6S Teknoloji · 12.11.2026 · 6 min read
In a hospital, a power outage is not merely a comfort problem, as it is in most buildings. An anaesthesia machine in an operating theatre, a ventilator in intensive care or the cold chain in a laboratory depends directly on continuity of supply. For that reason a hospital electrical installation is not designed around a single source but as a chain of links arranged in series, each one backing up the next. This article covers the architecture of that chain, redundancy concepts, medical location groups, the isolated IT system, and what should be recorded during commissioning and periodic testing. Generator fuel and run-hour tracking and UPS battery monitoring are covered in separate articles, so they are not repeated in detail here.
In a typical hospital, energy travels from the distribution grid to the medium-voltage switchgear and from there to one or more transformers. The main low-voltage switchboard at the transformer output separates the busbar feeding general loads from the busbar feeding safety services. The safety busbar is connected to both the grid and the generator through an automatic transfer switch (ATS). Behind the ATS sit UPS systems for loads that tolerate no interruption at all. The final link consists of the medical panels and the IT panels.
Redundancy is described in symbolic terms. N is the number of units needed to carry the load. N+1 means one spare unit is added to that number; if any unit fails or is taken out for maintenance, the rest continue to carry the load. 2N means a fully independent second copy of the whole system: two separate UPS groups, two separate distribution paths and dual-corded loads. Expressions such as 2N+1 describe a combination of the two approaches.
What matters is that redundancy exists not only in the number of units but also along the path the energy takes. In a design where two UPS units feed the same output panel, a fault in that panel or on a single busbar removes all of the redundancy. A design review should therefore ask, link by link, "if this element fails, which load drops?", and the answer should be visible on the single-line diagram. Load sharing between redundant units should also be monitored: a system built as N+1 may have quietly become N as loads were added over the years.
IEC 60364-7-710 sets out particular requirements for electrical installations in medical locations and groups those locations by the nature of contact between patient and equipment. In general terms:
The standard also classifies safety supplies by how quickly the source must take over. The shortest changeover classes cannot in practice be met by a generator and call for a UPS. For exact times and for which load belongs to which class, refer to the current edition of the standard and to national regulations, and state this mapping clearly, load by load, in the design documentation.
In Group 2 locations, circuits feeding life-support and surgical equipment are typically supplied through a medical isolated IT system. In this arrangement the secondary side of an isolation transformer is not earthed, so a first insulation fault does not trip the circuit and supply continues while surgery is under way. However, the first fault effectively turns the system into an earthed one, so it must be detected before a second fault develops.
That task falls to the insulation monitoring device (IMD). The IMD continuously measures the insulation resistance of the system to earth and, when the threshold is crossed, gives an audible and visual warning at a place where medical staff can see it. In practice, overload and overtemperature of the isolation transformer are shown on the same alarm panel. The operational weak point is when these warnings never leave the room panel: if the technical team on the night shift learns about an IMD alarm only when a nurse phones them, the monitoring chain is incomplete.
What proves that the critical power chain really works is not the design but the record. The following tests should be documented at commissioning and periodically afterwards:
Every record should include the date, the person who performed the test, the test duration, the load condition, any deviation observed and the action taken. Test intervals follow the manufacturer's instructions and the regulations in force. Paper forms work for audits but do not show trends; to see how the same test changes over the years, the measurements need to be stored as numerical data.
| Point | Quantity to Monitor | Why |
|---|---|---|
| Grid incomer | Phase voltages, frequency, phase sequence | Timestamps for outages and voltage disturbances |
| Transformer | Load, temperature, power factor | Knowing the remaining capacity when one transformer is out |
| ATS | Position (grid or generator), moment of transfer | Evidence that transfer and retransfer took place |
| Generator | Running status, run-time counter, fuel level, alarm bits | Readiness and matching against test records |
| UPS | Input and output values, load, battery status, bypass | Making unprotected operating periods visible |
| Medical IT panel | IMD warning output, isolation transformer load and temperature | Detecting a first insulation fault without delay |
| IT and server panels | Panel and PDU loads | Whether redundant paths are still truly redundant |
In 6S Teknoloji the field agent on site reads devices over Modbus TCP/RTU, M-Bus and BACnet/IP. From DSE generator control panels it collects fuel level, engine run time, mains-side and generator-side electrical values and alarm bits. On the UPS side, BACnet/IP reading uses a point map defined for the Vertiv Liebert EXL S1; there is no automatic discovery, and points are defined per model. As a result, the moment the grid dropped, the moment the generator took the load and the state of the UPS can be examined side by side on the same timeline.
If the connection is lost, readings are held in a local queue on site for 30 days and are not deleted until the server acknowledges them (ACK); on the server, the raw measurement series is kept for 13 months. The alarm engine sends e-mail notifications when defined conditions occur. In practice, this means the team can look back at how a test went using the stored measurements, and receives warnings by e-mail rather than only at the panel.
Related Content
Explore the system with your own data in a demo session.