Featured Product
This Week in Quality Digest Live
CMSC Features
Ryan E. Day
September Speaker Series in review
Ryan E. Day
Realigning a cornerstone of industry
David H. Parker
Practical implications for electronic distance measurement
Belinda Jones
Users of CMMs, industrial scanners, and portable metrology systems are urged to participate
Belinda Jones
Register for a day-long test-drive of 3DMU at no charge

More Features

CMSC News
New facility in Toronto area will showcase multiple Hexagon product lines
API division named ‘Top External Provider 2018’
Exact Metrology selected for project
Faster and more powerful than ever before
Accurate measurement out of the box
Engineering and design teachers will benefit from enhanced 3D scanning performance
Partnering with FARO Technologies
Structured-light 3D scanner for industrial design professionals

More News

CMSC

Point Cloud Measurements on a CMM Artifact Using a Laser Scanner

Published: Wednesday, March 28, 2018 - 10:38

Tight-tolerance part inspection, whether for industrial applications or national laboratory requirements, is generally performed using cartesian coordinate measuring machines (CMMs). High-data density measurement sets to characterize part geometries are achieved using tactile probing or dynamic scanning to ensure high-quality part inspection. Requirements such as high accuracy and high data density limit the class of measurement instrument that can produce the necessary and satisfactory results. High-accuracy measurements are limited to hard probing with low-data density, while high-data density may require an optical alternative with lower accuracy in the resulting measurement data. The question becomes: Is there an alternative technology which can yield a combination of these two sought-after qualities?

To address these issues, an articulating arm CMM (AACMM) with an integrated laser scanner is a frequently opted-for alternative. The research in this article investigates the accuracy, precision, and measurement capability associated with laser scanning to generate point clouds of common, unclassified CMM hemispherical shell artifacts used in a nuclear weapons laboratory.

Introduction

Hemispherical shell inspection efforts at the Los Alamos National Laboratory (LANL) have traditionally been performed using either custom-designed rotary contour gauges or off-the-shelf/semi-custom cartesian CMMs. Both measurement processes utilize a required high-data density collection for characterization of the inner and outer surfaces of the hemispherical shells. Due to the large number of points collected, workpiece setup, and other related protocols, the measurement times range from four to 12 hours. Cast hemispherical shells or “blanks” are normally inspected on cartesian CMMs prior to initial machining, which leads to long downtimes in the manufacturing process. With the ability to scan on the AACMM, the “in-process” workpieces can be measured more rapidly, which aids in a better characterization of the physical characteristics (i.e., geometric features and surface finishes) and quicker turnaround to enable a better product during final machining.1 Investigation into the capability of laser scanner measurements is carried out; more specifically, point cloud measurements of unclassified CMM hemispherical shell artifacts are taken using an AACMM with an integrated scanner.

Preliminary measurement test case: CMM test artifact (U)

As an initial test case to investigate the accuracy of point cloud measurements, a generally used metrology artifact, specifically manufactured at LANL, is utilized.2 This artifact is called a “CMM Test Artifact (U),” which is an unclassified representation of some geometric features commonly seen on actual production parts. The inner and outer profiles are combinations of 2D shapes (i.e., lines and circles) where a 3D representation is simply made by revolving around a central axis (as seen in figure 1a.)




Figure 1: Measurement process: (a) CMM Test Artifact (U) drawing, (b) measurement of the inner contour, and (c) measurement of the outer contour

The measurement process consisted of measuring the artifact in two different orientations; first, the inner contour (as seen in figure 1b) and second, flipping the workpiece 180° (as seen in figure 1c) to measure the outer contour. Each contour measurement is a separate point cloud in the measurement software, where the CAD model of the workpiece is imported for alignment and comparison. A scribe line on the workpiece (and CAD) was used to align the “clocking” rotational degree of freedom, thus minimizing errors associated with the fitting algorithms used to align the point clouds. The scribe line is also used for referencing measurements to the coordinate system of the part, thus minimizing errors during feature analysis. A fully aligned point cloud of the artifact can be seen in figure 2, where each of the contours is aligned to the CAD model.




Figure 2: Point cloud measurements: (a) inner contour, (b) outer contour, and (c) aligned point clouds on the CAD model

Once the measurement data is aligned, it is necessary to look at the differences between measured and nominal. With the large number of points, a colormap is the ideal data characterization tool for investigating local and global regions of the measurement data. From the colormap (as seen in figure 3), the inner and outer profile calculations were 0.047 mm and 0.097 mm, respectively. The inner contour profile is within specification but the outer is nearly twice the allowable specification limit. However, the outer contour had known irregularities (i.e., dings and scratches) which contributed to the large error in the profile result. To minimize the measurement error, as well as measurement uncertainty, future work will investigate optimization of the measurement data through software (i.e., filtering, fitting, etc.) and the measurement process (i.e., better fixturing, operator experience, etc.) as well. Overall, the results are very promising and the entire measurement process took less than 60 minutes to complete, with further experimentation be conducted for more viability.



Figure 3: Colormap of point cloud measurements: (a) inner contour and (b) outer contour.

Customer test case 1: machine tool evaluation artifacts

The preliminary measurement test case proved that point cloud measurements are a viable and feasible alternative, given task-specific circumstances (i.e., in-process, less-stringent tolerances, etc.) and appropriate circumstances. Another area of application is on artifacts utilized for machine tool evaluations, specifically artifacts manufactured as test pieces to prove the accuracy and precision of a specific machine tool.3 More explicitly, each artifact will be used as acceptance criteria for interim checking, but not for calibration since an uncertainty will not be stated.4 Nonetheless, a set of reference measurements from a higher-accuracy measurement instrument is always recommended when possible.

The machine tool evaluation artifacts are hemispherical but one is concave (inner contour) and the other is convex (outer contour), as seen in figure 4. The measurands are the surface profiles of the inner and outer contours which have a tolerance specification of 0.050 mm (± 0.025 mm). To gather reference measurements, each artifact is measured on a high-accuracy CMM, which has an expanded length uncertainty of U (L) = 0.8 μm + 0.002L μm/m at k = 2, assuming a uniform distribution. The probing strategy is a single probe, parallel to the CMM ram, with data collection via continuous scanning. An expanded uncertainty at k = 2, also assuming a uniform distribution, is determined to be U(THP) = 1.3 μm.



Figure 4: Machine tool evaluation artifacts: (a) concave—inner contour and (b) convex—outer contour

An important caveat to consider is the sampling strategy used for data collection of the reference measurements on the high-accuracy CMM. The strategy is a uniform approach, with latitudinal sweeps (i.e., pole to equator) every 2° and longitudinal moves every 10° (i.e., 36 individual scan paths). The same cannot be done for point cloud measurements, as they are random throughout the measurement process. From the high-accuracy CMM measurements, a data density of 1,620 data points per contour is collected. Given the low density of the sampling strategy and the infinite number of points required to constitute a feature, estimation of an uncertainty contribution is recommended. A statistical approach using the data density, the point coordinate error, and the mathematical minimum to constitute a feature is used5 but caution is invited in that this approach is insensitive to random sampling. Reference measurements resulted in surface profiles of 0.012 mm for the inner contour and 0.021 mm for the outer contour, both with sub-micrometer uncertainty. The point cloud measurements resulted in surface profiles of 0.061 mm for the inner contour and 0.081 mm for the outer contour, both being outside of tolerance. This makes sense for a few reasons: There is a very large difference in data densities between the two methods, so extremum values from the high-accuracy CMM measurements may not fully represent the surface of each contour, and the fitting algorithms for the point clouds have a significantly large data set to compute (average). The point cloud measurement values would likely decrease with optimizations (i.e., filtering, etc.) performed on the data.

An interesting result that arose from the comparisons is the lobing effect commonly seen in turned parts. The equatorial region of each part is representative of a circle or “2D cylinder,” which serves the purpose of a secondary datum. Reference measurements from the high-accuracy CMM showed the systematic lobing effect (undulations per revolution) rather noticeably, but surprisingly, the point cloud measurements also showed the lobing effect, yet less obviously, from the 3-jaw chuck (as seen in figure 5). With the addition of this result, the customer is satisfied with the measurement accuracy of the point cloud data as well as the significant decrease in measurement time of almost 75 percent, which reduces a significant bottleneck in the production process.


Figure 5: 3-Lobe effect from CMM measurements: (a) high-accuracy CMM and (b) AACMM

Customer test case 2: copper and stainless steel artifacts

Another type of customer request often seen is the validation of an outside vendor’s ability to manufacture unclassified artifacts or test pieces by measuring critical features. Although not a traceable validation due to the exclusion of an uncertainty statement from the measurement results, this is a validation with respect to a set of drawing specifications. Particular to these customer parts, high-accuracy measurements were of no interest, but noncontact point cloud data is requested. The parts are similar in nature to the previously described parts in the preliminary test (as seen in figure 6).



Figure 6: Customer parts: (a) copper artifact and (b) stainless steel artifact

The tabulated measurement results can be seen below in table 1. Measurement results on the stainless steel artifact agreed well with the nominal values, with deviations within the accuracy of the measurement instrument’s stated maximum permissible error (MPE) of 0.058 mm for scanning. With no information on the measurements instruments’ capabilities for measuring angles, it is assumed that the measurement deviation of –0.01° is within what is acceptable of typical angular tolerances (i.e., 0.1° for one-decimal values), as required on drawing specifications. As for the copper artifact, the measurement artifacts did not agree well with the nominal values, with the exception of the spherical, outside radius and conical angle, and were outside the measurement instrument’s MPE for scanning. The largest deviations came from the pole and conical heights of the copper part where deviations of -0.843 mm and -0.856 mm were calculated, respectively. Given the closeness in deviations (thus a correlation), the manufacturing process error is assumed to be significant. An independent check of the pole height with a calibrated height master gauge showed a small difference from the point cloud measurements, therefore justifying the assumption of a large manufacturing process error. The measurement results both validated the outside vendor’s manufacturing capability for the stainless steel part as well as identified an issue in the manufacturing process utilized on the copper part, thus meeting the customer’s requirements.


Table 1: Measurement results and deviations from nominal

Conclusions and future work

An investigation into the accuracy, precision, and capability of laser scanning to produce point cloud measurements on CMM artifacts is presented in the form of test cases. First a preliminary test case on the CMM Test Artifact (U) is presented with favorable results where the inner contour surface profile measurement was within tolerance but the outer contour profile is approximately twice the allowable tolerance. This is expected given the condition of the outer contour surface. Next, a customer artifact is measured where a set of high-accuracy CMM reference measurements were collected and point cloud data surface profile evaluations were compared. As expected, the point cloud results were larger than the reference results but this is likely due to significantly different data densities and maxima used in the fitting algorithms. Nonetheless, the point cloud did agree with the CMM data on the shape of the part. Lastly, a second customer test case for validation of an outside vendor’s manufacturing capabilities where the point cloud results for the stainless steel artifact were within the MPE for laser scanning. However, the copper artifact showed a large bias in the pole and conical heights, but this is determined to be error from the manufacturing process, not the measurement process.

In the immediate future, investigation into addressing the measurement uncertainty is a high priority to aide in establishing validated and traceable measurements. Possible methodologies would be task-specific uncertainty such the substitution method6 or Monte Carlo simulation7-8; quality of the point cloud in terms of noise, accuracy, density, and completeness9-11; or task-specific testing to isolate certain uncertainty sources.

References

1

Harding, K.; “Handbook of Optical Dimensional Metrology;” 1st Edition; CRC Press; 2013.

2

Trujillo, I.S., Montaño, J.D., Valdez, M.O., and Valdez, L.M.; “Pointcloud Measurements on a CMM Artifact Using an Articulating Arm CMM;” Los Alamos National Laboratory: LA-UR-16-25869; 2016.

3

Smith, G.T.; “Machine Tool Metrology: An Industrial Handbook;” 1st Edition; Springer; 2016.

4

ISO/IEC Guide 98-3:2008; “Uncertainty of Measurement – Part 3: Guide to the Expression of Uncertainty in Measurement;” www.iso.org; 2008.

5

Hocken, R.J. and Pereira, P.H.; “Coordinate Measuring Machines and Systems;” 2nd Edition; CRC Press; 2011.

6

ISO 15530-3; “Geometrical Product Specification (GPS) - Coordinate Measuring Machines (CMM): Technique for Determining the Uncertainty of Measurement – Part 3: Use of Calibrated Workpieces or Measurement Standards;” www.iso.org; 2011.

7

ISO 15530-4; “Geometrical Product Specification (GPS) - Coordinate Measuring Machines (CMM): Technique for Determining the Uncertainty of Measurement – Part 4: Evaluating Task-specific Uncertainty Using Simulation;” www.iso.org; 2008.

8

JCGM 101:2008; “Evaluation of Measurement Data – Supplement 1 to the ‘Guide to the Expression of Uncertainty in Measurement’ – Propagation of Distributions Using a Monte Carlo Method;” www.bipm.org; 2008.

9

N. Van Gestel, S. Cuyers, P. Bleys, J.P. Kruth, “A Performance Evaluation Test for Laser Lines Scanners on CMMs”, Optics and Lasers in Engineering, vol. 27, pp. 336-42, 2009.

10

C. Lartigue, A. Contri, P. Bourdet, “Digitised Point Quality in Relation with Point Exploitation, Measurement, vol. 32, pp. 193-203, 2002.

11

P. Bourdet, A. Contri, C. Lartigue, “Quality of 3D Digitised Points with Non-contact Optical Sensor”, Annals CIRP, vol. 50, no. 1, pp. 443-446, 1999.

Discuss

About The Authors

Lucas M. Valdez’s default image

Lucas M. Valdez

Los Alamos National Laboratory, R&D Engineer

Mario O. Valdez’s default image

Mario O. Valdez

Los Alamos National Laboratory, R&D Engineer

Joshua D. Montano’s default image

Joshua D. Montano

Los Alamos National Laboratory, R&D Engineer

Ivan S. Trujillo’s default image

Ivan S. Trujillo

Los Alamos National Laboratory, Student Intern