STATISTICALLY CAPABLE MACHINING
In a manufacturing age of limited resources and rising costs, machining decisions made on probabilities, as opposed to limits, can offer a significant benefit. By implementing modern statistical process control, machinists can ultimately improve quality and reliability, and reduce operating costs.
Modern manufacturers compete under rising costs with limited resources. For this reason, machining decisions made on probabilities, as opposed to limits, can offer a significant benefit. By implementing modern statistical process control (SPC) that gathers and analyzes data to determine which machining process is most suitable for a workpiece, machinists can ultimately improve quality and reliability, and reduce operating costs.
Optimal development of SPC requires consideration of the critical characteristics of the part, the allowable range, part setups and fixtures, and the overall precision feasibility of the process. When applied correctly, the SPC approach has led to enhanced confidence in many aspects of electronic, automotive, aerospace and medical equipment used worldwide today.
STATISTICALLY CAPABLE PROCESSES
A statistically capable process is one where the chance of producing a part characteristic out of tolerance is quite remote. The process is measured by a best-fitted Gaussian, or normal distribution, bell curve to analyze the process results data and see how many standard distributions fit into a tolerance.
The expectation with statistically capable processes is that the tolerance should be 6, 8, 10 or 12 standard deviations from the nominal size, corresponding to capability levels (Cp) of 1, 1.33, 1.67 or 2. For example, a part characteristic that has 1.33 Cp has only one chance out of about 16,000 of being out of tolerance if it is targeted correctly.
Customers are typically machining parts with hundreds of characteristics that demand enhanced precision and Cp levels of 1.33 and 1.67. As a result, part conformance is key, which is why SPC is critical to the machining operation.
FOCUS ON CRITICAL CHARACTERISTICS
Most parts have dozens of features and each has multiple characteristics, such as diameter, length, cylindricity, position and surface finish. If any one of these characteristics is out of specification, then the feature and thus the entire part is non-conformant.
It would seem that in order to ensure part conformance, manufacturers would have to drive the process to a very high Cp level. In reality, the actual Cp level of each part characteristic is unique. For example, for most open-tolerance features the Cp level can easily be greater than 3 by using modern CNC equipment and tooling. This means that the chance of being out of tolerance is very small for most characteristics.
As a result, it is more realistic to estimate part conformance by considering the critical characteristics, those features that are more difficult to manufacture because of their required precision, i.e., tolerances less than 0.1 mm or 0.004 in, or finishes smoother than 1.6 μm Ra). Critical characteristics ultimately have the most influence on SPC, so it is important to identify and list this data before making process decisions on a part. There are also usually only 30 or fewer critical characteristics on a part, making this a much better measure.
When developing a process and making decisions on acceptable part characteristics, manufacturers should consider the operations being performed and the equipment being used that will yield the desired results produced within a reduced permissible range.
In many cases, manufacturers may need to provide for some means of adjustment that may be related to the size of the features or to the position of a feature. Either way, manufacturers should ensure that the adjustment method being used is much finer than the permissible range of the process. Once the process is targeted, it might also be necessary to provide some means of automatic feedback to keep the process targeted or to lock out adjustment to prevent errors.
The process capability index (Cpk) is a measure of both the accuracy and the precision of the process and is a helpful calculation when making process decisions:
Cpk = Cp x (1-k), where k = 2 x abs (nominal – mean) / total tolerance
Usually 25 to 30 workpieces are run off to verify a process (the quantity to be run depends on the cost of the parts and the cycle time). From this limited number of results, Cp and Cpk levels are calculated. To be reasonably confident that a set of results will achieve a specific Cpk index, the range of result needs to be about 35 percent of the total tolerance and the average of the results targeted to less than 10 percent of the tolerance away from the nominal. We can convert tolerances to ranges using the following guidelines:
- 1.00 Cpk ˜ 45 percent
- 1.33 Cpk ˜ 35 percent
- 1.67 Cpk ˜ 25 percent
- 2.00 Cpk ˜ 20 percent
These general rules help to guide thinking in the planning stages and can also be an indicator of potential problems when looking at actual results during the debug stage.
CHOOSING THE CORRECT NUMBER OF SETUPS
The number of setups in an operation also has a significant impact on the manufacturing process and should always be evaluated in the planning phase. Although additional setups are many times necessary, the best way to reduce costs and increase quality is by keeping setups to a minimum whenever possible.
Added setups create the possibility for inconsistency. Each time a part is physically moved to another setup creates potential inaccuracies in the clamping or the location that become undesirable variance. Physical moves can also cause potential loading errors and handling damage and allow contaminants or chips to get in-between the location surfaces, all of which can reduce part quality.
Additional setups are also highly expensive. The cost of the fixtures themselves, fixture maintenance, the cost of labor or robotics to move the part (material handling) and the cost of additional inventory due to work-in-progress all contribute to the high cost of additional setups.
Many abnormal variations can be attributed to the fixture. Fixture planning at the onset can help manufacturers avoid these variations during the debug phase, enhancing repeatability and minimizing distortion. Fixture planning involves visualization of the locators and clamping that will be used for each operation, even in the early process planning stages.
When planning, the temptation to use more primary locators many times arises. Manufacturers should instead utilize a principle of 3-2-1 to minimize the potential for distortion, contamination and other sources of variance. In order to minimize deflection in the machining process, manufacturers can also plan placement of locators to directly oppose clamps and run through solid part material whenever possible.
On high-thrust operations where the locators cannot directly oppose the cutting forces and the casting is not critical, manufacturers can consider serrated, carbide or surface-coated locators and clamps. For precision parts, high-speed and metal-cutting methods will reduce cutting forces. The quality, robustness and efficiency of the process will make up any extra cut time.
When faced with tighter or restrictive tolerance ranges, manufacturers can use crowder devices to push the part against the locators while clamping, preventing the part from being moved away from the locators by motion of the clamps. Adding loading guides can also improve part quality by eliminating assignable causes due to damage to locating surfaces.
In designing a statistically capable process, stack-up analysis and process feasibility studies can help to guide manufacturers between the choices.
A stack-up analysis summarizes what tolerances are allowed for each particular process. To conduct a stack-up analysis, manufacturers draft out each process method, as there are always a number of ways to make a part. The processes under consideration are listed out by identifying the machine type, how the part will be located and clamped, and which features will be machined in each operation. Features that cannot be machined should also be identified relative to their datum in the same operation. Then, for each proposed operation, the part tolerances and stack-up situations are outlined.
The result of a stack-up analysis and process summary is a list of cuts and the required part tolerances for each. When making comparisons from the list, the process that has the greatest average tolerance generally has the advantage. This analysis will also expose features that require an exceedingly small tolerance, making the process risky and best avoided.
So that manufacturers have confidence that their processes can produce the tolerances identified in the stack-up analysis, a process feasibility study should also be conducted for each of the steps in the process. This study estimates what is feasible for the tooling, fixturing and machining equipment.
Real-time influences such as machine repeatability, clamping repeatability, temperature effects and tool wear are important considerations and should be taken into account when designing a statistically capable machining process. Thinking about each critical feature in this detail is very useful in making sure the correct measures are taken. This awareness leads to better decision-making in the planning stages and avoids major issues during the process debug stage.
To follow the right track for an effective statistically capable process, manufacturers must question each potential process involved in machining a part. By identifying critical features and allowable ranges, reducing the number of setups, planning for fixtures and determining the overall feasibility of the process, the best method becomes apparent.
Armed with this information, competitive manufacturers are one step closer to a statistically capable process, which can ultimately help them achieve quality goals and reduce unnecessary costs.
Makino, Inc., 7680 Innovation Way, Mason, OH 45040-8003, 513-573-7200, www.makino.com.