Building Management System (BMS) in Guwahati.

The first choice is scope depth, and it runs from monitoring-only through supervisory control to full direct digital control. Monitoring reads what the plant is doing and raises alarms. Supervisory control schedules and adjusts equipment that keeps its own local controls. Direct digital control replaces those local controls entirely. Each step upward buys finer control and adds a dependency: the building can no longer run well without the BMS, and the team who operate it become part of the design.

The second choice is integration breadth — whether the system supervises HVAC alone or extends across energy metering, pumps, lighting schedules, water systems, lifts and the critical-alarm estate. Breadth is where a BMS earns its keep, because most of the value is in correlation: a chiller working harder than the outside conditions justify is invisible on a chiller screen and obvious next to a metering trend.

The third choice is the one that decides whether the system survives its second year: who operates it. A BMS specified for a facilities team that does not exist becomes an expensive alarm annunciator that nobody acknowledges. The honest question at design stage is how many people will sit in front of this, how often, and what they are expected to do — and then the interface, the alarm priorities and the reporting are built for those people rather than for a demonstration.

Interoperability is the standards question that matters. Open building protocols exist so that plant from different manufacturers can be supervised from one place, and the practical test is not whether a vendor claims openness but whether the specific points you need are exposed on the specific equipment you are buying. A chiller may speak an open protocol and still withhold the compressor-stage data the energy analysis depends on.

That makes the points list the governing document of the entire discipline. It enumerates every value the system reads or writes, its type, its units, its source device and whether it is monitored, controlled, trended or alarmed. Written properly at tender stage, it makes competing bids comparable and it makes commissioning testable. Written afterwards, it becomes a record of whatever the installer happened to connect.

Metering deserves its own line in that document. Energy management is only as good as the meter tree beneath it, and a building with one incoming meter can report a total and nothing else. Sub-metering by tenancy, by plant group and by service is what turns a BMS from a monitoring screen into a management tool — and it is an electrical-package decision, taken long before the BMS contractor is appointed.

Point ownership is the coordination item that causes the most site disputes. For every point on the list, somebody must supply the sensor, somebody must supply the field wiring, somebody must supply the interface card, and somebody must terminate it. When the mechanical, electrical and BMS tenders each assume one of the others is doing it, valves arrive without actuators and chillers arrive without the communication card that the integration assumed.

The second item is control-panel and plant-room space. BMS field panels need locations, power, containment routes and a network path back to the head end, and those are shaft and plant-room drawing decisions rather than afterthoughts. In a retrofit, the plant room is already full and the panel location becomes a genuine constraint on what the system can economically supervise.

The third is the network. A BMS carrying live plant data over a building network shares infrastructure with everything else in the building, and the segregation, addressing and access arrangement for it has to be agreed with whoever owns that network. Deciding this after both systems are installed produces either an insecure flat network or an integration that quietly does not work.

Humidity is the defining local variable for control strategy. A control loop tuned to temperature alone will hold a comfortable dry-bulb reading and let latent load run away, and occupants experience that as a building that feels damp while the display insists everything is correct. Guwahati's humidity profile makes dehumidification an explicit design objective rather than a side effect of cooling, and a BMS that cannot see humidity cannot manage it.

Power interruptions produce the second characteristic failure: controllers that restart into an undefined state. Without a defined restart sequence, plant comes back in the wrong order, setpoints revert to defaults, and schedules that were adjusted locally are silently lost. The restart behaviour of every controller is a specification item here in a way it is not in a city with a stable supply.

Monsoon condensation attacks field devices that were selected for a drier assumption — duct sensors, outdoor-air sensors and anything in an unconditioned riser. Enclosure ratings and sensor placement need to reflect the local envelope, and the cost of getting it wrong is a trend line that quietly stops being true while the system continues to display it.

The most common failure of all is organisational rather than technical: alarms that nobody triages. A system that raises four hundred alarms a day trains its operators to dismiss all of them, including the one that mattered. Alarm rationalisation — priorities, suppression during known transitions, and a defined action per alarm class — is part of the engineering, not a tuning exercise for later.

Point-to-point verification of the whole list — every point read at the device and confirmed at the head end, with the record signed off line by line. A points list marked complete without that record is an assertion rather than a test.

Each control sequence demonstrated against the narrative, including the failure cases: what happens on sensor failure, on communication loss, and on restart after a supply interruption. Trends verified as actually logging and actually retaining. Alarms tested to confirm they raise, route and clear, and reviewed to confirm the volume is one a human can act on.

Handover carries as-built points list and graphics, the controller programming files, the control narrative, trend and alarm configuration, licence documentation and a signed training record. Then a seasonal review, because a building commissioned in one season has not yet been asked to hold conditions in the other — and in this climate the monsoon is the test the winter commissioning never ran.