SYSTEM AND METHODOLOGY FOR TOLERANCE CONSISTENCY IN COMPONENT ASSEMBLY
BACKGROUND OF THE PRESENT INVENTION
Field of the Invention
The present invention is directed to improvements in analysis tools to evaluate the effects of tolerances in component designs.
Description of the Related Art
With the advances in computing power over the last few decades, the power of the computer has been harnessed in a variety of industries, simplifying many mundane and/or complicated tasks. One such industry is component assembly, e.g., automobile, computer and any other mechanical assembly of parts into a device.
As is well understood in the art, the mass production of the component parts for assembly on an assembly line is a standard and efficient methodology for most all consumer products. One flaw in this approach, however, is the potential for variability of the component parts, i.e., where one or more parts are deleteriously outside of a range of tolerance, resulting in potential or actual malfunction of the assembled device. For example, various alignments of parts within an automobile during the assembly process may be required, and misalignment due to component fault may result in catastrophic failure later. Similarly, component errors within a chemical or nuclear power plant could cause damage on a large scale .
In an effort to curb such dangers, engineers model the construction of the assembly and employ various techniques to monitor the potential for error. Present methodologies define discrete component tolerances and study the effect of that one tolerance separately. Where multiple tolerances are concerned, a corresponding multiple number of discrete studies are performed to determine the effect of the respective tolerance on the entire assemblage. A modification to one of the studies, however, has serious repercussions in that each of the other tolerance studies must be recalculated in view of the change, however minimal. Failure to re-run the studies with the modification could result in an invalid analysis. It should be clear that the manual maintenance of coherence between a multitude of individual studies on each part of an integrated assembly is a complicated task prone to error, predominately human error.
There is, therefore, a present need for a tolerance analysis tool for enabling engineers to evaluate during the design and subsequent stages the one-dimensional effects of a variety of interlinked component tolerances and assembly process variations on the quality of manufactured products. It is, accordingly, an object of the present invention to provide a tolerance analysis tool and methodology to enable engineers to quantitatively analyze and evaluate the capability of designs, ensuring that the assemblies interconnect properly while simultaneously meeting all of the build objective (design goal) criteria.
SUMMARY OF THE INVENTION
The present invention provides a methodology and system for optimizing tolerance restrictions between design studies, facilitating rapid component assembly design and production schedules, while maintaining multi-tolerance restrictions.
A more complete appreciation of the present invention and the scope thereof can be obtained from the accompanying drawings which are briefly summarized below, the following detailed description of the presently-preferred embodiments of the invention, and the appended claims.
BRIEF DESCRIPTION OF THE DRAWINGS
A more complete understanding of the method and apparatus of the present invention may be obtained by reference to the following Detailed Description when taken in conjunction with the accompanying Drawings wherein:
FIGURE 1 illustrates a screen configuration as used in the present invention;
FIGURE 2 illustrates a project assembly listing as used in FIGURE 1; FIGURE 3 (shown in two parts as FIGURES 3A and 3B both referred to herein collectively as FIGURE 3) illustrates a methodology implementing the principles of the present invention;
FIGURE 4 illustrates an exemplary assembly with tolerance restrictions for use in describing the present invention;
FIGURE 5 illustrates an exemplary study of the assembly and tolerance restrictions in accordance with
the principles of the present invention for the assembly shown in FIGURE 4;
FIGURE 6 is a sensitivity report prepared in accordance with the teachings of the present invention illustrating respective contributions of the component parts of the assemblage to the tolerance computations of the assembly shown and described in connection with FIGURES 4 and 5;
FIGURE 7 illustrates a statistical report concerning the tolerances for the assembly shown and described in connection with FIGURES 4-6 pursuant to a worst case analysis;
FIGURE 8 illustrates a statistical report concerning the tolerances for the assembly shown and described in connection with FIGURES 4-6 pursuant to a root mean square analysis; and
FIGURE 9 illustrates a statistical report concerning the tolerances for the assembly shown and described in connection with FIGURES 4-6 pursuant to a Monte Carlo analysis.
DETAILED DESCRIPTION OF THE PRESENTLY PREFERRED EXEMPLARY EMBODIMENTS
The present invention will now be described more fully hereinafter with reference to the accompanying drawings, in which preferred embodiments of the invention are shown. This invention may, however, be embodied in many different forms and should not be construed as limited to the embodiments set forth herein; rather, these embodiments are provided so that this disclosure will be thorough and complete, and will fully convey the scope of the invention to those skilled in the art.
As noted, the need for an improved mechanism to dynamically interlink studies and tolerances between studies is manifest in any design and manufacturing environment. The present invention is a tool that enables engineers to evaluate, optimize and validate assembly build quality relative to piece part tolerance and assembly process variation. The system and methodology of the present invention can also be used early in the development phase before design solidification and hard tooling procurement, thereby minimizing or eliminating costly tweaking delays at ramp-up, as well as long-term variation problems.
As will be noted in more detail hereinafter, the system and methodology of the instant invention can predict the quality of an assembly based upon predetermined tolerance values or determine the component tolerances required to meet a given assembly build objective, i.e., determine roll-up and roll-down. The present invention also automatically optimizes all tolerances in a given model. As illustrated and described, the modeling utilities of the present invention are easy to use, e.g., menu-driven, reducing modeling time and effort, and generate easy-to- interpret reports. The various improvements of the present invention will now be described in more detail.
Variation analysis, as is understood in the art, involves mathematically predicting the resultant effect of a piece part and subassembly tolerances along with assembly process and fixturing variation on a particular build objective of the assembly. An assumption in the present invention is that all of the tolerances deviate in one direction and that these one- dimensional deviations accumulate, i.e., create a tolerance stack, where the net effect is considered. It
should be understood that the phrase "tolerance stack" refers to the mathematical calculations required to determine the net effect of the deviation caused by contributing tolerances. Build objectives are design tools or assembly requirements that must be met by the design in production. The system and methodology of the present invention is directed to dimensional build objectives relating to a given assembly's fit, finish and function, e.g., a gap between a car door handle and bezel or shims within a hinge opening, as described and illustrated hereinbelow. Two scenarios are handled: roll up and roll down. If the tolerances are known for individual piece parts, then the present invention uses this information to determine the build objective by using a roll-up analysis. On the other hand, if only the build objective is known, the present invention can allocate the allowable build objective tolerance to the contributing tolerances using a roll-down analysis. Several types of analyses may be performed. For example, a worst case analysis simply sums all of the tolerances in the assembly system in a linear direction and predicts the maximum variation expected for a particular build objective. In a Root Sum Square (RSS) analysis, the square root of the sum of the squares of individual tolerances is calculated to predict the build objective variation. A Modified Root Sum Square
(MRSS) analysis is similar to a Standard RSS analysis except that a constant, generally designated as "k", is added to the RSS equation to provide a more accurate picture of what is actually happening in the assembly process. Finally, in Monte Carlo simulations, numerous virtual tolerance stacks are created. For each stack, the component tolerance values and assembly process
variations are applied randomly (per the specified statistical distribution) within the acceptable tolerance zone using a random number generator, and the random values for each virtual tolerance stack assigned to each tolerance are added up to determine simulated values of the build objective. Ultimately, the results are compared to the desired build objective specification limits and relevant statistics computed therefrom. Each of these respective techniques will be described in more detail hereinbelow.
With reference now to FIGURE 1 of the Drawings, labeled as such in the accompanying text, there is illustrated an improved screen format for practicing the principles of the present invention. The windowed configuration illustrated is preferably practiced in conjunction with an assembly build analysis tool that enables engineers to evaluate the effects of component tolerances and assembly process variations on the quality of manufactured products. With reference again to FIGURE 1, an overall screen, designated by the reference numeral 100, such as may be displayed on a computer monitor or other display device, is subdivided into three windows. At the left, there is a project portion split into a project assembly window 105 and a project study definition window 110. At the right, there is a study review window 115 for reviewing the particular study defined in the project assembly window 105 and the study definition window 110, e.g., by double clicking on the study shown in the project study definition window 110.
In general, subassemblies , tolerances, parts and subassembly-tolerances are created and defined in the project assembly window 105. The relationships
constitute an Assembly Flow Diagram, as described in more detail in connection with FIGURE 2. As illustrated in FIGURE 1, the project assembly window 105 contains information on three parts, i.e., Part 1, Part 2, and Part 3. The project assembly window 105 provides visual feedback regarding the relationships between parts, subassemblies and the tolerances. Through use of the study definition window 110, any number of studies may be defined and evaluated for a given model without having to recreate assemblies, tolerances, etc., each time a new study is created, e.g., Study 1 and Study 2 , ' as illustrated. Since each study is unique and, once created, has no relationship with the original study, the flexibility of creating duplicate studies, particularly having only minor differences therebetween, saves considerable user time and effort. Upon defining relationships and studies in the project assembly window 105 and the study definition window 110, studies can be created and visualized in the study review window 115, e.g., by double clicking on one of the study icons within the study definition window 110, e.g., study 1, as highlighted. As shown in FIGURE 1, Study 1 includes range and offset gap data for Part 1, along with respective roll up and roll down for each tolerance. To include a given tolerance in a study, the user of the software tool of the present invention simply selects and drags the given tolerance, e.g., tolerance 4 for Part 3, or other information from the project assembly window 105 into the study review window 115, forming another tolerance branch from Part 1.
Some general rules for the transference of part or tolerance information between windows include: tolerances can be moved to any part. However,
tolerances cannot be moved to subnodes or other tolerances . Parts may be moved to any other parts to form a subassembly. Parts, however, may not be moved onto tolerances or subnodes. Likewise, subnodes cannot be moved onto tolerances, but can be moved onto subassemblies, which is a part that has another part defined thereon.
With reference now to FIGURE 2 of the Drawings, there is illustrated an exemplary project assembly window, generally designated by the reference numeral 205, which illustrates relationships between parts and subassemblies, and defines tolerances. The project assembly window gives accurate and instantaneous visual feedback regarding the relationships between parts, subassemblies and the defined tolerances. This feedback takes various forms, some of which are illustrated in FIGURE 2. For example, in the project assembly window 205, a part (Mirror Flag) 210 is defined along with tolerances 210A and 210B, as illustrated, which are user locked. Another part (End Item door trim) 215 contains subassembly tolerances, which may be highlighted, e.g., using a different color, to differentiate such tolerances from other tolerances. By virtue of the hierarchical arrangement, varies other parts and respective tolerances are defined in relation to the part 215, e.g., parts 220, 225, 230, 235 and 240. Part 225 has a tolerance value 225A defined and value locked by the user, and part 235 has a user defined tolerance 235A also locked by the user. Moving hierarchically upward, dependent from Assembly, another part (Sheet Steel variation) 245 includes a user-defined subnode 246A.
By virtue of the user-friendly interface and icon usage, manipulation of the various tolerances,
subassembly tolerances, parts and subassemblies is greatly facilitated, e.g., using powerful computer actions such as copy, paste and move. In this manner, a model may be edited and updated constantly as it is being built. The auto updating capability of the instant invention ensures that changes made to any tolerance fields are automatically updated in all studies whether the change is initiated in the project study window or the project assembly window. With reference to FIGURE 3 of the Drawings, there is illustrated a methodology for implementing the principles of the present invention. In any project, various objectives are ascertained to achieve goals, e.g., the mass production of a component or inter- related components . Some of these goals may be intertwined and tolerances between these disparate goals may vary. As discussed, the present invention provides a methodology whereby various tolerances for multiple goals may be made consistent. The methodology, generally designated by the reference numeral 300, begins with a pooling of all the various tolerances that are defined in a model (step 302), e.g., tolerances Tl, T2 , T3 and T4. At step 304, a determination is made whether there are any enabled tolerances (ETs) , e.g., a given tolerance may be enabled or disabled for the analysis. If no tolerances are enabled, then the process ends (step 306) since there is nothing to optimize. If yes, a determination is then made whether there are any non-contributing tolerances (NCTs) (step 308) . For example, if a number of tolerances are defined in a study but some are not used, the unused ones are non-contributing and may be ignored (step 310) .
Control in either event passes to step 312, where a determination is made whether there are any studies where the Roll Down method is not defined. If yes and undefined Roll Down methods are present (step 314) , then the undefined methods are ignored; otherwise, a determination is made whether there are any studies with allocation errors (step 316) . If tolerance allocation errors are present in one or more studies, then those studies are ignored (step 318) . A roll down method is then performed for all the studies in question (step 320) . A first determination is whether there are any locked tolerances (LTs) , i.e., a tolerance that must remain consistent and not varied (step 322) . For example, studies 1 and 2 may each have tolerance Tl defined therein, but with the Tl in study 1 having a different range of tolerance than that in study 2. The value of Tl is locked in both studies as per the definition. If the determination (step 322) indicates the presence of LTs, then those LTs are already optimized and may be ignored (step 324) .
In either event, control passes to step 326, where a determination is made whether there are any unlocked contributing tolerances (CTs) that occur in multiple studies, e.g., Tl in studies 1 and 2. If not, then no further optimization is required and the methodology 300 is done (step 328) ; otherwise, control passes to step 330 where all of the repeating CTs are sorted. A determination (step 332) is then made whether, among the sorted CTs, there is a repeating CT with a unique minimum range, i.e., a CT having the narrowest restrictions, e.g., a range of 4.9 to 5.1 mm as opposed to other CTs with broader ranges. If there is one, that CT is then locked (step 334) at that most restrictive range or tolerance. At this point, the
methodology 300 must do recalculations in all of the studies, i.e., re-evaluations whether this narrower tolerance works in the other studies (step 336) . Control thereafter transfers to step 322, as discussed above .
If there are repeating CTs with the same range
(step 338) , the best CT must be determined from the multiple minimum tolerances. For example, if Tl, T2 and T3 all have the same tolerance, each must be analyzed in seriatim by locking the Tx tolerance (step
340) , performing roll down techniques on all studies
(step 342) , calculate the respective sum of squares of the various tolerances defined in the model (step 344) , unlock the respective Tt (step 346) increment to the next tolerance TL +1 (step 348) and determine if done, i.e., whether L=n+1, the terminating tolerance (step 350) . If not the last tolerance, the iterative loop repeats for that next tolerance at step 340, as discussed. Of all the tolerances analyzed among the group of minimum tolerances, a determination (step 352) is made whether there is a tolerance having a maximum summation value (Si) . If so then the corresponding tolerance T( is chosen and locked (step 354) . If no unique Si exists and multiple summations are equivalent, a particular tolerance is preferably chosen at random (step 356) . Control then returns to step 336 so that the effect of these lockings can be evaluated within the studies .
As noted in the methodology shown and described in more detail in connection with FIGURE 3, another feature of the present invention is optimization. When build objective driven studies (or roll Down studies) are performed, it is possible for a particular tolerance to act as a contributor in more than one
study and attain different values. The design engineer, however, must specify a unique value for the tolerance for the drawings. Typically, the lowest value attained by the tolerance is used for completing the specifications. In reality, however, this value may not be the optimal tolerance value.
Since every tolerance can have only one unique value, the optimizer methodology of the present invention maximizes all of the tolerances that appear in all studies while making sure that no build objectives are violated. In the software tool of the present invention, the icon or indicia used for an optimized tolerance may be highlighted, boxed or otherwise marked. With reference now to FIGURE 4, there is illustrated a model of two shims in a hinge opening. The opening, generally designated by the reference numeral 410, dimension has a range of 20.5 ± 0.2mm and the respective shims, generally designated by the reference numerals 420 and 430, have ranges of 18.0 ± 0.2mm and 2.0 ± 0.2mm, respectively. The nominal design gap, generally designated by the reference numeral 440, is 0.5mm and the build objective is for the gap to be between 0 and 1mm. In this example, three parts must be created: the hinge the shims fit into and then the two shims. In other words ,
Project
Assembly
Hinge Shim 1 Shim 2
Next, tolerances must be defined on the various components. For example, on the Hinge, a tolerance here is hinge opening size = 20.5 +/- 0.2mm, where a nominal value is 20.5 (the desired value) and a roll up range of 0.4 (the entire tolerance range). Likewise, Shim 1 has a shim thickness = 18.0 +/1 0.2mm, where the nominal value is 18.0 and the range is 0.4. With Shim 1 defined, the software tool of the present invention allows selection of the defined part, i.e., Shim 1 in the project assembly window 105 of FIGURE 2, and dragging the icon thereof into the study review window 115 under assembly. After renaming the duplicated record as Shim 2, another tolerance is defined thereon, i.e., shim thickness = 2.0 +/- 0.2mm, nominal = 2.0 and range = 0.4.
The resulting study, labeled gap size, shown in the project assembly and study definition windows 105 and 110, respectively, is:
Pro ect Assembly
Hinge
± Hinge Opening Shim 1
± Shim Thickness Shim 2
± Shim Thickness
Study
Gap Size
In other words, the Hinge Opening node is a child of the build objective root node (Hinge) and the respective shim thickness nodes children of the Shim nodes .
Double clicking on the Gap Size icon in the study definition window 110 portion results in a representation of the study in the study review window 115, generally designated by the reference numeral 500 and as illustrated in FIGURE 5. The root node, generally designated by the reference numeral 505 specifies the build objectives, i.e., the gap size be from 0 to 1.0mm or 0.5 +/- 0.5mm. It should be understood that to ensure the calculations are carried out correctly, the nominal values for the shims are negative and the nominal values for the hinge opening positive. The child nodes are for the Hinge Opening tolerance, first shim thickness tolerance and the
second shim thickness tolerance, respectively generally designated by the reference numerals 510, 515 and 520. Performing a worst case analysis on the aforementioned example, e.g., by clicking on a worst case icon on a toolbar of the software tool of the present invention, results in the addition of the respective tolerances in the child nodes, i.e., +/- 0.2 x 3 = -0.6 to 0.6. It should, of course, be understood that a worst case analysis does not consider the type of distribution of individual tolerances, and the tolerances cannot exceed their specified limits. Mathematically, this analysis assumes all individual tolerance values to simultaneously be at one of the extreme limits. As is understood in the art, a worst case analysis is commonly used for critical mechanical interfaces where any possible tolerance problem may have serious consequences. Worst case analysis ensures that parts will always assemble properly, albeit at the cost of high manufacturing costs due to tight individual component tolerances. As is clear from the build objective criteria of the example, this result is out of the range of the desired -0.5 to 0.5 in the specification.
The software tool of the present invention allows the user to visually ascertain the sensitive areas within the analysis in graphical form, e.g., a bar graph showing the relative contributions of each tolerance to the build objective. For example, regarding the aforedescribed opening-shim scenario, FIGURE 6 illustrates that each tolerance has an equal contribution to the overall build objective, i.e., 33.33% each. Similarly, a statistical analysis of the calculated distribution over the specified control limits can be generated, as illustrated in FIGURE 7,
calculating the percent out of spec for the calculated distribution, after specified control limits can be generated (under a normal distribution assumption for the result) . As illustrated in the instant example, 0.972% of assemblies will be built out of spec in this wort case analysis.
A Root Sum Square (RSS) analysis calculates the square root of the sum of the squares of individual tolerances to predict the build objective variation. Mathematically, this calculation is represented by:
= Build Objective Variation
where Ts are the tolerances (equal bilateral) affecting a particular build objective and n is the number of tolerances affecting the build objective. In this scenario, it is understood that all individual tolerances are independent and follow a normal distribution, and that the probability of individual piece part tolerances coming in at their worst case tolerances simultaneously in forming an assembly is almost zero. In other words, production lot runs of the component parts used in the assembly are thoroughly mixed and the parts are selected at random, minimizing the chance for multiple deleterious part variations being selected for a given assembly.
In the above example, computation of the tolerance range using RSS results in a range of -0.3464 to 0.3464, which is well within the desired tolerance of -0.5 to 0.5. Shown in FIGURE 8 is a statistical report of the calculated distribution over the specified control limits, as in FIGURE 7. In this example,
however, all of the assemblies will be built within specification for the RSS analysis.
As the numbers of contributors increase in an RSS study, the variation of the output will approach normality regardless of the shape of the distribution of the individual contributors. If the individual contributors are all normally distributed, then the resultant will have a normal distribution regardless. However, when one of the contributors is defined by a non-normal distribution that is much larger than the other contributors in the stack up, the variation will exhibit normality only if there are a sufficiently large number of smaller tolerances in the stack up to help offset the effect of the non-normal contributor. By use of a correction factor in the RSS model, however, these errors can be overcome and a Modified RSS study may yield more realistic results. A typical "k" value is within the recommended range of 1.4 to 1.7, which by modifying the range of each tolerance contributor is more consistent with real world applications. In other words,
^MRSS = k TRSS
where TMRSS is the total predicted assembly tolerance (equal and bilateral) variation using MRSS and k is the correction factor.
In the instant example, using a k value of 1.50, the root node tolerance range is -0.5196 to 0.5196, which is only slightly outside the desired build objective. A statistical report shows that 0.119% of the assemblies will be built out of spec for this MRS analysis .
As is understood in the art, Monte Carlo simulations are numerical methods for solving mathematical problems by random sampling. This longstanding technique may be used in a wide variety of applications, including tolerance analysis in the instant invention. Each tolerance used in a Monte Carlo simulation must be assigned a theoretical distribution, the type of which can be determined by the previous history of the process or observations made on a real system. Without any information on a process, the tolerance from that process may be assumed to follow a Normal distribution. Random values following the specified distribution and falling within the specified tolerance range are then assigned to each tolerance. Because these "random" numbers are reproducible via an algorithm, they are called pseudorandom numbers, i.e., not truly random.
As is understood in the art, the first step in a Monte Carlo simulation is the generation of a random number between zero and one following a Normal distribution. This number is then mapped to a Cumulative Distribution Function (CDF) of the distribution type specified by the user, and the CDF value is in turn mapped into a Probability Density Function (PDF) . Naturally, the more simulations that are run better approximates the true results . A sample size of about 5,000 is recommended although the user may input into the software tool of the present invention number between 30 and 100,000. The seed value introduced into the "random" number generator should not matter so long as the model is statistically stable. A default seed value of one may be used to initiate the algorithm.
In the instant example, with a seed value of 1 and 5,000 simulations run, results in a asymmetric range of -0.3478 to 0.3504, which is well within the desired build objective of -0.5 to 0.5. The Monte Carlo analysis, being statistical in nature, is best viewed in a statistical report, such as that shown in FIGURE 9, As noted in the figure, in addition to calculating the percent out of spec for Monte Carlo, the present invention also calculates a variety of other statistics for the simulated distribution.
In addition to combining predetermined tolerances to calculate the resultant effect on a build objective, the system and methodology of the present invention may also be employed to start within a desired build objective and calculate the maximum possible tolerance values for all of the children (nodes) of the build objective. In the instant example, the desired gap build objective is 0.5 +/- 0.5 and the hinge opening size is restricted to 20.5 +/- 0.2. The maximum tolerance on the shims is then determined using a Roll
Down calculation. With a range of 0.4 mm, a roll down analysis results in shim thickness tolerances of - 0.3241 to 0.3241 mm for each shim. It should be understood, of course, that worst case and MRSS techniques could also be used. By virtue of freezing or restricting the tolerance on the hinge opening size, the contribution of that component to the build objective decreases to 16%, leaving the shims each with 42% of the sensitivity contribution. It should be understood that through use of color, icons and other visual feedback numerous layers of information may be imparted to a user, e.g., a highlighted study such as Study 1 in FIGURE 1 may have a box outline, another color or other indicia.
It should also be understood that the system and methodology of the present invention may be used with both dimensional and geometric tolerances. For example, in some cases a stack may be constructed where several tolerances are not facing in the same direction. As such, if a tolerance is skewed at an angle relative to the direction of the stack, a Trigonometric Factor can be used to reduce the tolerance in an active study. For example, if a tolerance were skewed 45° from the direction of the stack, then a trigonometric factor of sine (45°) or 0.707 would be used to adjust the tolerance.
The previous description is of a preferred embodiment for implementing the invention, and the scope of the invention should not necessarily be limited by this description.