Machine design and equipment design come in many different shapes and sizes. While some are more akin to flavor-of-the-month approaches, there are design philosophies that are very useful in the appropriate environment.
For a high volume consumer product, especially one with “no user serviceable parts inside,” the philosophy of “design for assembly” might be useful in simplifying the assembly process with snap-together parts, reducing overall part count and meeting cost targets.
Processes that include casting, fabrication, heat treatment, machining and finishing would call for a different design approach. “Design for manufacturability” would fit the bill.
On the other hand, many projects require matching the equipment and system performance to the operating needs, which provides the long-term, most cost-effective system to meet the defined mission. We call this “design for reliability.”
According to the Center for System Reliability (CSR), “Reliability should be designed and built into products at the earliest stages of product development. As most of a product’s lifecycle cost has been locked in by the time its design is complete, design for reliability is the most economically sound approach to take.”
Reliability doesn’t just happen. It’s the result of careful planning and effective execution. We can measure it using overall equipment effectiveness (OEE), or the combination of quality, performance and availability. To reach target production, a machine must be operating (available), must be operating within specification (quality) and must be operating at the required cycle rate (performance). The goal of our design effort is simply to meet those requirements and maximize OEE.
The mission must be fully defined to achieve maximum performance and efficiency upon completion. Variables, including throughput, power consumption, environment, materials being handled and required lifetime, must be accounted for. If the lifetime is relatively short, a sealed bearing requiring no maintenance might be selected. If a piece of machinery must last for 20 years, the maintenance and replacement must be accounted for, as well as the relative life of alternate materials. A schedule for inspection, measurement and a feedback loop must be defined to ensure adequate warning of developing problems. There are situations in which a unit that fails predictably but can be repaired quickly with standard parts might be more effective than a unit that fails less often but needs special skills and parts to repair. The question is how we define the reliability goal. We might be able to deal with one hour of downtime a week to replace a worn impeller, but not two days every three months to replace a more complex and reliable part.
The design and review process helps ensure the following questions are answered:
- Can the system be constructed?
- Can the system be operated?
- Can the system be maintained?
- Can the system be updated for changes in requirements?
- Can this all be accomplished within the budget constraints?
Designing-in inspection capabilities and indicators promotes reliability. When an operator or maintenance technician can’t easily and safely access the equipment, a check mark for completion is likely to appear, whether the task is completed or not.
An example is the need to do routine inspection on a belt drive. The old design used a sheet metal shroud over the belt pulleys. The machine had to be stopped and the guard removed to do this simple check. Because operations didn’t want to shut the system down, the check was often skipped. Belt failure then occurred at inconvenient times, resulting in conflicts with operation and maintenance.
Using design-for-reliability principles, a new shroud was made using an expanded metal front. This was painted black, and the belt and pulleys can now be inspected with the system running using a strobe lamp. The inspections are completed every time, and belt changes are scheduled before failure occurs. Because the system doesn’t have to be stopped to be checked, operations now actually encourage inspections. Reliability happens.
If you can’t measure it, you can’t control it
Designing systems having feedback mechanisms is a critical element for long-term success. Regardless of the information source — failure logs, failure analyses or a computerized maintenance management system (CMMS) — this organization can compare reliability of the same or similar equipment to a performance standard. Early warnings of problems or confirmations of successes are equally important. We don’t need to mess with perfection.
The earlier in a project we start to design for reliability, the easier the process will be. This includes operation, inspection and maintenance practices, not just the hardware/software. By considering how a system is to be operated and maintained, we can eliminate problems such as inaccessible control positions, confusing controls and hidden maintenance tasks. An oil reservoir might need to be moved so the level is visible or the order of a row of switches rearranged so they are sequential. Small changes can avoid large problems later.
For example, because of bad design, the controls for seal water valves at a water treatment plant were located 25 ft from the pump/motor starter panel. In addition, the path to reach the seal water controls wasn’t direct. Although the operators had been told the importance of seal water flow, no one had actually told them why seal water flow had to be started before the pump. Only after a series of costly seal failures did the training get changed. One could call this “design for unreliability.”
How does one approach a design project to maximize reliability?
Albert Einstein said a scientific theory should be as simple as possible, but no simpler. He saw the need to add complexity to a number of models that didn’t work well and made history.
This is the way many projects develop. They start simple but become increasingly complex as deficiencies are eliminated. The result is often overcomplexity and unreliability. Looking at the history of the automobile, we see vehicles going from simple and unreliable to complex and unreliable. A number of years back, the automakers realized they had to fix excessive warranty problems because of both the cost and the negative publicity. It took Ford, in cooperation with many suppliers, several years to develop an engine that would run for 100,000 miles with no attention except routine oil and filter changes. Many components had to undergo radical change. Just one component of those engines, the spark plug, had to go through several development cycles of material, geometry, construction and spark energy changes to finally achieve the goal.
The advantage with the automobile is the high volume of parts that amortize the development cost, and payback of pennies per unit adds up over time. For industrial systems we don’t often have the same luxury. The result is often reduced development and lowered goals. The root of the problem is that customers don’t want to pay extra for reliability. The challenge then of design for reliability is to find the root causes of problems and eliminate each problem without introducing additional problems or raising the overall cost. In other words, we need the cost savings to pay for the development.
One technology that ABB has been working on is variable speed drive systems. The workhorse has been the DC motor with tachometer feedback to provide the best combination of power, torque and speed control. DC motors have commutators and brushes that comprise a sensitive mechanical and electrical interface, which is adversely affected by corrosion, contamination and wear. Brushes need to be replaced regularly because non-optimum operating conditions can lead to rapid commutator wear. The DC motor also is expensive to manufacture. For many applications, an AC induction motor can be used, but it has limitations in its inherently nonsynchronous operation. The AC induction motor has the great advantage of being simple, reliable and low-cost. Unfortunately, simplicity comes with a performance limitation. To find an optimum solution, ABB has done extensive development with permanent-magnet rotors. The new designs solve both performance and reliability issues with a system that runs in a synchronous mode and has torque and power curves much closer to the needs of typical applications. In fact, the permanent-magnet technology can enable a motor to be directly coupled without the need for a gearbox, and the result is further increased reliability, lower cost and reduction in maintenance.
This new package took extensive design for performance to get the right combination of materials, geometry and structure; design for manufacturability and assembly to be able to make and build the motors; and design for reliability for long and stable operating life. In particular the magnets themselves had to meet stringent requirements on the effects of age and use to ensure the motors didn’t lose performance over time. Hidden breakthroughs like this don’t make headlines — in fact they are more often than not top secret — but are key to the long-term performance and reliability of the new technology.
The tools that enhance reliability include optimization, prediction and analysis. First, optimize the choice of actions for the maximum return on investment. This ensures your job continues. Then estimate the probabilities of failure. Find which components are stressed most and which items are most likely to be damaged in use or repair. Finally, test and analyze, develop fault trees and identify potential failures before they occur. Now we should have a design that meets the performance objectives. No one promised it would be easy.
Making predictions always is risky; the longer the timescale and meager the data, the higher the risk. A glance out of the window will tell you if it’s raining, but not whether it will rain next week. For a reliability prediction, we need to observe how close the prediction fit the actual performance and then update our model. The whole point of prediction is to make the best estimate of system performance before it reaches failure. We base the model on existing components and similar systems.
Reliability means different things to different people. To a deep-sea fisherman, engine reliability centers on the engine operating for several months under adverse conditions without breakdowns. To a drag racer, it means running for five seconds at full power without the engine exploding. For each project or mission, the variables will be different, but the underlying principle is the same. To achieve enhanced reliability, we must have a process of feedback and evaluation to improve the system continuously.
These needs involve close interaction to check that controls are logical and components needing maintenance are accessible. We can’t limit the work to making the hardware and software. Training, instruction manuals and operating procedures also are part of the project. The drag racer is concerned with being able to strip and rebuild his engine quickly while the boat captain is concerned that he never has to touch his, but the end result is the same: they both want to complete their mission. Design for reliability is all about successfully completing the mission.