Insulation Monitoring Devices (IMD) for IT Earthing Systems

What is an insulation monitoring device (IMD)? An IMD continuously measures the insulation resistance between live conductors and earth in an unearthed IT system and issues an alarm once that resistance drops below an adjustable threshold, per IEC 61557-8. A single earth fault in an IT system doesn't trip a breaker or create a shock hazard by itself, so without continuous monitoring that fault can sit undetected until a second fault on a different phase turns it into a dangerous phase-to-phase path through earth. This article covers the measuring principle, how alarm thresholds and fault location are set, where IT systems and IMDs are required — hospitals, process plants, ships — and how the IMD's output relay fits into the rest of the panel.

Why an IT System Needs Continuous Insulation Monitoring

IEC 60364 recognizes three earthing systems: TN, TT, and IT. In a TN or TT system, an earth fault produces enough fault current to open an overcurrent device or an RCD within milliseconds. In an IT system, the source has no direct connection to earth, or connects through a high impedance, so a single fault-to-earth is current-limited by that impedance and by the network's own distributed capacitance. The fault current stays too low to trip anything and too low to create a dangerous touch voltage on its own.

That's the operational upside: a process line, an operating theatre, or a ship doesn't lose power because one cable nicked its insulation against a conduit. It's also the risk. Nothing forces that fault to get fixed. Leave it long enough and a second, independent fault on a different phase closes a real phase-to-phase loop through earth, full fault current, with nothing left to limit it. The IMD exists to close the gap between those two events: find the first fault while the system is still running, and get it cleared before the second one shows up.

How an IMD Measures Insulation Resistance

An IMD doesn't wait for a scheduled megger test. It injects a small measuring signal, DC or low-frequency AC depending on the model, between the network and earth while the installation stays live and loaded, and calculates the resulting insulation resistance from that injected voltage and the current it drives through any fault or leakage path.

A real installation always carries some resistance and capacitance to earth from cable insulation, EMI filters, and surge suppressors, even with zero faults present. Better IMDs separate that background from an actual fault (the AMP measuring method is the common approach) so a long cable run or a VFD's input filter doesn't trigger a false alarm on a healthy network. Set the threshold without accounting for that background, and the first thing that happens after commissioning is a nuisance alarm on a system with no real fault.

Alarm Threshold, Response Time and Fault Location

The alarm threshold is adjustable in kilohms, and where it gets set depends on the network, not on a factory default. A short, lightly loaded IT network with little cable capacitance can run a high threshold and stay quiet; a long run feeding several VFDs sits at a lower healthy baseline resistance from day one, and a threshold set without accounting for that baseline alarms constantly on a system with no fault at all. Commission the network, read the healthy baseline, then set the threshold below it with margin.

Response time works the same way. A larger network with more distributed capacitance needs a longer, filtered measurement to settle on a stable reading; push the response time too fast on a large network and the reading swings enough to alarm on its own noise. The base IMD reports only that a fault exists somewhere on the network. Locating which of several outgoing feeders holds it takes an add-on fault-location module, current transducers on each feeder, and coordination with the IMD's test signal so the whole system reads one consistent fault instead of several conflicting ones.

IMD in Medical IT Systems

Group 2 medical locations, operating rooms and other rooms where interrupting power mid-procedure is not an option, run on a medical IT system for exactly this reason. A single earth fault on connected equipment doesn't cut power to a ventilator or a surgical light; the IMD alarms staff and maintenance, and the equipment keeps running until someone finds and clears the fault on a schedule that doesn't involve a patient on the table. This is the specific case IEC 60364-7-710 addresses for hospital electrical installations.

What we see in the field: the corridor alarm panel outside an operating suite is often the only place staff ever see the IMD's output, and it's easy for that alarm to sit ignored for weeks if nobody owns the response process. The relay is only half the system. The remote display, the escalation path, and someone assigned to act on it matter just as much.

IMD in Industrial and Marine IT Systems

Process plants with continuous operations follow the same logic as a hospital: an unplanned trip on a first earth fault costs more than tolerating it until the next maintenance window, especially on VFD-heavy lines where nuisance earth faults from cable capacitance and filter leakage are common. Ships apply IT distribution for a different reason. The hull is the return path risk, so isolating the network from the hull and monitoring against it as the earth reference keeps a single insulation fault from turning the hull itself live. Dedicated insulation-monitoring ranges, led by manufacturers like Bender and ABB, cover both cases with the measuring principle described above, scaled to network size and to how many feeders need separate fault location.

Selecting and Wiring an IMD: Output Relay and Panel Integration

The IMD's output is a change-over relay, normally-energized and fail-safe like the other monitoring relay types covered in types of monitoring relays: it de-energizes on alarm or on loss of its own supply. That output drives a remote alarm, a BMS or PLC input, or an audible/visual annunciator. It should never drive the main breaker directly; tripping the supply on the first fault defeats the entire reason for choosing an IT system in the first place. Anyone new to how a monitoring relay's threshold-and-output logic works in general can start with how a monitoring relay works before specifying an IMD.

Selection comes down to matching the device to the network: rated system voltage and frequency range, DC/AC signal compatibility on networks with VFDs or rectifier loads, an alarm threshold range wide enough to sit below the network's real baseline with margin, and remote display or communication if the panel isn't attended around the clock. Plants running several separate IT sub-networks need either separate IMDs or a coordinated multi-channel unit; two IMDs testing the same network at once read each other's test signal as noise. Where a second fault does trip something downstream, that's usually a standard breaker or a contactor acting on a protection device, not the IMD itself. For the wider category this device belongs to, see the monitoring relay engineering guide, and for the stocked range of monitoring and control relays IT-system electricians typically keep on hand, check the collection directly.

Conclusion

An IMD is what makes an IT system's core advantage, staying online through a first earth fault, actually safe instead of just convenient. Set the alarm threshold and response time from the network's real baseline, treat every alarm as a fault to clear rather than noise to silence, and keep the output relay driving an alarm rather than a trip. That's the whole job: catch the first fault before a second one turns it into an incident.