Where does it come from? For those of you getting pummeled by your management for these answers now, you already know. For those of you contemplating a new system in the next year, consider that the CMMS, database engines, and servers all cost money. A common way to justify the cost of all this additional power in the maintenance activity is to talk about the cost savings that the CMMS can provide. In other words, by installing this software, I save X dollars to the company.

CMMS marketing and sales forces are good at providing statistics to back up what they claim their package can save you. Bob, if you properly implement our package Mainten-X 2000, you can save upwards of $200,000 per year! Management and bean-counting types buy into this as a good justification for any procurement that can cost what a CMMS does. However, what happens a year later when the cost of the operation is not $200K less? Or worse, as I've heard in one operation; the accounting folks slashed the budget by $200K assuming that those savings were real.

While I generally find good cause to abuse the sales forces of most CMMS vendors, often times they do tell the truth (in whole or part) when it comes to the cost savings potentials. CMMS can save substantial money when properly implemented. The key words here are, properly implemented. Also assumed is that you can and do measure the impact of the CMMS implementation—often the cost savings impact is buried, obscured, or forgotten.

There are two types of savings impact that can come from implementing a CMMS.

Intrinsic cost savings are the inherent savings you get the day you install the software and start using it. It is driven off features that your current package did not have, or is dramatically better than what you were doing before. In one company they used three big binders to manually track work orders, all stamped with dates, times, etc. Any reporting they did they manually compiled. Installing a CMMS saved them hours of time.

In some cases these intrinsic savings come from simple processing advantages with newer technology. Company K's old CMMS would take five hours a week to generate PMs, all of which had to be manually sorted. A new software package was able to do all of this significantly faster, in one-half an hour, and pre-sorted everything on-line before printing them by supervisor.

When you implement your CMMS, these savings are the easiest to notice. In reality, however, they often result in the smallest amount of dollar savings against your return on investment. Where an implementation effort can falter is in trying to capture all of these savings—since the cost to capture them can often outweigh the savings themselves. If clerk Judy saves 15 minutes a day from sorting work orders, is there merit in spending two hours documenting it? The rule of thumb is that the savings in hours per month has to be two times the amount of time to document it once in order to take the time to capture these savings.

Implementation driven cost savings are where a CMMS can actually start paying back a company for its investment. It can, in many respects, be easier to measure because it forces you to understand the cost of doing a process now and the impact of making a change to this process.

These are the savings that are driven by a CMMS implementation effort that looks at altering how work is managed and performed as a result of having a new tool (the CMMS) that has features and functions that allow better control and management.

An example of this might include something such as tracking equipment downtime. If your new CMMS has the means to do this and your old one did not, it's going to require some time to set up (as part of implementation). It's also going to require some retooling of your processes (we've never asked the guys on the floor to track this in the past). Then comes the issue of what you're reporting and what you're going to do with it. When you find out that one piece of machinery is causing 20 percent of your reactive maintenance costs, and that you can alter the PM schedule to reduce that dramatically, there is a cost savings—both from the loss of downtime and the cost of the emergency maintenance runs.

This is where a CMMS can save you real money toward the return on investment. These are not little dollars either, but often are long-term steps and actions that generate a substantial savings for your company. MT

Blaine Pardoe is a principle in Enterprise Management Systems and a highly regarded expert in the field of technology learning, CMMS implementation, and the maintenance industry. He is the author of the best-selling book, Cubicle Warfare, and numerous novels and is a frequent contributor to MAINTENANCE TECHNOLOGY.

Robert C. Baldwin, Editor

Now, with the calendar rolling over to 2000, it is time to move forward and make some changes.

Using some seminarese: It is time to walk the talk. If you don't, and you do what you always did, you'll get what you always got. To think the outcome will be otherwise is one definition of insanity.

What needs to be changed? I'm sure you have plenty of ideas: better scheduling, more training, condition based maintenance, effective planning, etc.

Where do you start? Start with yourself and your job. You won't know what to do in every instance, and that's OK. The important thing is to get started. There is plenty of help available at conferences and seminars, from professional societies, on the Internet, and in technical business magazines. In this issue, we serve up some information on using reliability centered maintenance, help in failure analysis, and recommendations for improving your condition monitoring program.

If you put forth some effort, you should be able to make a difference. The amount of difference could be surprising. In some of my resent conference presentations, I've mentioned the Law of Tipping Points, referring to a characteristic of the new wired economy when a process or technology gains momentum and begins changing the way we work. You could cross the tipping point to start change of significance in your company.

In his forthcoming book, The Tipping Point: How Little Things Can Make a Big Difference, Malcolm Gladwell explains that tipping point comes from the world of epidemiology. It's the name given to that moment in an epidemic when a virus reaches critical mass. It's the boiling point.

Gladwell, a staff writer for The New Yorker, points out in book excerpts put up on his web site that change often happens all at once, and little changes can make a huge difference, like one child bringing a virus into a classroom. He says ideas and behaviors and new products move through a population very much like a disease does.

If Gladwell has it right, we have the potential to infect our company with proactive equipment maintenance. Perhaps we can provide the tipping point for a national epidemic of effective enterprise asset management. MT

Commercial CE-based handheld personal computers are useful for a range of industrial applications - both for mobile computing and embedded applications.

Trends
The need for devices to program and troubleshoot industrial electronic equipment has existed as long as the equipment itself. Most companies have offered various portable or handheld devices that are easily transportable. These devices typically have been task- or target-specific, and limited in usability and productivity features.
Commercial CE-based handheld personal computers (HPC) provide a platform for maintenance and troubleshooting of industrial equipment. HPCs have instant on capabilities, and the RAM- or ROM-based software eliminates the need for disk drives—allowing users to go to work without waiting to boot up.

They are similar in size to a typical multimeter tester carried by many electricians, and will easily fit in a pocket or tool pouch. They have small, but clear and readable, displays and full-function keyboards. Most have enough memory to host several applications and can be expanded with low-cost flash memory.

On-board serial ports allow interfacing to most types of industrial equipment and provide an important additional feature—auto synchronization with desktop computers. This allows automatic archiving of changes made on the plant floor by connecting to the serial port of a desktop unit and eliminates the need to manually record changes or move files.

One example of this technology is the Rockwell Software RSPocketLogix, which provides users with a familiar look and feel and HPC-specific features such as display zooming and navigational tabs for tasks most used in troubleshooting and debugging. The software is designed to interface to multiple Rockwell Automation controller families from a single program while providing flexibility and value to the customer. In the future, engineers or maintenance persons with a single HPC will be able to work on most Allen-Bradley controllers, drives, and DeviceNet devices.

Because Windows CE can be embedded on industrial devices, users soon will be able to extend the same troubleshooting and maintenance software capabilities to their dedicated fixed machine interfaces.

Troubleshooting drives
Troubleshooting drives is already possible with new software running under Windows CE. In the past, monitoring and troubleshooting drives on the factory floor involved carrying a laptop computer from drive to drive. Engineers had to be especially careful that the laptop had the appropriate card and connection devices to interface with the drives communication module. The process was labor intensive and time consuming compared to current tools. Today, CE-based software running on an HPC can make troubleshooting and configuring drives on the factory floor much easier by performing online programming and troubleshooting, creating a highly portable, simple method to configure and monitor drive parameters.

Software-based control
Recently, PC-based control—defined as combining industrial PC hardware with control software, often called a soft PLChas become increasingly popular in applications requiring large amounts of memory, an open platform, or alternative programming languages. Today, standard Windows NT is the operating system of choice for PC-based control because it allows users to make the most of the primary advantages of a PC environment—integration and customization.

Because of its open and modular approach to control systems, Windows CE can better serve small, cost-effective dedicated systems. This means that PC-based control users will be able to more closely match the right level of control with specific application needs.

Windows CE also answers a dilemma for users with applications requiring hard real-time. Before the advent of CE, users requiring hard real-time control had been forced to choose proprietary PC-based control systems, specifically those based on real-time extensions, or in some cases chose to stay with traditional PLC control. Windows CE will allow hard real-time control because it is the option that does not sacrifice the integration benefits of an open platform.

Easing communications
As manufacturing begins to migrate toward Windows CE as the standard for operating systems, communication across all levels of operation will make it much easier to integrate the factory floor to the top floor. The barriers that once inhibited or restricted the flow of information between machine-level applications and supervisory applications that run on Windows 98 or NT will be reduced.

Windows CE will allow an application to benefit from the flexibility of open technology and the price advantage and simplicity of an integrated electronic operator interface (EOI) device.

As Windows CE technology becomes available on industrial computers, the end-user will be able to incorporate a variety of human machine interface (HMI) and control software packages into the application while still maintaining the simplicity of an integrated product—in effect, combining the best of both types of operator-interface applications.

Additional mobile applications
Windows CE is a good choice for mobile data collection applications. The full-featured operating system provides on-board computing power for both traditional statistical process control and specialized or custom quality control algorithms. Its rich graphical environment can produce graphical data displays and graphically driven user interfaces including work instructions, CAD information, and inspection plans for complex or intricate data collection tasks.

The compact nature of CE allows devices to be more portable with reduced complexity so training efforts and resources can be focused on the quality program rather than on using the data collection tools. The power-conscious components designed for Windows CE allow a device to operate for an entire shift without recharging or exchanging batteries. Users also have found that batteries tend to last longer in handheld devices, are more easily recharged, and are easier to swap out, making extended troubleshooting and monitoring of devices even easier.

Finally, the open architecture of CE has generated broad-based appeal. Direct Win32 support facilitates code reuse from other data collection or specialized applications. A portable data collector with an open Windows CE architecture is even more attractive because of the wealth of third party applications and accessories. Now, a single unit not only can collect data of all kinds, but can provide other digital functionality including e-mail, network client, and wireless messaging. MT

Information supplied by Rockwell Software, 1201 South Second Street. Milwaukee, WI 53204, USA. (414) 382-2000 ; Internet http://www.rockwellautomation/rockwellsoftware

This Ingersol 7-axis spar mill is one of 11 identical mills on the production line at the Frederickson facility

This story begins in the early 1960s when the Type Certification process for the 747-100 airplane was initiated by the Federal Aviation Administration (FAA). The process required that Boeing define an acceptable preventive maintenance program for the 747-100. The FAA initially envisioned this program to be three times more extensive than the 707 program under the rationale that the 747 would carry three times more passengers. United Airlines (UA), one of the first of two buyers, and Boeing realized that such a requirement was so costly that the airplane could not operate in an economically viable fashion. This problem was amplified by the fact that the existing 707 maintenance philosophy was built on the premise that an airplane periodically wore out, and required a major (and costly) overhaul in order to retain airworthiness stature.

UA and Boeing decided to return to ground zero with a clean sheet of paper, and to challenge the validity of the wearout premise. Tom Matteson, vice president, maintenance planning, and his maintenance analysts at UA played a key role in this evaluation, and were able to use their extensive historical database to prove that, in reality, only about 10 percent of the nonstructural equipment in their jet fleet showed an end-of-life or wearout characteristic. As a result, they structured a common sense decision process to systematically determine where, when, and what kind of preventive maintenance (PM) actions were really needed to preserve airworthiness.

This new look evolved into what is known as the Maintenance Steering Group (MSG). The MSG defined an acceptable and economical PM program for the 747-100, which received FAA approval. This MSG process became the standard for commercial aviation and still exists today. In 1970, the process was labeled reliability-centered maintenance or RCM by the Department of Defense (specifically, the Navy), and is now a recognized maintenance process which is practiced in many industries worldwide.

The airplane design side of BCA, which was instrumental in the development of RCM, did not communicate this finding to the production side of the house. Rather, it took more than 30 years for RCM to grow its roots in other industries before the Boeing facilities maintenance people learned about RCM in their best practices investigation. This is what brought RCM home to Boeing.

Organization and approach
Within BCA, six major regions containing all of the commercial airplane production plants have been combined under one facility organization for maintenance purposes. This organization is known as Facilities Services. The six regions are located in Washington State, Oregon, Kansas, and California. The manufacturing centers in these regions are the main customers for the organization's services.

Region leaders report to the vice president of facilities services, and they have group and team leaders as their management structure. While there is autonomy within each region, for the most part, each of the various teams consists of mechanical and electrical craft personnel, maintenance analysts, and reliability engineers. Additional core resource groups include planning as well as equipment and plant engineering capabilities.

A key element of the Facilities Services strategy was the establishment of an Asset Management Initiative in conjunction with production customers. This initiative is a formal partnership with production with the stated objective to significantly improve how BCA manages all of its assets in order to achieve lean manufacturing, process improvements, and dependable measures of asset utilization.

Facilities Services has developed a long-range strategy to optimize the application of its resources, i.e., people, material, equipment, and specialty tools. Customer (production) involvement in its execution is essential to assure that the plans align with the customer's business commitments. To achieve these goals, various state-of-the-art maintenance concepts have been deployed through an Advanced Maintenance Process program (AMAP).

BCA is investing in today's best maintenance practices, such as tactical planning and scheduling (including a new computerized maintenance management system), a continuing program of craft skills development, advanced preventive and predictive maintenance technologies, TPM, and RCM. There is also an equal focus on safety and regulatory issues. One essential element of AMAP is the inclusion of the reliability parameter. Within Facilities Services, reliability is defined as asset availability and performance, and thus encompasses every aspect of what the organization does on a day-to-day basis to prevent the loss of critical facility systems while preserving, first and foremost, employee safety.

In a review of industry best practices, it was learned that the RCM process used on the airplanes has been employed increasingly throughout U.S. industry as an effective decision technique to optimize the application of maintenance resources. The RCM focus has provided dramatic results in reducing equipment corrective maintenance actions and loss of system availability (i.e., less downtime). Further, when different RCM solutions on the market were evaluated, it was discovered that the most effective programs used the classical RCM process that followed the original airplane methodology. After it was decided to use the classical RCM process, the first pilot project was the Spar Mills at the Wing Responsibility Center in Frederickson, our newest commercial factory located south of Seattle.

Plant and equipment overview
This project was conducted at the Frederickson production facility. Two major production capabilities are located here: one is dedicated to building the vertical and horizontal composite tail sections for the 777 airplane; the other is dedicated to producing the wing spar and skin kits that are used in all airplane types currently produced by BCA at its Renton and Everett, WA, final assembly lines. It is this latter facility where we conducted the first RCM pilot project.

The spar and skin facility is a 550-person organization with an annual operating budget of about $300 million. The Facilities Services personnel are composed of skilled machinists, mechanical and electrical technicians, numerical control (NC) specialists, equipment and reliability engineers, and other support personnel. This facility opened for production in April 1992. It has 21 acres of floor space that produce both spar and skin wing sections in continuous aluminum pieces up to 110 ft in length. The spar production line includes automated stringer handling, overhead crane, spar mill, drill router, deburr, paint, chip collection, shot-peen and bending/forming systems.

For the RCM project, the spar mill was divided into three subsystems: cutting, control, and auxiliary support. The cutting subsystem was chosen for a detailed RCM analysis.

The RCM team
Based on past success, the RCM team was composed of craft personnel (operator, mechanic, electrician), supported by a reliability engineer and maintenance analyst. The RCM consultant, Mac Smith, was the team trainer and facilitator. This combination of experience and hands-on knowledge of the spar mill was the key to project success since their contributions were reflective of the technical details of the equipment, how it was used in the daily operations, and how it applied to the RCM methodology.

Initially, fear of change, suspicion of goals, ingrained viewpoints, skepticism about the new process, and individual agendas were all present. But, as the project moved on, the team felt a real sense of satisfaction because they were direct participants in an opportunity to provide meaningful value added content to their daily work.

Classical RCM
Four basic features define and characterize the RCM process:

• Preserve system function.
• Define functional failures and specific component failure modes that can defeat required functions.
• Prioritize the importance of the failure modes.
• Select applicable and effective PM tasks for high-priority failure modes. Applicable tasks include those that will prevent, mitigate, detect the onset of, or discover hidden equipment failure modes. Effective tasks are the lowest cost tasks among competing options.

The RCM system analysis process consists of seven steps.

1. System selection and information collection. In a large facility or plant, certain systems tend to have a higher impact on plant operation and maintenance than others (higher maintenance cost, more forced outage contributions, etc.). Boeing focused the effort by selecting the systems that have the highest potential for improvement of the maintenance process, where the best return from an RCM program can be obtained. This selection essentially follows the 80-20 rule where 80 percent of the unexpected costs and production losses derive from 20 percent of the systems.

2. System boundary definition. Each system selected must have well-defined physical boundaries to prevent overlaps with adjoining systems or voids in the analysis, and to precisely define what moves back and forth across the boundary in step 3.

3. System description and functional block diagram. Complex systems are usually subdivided into two or three functional subsystems. Detailed descriptions and component equipment lists are developed for each subsystem. The system is represented in a functional block diagram and all in and out interfaces at the system boundary are carefully listed. The out interfaces become the key data for understanding the functions that must be preserved.

4. System functions and functional failures. The information developed in steps 2 and 3 provides the basis to precisely define the system functions and then the functional failures associated with each.

5. Failure mode and effects analysis (FMEA). The functional failures and components for each subsystem are arrayed in a matrix to reveal intersections where a potential component failure could produce the functional failure. This matrix is used as a road map to develop the FMEA for each critical point in the matrix.

6. Logic (decision) tree analysis. Any component failure mode that produces a system or plant level effect in the FMEA is tagged for logic tree analysis (LTA) and further refinement as to its significance and priority. By using three simple yes or no gates, each failure mode is categorized as safety related (Category A), outage related (Category B), or minor economic disadvantage (Category C); each is also designated as evident or hidden (Category D) relative to the normal operations process. As a general rule, all Category C failure modes are designated as run to failure, and the focus becomes the Category A and B failure modes.

7. Task selection. The safety and outage component failure modes from the LTA are evaluated to select an applicable and effective PM task. All failure modes that had been designated as run to failure (RTF) are put through a second evaluation called a sanity check before any final RTF decision is reached.

Analysis results
The existing PM program at project initiation was essentially a one-shot overhaul activity scheduled to be done on 9-month intervals. It was discovered that over a 5-yr period since their installation, this interval exceeded 9 months about 80 percent of the time, and extended to 18 months or more 40 percent of the time. This resulted in an excessive trouble call history that had then caused major downtime problems, elevating the spar mill to one of the 80-20 systems at Frederickson.

A statistical overview of the team's analysis and results for the cutting subsystem are shown in Figs. 1-3.

RCM Systems Analysis Profile Spar Mill #6 - Cutting Subsystem

The systems analysis profile (Fig. 1) provides a feel for the scope of the project that took 32 days to complete, and was spread in one-week efforts over a six-month period. Perhaps the biggest surprise was that almost half of the failure modes (42 percent) were hidden if they occurred, thus the operator was unaware that any problem was developing with the spar mill until later, when the consequence finally would show, often in a detrimental way. Not surprising was the fact that most of the failure modes were high on the criticality list, and could cause personnel injury or downtime. The team made 197 decisions on what to do with the 172 failure modes (some failure modes received multiple PM actions).

RCM-PM Task Type Profile Spar Mill #6 - Cutting Subsystem

Fig. 2 shows the makeup of the RCM results in the task type profile, and its comparison to the existing maintenance system. The outstanding point is that the number of failure modes receiving no attention currently was reduced by one-third, and the nonintrusive actions with condition directed (including predictive maintenance) and failure finding tasks were increased by a factor of four.

The real impact of the RCM results is seen in Fig. 3, the task similarity profile, where the similarities and differences between the RCM and current PM tasks were examined. Because the analysis showed where the critical failure modes were located, appropriate PM actions were developed where needed and PM work that was not needed was eliminated. Overall, the RCM results changed 71 percent of the current maintenance practices on the spar mill. Because there are 11 spar mills at Frederickson, the multiplying effect is quite dramatic.

RCM-PM Task Familiarity Profile Spar Mill #6 - Cutting Subsystem

Finally, the intensity of the RCM process enabled the team to discover a number of nonmaintenance related findings, items of interest (IOI), which also provide a significant portion of the benefit achieved. There were 36 IOIs recorded that affected design, operations, reliability, safety, and logistics.

Implementing the results
Implementation of the PM task findings that were developed in step 7 of the system analysis process proved to be challenging, because:

• A general understanding of the RCM process and a buy-in to the results of the analysis from a broad group of personnel who were peers of the team members was needed for implementation. Nothing new on a factory floor is ever successfully deployed by simply announcing that it will happen!
• Several of the new tasks required a more direct participation on the part of the operators, and this had to be carefully integrated with the production shift supervisors.
• With the substantive changes being made to current procedures, time was needed to develop several new procedures and coordinate their review and approval with all affected parties.
• Several of the condition-directed tasks required some exploratory work to ascertain their suitability for the failure mode(s) in question. Some of this work is still on-going.

All of the above required extensive communication across organizational lines and among the disciplines that are resident in the production and maintenance work force. Involvement of production personnel, and a close integration of their viewpoints and experiences with Facilities Services was a key ingredient in the entire project.

Currently, 21 RCM-based PM procedures have been deployed on the spar mill. These procedures essentially encompass all of the analysis findings, except for a few condition-directed tasks still under evaluation. The new PM format being used includes additional descriptions of the work to be performed plus references to the specific failure modes and failure causes that triggered the PM tasks. Deployment to the factory floor was done step by step to introduce the shift from traditional to RCM tasks without disruption. Again, open communication was essential to successful implementation. Giving honest and positive feedback to the questions that were asked was crucial to creating a positive paradigm shift.

The IOIs are in the process of being evaluated and, where appropriate, implemented. To date, several have been accomplished, including:

• Pressure wash of the entire machine has been eliminated in favor of selective washing of a few components. This has virtually eliminated severe corrosion and chip contamination damage caused by the pressure wash.
• Spindle vibration analysis is being closely correlated with the as-produced part quality (tolerance) to obtain the maximum spindle life before changeout.
• A&B axis rack covers have been removed, since they trap chips and cause pinion seal damage, rather than prevent chip entrance to the racks.
• All 7 axis drive motors will be replaced with brushless motors, eliminating five specific failure modes of concern.
• Counter-balances have been added to all W and Z axes to eliminate failure of the thrust bearing.

Return on investment considerations
The objective of the RCM program is to focus PM resources to reduce costly corrective maintenance actions and resulting loss of machine uptime. While no hard measurements are yet available, the following observations can be made:

• While the PM program has changed significantly, its cost is virtually unchanged. Costly time directed tasks have been replaced by expensive condition directed and failure finding tasks, and the task frequencies have been extended.
• With the program now focusing on the critical failure modes, a reduction in unexpected corrective maintenance actions (trouble calls) of at least 50 percent is expected. Downtime should also decrease by at least 50 percent.
• From preliminary analyses, IOIs have the ability to produce annual savings in excess of $3 million when implemented. Of those incorporated to date, annual savings in excess of $500,000 are expected.

Future directions
With the experience gained and success achieved at Frederickson, two additional RCM pilot projects at Everett and Wichita have been initiated and recently completed. These projects were performed on a Cincinnati 5-axis router and Modig extrusion mill, respectively, and are currently in the transition to implementation. ROI benefits simlar to the spar mill are expected to accrue from these projects.

Currently, six other RCM projects are in progress, two at Everett, WA, and four at Wichita. Several additional RCM progress are contemplated in the schedule for 2000.

The RCM Living Program will be applied to all completed projects in order to periodically update the PM tasks as may be required and to effectively measure the results of the RCM program.

The classical RCM process will continue to be used on critical systems because the actual benefits have exceeded our original expectations. MT

Dennis Westbrook is Maintenance Process Focal and Robert Ladner is Facilities Maintenance Analyst, Fredrickson Site, for Boeing Commercial Airplane. Anthony M.(Mac) Smith is principal at AMS Associates, San Jose, CA. The authors can be contacted by email at dennis.westbrook@boeing.com, robert.ladner@boeing.com, and amsassoc@aol.com

As with any failure analysis, finding the root cause of damage to gearing often requires a lot of detective work. You may need to review the service history and interview witnesses or employ technical tools such as vibration analysis and oil analysis. However, the cause of failure cannot be determined without a complete inspection of the condition of the gear teeth themselves. An understanding of the failure modes indicated by the condition of the teeth, when combined with knowledge of the operating conditions and maintenance history, will permit developing methods to avoid similar failures in the future.

Gear Tooth Profile Terminology: Tooth mesh changes from sliding ro rolling action at the pitch diameter and then back to sliding during gear rotation.

Getting into gear
In order to analyze and interpret gear failures, it is helpful to consider some of the terminology and practices commonly used in the gear industry. The accompanying drawing shows a few of the common terms used to describe gear tooth profiles.

Gear quality ratings are established by the American Gear Manufacturers Association (AGMA). Quality levels are driven by the application requirements. In some basic applications, AGMA 4 or 5 quality gears may suffice, while other more demanding applications may require an AGMA 12 or 13 gear; aircraft transmissions may require AGMA 14 or 15 accuracy. The case hardened and ground gears used in many high-capacity gear drives today are generally at least an AGMA 11 quality level. The differences between quality levels are progressive, somewhat like the Richter earthquake scale, where the difference between one level and the next is substantial. This can cause problems if an attempt is made to reverse-engineer a replacement gear without knowledge of its quality level. Replacing a gear with one of lesser quality may have disastrous effects on gear life.

Service factors play an important role in selecting the proper gear drive for the application. Manufacturer catalogs list typical service factors for various types of applications. In a speed reducer, the ratings are applied to each gear set. A multi-stage reducer will be limited by the lowest rated gear set, which will usually be the low-speed gear set of a typical industrial gear drive. This gear set also transmits the most torque.

Things to be aware of when reviewing an application for possible causes of failure include the possibility of design error in specifying the original gear set. As an example, the speed reducer on a mixer might be sized adequately for operation but not for startup if the mixer is full and therefore requires considerably more power to overcome the inertia of the load. If this happens, the high speed pinion shaft could deflect, which may cause the gear teeth to run misaligned and overload them. Not only does this accelerate wear, but it can force the oil out of the gear mesh and cause several types of failure.

The primary way to check design and manufacturing errors is to review the inspection charts, specifications, and other information from the manufacturer, then compare them with the requirements determined by reviewing the actual application parameters. The original design may have been satisfactory, but subsequent changes in the application could cause it to be inadequate.

Evidence Of Wear Failure Fig. 1. Moderate wear Fig. 2. Abrasive wear Fig. 3. Corrosive wear Fig. 4. Scoring

Why gears fail
One person's failure may be another's break-in. The difference between wear and failure can be simply a matter of time. If a gear fails in 25 years, it did its job. If it fails in 25 minutes or 25 hours, there's a serious problem.

When gears mesh, they roll only at the pitch line, as noted in the drawing. Above and below this line, the sliding action that occurs causes inherent wear that can lead to failure. Gear teeth also flex as they go in and out of mesh. Therefore, they have to be soft enough to deflect and give without breaking. Yet a hardened gear has higher capacity ratings, so most gears are heat-treated to harden them to the degree necessary for the application.

Gears may be either through-hardened or case-hardened. Through-hardened gears are put through a heating and controlled cooling process as a unit, so the hardness is the same throughout the gear. These gears are usually below 390 Brinell in hardness, above which conventional machining becomes difficult or impossible. Case- hardened gears are hardened only on the surface of the gear teeth, to a predetermined depth, to about 58 to 62 Rockwell C, or roughly as hard as a bearing race. The increased hardness improves the gear's durability rating by providing greater resistance to pitting and greater strength, or resistance to breakage.

From one point of view, causes of gear failure may include a design error, an application error, or a manufacturing error. Design errors include such factors as improper gear geometry as well as the wrong materials, quality levels, lubrication systems, or other specifications. Application errors can be caused by a number of problems, including mounting and installation, vibration, cooling, lubrication, and maintenance. Manufacturing errors may show up in the field as errors in machining or heat treating.

AGMA recognizes four main modes of gear failure, plus a fifth that covers everything else. They are wear, surface fatigue, plastic flow, breakage, and associated gear failures.

When a gear is suspected of showing signs of failure, if possible it should be examined periodically over time. Recording contact patterns or taking photographs at intervals will aid in comparison and help determine whether the condition is progressive. Keep in mind also that failure never occurs as an isolated event. Two or more failure modes may occur simultaneously or in succession, and the eventual failure mode may be different from the root cause.

Wear Failure
Wear, the first failure mode category, occurs when metal is worn away from the contact areas of the gear teeth in a more or less uniform manner. Some wear is normal, but there are several degrees of wear and many ways in which wear can occur.

Polishing is a slow process of wear in which metal-to-metal contact during operation causes a very smooth surface to develop on the gear teeth. It is most common during slow-speed operation, where the lubricant film is too thin, and the gears are operating near the lubrication borderline. Normally, this condition does not cause a problem unless continued wear prevents the gears from reaching the design life of the equipment. Once the gears are polished, further action can be reduced or prevented by using a higher viscosity lubricant or lowering the lubricant temperature. Other possible remedies include reducing the transmitted load or increasing the operating speed to provide a better oil film.

Moderate wear (Fig. 1) shows up as a contact pattern in which metal removal occurs from both the addendum and dedendum tooth surfaces, and the operating pitch line remains as a continuous line. This may be caused by lubricant contamination but is often unavoidable due to limitations of lubricant viscosity, gear speed, and temperature. It may occur normally throughout the design life of a gear set, particularly when gears operate near boundary lubrication conditions. Increasing oil film thickness, either by cooling the lubricant, using a higher viscosity lubricant or operating at higher speeds, can sometimes reduce normal wear. Replacing a splash-fed lubrication system with a filtered positive-spray system may improve lubrication by removing particles and delivering a more consistent supply of oil to the working surfaces. Further solutions include reducing the gear loading and changing the gear geometry, materials, or hardness.

Extreme wear may appear as the same kind of contact pattern and pitch line visibility that occur with moderate wear, but the progression rate is much faster. Here, a considerable amount of material may be removed uniformly from the gear tooth surfaces, and the pitch line may show signs of pitting. Extreme wear will cause failure to occur before the design life of the gear set is reached. It may cause enough damage to the tooth profile that the resulting high dynamic loads will further accelerate the wear. Causes of extreme wear include a lubricating film too thin for the tooth load, fine abrasive particles in the lubrication system, and severe vibratory loads. Shaft seals and air-vent filters, properly installed and maintained, may help reduce wear. Other solutions include oil cooling, higher viscosity lubricants, higher speeds, reduced loads, and possibly reduced vibratory loads if the application permits.

Abrasive wear shows up as a lapped surface, with radial scratches or grooves on the tooth contact surfaces. When this occurs shortly after startup of a new installation or on any open gearing, particles in the lubricating system are generally the cause. These may include metal particles from the gears and bearings, weld spatter, scale, rust, and sand, dirt, or other environmental contaminants. Fig. 2 shows severe abrasion. Careful cleaning of the gearbox and lubrication system before use can minimize abrasive wear. With a circulating lubrication system, adding a filter or using a finer replacement filter will help reduce this type of wear. Regular oil changes will help for splash-lubricated drives, and higher viscosity oil also may help protect either type of system with a thicker oil film that will keep the finer particles from scratching.

Corrosive wear (Fig. 3) is visible as surface deterioration, caused by the chemical action of active ingredients in the lubricant. These may include acid, moisture, foreign materials, and extreme-pressure additives. During operation, the oil breaks down and allows corrosive elements present in the oil to attack the gear contact surfaces. This action may affect the grain boundaries and cause fine, evenly distributed pitting. Checking the oil for breakdown and changing it at regular intervals can help minimize corrosive wear. Lubricants with high antiscuff, antiwear additive content must be observed even more carefully because they are chemically active. Gear units that are exposed to salt water, liquid chemicals, or other foreign materials should be sealed from their environment.

Scoring may be moderate, localized, or destructive. It can be caused by failure of the lubricant film, usually from overheating in the mesh area, as well as by misalignment, deflection, and uneven temperatures or loads. The resulting metal-to-metal contact produces alternate welding and tearing that quickly removes metal from the gear surfaces. Moderate scoring shows up as a characteristic wear pattern, often in patches on the addendum, dedendum, or both. Radial tear marks usually appear more prominently in softer areas. Upon closer examination, the frosty appearance shows that the rotation has caused the metal to weld and tear apart (Fig. 4). Localized scoring is similar to moderate scoring but takes place in concentrated portions of the contact areas of the gear teeth, rather than spreading across their full face width.

Destructive scoring or scuffing shows definite radial scratch and tear marks, and material may be displaced radially over the tips of the gear teeth. Excessive material may be missing from above and below the pitch line, causing the pitch line itself to stand out prominently. At this stage, the gear is unfit for further service.

Reducing the temperature in the mesh area can prevent moderate scoring. This can be accomplished by reducing the load, gear speed, or inlet oil temperature. Other solutions include use of a lubricant with extreme-pressure additives, plating a solid lubricant on the contact surfaces, or honing.

Localized scoring is more likely to result from misalignment factors than moderate scoring. A wear pattern that shows load concentration near one end of the teeth indicates possible misalignment or helix angle error. This results in one portion of the teeth carrying more load than the lubrication film can support. Eliminating the causes of uneven loading can prevent localized scoring. These may include nonuniform gear case deflection, excessive shaft deflection, out-of-parallel bores in the casing, or helix angle errors. Uneven temperature gradients also may cause localized scoring and should be remedied by changing the amount of cooling oil applied to the mesh or the way in which it is applied.

To eliminate destructive scoring (Scuffing), it is necessary to attack the source of the excessive heat that causes the lubricant to break down. Extreme-pressure additives are one way to help the lubricant stand up to the load, speed, and temperature conditions. Special high-viscositycompounded gear oil or synthetic fluids with anti-scuff additives also will help prevent scoring. In extreme cases, the gearing may have to be redesigned to reduce surface stresses, pitch line velocity, and oil temperature of the gears.

Tip and root interference is another type of scoring, usually resulting from improper design and manufacture. Metal removal will be seen near the root of the gear tooth profile while other portions of the contacting face will appear undamaged. The tip of the gear or pinion may look abraded, with tear marks in the direction of rotation. With high speed gears, scoring at start-up is considered failure, and the gears should be replaced after correcting the cause of scoring.

Evidence Of Surface Fatigue Failure Fig. 5. Pitting Fig. 6. Destructive pitting Fig. 7. Spalling Fig. 8. Micropitting Fig. 9. Micropitting magnified Fig. 10. Case Crushing

Surface fatigue failure
Surface fatigue can be noticed by the removal of metal and the formation of cavities. These may be small or large and may grow or remain small. It occurs when the gear material fails after repeated stresses that are beyond the endurance limits of the metal. Here are the main types of surface fatigue, their causes, and cures.

Pitting failures depend on surface contact stress and the number of stress cycles. Initial pitting (Fig. 5), with areas of small pits from 0.015 in. to 0.030 in. in diameter, occurs in localized parts of the gear teeth that are over-stressed. It is sometimes called corrective pitting because it tends to redistribute the load by progressively removing high contact spots, and often stops once the load has been redistributed. Continued operation may polish or burnish the pitted surface and improve its appearance. Pitting can be monitored by periodically putting some bluing on the affected area, then applying some cellophane tape to lift the pattern and put it in a notebook. Comparing the impressions over time will tell whether the pitting has stopped. While accurate manufacturing control of involute profiles is the best method of preventing pitting, a careful break-in at reduced loads and speeds once the unit is installed also will help minimize pitting by improving gear tooth contact.

Destructive pitting (Fig. 6) appears as much larger pits than initial pitting, often in the dedendum section of the gear teeth. These larger craters usually are caused by more severe overload conditions that cannot be relieved by initial pitting. As stress cycles build up, pitting will continue until the tooth profile is destroyed. To correct the cause of destructive pitting, the load on the surface of the gear needs to be reduced below the material's endurance limit, or the material hardness needs to be increased to raise the endurance limit to where pitting will not occur.

Spalling (Fig. 7) resembles destructive pitting, except that the pits may be larger, quite shallow, and irregularly shaped. The edges of the pits break away rapidly, forming large, irregular voids that may join together. Spalling is caused by excessively high contact stress levels. Remedies include reducing contact stress on the gear surface or hardening the material to increase its surface strength.

Both spalling and destructive pitting are indications that the gears do not have sufficient surface capacity and should probably be redesigned if possible.

Micropitting is a type of contact fatigue that appears as frosting or gray staining under thin film conditions (Fig. 8). The surface acquires an etch-like finish, with a pattern that sometimes follows the slightly higher ridges left by cutter marks or other surface irregularities. It usually shows up first on the dedendum section of the driving gear, although it may begin on the addendum section as well. When viewed under magnification (Fig. 9), the surface is seen as a field of very fine micropits under 0.0001 in. deep. Causes include high surface loads and heat generation, which thins the lubrication film and leads to marginal lubrication. Improving the surface finish is an effective remedy, through either manufacturing techniques such as hard honing and grinding or a careful break-in cycle. These techniques help lower heat generation by improving conformity of tooth contact and equalizing load distribution. Reducing the lubricant temperature and surface loading will also minimize frosting. Sometimes, frosted areas that appear initially will slowly be polished away during subsequent operation if loads and temperatures are not excessive.

Case crushing occurs in heavily loaded case hardened gears, including those that are carburized, nitrided, or induction hardened. It is a subsurface fatigue failure that occurs on material where the case is substantially harder than the core, when surface contact stress at high cycle levels exceeds the materials endurance limit. Case crushing may appear similar to pitting, if some damage occurs on contacting surfaces. However, it often occurs as longitudinal cracks on the surface of only one or two teeth, and long pieces of the tooth surface may break away (Fig. 10). The case material may appear to have chipped away from the core in large flakes. Case crushing occurs when cracks form because stresses in the subsurface area exceed the strength of the core material. High residual stresses may contribute to this effect. The cracks move toward the case-to-core boundary and then to the gear surface, where they may eventually cause large pieces of material to fall off. To prevent case crushing, it may be necessary to in- crease the depth of the case hardening and possibly the hardness of the core material. Changes in the material, heat treatment process, or the design itself may be necessary.

Evidence Of Plastic Flow Fig. 11. Rippling Fig. 12. Ridging

Plastic flow failure
Plastic flow is a surface deformation that occurs when high contact stresses combine with the rolling and sliding action of the meshing gear teeth to cause cold working of the tooth surfaces. Although usually associated with softer materials, it also can occur in heavily loaded case hardened and through-hardened gears. Plastic flow generally takes one of three distinct forms.

Cold flow, rolling, and peening can be identified through evidence of metal flow in the surface and subsurface material. The surface material may have been worked over the tips and ends of the gear teeth, resulting in a finned appearance. Tips of the gear teeth may be heavily rounded over, and a matching depression may appear on the tooth surface. Cold flow occurs under heavy loads and high contact stresses, as the rolling and peening action of the meshing gear teeth cold-works the surface and subsurface material, pushing or pulling it in the direction of sliding. Continued operation during this deterioration increases dynamic loading and results in a dented, battered appearance on the surface, much as if it had been hit with a ball peen hammer. To eliminate the problem it is necessary to reduce contact stress and increase hardness of the contacting surface and subsurface materials. Increasing the accuracy of both tooth spacing and profiles will help reduce dynamic loads, and any mounting deflections or helix angle errors should also be corrected.

Rippling is a regular, wave-like formation that occurs at right angles to the direction of motion and has a fish scale appearance (Fig. 11). It is most common on hardened gear surfaces and is generally considered a surface failure only when it has progressed to an advanced stage. It usually occurs in slow speed operation with an inadequate oil film thickness. High contact stresses during repeated cycles may then roll and knead the surface, causing it to ripple. Rippling can be prevented by case hardening the tooth surface, reducing the contact stress, increasing oil viscosity, and using an extreme-pressure oil additive.

Ridging is a definite series of peaks and valleys that occur across the tooth surface in the direction of sliding (Fig. 12). It occurs when high contact compressive stresses and low sliding velocities cause plastic flow of the surface and subsurface material. It is frequently found on heavily loaded worm gear drives, as well as on hypoid pinion and gear drives. Remedies for ridging include reducing contact stress, increasing material hardness, and using a more viscous lubricating oil with extreme-pressure additives.

Breakage failure
Breakage is the fracture of a whole tooth or substantial part of a tooth. Common causes include overload and cyclic stressing of the gear tooth material beyond its endurance limit.

Bending fatigue breakage starts with a crack in the root section and progresses until the tooth or part of it breaks off. It can be recognized by a fatigue eye or focal point of the break. The break area itself usually shows signs of fretting corrosion and smooth beach marks that resemble patterns in the sand on a beach. A small area will probably have a rough, jagged look where the last portion of the tooth broke away. Most such failures result from excessive tooth loads, which cause repeated root stresses that eventually exceed the endurance limits of the material. Stress risers, such as notches in the root fillet, hob tears, inclusions, small heat treating cracks or grinding burns, may aggravate this condition. To remedy this condition, root fillets can be polished and shot-peened. Material should be properly heat-treated to minimize residual stresses. If redesign is necessary, use a full-fillet radius tooth, which is less prone to breakage and has greater capacity than a tooth with too small a fillet radius.

Overload breakage appears as a stringy, fibrous break that has been rapidly pulled or torn apart. In harder materials, the break will have a finer stringy appearance. The eye and beach markings found in fatigue breakage will be missing. This type of breakage is caused by an overload that exceeds the tensile strength of the gear material. Typical overloads that lead to such breakage include a bearing seizure, failure of driven equipment, foreign material passing through the gear mesh, or a sudden misalignment. Since the failure is usually the result of some unpredictable occurrence, it is difficult or impossible to prevent. If possible overloads are anticipated, torque-limiting couplings may provide some protection.

Random fracture can occur in areas such as the top or the end of a tooth, rather than the usual root fillet section. These failures are typically caused by stress concentrations from such things as minute grinding cracks, foreign materials in the gear mesh, or improper heat treating. Little can be done to prevent random fracture, except at the design and manufacturing stages. However, maintaining cleanliness of the lubricant can help prevent one cause.

Evidence Of Associated Gear Failure Fig. 13. Quenching cracks Fig. 14. Grinding cracks Fig. 15. Rim and web failures Fig. 16. Electric current damage

Associated gear failures
Associated gear failures usually are caused by improper processing, environmental conditions, or possibly by accidents. To minimize many of these failures, any gear that is repaired and heat treated should be checked by magnetic particle inspection before being put back into service to be sure no cracks have developed. Whenever repairs are made to any gearing, at the very least, a dye penetrant inspection should be performed to check for cracks.

Quenching cracks may appear across the top land of a tooth, in the fillet area, or randomly at the tooth ends, although they may not become visible until after they have been used for a short time (Fig. 13). They are caused by improper quenching or uneven cooling during heat treatment, which causes excessive internal stresses. Prevention of quenching cracks calls for a thorough review of heat treating procedures, as well as an inspection of the equipment used.

Grinding cracks (Fig. 14) usually show up as a definite pattern, either as a series of short cracks that are parallel to each other or with the appearance of chicken wire mesh. Usually, they are between 0.003 in. and 0.005 in. deep, with the parallel type being deeper than the chicken wire pattern. Causes include improper heat treatment or a metallurgical structure that is prone to cracking. To prevent this cracking, the grinding procedure should be reviewed. Feeds and speeds may have to be reduced to lower the heat developed during grinding. The metallurgy of the gear material also should be examined to choose an alloy and heat treatment that will not tend to crack during grinding.

Rim and web failures tend to start between two teeth and propagate through the rim and into the web (Fig. 15). These failures are common on highly loaded thin rim and web sections. Causes include stress risers from holes in the web as well as from web vibrations. Remedies include increasing rim or web thickness, depending on failure mode, and eliminating stress risers such as grinding marks, tool marks, and sharp fillets. Rim and web failures also may be caused by vibrations, which can be minimized by damping or by redesign to change the natural frequencies of the gear.

Electric current damage shows up as tiny pits occurring in a well-defined pattern that is distributed uniformly along the gear surfaces (Fig. 16). They can be further identified by their smooth, molten appearance and lack of any fibrous appearance. This damage results from electric current passing through two lightly contacting surfaces, either from arc welding or from electric equipment such as motors or electrically actuated clutches. The remedy is to insulate the electrical equipment or relocate the grounding wires properly. Welders and maintenance workers should be made aware of proper grounding procedures.

Determining the real cause
A complete and accurate assessment of the cause of any gear failure requires a knowledge of the basic gear failure modes, their causes, and possible remedies. All available information on operating conditions, performance history, and maintenance details will help to point to the specific cause and to develop solutions to prevent future failures. The purpose of this article is provide a basic knowledge of the terms used in gear failure analysis and to promote accurate communication when determining the cause of failure and how to prevent future problems. In the majority of cases a single failure mode is not evident. The initial failure damage may be obscured by subsequent damage. To determine the specific mode and cause of the initial failure, the assistance of an experienced gear failure analyst may be required. MT