Most factories don’t have sufficient access to data that is related to their real-time production. The dilemma this entails can be illustrated in the following example: A factory manager oversees 50 CNC machines and has set a key performance index (KPI) of 200 manufactured pieces per week. The line manager instructs the operators to produce the required quantity and record the production data manually. The recorded data is passed on daily to an assistant who puts it into an enterprise resource planning (ERP) system manually. After one week, the line manager learns to his surprise that only 80 pieces have been produced— far below the set KPI. This raises the question of how managers can readily access real-time production data so that they can take faster action and not wait one week, for example, to restore production.
A lackluster production usually echoes back to any of the three elements of OEE. From the above example, it can be inferred that any machine that shuts down unexpectedly will go undetected for some time due the absence of a proper machine condition monitoring system. Therefore, the availability rate is low due to a long downtime. The challenge is to provide a reliable communication network to capture essential production data and reduce the downtime of the machine.
Solutions to improve the availability rate of machines
The crux is to interpret the working status of the machines at all times during production to reduce downtime and improve network availability in order to ensure maximum uptime.
For many, it is perplexing how real-time production data can prevent unexpected downtime. The simple answer lies in predictive maintenance—for which you need real-time production data. During production, machines generate different types of data, such as machine vibration, motor current, tool level, coolant level, and many more. Based on this data, machine maintenance engineers schedule maintenance tasks (predictive maintenance) to avoid any unexpected machine downtime. However, the data presents itself in different forms.
One is streaming data, which is transmitted in large volumes and requires preprocessing before it is sent to a back-end system. The other is status data, which is transmitted in small volumes and via a transparent method without any preprocessing. Thus, the system has to use different methods to collect both status and streaming data. For status data, the best way is to use transparent data collection. For streaming data, the best method is to use front-end data processing to reduce the amount of data sent so that only valuable data is sent to the backend system. Downsizing is necessary because there will be too much streaming data to transfer all the raw data to a back-end system. In terms of front-end processing, a device with a programmable platform that is designed to handle different formats of streaming data is the best fit for the job. Sometimes, a math function is also required in the programming platform to cope with streaming data. If users only need to process simple status data, then an I/O gateway or protocol gateway offers the best solution.