WORLD PULP&PAPER 35 Edge computing is defined as solutions that facilitate data processing near the source of data generation, usually on site usage and environment has the biggest impact on their reliability: • Drives are typically specified to operate in up to 40°C ambient at full load. Even as they work at high temperatures, their expected lifetime and reliability will be heavily reduced in such conditions. Optimal ambient conditions are low humidity, and maximum 25°C ambient. Monitoring of environmental conditions based on available drive signals helps to maintain electrical room temperatures at optimal levels. • Drives are typically air cooled with fans, both externally and internally. The lifetime of a cooling fan is typically 3 to 10 years. Continuously monitoring the cooling fans helps to reduce the risk and helps to plan the right timing of cooling fan replacements. • Drives’ internal components (such as semiconductors and capacitors) impact their life cycle. Certain conditions such as cyclic loads can significantly reduce these components’ (and therefore the drives’) life cycle. Monitoring algorithms are used to identify the conditions that can shorten the drives lifespan. Improvements in control or conditions can then be applied in advance to extend the expected drives lifespan. Communication The local communication to equipment is based on industry-standard Ethernet protocols (such as the IEC 61850 standard regulating the communication of the electrical power grid), which facilitates open systems, flexible architectures and ensures interoperable systems for a future- proof investment. Today’s protection relays are communicating over the IEC 61850 protocol, which enables efficient data collection. In addition to control and monitoring signals, maintenance related signals are also transmitted using Modbus TCP – an Ethernet based protocol used to connect breakers and variable speed drives. (Figure 3 shows the communication architecture). Edge Computing Edge computing is defined as solutions that facilitate data processing near the source of data generation, usually on site. Equipment normally communicates data directly to edge servers. Edge data storage and processing power needs to be dimensioned for the right balance of data resolution and intermediate data storage. High-resolution data is used for trouble-shooting purposes and can be stored for a few months. Averaged or compacted data becomes useful after months or years of operation. In this particular case, an average of 40 signals are collected from each electrical device every second: • Currents and voltages • Power • Speed and torque • Internal controller signals A sampling interval of minute, hour or day is applied for slow changing signals: • Temperature • Service counters • Wear indicators Figure 3. Metsä Fibre’s condition monitoring architecture for its electrical system.