Method for balancing an article for rotation
View Patent ↗A method for statistically analyzing the operation of a machine for balancing articles for rotation so as to reduce measurement errors and thereby enhance productivity and quality including the steps of collecting raw data during production balancing of articles for rotation, and statistically analyzing the data to determine appropriate process control limits for prompting investigative or corrective action if a process control limit is exceeded.
1. A method for statistically analyzing the operation of a balance system for balancing articles for rotation so as to reduce measurement errors comprising:
a) collecting raw data during balancing of articles for rotation, including, for multiple articles:
i) measuring a first imbalance of the articles in a first position; and
ii) measuring a flip imbalance of the article in a second position, rotationally displaced from the first position;
b) calculating a midpoint of the first imbalance and the flip imbalance for each of the multiple articles;
c) statistically analyzing the midpoint data; and
(d) in response to step (c), performing a corrective action to reduce measurement errors.
2. The method according to claim 1 , wherein step c) further comprises the step of determining process control limits so as to discriminate between midpoints that are out of statistical control from midpoints that are within statistical control.
3. The method according to claim 2 , wherein step a) further comprises the collection of the first imbalance and flip imbalance data in polar vector form; and
step b1) further comprises the steps:
b2) converting the first imbalance and flip imbalance data to Cartesian coordinates by computing the X and Y components of each polar vector; and
b2) calculating a mean for each pair of initial and flipped data points by averaging the X and Y components thereof to generate an X_Midpoint and a Y_Midpoint.
4. The method according to claim 3 , wherein step c) further comprises the steps:
c1) calculating a sample mean Xbar for the X_Midpoints and a sample mean Ybar for the Y_Midpoints of the data sample;
c2) calculating a sample standard deviation s 1 for Xbar over all samples and a sample standard deviation s 2 for Ybar over all samples;
c3) estimating a common standard deviation s based upon the sample standard deviations s 1 and s 2 ;
c4) calculating at least one of a warning limit R WL and a control limit R CL based upon the common standard deviation s; and
c5) generating at least one of a warning limit circle having a radius equal to the warning limit R WL and a control limit circle having a radius equal to the control limit R CL , centered at an origin of a polar coordinate graph.
5. The method according to claim 4 , wherein step c) further comprises, following step c5), the steps:
c6) conducting single pass editing of midpoints to eliminate outlying midpoints to leave an adjusted sample of midpoints; and
c7) reconstructing the at least one of a warning limit circle having a radius equal to a reconstructed warning limit R WL based upon the adjusted sample and a control limit circle having a radius equal to a reconstructed control limit R CL based upon the adjusted sample.
6. The method according to claim 1 , wherein step c) includes the steps:
c1) calculating a sample mean Xbar for the X_Midpoints and a sample mean Ybar for the Y_Midpoints of the data sample;
c2) calculating a sample standard deviation s 1 for Xbar over all samples and a sample standard deviation s 2 for Ybar over all samples;
c3) estimating a common standard deviation s based upon the sample standard deviations s 1 and s 2 ; and
c4) calculating a Maximum Acceptable Line-Out value based upon the formula MALO=FullBalanceSpec−(R BIAS +√{square root over (2*2.4 5)}s), where FullBalanceSpec is the total balance tolerance and R BIAS is the magnitude of the bias.
7. The method according to claim 1 , wherein step a) includes, before step a) i), the steps:
A) measuring and recording an amount of initial imbalance (before any correction weights are applied) that is present in the articles; and
B) if the initial imbalance exceeds a pre-determined limit, taking corrective action to bring the articles to within the pre-determined limit of imbalance, whereby the articles are then ready for measurement of the first imbalance thereof in step a) i).
8. The method according to claim 1 , wherein step c) further comprises the step of analyzing patterns that are created by the distribution of the midpoints on a polar coordinate graph so as to provide an understanding of the data by determining the slope of a least squares regression line through the original plotted location, and transforming the plot of the data so that the data is centered at the origin of an Cartesian coordinate graph with a slope equal to zero, whereby the X and Y components of the midpoints may be easily analyzed to determine if the plot exhibits a significant departure from circularity.
9. The method according to claim 1 , wherein step c) further comprises the step of performing measurement error analysis so as to measure an amount of error that is observed by the balancing system.
10. The method according to claim 9 , wherein the balancing system includes a balancing machine and a fixture for operatively connecting the articles to the balancing machine, a setup being defined as an instance of operation of the balancing system from the time of an instance of installing and balancing a particular fixture on a particular balancing machine, and calibrating the machine, until the fixture is removed or the machine is re-calibrated, step c) further comprising estimating the variance of midpoint data collected during a setup, V M , according to the following formulas:
V
MX
=
{
∑
s
[
∑
m
(
X
sm
-
X
_
s
)
2
/
(
M
s
-
1
)
]
}
/
S
,
V
MY
=
{
∑
s
[
∑
m
(
Y
sm
-
Y
_
s
)
2
/
(
M
s
-
1
)
]
}
/
S
,
and
V
M
=
(
V
MX
+
V
MY
)
/
2
where
S is the number of setups,
M S is the number of midpoints in the s-th setup;
m is the index for midpoint;
V MX is the estimate of the variance of the X components of midpoint data collected within the s-th setup,
V MY is the estimate of the variance of the Y components of midpoint data collected within the s-th setup,
X sm is the m-th X_Midpoint within the s-th setup,
X_Midpoint is the X component of a midpoint,
X s represents the average X_Midpoint for the s-th setup,
Y sm is the m-th Y_Midpoint within the s-th setup, Y_Midpoint is the Y component of a midpoint, and Y s represents the average Y_Midpoint for the s-th setup.
11. The method according to claim 9 , wherein the balancing system includes a balancing machine and a fixture for operatively connecting the articles to the balancing machine, a setup being defined as an instance of operation of the balancing system from the time of an instance of installing and balancing a particular fixture on a particular balancing machine and calibrating the machine, until the fixture is removed or the machine is re-calibrated, step c) further comprising estimating the common between-setup variance of midpoint data, V s , according to the following formulas:
M
S
Bar
=
(
∑
S
M
S
-
∑
S
M
S
2
/
∑
S
M
S
)
/
(
S
-
1
)
,
V
SX
=
[
∑
S
[
(
M
S
)
(
X
S
_
-
X
_
)
2
]
/
(
∑
S
M
S
-
M
S
Bar
)
]
,
V
SY
=
[
∑
S
[
(
M
S
)
(
Y
_
S
-
Y
_
)
2
]
/
(
∑
S
M
S
-
M
S
Bar
)
]
,
and
V
S
=
(
V
SX
+
V
SY
)
/
2
where
M s Bar is the average number of midpoints in a setup,
X s is the average X_Midpoint for the s-th setup,
X_Midpoint is the X component of a midpoint,
Y s is the average Y Midpoint for the s-th setup,
Y_Midpoint is the Y component of a midpoint,
X is the average X_Midpoint for all setups,
Y is the average Y_Midpoint for all setups,
V sx is the estimate of the common between-setup variance of the X components of midpoint data,
V SY is the estimate of the common between-setup variance of the Y components of midpoint data,
M s is the number of midpoints in the s-th setup, and S is the number of setups.
12. The method according to claim 10 , step c) further comprising calculating the components of variation due to Repeatability and Reproducibility, R&R, according to the following:
M
S
Bar
=
(
∑
S
M
S
-
∑
S
M
S
2
/
∑
S
M
S
)
/
(
S
-
1
)
,
V
SX
=
[
∑
S
[
(
M
S
)
(
X
S
_
-
X
_
)
2
]
/
(
∑
S
M
S
-
M
S
Bar
)
]
,
V
SY
=
[
∑
S
[
(
M
S
)
(
Y
_
S
-
Y
_
)
2
]
/
(
∑
S
M
S
-
M
S
Bar
)
]
,
V
S
=
(
V
SX
+
V
SY
)
/
2
,
σ
M
2
=
V
M
,
σ
S
2
=
V
S
-
σ
M
2
/
M
S
Bar
,
σ
M
=
σ
M
2
σ
S
=
σ
S
2
EV
=
5.15
σ
M
,
AV
=
5.15
σ
S
,
and
R
&
R
=
EV
2
+
AV
2
where
M s Bar is the average number of midpoints,
X is the average X_Midpoint for all setups,
Y is the average Y Midpoint for all setups,
V sx is the estimate of the common between-setup variance of the X components of midpoint data,
V sy is the estimate of the common between-setup variance of the Y components of midpoint data,
V s is the common between-setup variance of midpoint data,
σ M 2 is the midpoint to midpoint variance,
σ M is the standard deviation of the midpoints,
σ s 2 is the setup to setup variance,
σ s is the standard deviation of the setups,
EV is Equipment Variation, and
AV is Setup (Appraiser) Variation.
13. The method according to claim 12 , wherein R&R is related to Percent of Tolerance according to the formula
%R&R32 100 (R&R/Total Tolerance), where
Total Tolerance =(maximum allowable imbalance)×2/√{square root over (2)}.
14. The method according to claim 12 , further including calculating of the portion of the maximum allowable imbalance that is taken up by Bias expressed as a percentage, Bias Percent of Tolerance, utilizing the following formulas:
N
=
∑
S
M
S
,
X
Bias
=
∑
i
=
1
N
X_Midpoint
/
N
,
Y
Bias
=
∑
i
=
1
N
Y
-
Midpoint
/
N
,
and
R
Bias
=
(
X
Bias
2
+
Y
Bias
2
)
,
%
BIAS
=
100
(
R
BIAS
)
/
(
Total
Tolerance
)
where
N is the total number of midpoints,
X BIAS is the X component of Bias,
Y BIAS is the Y component of Bias, and
Total Tolerance=(maximum allowable imbalance)×2/√{square root over (2)}.