What’s driving a real-world standardized shop-floor communications standard?
As factories become more automated, more elements within manufacturing could go wrong if maintenance is not performed correctly. Even so, the risk of failure is seldom as much of a barrier to the adoption of automation as it was in the past. The associated costs of the maintenance regime, however, including line downtime for routine maintenance work, the potential cost of breakdowns, and the potential need for investment in a storeroom full of spare parts are significant. In today’s high-mix production environment, tools that will help drive down the cost of maintenance while reducing the risk of equipment failure are needed, which would eliminate several forms of waste and be driven by the Internet of Manufacturing.
As an honest analysis of the cost of automation would show, the cost of poor maintenance is often surprisingly high. The headlines used to justify the adoption of automation often focus on comparing the purchase cost, or lease cost, with the cost of labor that is being replaced. In reality, the need for another kind of labor is created, one with an engineering skill-set, rather than an operational one. Machines need programming and careful planning because the flexibility of automated machines can be limited. Then there are maintenance costs, the painful part of which is the time the whole line is down whenever any of the machines in the line need specialist attention for routine maintenance such as calibration, check and adjustment, lubrication, etc. More painful still are the unplanned events, where a breakdown of a certain part of the machine stops operation.
The third and most painful element of all is the hidden growth in the risk of defects as machine parts wear. Even if errors are not detected during assembly, a combination of elements can cause quality issues which cannot be easily tracked to a single origin or cause. Ensuring each automated process is well-maintained will result in smooth operation without loss of productivity or quality. But even when knowing this, maintenance is often regarded as something that could be compromised in a resource- and cost-constrained shop floor because there is usually no immediate and apparent consequence. As a result, it is a key aspect of good production management.
Leading operations adopt a preventative maintenance regime whereby, after a certain amount of time, certain maintenance jobs are undertaken to provide assurance of continued performance and good quality. The issue with this approach is it is unlikely the point at which the maintenance was performed was the point at which it was needed because of the simplistic nature of maintenance planning.
Example: servicing a car. Many car companies used to insist on annual service, although some cars may be driven very few miles in a year, and others may be driven almost constantly. So companies introduced mileage as a guiding factor for service, for example, with maintenance every year or every 20,000 miles, whichever comes first. Some aspects of service are time-based, such as the use-by dates of some fluids like brake fluid, whereas some items are use-related and depend more on the distance traveled, such as brake pads. Even this is not accurate because brake pads wear much differently when driving around town compared to driving for long periods on the highway. The style of driving is also significant. Makers of high-end models have gone a long way toward measuring and predicting the different types of maintenance required. All components exist within the physical car, and they are all connected to the car’s computerized infrastructure. Exact wear on most key parts can be calculated. The detail would be rather overwhelming for the average motorist, so we only see the simple service instructions and not the calculations that go on behind it.
The Internet of Manufacturing will bring the opportunity to do the same performance and work-related maintenance calculations in PCB assembly factories. Many individual components of an SMT operation can be classified into different categories for maintenance purposes based on the amount of work that can be measured or at least estimated.
The actual workload of all significant items within an automated process can be measured according to the amount of work done, as well as consideration for process performance. The extension of the calculation is to forecast the workload using schedule information in the short-term and an estimate of maximum work-assignment over longer periods. In this way, the exact date and time for each detailed maintenance requirement need can be assessed so that each task is done no sooner than required, which would be a waste of line time, engineering time and spare parts. And, it certainly would not be done too late, which would risk the quality and productivity of the machine if the wear became significant or there were a breakdown event.
The information for these calculations comes from the ability to model the machine’s key components’ operation and understanding the mean time between failure (MTBF) for each. Every component part can be uniquely identified so that, in the case of feeders, work can be accumulated correctly no matter on which machine the feeder is used. The performance and maintenance record is associated with a specific part, which can be of interest when comparing different suppliers or types of nozzles, for example. The information needed by the Lean maintenance computerization consists of the machine program, which defines the amount of operational cycles of each type, in addition to real-time data from the machine during run-time, so the exact number of cycles of each type can be accumulated.
A standardized “Internet of Manufacturing” communication capability is necessary to enable this computerized maintenance process because information from many different types of machines and automated processes on the shop-floor will need to be monitored and managed together to provide a complete Lean maintenance solution for the factory.
As information is gathered and processed, there are two key ways in which it can be used. First, when maintenance is performed, it can be intelligently planned. Linking maintenance information to a finite production plan schedule can find the times when the maintenance jobs would have little or no effect on planned production. Instead of a “major service” that may stop the machine for hours, the whole maintenance regime can be split into smaller components executed during the period when there is a line changeover, so that many of the maintenance jobs do not contribute to line downtime. This helps increase the availability and productivity of the machine.
Another use of maintenance computerization is the ability to predict when critical spare parts are needed. Some less frequently needed machine spare parts also tend to be the most expensive. However, the risk of not having a key spare part onsite may mean that, in the event of a breakdown, a whole production line can be down for many days or even weeks as said part is obtained. On the other hand, having a stock of every conceivable spare part onsite represents a significant investment when many of the spare parts may never be needed.
Knowing the expected life of key items and being able to measure the work done means the time for replacement can be accurately predicted. The replacement parts can be ordered in good time with significantly less need to keep spare parts in the factory. Of course, unexpected failures can occur that may need to be catered for. The expected life of a part needs to include real-world performance and reliability aspects, so the information can be built into the logic of the modeling of the process by the maintenance system. Key wastes are removed: the waste of investment in spare parts earlier than needed and the space for the storage of the spare parts.
One of the key elements that makes Lean maintenance possible is the availability of live and accurate information from the machines, in addition to engineering information about products built, most likely in the form of machine programs, and the ongoing production plan specific to each machine. A model of the operation of the machine is required to define the working parts of the machine that may require maintenance and to then translate the machine instructions into accumulation of work. Individual spare parts will have their own profile definitions that may differ from each other where multiple vendors are possible. The whole Lean maintenance calculation uses this working information to create a pull signal, which not only determines the best time for maintenance work but also applies the pull for spare parts.
Adoption of the Internet of Manufacturing to deliver real-time engineering and planning information to the Lean maintenance computerization is essential. This is another driver of a real-world standardized shop-floor communications standard. A practical and achievable Lean maintenance regime should be high on the agenda for any forward-looking company this coming year.
firstname.lastname@example.org. His column runs bimonthly.is marketing development manager, Mentor Graphics (mentor.com);