0.00
SC
24 13.3 850 100 27 1520 0.57 88.48 0.14 0.00
DF250 (0.5 mmol/L)
y T
NU
Trial U (cm/s) Q (mL/s) (cm) (ºC) n d32 (mm) Re σ (mm) SE(mm)
25 7.0 650 49 26 74 0.34 27.09 0.09 0.01
26 9.1 850 49 26 185 0.48 50.33 0.17 0.01
MA
27 8.5 650 79 24 303 0.46 42.87 0.16 0.01
28 11.2 850 79 24 1030 0.57 69.63 0.20 0.01
29 9.9 650 98 23 1035 0.52 55.34 0.18 0.01
ED
Table 1) Summary of Experimental Results. U = terminal velocity, Q = volumetric flow rate, y = distance
PT
between the camera and to the bottom of the column, T = temperature in the column, n = number of measurable
bubbles, d32 = Sauter mean diameter in the swarm, Re = Reynolds number, σ = standard deviation away from the
Sauter mean, SE = σn-½ = standard error of the mean.
CE
AC