SPC and Thread Manufacturing

Almost seventy years after its introduction, Statistical Process Control and other types of process control mechanisms finally have become accepted as an integral part of modern manufacturing. Unfortunately, despite well-documented successes in other areas of manufacturing, the full potential of these aids to efficient production of threaded components has not been realized on a widespread scale. While it is difficult to isolate all the underlying reasons for this unenthusiastic response, among the primary causes are the difficulty in determining the key process variables related to threading operations and the widespread utilization of thread inspection systems incompatible with the basic philosophies of process control.

Paramount among the prerequisites for the establishment of a successful Statistical Process Control Program for thread components is the selection of the Thread Inspection System. The ability to properly target the manufacturing process is a necessity to any SPC-driven system, and is absolutely critical for efficient production of threaded components. Further, the inherent difficulties in threading operations virtually demand an inspection system with the ability to differentiate between variation directly related to size or variation related to the interdependence of size and form. Without this capability knowledge and understanding of the threading operations is limited. Given these basic requirements, the incorporation of Variables Indicating Thread Inspection into Process Control System is a virtual necessity. Attribute data is simply inadequate for Statistical Control of threading operations.

Further, even Variables Thread Inspection Systems should receive careful scrutiny prior to utilization for process control. While ASME B1.3M may state “acceptance by any one gage specified for a characteristic shall be the criterion for acceptance of that characteristic,” practical realities refute this over-generalization. While the Standard may imply technical equivalency, there are significant technical differences between the Functional Tri-Roll and Functional Segment. The design limitations of the annular Functional Roll conceals the full extent of deviations in critical thread elements and characteristics such as Helical Path Deviation, 180O Out-of-Roundness, and Oversize Minor Diameter. The Segment profile, which incorporates a true helical path, eliminates these problems.

No initial evaluation of the process and its inherent capability is complete without consideration of the final acceptance criteria. Statistical Process Control or any other process control system retains its validity only when a viable index for the assessment of process variation is firmly in place. In most instances, this index is based on dimensional tolerances. If the standards and specifications defining these tolerances and allowances are erroneous, misleading, or incomplete, the entire concept of SPC loses all meaning. Careful evaluation of the governing standards and specifications is a necessity when developing a System for the orderly control of threading operations. An illogical structure with standards and specifications inevitably creates a unresolvable conflict between acceptance and Statistical Process Control. Fortunately, the inclusion of default inspection levels similar to System 22 in specifications such as MIL-S-8879C, MIL-S-7742D, and MIL-S-1222H eliminates much of this confusion. Internally-generated engineering documents of the Ford Motor Company and General Motors also invoke System 22 thread inspection.

Compounding the difficulties in the development stages of the program is a lack of comprehensive guidelines for the integration of process control mechanisms into the manufacture of threaded components. While there is wide variety of excellent resource material dedicated to process control in general and Statistical Process Control in particular, the only documents specifically addressing the unique aspects of the threading process are either limited in scope or are dedicated to a single parochial interest. Organizations such as the Ford Motor Company and General Motors have done an admirable job in the education of their supply base, but other similar efforts are few and far between. In a surprising development, despite the efforts of a minority dedicated to the clarification of Statistical Process Control, the ASME B1 Screw Thread Committee recently discontinued its efforts to clarify many of the process-related aspects of controlling the threading process. Hopefully, the single-interest political winds in the committee will change, and ASME B1 will re-assume its rightful posture of actively promulgating standards and documents which serve the general interest of industry.

Upon completion of the initial evaluation phase, intense scrutiny can be focused on the process itself. At this point, it is critical to identify the key process variables directly related to thread manufacture. Unfortunately, the complex geometry defining the screw thread defies easy solutions and approaches. Unlike many characteristics in manufacturing operations, variation in the screw thread is not limited to simple variations in size. While variation in the Pitch Diameter Size is significant, the Functional or Effective Size is also a key process variable for the efficient control of threading operations. Defined in FED-STD H-28 as “the cumulative effect of all profile variation,” Functional Size shares the tolerance limits defined for the Pitch Diameter Size and includes the entire effect of both Lead and Flank Angle Error on the Pitch Diameter Size. Differences between these different interdependent dimensions reflect the total of all Thread Form Error in the process.

Compounding the difficulty in selecting the key process variable for the statistical control of thread manufacturing is the relative importance of these two different but inter-related expressions of thread geometry. From an engineering perspective, Pitch Diameter assumes critical consideration due to its central role in ultimate performance. Virtually all mathematical calculations for the strength of threaded connections are dependant upon the Pitch Diameter Size in some respect. Conversely, Functional Diameter Size is equated with the ability to assemble. While these considerations are not central to the ability to statistically control the threading process, they are cornerstones in the overall Total Quality tenet of customer satisfaction, and should receive careful review and consideration.

The central dilemma is apparent: control Pitch Diameter and assure optimal performance yet risk selective assembly, or control Functional Diameter Size and assure assembly while sacrificing long-term reliability. In many instances, this paradoxical impasse is resolved by considering both Pitch Diameter Size and Functional Size in any process evaluation. Both are considered key variables, and are used simultaneously in the control of the process. Although somewhat burdensome, this approach has unquestionable appeal, and is especially helpful in establishing initial capability.

A key characteristic often utilized during the determination of process capability is the differential between the Functional Size and the Pitch Diameter Size. As stated earlier, the total magnitude of all thread form variation is reflected in the differential. Careful analysis of the ultimate causes of the difference between the Functional and Pitch Diameter Size can isolate problem areas in machine set-up and tooling. Diligent monitoring of the differential during the production run can detect break-downs in the overall process or individual components of the process. If only the Functional Size is used as a control mechanism, it is impossible to isolate excessive form error either in the initial manufacturing set-up or as a result of process and tooling degradation. All process control systems are based on the elimination of variation and its causes. Any inspection system that cannot determine the difference between Tiffany’s jewelry or outright junk creates a questionable foundation for the orderly detection and control of process variation.

As process knowledge increases, more and more data directly related to capability becomes available. Unfortunately, the vast majority of threading operations are subject to less than optimal conditions. Despite the best intentions, it is virtually impossible to eliminate the effects of Lead Error or Angle Error as a source of variation. As a result, the differential between the Pitch Diameter Size and the Functional Size is relatively large relative to the Pitch Diameter Tolerance, often exceeding 50% of the total tolerance. In these instances, practical considerations must prevail. Suspending production until problems are solved and variation eliminated is not an option. Deliberate violation of established tolerances invites eventual rejection, and may antagonize customer relations. Instead, compromises and accommodations directly related to the capability of the process should be initiated. In the short term, among these considerations are increased surveillance of the process, less than optimal production runs, and extensive analysis of improvement possibilities. Long-range systematic improvements might include upgrades in machinery and tooling, extensive operator training, and changes in management goals and orientation. In any event, the quest for continuous improvement will establish a firm basis for a successful process monitoring and control system.

As the program for the Statistical Control of the threading process evolves, the desire for simplicity without sacrifice of critical information becomes paramount. At this point, standard SPC techniques have been utilized to identify and isolate random sources of variation. As the process reaches the point of statistical control, there is sufficient information about the threading process to refine not only the process itself, but also in the means and methods of control. At this point, an examination of the data may suggest that there is an insignificant or small differential between the Pitch Diameter Size and the Functional Diameter Size. In other words, the effects of Lead Error or Angle Error on the process have been minimized. Given this process knowledge, it is possible to conclude that control of a single variable or characteristic will assure control of the interdependent characteristic. While this assumption may be statistically valid and will certainly simplify the control mechanisms, it is critical that any change in mechanisms utilized to monitor and control the process are firmly based on internally-generated empirical data.

In some advanced systems, control of an independent variable critical to the success of the threading process has shown great promise and benefit. Through the control of blank diameter prior to the actual thread fabrication, the thread forming operation has shown significantly reduced process variation. While this method of control is extremely useful as a part of the overall statistical control of threading operations, it is not a substitute for control of the actual process.

Selecting and locating the process on a target is another prime consideration when developing a statistical process control system for thread manufacture. Generally constructed in parallel to selecting key process control variables, this step is critical in the ability of the process to meet specifications or customer requirements on a consistent basis. Once again, the presence of the interdependent variables of Pitch Diameter Size and Functional Diameter Size and the inevitable differential presents a challenge. Despite the refinement of a statistically controlled process, it is almost impossible to eliminate all process error from threading operations. In other words, the differential creates a separation between the Functional Size and the Pitch Diameter Size. As a result, it is entirely normal and even advantageous to target the Functional Diameter Size at a different location than the Pitch Diameter Size. Given normal circumstances in the thread manufacturing environment, the optimal target value for control of the Pitch Diameter Size is generally at a point between the process mean and the Minimum Material Limit, while the optimal target for the Functional Diameter Size is between the process mean and the Maximum Material Limit. To assure process stability as well compliance to expectations of performance and assembly, the entire focus of process improvement should concentrate on the elimination of the effects of the differential between these different but inter-related expressions of thread size.

The specific indices defining the process capability of the fabrication operation have also created controversy and confusion. CP, a statistical index defining the process capability without regard to process centering, and CPK, a statistical index describing the process capability with respect to the mean of the tolerance, are both widely used in a wide variety of SPC programs. However, the inherent bias of CPK with regard to the necessity to target the process on the specification mean negates much of its value for the control of threading operations. Unless specific compromises and clarifications regarding use of CPK are built into the system, CP may be a more valuable index than CPK when expressing capability of the thread manufacturing process.

In those circumstances where the use of CPK is encouraged or even requested, the index does retain validity if its inherent bias to the process mean is tempered through the introduction of other indices of process capability. As stated earlier, it may be desirable to target the process at some other location within the specification limits. Operations which are affected by the effects of subsequent operations such as the application of additive finishes as well as heat treating are typical examples of this type of “off-centering” targeting. This method of process targeting will yield a CPK which indicates that there is a probability that a significant portion of manufactured output will exceed that specification limits. Yet depending on the application, this condition may indicate optimal process targeting. In these instances, it may be advantageous to compare CPK to CP. If the CP indicates a high degree of process capability, the necessity to “target the mean” is minimized. The process can be centered at a point ideally suited to efficient production, and the degree of centrality around the optimal target expressed by an intentional offset of the CPK. In these instances, while CPK does not represent optimal process distribution in the traditional sense, it does maximize the capability of an individual manufacturing operation. An in-depth knowledge of the manufacturing process including the capability for the control of all possible process variation is integral in the development and continuation of this type of program.

The development of an effective and cost-efficient program for the manufacture the threaded components based on Statistical Process Control parameters is challenging. However, experience has demonstrated that the rewards are worth the cost. Today, hundreds of diverse companies are reaping the benefits of the latest miracle in modern manufacturing. The systems are in place and are performing admirably. Customer expectations are exceeded on a daily basis. Still other companies are in the process of investigating the introductory parameters and laying the basic framework for the establishment of advanced process control systems. However, there are too many organizations who refuse to accept change. Their empty gestures condemn them to mediocrity and extinction.