Proceedings of the 3rd Annual World Conference

of the Society for Industrial and Systems Engineering,

San Antonio, Texas, USA
October 20-22, 2014

Analysis and Modeling of Roundness Error in the Design Process of a Measuring

Machine
J. M. Daz-Mendoza, J. Molina, L. Rico, and L. Vidal
Universidad Autnoma of Cd. Jurez
Henry Dunant 4016, Zona Pronaf
Cd. Jurez Chih. Mxico, C.P. 32310
Ph (52656-68848439) Ext. 4644
Corresponding author's Email: juan.diaz@uacj.mx
Abstract: Roundness is a geometric form tolerance required for circular or cylindrical parts used in a great amount of
mechanical assemblies. This form tolerance is controlled in order to assure the correct performance and lifetime of an
assembly. The design and construction of a roundness machine requires the ability to measure, analyze and validate
roundness on parts, to assure the measuring capability. The analysis applied in this paper uses two international methods to
determine roundness error: minimum zone circle (MZC) and least square circle (LSC); The calibration of a Roundness
Standard of the National Bureau of Standards from USA was used to elaborate the study. The referred methods use polar
coordinates data as primary source. The analysis performed using these methods help to determine the preliminary machine
error in measuring roundness.
Keywords: Roundness, Geometric Characteristic, Least Square Circle, Minimum Zone Circle, Polar Coordinate, Error
Evaluation, Measuring Standard, Coordinate Data

1. Introduction
The measuring process is a key factor to determine if product dimensional and geometric features are inside
specified tolerances. The components design process is critical to product performance, whether is on a prototype phase or
trial run prior to production. Since product performance is a key factor, dimensions and tolerances are relevant to meet the
requirements in the final design. Roundness geometric tolerance is critical to circular machined parts (Lei, 2011), which has
been a feature discussed in several researches (Dhanish, 2002). (Dhanish, 2002) also points out that Least squares has been
broadly used for the calculation and analysis of roundness. However, ISO 1101 establishes that MZC is the required
methodology to estimate roundness error.

1.1 Minimum Zone Circle (MZC)

According to the standard ISO 1101, roundness is defined as: two concentric circles limit the tolerance zone in the
considered cross-section, with a difference in radii of t. (DIN, 2008). However, the method to calculate the roundness error
is not given by ISO (Dhanish, 2002). The description of the minimum zone circle (MZC) is as: minimum zone deviation
form from ISO 1101 standard (Lai & Chen, 1996) and (Moroni & Petr, 2007). Moroni also indicates that the method to
estimate MZC is a non-linear problem, which calculation evolves with some difficulty. In Figure 1-a is shown the geometric
interpretation of MZC (DIN, 2008). There has been work done to estimate MZC applying optimization and graphical
methods to obtain the tolerance value. However, among these analyses, as mentioned by (Moroni & Petr, 2007), several of
the methods transform the non-linear problem into a linear system.

1.2 Least Squares Circle (LSC)

The least square method has been used to determine the best fit for a line, circle or higher degree curve. It has been
applied in several areas such as: statistics, finance, economics engineering and others. However, the least square method has
been used longer than the MZC to determine the least square circle (LSC), since its mathematical roots is the Gauss
minimum square method. LSC has been used in coordinate measuring machines (CMM) and probe machines. Nevertheless,
LSC has a limitation to estimate the mean best fit for a linear function or nonlinear function from a set of points. This does

ISBN: 97819384960-2-8

340

Proceedings of the 3rd Annual World Conference

of the Society for Industrial and Systems Engineering,
San Antonio, Texas, USA
October 20-22, 2014
not comply with the ISO requirements, since it estimates the vertical offset distance among data points collected to fit a
circle. Figure 1-b shows LSC determination.

Figure 1. a) Left. Minimum Zone Circle (Wentao, 2012); b) Right. Least-square Circle (Reeve, 1979)

1.3 Machine Characteristics

In metrology laboratories of metal-mechanics industry, roundness measuring is required for a lot of components
such as: shafts, nozzles, needles, cylinder bores and others. Understanding and measuring correctly roundness is key to
machine and product design processes. This is in an incentive to develop certain knowledge regarding roundness
measurement machines either in the mathematical as in the design mechanical approaches. Therefore, the metrology research
group has considered designing a machine of this type, with primary focus on academic applications in small shafts and
similar parts. The machine design evolution will go from basic mechanical measuring of roundness to other geometric
features such cylindricity, straightness and others. The machine in the final design will integrate mechatronics systems.
Several of the characteristics that will be discussed in this paper are related to roundness evaluation, and are based in two
methods.

2. Methodology
The ability to perform the evaluation of roundness error in the design process of the machine is determinant to
establish the capability of the measurement roundness machine before the design conceptual phase is set. Since both
methods, LSC and MZC have been used to measure roundness; the paper approach collects data from a master pin, and
develops the calculation for both to determine the roundness error in a conceptual machine mechanism. The analysis uses two
best fit approaches of MZC to find the minimum error. The data collection points is developed considering three features in
the device mechanism which consist in a base with a high precision bearing mount, mounting chuck for holding, and the part
under measurement. The set up is aligned in the CMM to assure the adequate framework datum reference (zero alignment,
base plane and perpendicularity among them.

2.1 Least Square Circle (LSC)

Least square Circle (LSC) is based in the minimum least square method from the Gauss approach, since is the most
used tool in several papers (G.L Samuel, 2000), (Lei, 2011), (Huang, 2001). The method is based in the vertical offset of
data points that fits in a mean minimum circle. We applied the basic approach for LSC using the generic Gauss method
(Bullock, 2006) and (Reeve, 1979) where the base is shown in equation (1).
( )2 + ( )2 = 2

(1)

341

Proceedings of the 3rd Annual World Conference

of the Society for Industrial and Systems Engineering,
San Antonio, Texas, USA
October 20-22, 2014
The equation will be formulated to determine the minimum distance from ( , ),since ()/ = 0 the function will
be differentiate and equal to zero. However, the function has to be partially differentiated for each variable and transformed
from a non-linear function into a linear function using equation (2).
=
=

(2)

+ = 1/2( + )
+ = 1/2( + )

(3)

Considering that S = (g(ui , vi ))2 , where g(u, v) = (u uc )2 + (v vc )2 and = R2 (Bullock, 2006),

differentiating S for each u,v and , a solvable linear equations system may be established to determine the circle center and
radius which is subtracted from the master pin diameter and find the roundness error. The formulas are in equation (3).

2.2 Minimum Zone Circle (MZC)

The MZC has been analyzed in many studies since there is no specific method to estimate it. In this paper, the loop
controlled algorithm (Geovanni Moroni, 2007) is applied. This algorithm iterates quickly and provides an accurate MZC
value. This approach combines a brute algorithm and an optimization process by selecting randomly a set of points of
measured circles. There are two considerations: at least two points should be from the outer circle and two points from the
inner circle. Each iteration selects continuous points randomly. However, in this study the selected points are orthogonal
between each other to have a clear circle selection. The study uses a matrix construction from the selected points that
estimates a new circle center and then calculates a radius. The equations were generated from the arrangement in equation (4)
(Geovanni Moroni, 2007), which is random. This arrangement is then used to formulate the linear equations expressed in
equation (5).
1,2 3,4 [ = 1]
1,3 2,4 [ = 2]
1,4 2,3 [ = 3]

(4)

2(2 1 ) + 2(2 1 ) = 22 + 22 (12 + 12 )

2(4 3 ) + 2(4 3 ) = 42 + 42 (32 + 32 )

(5)

The system has two variables which are and which are the new points for the coordinate center (Geovanni Moroni,
2007). Once solution is obtained radii can be calculated for each of the points and the center found, this will help to
determine MZC using equation (6) (Wentao Sui, 2012).
= ( )

(6)

Determining the circle center becomes the key factor in the paper method. In this study the iteration process was
performed only two times as described above.

2.3 Data Collection

A reference master pin is used for the study, it is measured mounted on a base, which is set up in a CMM to collect
coordinated data points. The units used in the study were millimeters. Twelve data points (, ) were collected and are listed
in Table 1.
Table 1. Coordinate Measured Points
Data

1a

2a

3a

4a

5a

6a

-6.9485

-3.2283

2.0136

4.8936

7.5549

7.6781

5.9796

0.8106

-1.2392

-4.5095

-7.8876

-7.7455

-3.9766

-7.3287

-7.8904

-6.3519

-3.0382

2.3408

5.3767

8.0185

8.0464

6.7212

1.3788

-2.0504

342

Proceedings of the 3rd Annual World Conference

of the Society for Industrial and Systems Engineering,
San Antonio, Texas, USA
October 20-22, 2014

3. Results
The roundness analysis was performed using Table 1 set of data to calculate the error using LSC and MZC. The
result shows that the roundness obtained by LSC is higher than MSC result. This, correlates with other studies done by (Lei,
2011), (Dhanish, 2002), (Geovanni Moroni, 2007) and (Huang, 2001).

3.1 Least Square Circle (LSC)

The study result shows an error equal to 0.05438 mm. Table 2 shows the values and coefficients estimated for
equations (4) and (5).

Table 2. Equations System Values Coefficients

1
2
3
4
5
6
7
8
9
10
11
12

-6.7295
-3.0093
2.2326
5.1126
7.7739
7.8971
6.1986
1.0296
-1.0202
-4.2905
-6.7295
-3.0093
2.2326

_ _

-4.0805
-7.4326
-7.9943
-6.4558
-3.1421
2.2370
5.2729
7.9147
7.9426
6.6174
-4.0805
-7.4326
-7.9943

27.4593
22.3666
-17.8481
-33.0058
-24.4260
17.6655
32.6844
8.1491
-8.1029
-28.3916
27.4593
22.3666
-17.8481

= 2

45.2859
9.0558
4.9846
26.1388
60.4338
62.3645
38.4228
1.0601
1.0408
18.4082
45.2859
9.0558
4.9846

= 3

-304.7510
-27.2514
11.1287
133.6379
469.8072
492.4994
238.1685
1.0915
-1.0618
-78.9803
-304.7510
-27.2514
11.1287

= 2

16.6501
55.2428
63.9080
41.6767
9.8725
5.0039
27.8029
62.6417
63.0841
43.7893
16.6501
55.2428
63.9080

= 3

-67.9398
-410.5949
-510.8968
-269.0544
-31.0198
11.1936
146.6008
495.7870
501.0486
289.7693
-67.9398
-410.5949
-510.8968

= 2

-112.0464
-166.2412
142.6821
213.0770
76.7478
39.5167
172.3398
64.4969
-64.3573
-187.8774
-112.0464
-166.2412
142.6821

= 2

-184.7870
-67.3076
-39.8480
-168.7459
-189.8860
139.5062
202.5979
8.3904
8.2664
121.8138
-184.7870
-67.3076
-39.8480

The coefficients equations were built from Table 2. In a excel sheet, the values for uc and vc were determined to find the
new center coordinate and to calculate the radius R. The systems equations are:
382.65049 + 2.98732 = 93.95254
2.98733 + 395.93838 = 35.04440

These system equations (Table 3) provided the values of and to estimate the new center and the radius.
Table 3. Initial Data Points

0.24624

-0.09037

0.02722

0.01348

64.9512

8.05923

16.11846

These results show the calculated diameter of the reference master pin which is 16.118462 mm. The stated pin value is
16.0097 mm, the roundness is calculated with the difference between both radii:
= 8.05923 8.00485 Then = 0.05438

343

Proceedings of the 3rd Annual World Conference

of the Society for Industrial and Systems Engineering,
San Antonio, Texas, USA
October 20-22, 2014

3.2 Minimum Zone Circle (MZC)

In this paper, the method to estimate MZC, is the loop controlled algorithm (LCA) proposed by Moroni (Geovanni
Moroni, 2007) . This method has advantages as, fast calculations, and requires only two iterations a preliminary roundness
value was determined. Table 4 has the arrangement and four set of points on each iteration were use calculate two orthogonal
diameters considering that one is an outer diameter and other is the inner diameter.

Table 4. Data Points Expanded in Power 2

1a

2a

3a

4a

5a

6a

48.2816

10.4219

4.05458

23.9473

57.0765

58.9532

35.7556

0.65707

1.53562

20.3355

62.2142

59.9927

15.8133

53.7098

62.2584

40.3466

9.2307

5.4793

28.9089

64.2963

64.7445

45.1745

1.9011

4.2041

In the first iteration, the set of points x1 , x1a , x4 and x4a are used to create the equations system for the first iteration
results.
25.8562 + 18.7066 = 0.56951924
9.4031 + 13.0731 = 0.80287743
Solving this system the results for ai and bi give the data to calculate the new circle center and a first radius which
values are in Table 4. In the second iteration, the set of four points: x3 , x3a , x6 and x6a were randomly selected and then
resulted the equation system to be solved.
6.5056 + 31.8736 = 0.03282752
30.8472 8.7824 = 0.23565384
The solution for ai and bi in this second equations system, a new circle center and a radius are calculated. Table 5 shows
the both results.
Table 5. Iterations Results
Variables

First iteration results

0.0075

0.0005

-0.21152

0.10435

64.31515

8.01967

16.03935

Second iteration results

0.04371

-0.02997

-0.1753

0.07388

64.46688

8.02913

16.05825

The results were estimated using LCA from Moroni & Petrs work, provided two that initial diameters were used to
estimate the roundness. Then, using equation (7) a direct t value is found.
= (8.019672892 8.029127315) = 0.00945442
The estimated MZC value is based on the collected data points as well as for LSC. The reference master pin together
with the calculated LSC and MZC are shown in Table 6.

Table 6. Comparison Roundness Results

Reference
Master pin
LSC
MZC

(value in mm)
8.00485
8.05923
8.01967

344

(error in mm)
0
0.0544
0.0140

Proceedings of the 3rd Annual World Conference

of the Society for Industrial and Systems Engineering,
San Antonio, Texas, USA
October 20-22, 2014

4. Summary and Conclusions

The measuring, analysis and modeling of roundness performed for the roundness machine using LSC and MZC
methods, verifies the preliminary capability. As is shown in the results, LSC gives a mean circle diameter which is compared
to the master pin value, giving an estimated roundness error of 0.0543 mm which is higher than the MZC value. The study
also shows that the Moroni algorithm is very objective and provides acceptable results. Thus, the following conclusions may
be considered: LSC provides a fast initial reference, however, it does not meet ISO requirements; LSC also may be modified
using different sets of data points to estimate roundness; MZC is friendly and provides good roundness results even though
only a few iterations were made; Further work is necessary to a better understanding of the LCA to estimate MZC for
roundness value.

5. References
Bullock, R. (2006). Least-sqaures Circle fit. Retrieved from http://www.dtcenter.org/met/users/docs/write_ups/circle_fit.pdf
Dhanish, P. (2002). A simple algorithm for evaluation of minimum zone circulariry error from coordinate data. Machine
tools & Manufacture , 42, 1589-1594.
Samuel, G. L., & Shunmugam, M. S. (2000). Evaluation of circularity from coordinate and form data using computational
geometric techniques. Precision Engineering, 24(3), 251-263.
Moroni, G., & Petr, S. (2008). Geometric tolerance evaluation: a discussion on minimum zone fitting algorithms. Precision
engineering, 32(3), 232-237
Huang, J. (2001). A new strategy for circularity problems. Precision Engineering , 25, 301-398.
Lai, J. Y., & Chen, I. H. (1996). Minimum zone evaluation of circles and cylinders. International Journal of Machine Tools
and Manufacture, 36(4), 435-451.
Lei, X., Zhang, C., Xue, Y., & Li, J. (2011). Roundness error evaluation algorithm based on polar coordinate
transform. Measurement, 44(2), 345-350.
Reeve, C. P. (1979). The calibration of a roundness standard. US Department of Commerce, National Bureau of Standards.
DIN, E. 1101 (2008) Geometrical product specifications (GPS): geometrical tolerancingtolerances of form, orientation,
location and run-out. Beuth, Berlin.
Wentao Sui, D. Z. (2012). Four methods for roundness evaluation. Physics procedia , 24, 2159-2164.

345