How is the equipment asset management system built? Status monitoring and WPC
RFID Knowledge

How is the equipment asset management system built? Status monitoring and WPC

PublishedShanghai Zejie Smart Technology Co., Ltd.

A night shift, so I can re-examine the equipment management.

Last winter at a precision machining shop, the spindle bearing in an imported five-axis machining center suddenly seized. The maintenance technician took 40 minutes to get from the warehouse to the machine, and the replacement took two days to arrive. The entire line was down, deliveries slipped, and losses approached CNY 200,000.

The retrospect: the bearing had vibration abnormalities a dozen days before the failure, but no one was reminded; the protection plan was still on a monthly fixed-cycle schedule, completely disconnected from the actual state of the equipment. This thing made me realize that the whole life cycle of equipment asset management is not an advanced idea, but a real connection between "state monitoring" and "vivivial-motion" -- monitoring data automatically triggers worksheets, and the results of this worksheet are a health file on anti-nursing devices. Shanghai has validated a relatively clear route after Shanghai technology landed in the system at various factories.

First, answer the central question: What is this system doing?

The whole life cycle of equipment asset management, essentially, is to get all the actions of a device from "purchase entry" to "disposal disposal," which are data available, rules and results. Status monitoring is eyes, WIPs is hands and feet -- eyes see problems, hands move immediately; if only a bunch of vibrating curves were collected, temperature trends were not implemented, monitoring was worthless.

The common error is that you spend a lot of money on sensors and collection gateways, and the data goes to the platform without anyone looking at it, and the platform becomes "data grave." To avoid this, the four questions that are "what's monitored from the start, what's triggered, who's assigned, how many hours to close the loop," are clear about, then select hardware and software.

Step One: Create the Equipment Register and Condition Records

And many factories have sensors on their way up, but one thing is ignored -- the basic information about the equipment is not right. What's the name of the device, which line it belongs to, who's in charge of routine inspections, when was last overhaul, how many hours -- no data, monitoring data coming and never compared to anyone, compared with history?

The first step is to take two or three weeks of the branding, BOM, and purchase records, which are integrated into a system of accounts for all critical equipment in the workshop. It is best to follow the three-tier structure of "Equipment-Subsystem-Key Components" (System) and organize it by a number control bed that can be divided into main electrical axes, hydraulic systems, knife stores, cooling cycle subsystems, and then removes the bearings, electric machines, encoders under the main axis.

The second step is to create a health profile for each key device, and then to put up the parameters of the exit plant, all maintenance records, cause of failure, and an alternative list. This step looks cumbersome, but when you do predictive maintenance, without this bottom map, the results of the model are not going to be able to land.

Step two: state monitoring should be graded, not one cut.

Not all equipment requires vibration sensors and online monitoring. The plant has hundreds of devices, which are costly to monitor online on a full-time basis, and not very profitable. It is more practical to grade them by "failure effects x frequency" and divide them into A, B, C categories.

Category A equipment is critical to the long-cycle maintenance of a single-stage, high-value, single-axis plant that directly affects the entire production line. The type of device suggests online monitoring, vibration, temperature, several current signals simultaneously, sampling as high as possible, and an axis component usually takes more than 1 kHz to catch early impact.

This type of equipment is used for periodic portable inspections, which are carried out on a one-to-two-week basis with a point-checker to monitor the data entry platform for trend analysis.

C-type equipment is low value, redundant, broken down by direct replacements like regular lighting, fans, simple work. Such devices are routine visual inspection, and there's no need to budget for sensors. When the hierarchy is clear, budgets and energy can focus on blades.

Step three: How did WIP move from "planned overhaul" to "demand response"?

The traditional protection model is time-bound, no matter how the equipment is in state, and it's scheduled maintenance, which has the advantage of not leaving out big items, and the disadvantage is that it's obviously doing "over-maintenance" -- often with the axes being removed before they reach their lifetime, both wasting spare parts and increasing the risk of failure from the dismantling itself.

The real place where state monitoring works is to make the "time-based" transition slowly to "state-based." The approach is to set warning and alarm thresholds for each key point. Early warning values represent a deterioration, but not yet at the risk line, and systems automatically alert; alarms are about to be problematic, and the system automatically sends a job order to its counterpart.

A more typical example. The gear tank vibration acceleration sensor in a wind field, if the range of a frequency band is increased by more than 15 per cent over three weeks in a row, is sent to the Chief of Transport; if the single-day instant value exceeds the alarm threshold set out in ISO 10816, it triggers an emergency work order and requires an inspection within four hours. This is how monitoring and protection moves.

And after the singles, there's another key action -- the results have to be written back. The maintenance staff went to the scene and did something, changed parts, decided what the failure was, and this information had to be filled in, and automatically placed in a health file of the device. There was more data in the files, and future prediction models were getting better and better.

Step Four: Use Data to Inform Purchasing and Disposal Decisions

Many plant equipment management systems only deal with the "in use" section, where the pre-procurement option is completely disconnected from the end-of-life disposal. One device is either durable or one brand has the most difficult spare parts to buy, and the cost of maintenance is much more than a data support for the procurement department.

The last link to the life cycle management is to get through the whole cycle of equipment from plant entry to end-of-life. Each device has an "id ID" (usually a unique asset code), coded in connection with the procurement contract, preservation records, data consumption, number of failures, cumulative stoppages. These data can support several very practical decisions:

Several common pits and responses to land

The first is a nice sensor, but the communication protocol is not. A plant often has both old PLCs that support Modbus, new equipment that supports OPC UA, and sensors that can only be modelled on 4-20mA signals. The platform side plans ahead to collect gateways, convert different protocols into uniform formats, or data is all uncoded.

The second pit is the protection team's resistance system. The most offensive part of the workers is "one more software and one more process," so that the worksheet design is extremely relevant to the field -- mobile phone scans can read the device, photo-shots can upload a malfunction description, and then a ticking instead of filling in a dozen fields after repair has been completed. These details determine whether the system really works.

The third is that no one is responsible for data governance. After three months, the platform starts to have problems with "sensor offline without warning" and "decoding" and "remaining records without a backlash," and then goes back to unsubscribed status. It is recommended that a data liability (even in part-time positions) be appointed from the outset, and that data integrity and consistency be checked monthly.

I'll give you a couple of suggestions for the factory that's ready to do it.

Do not attempt to digitize the entire plant at once. First, choose a line or a workshop as a pilot, for three to six months, to run the process and bring out the team before copying it to other regions.

The choice of a model is based on "openness" rather than a much smaller set of functions. Whether the platform supports access to new equipment, export raw data and connects with ERP or MES is relevant for the next three years.

If there is no dedicated industrial Internet team in the house, you can consider working with the implementers who have experience of plant landings. When doing such projects, Shanghai-kitsu usually spends a week or two to talk with the maintenance team leader and the technologist, and ask "Who needs to read what data in what circumstances," and then go to system logic -- that's not the order that goes upside down, technology works for processes, not the other way around.

The whole life cycle management of equipment assets, which sounds like IT projects, is essentially a change management project. It's not normal to take the state monitoring and WUC, to make a device "see" from the first day of its introduction, until the failure really comes.