How do IAMS choose? Workworthiness and data acquisition requirements
What exactly does the Industrial Asset Management System (IAMS) do?
The core of the Industrial Assets Management System (EAM, Enterprise Administration Management) is to manage the entire life cycle of equipment: from procurement, accounting, inspection, maintenance, spare parts until their end of life, every step must be marked. A qualified system must cover at least the equipment desk accounts, the worksheet management (CMMS), the preventive maintenance (PM), the spare parts inventory and data analysis five modules. If you have more than 50 plant types, with annual maintenance costs above 3 per cent of the value of production, it is necessary to go on to a system that would otherwise be able to sustain a single line of damage by Excel and paperwork, but the problem-trace and maintenance cost accounting will become increasingly confusing.
Why is the first of the chosen type?
Many candidates look at the list of functions first, but the first step is to see how well they fit.
- High temperature environments (e.g. metallurgical, glass industries), work temperatures of sensors and gateways are supported -40°C to +85°C, with direct adverse effects on ordinary commercial products.
- For powerful vibration scenes (e.g., mines, pressurized workshops), the collection equipment requires an anti-earthquake design and it is recommended that technical-grade products conform to IEC 60068-2-6 standards be selected.
- Dust and damp (e.g., cement, textile workshops) with a protection level of at least IP65 and an environmental recommendation IP67 for outdoors or water vapour.
- For blast-proof areas (e.g., chemical, oil and gas) hardware with appropriate blast-resistant level certification (Ex d IIB T4 or above) must be selected.
If the vendor system only shows a beautiful chart in the office without providing any field cases from the corresponding industry, it is asked a question mark when selecting.
Hardware compatibility is easily underestimated.
The other is often ignored. Old PLC (programmable logic controller, small machine in the plant that controls the operation of equipment) may support only Modbus RRU or RS 485, while new equipment is more used as Profinet, EtherCAT or OPC UA. The system needs to draw up a list of field equipment before bidding, checking the support of the protocol on a case-by-case basis.
How much data is needed to collect it?
Data collection is not as good as it is, but rather enough to support decision-making.
Sampling frequency matches the device's characteristics.
Sampling slow signals such as temperature and liquid level once every 1–10 seconds is enough. Fast signals such as vibration and current harmonics need at least 1 kHz to capture characteristic fault frequencies. For example, motor bearing inner-race spalling typically produces characteristic frequencies of 500 Hz to 2 kHz. At a 100 Hz sampling rate, the fault signal will be lost in the noise.
Data dimensions should be overwritten by "Mechanic Plus State".
Light has insufficient operating parameters (electricity, voltage, flow), preferably with vibration, temperature, sound-like states. After multisource data integration, predictive maintenance models can increase by 20 to 40%. This is the conclusion repeatedly validated by several industry research institutions between 2019 and 2022.
Edge computing capacity determines real-time response
If the site requires a msasecond alarm stop, data must be processed at the edge gateway (locally calculated box on the side of the device) before uploading the cloud, otherwise the network delay will drag the response time to over hundreds of milliseconds. The suggested type allows the supplier to demonstrate local storage and recovery mechanisms in case of a grid breakup, which is a hard indicator for reliability.
The most easily stomped pits in the selection.
Just focus on the functional list, not the ability to implement.
The functions written by the vendor PPT are often optimal and are actually dependent on the engineering experience of the implementation team. Ask if the other implementers have any business presence experience, more important than the number of functions.
Neglect of data governance and standards
The basic criteria for the device code, dot naming and failure classification are not well established, and data into the system is a pile of random codes.
Underestimation of the cost of future transportation
The system is only on line, and the iterative model, point-up expansion, and staff training require continuous input. It is recommended that the scope of annual transport services and response time for suppliers be specified in procurement contracts to avoid problems of mutual inversion.
How do you verify that the system is on the line?
The acceptance and inspection cannot be based on functional runaways, but on three indicators:
- Data completeness rate: Effective data coverage of critical equipment should be above 95 per cent, with no less than 30 days for continuous collection.
- Worksheet closed loop rate: the full process closure rate should be above 90 per cent from failure report to service completion and acceptance of the archive.
- Forecast accuracy rate: The failure prediction model has an early warning accuracy of over 70 per cent to be useful, with less than 50 per cent largely in the form of a setup.
If all three indicators are met, the system will be truly landed or simply replaced by a more expensive spreadsheet.
Now, back to the selection: first we'll start with the process and the protocol environment, then we'll be able to determine the frequency of data collection, dimensions and real-time requirements, and finally look at the implementation capacity and the delivery system of suppliers. And this three steps, and the risks of the system selection will be reduced by half, and the rest is business negotiation and project management.

Ms. Lei
Mr. Pang