The Level 2 method for power and cooling: logs, recordings, trends and owners

Collect and align the records of the scanner, the UPS, the chiller, a power analyser and the room loggers on one timeline, separate the first cause from its consequences, and hand the proven fault to the owner of the failed link with evidence that settles the question.

Where Level 1 stops and Level 2 starts

Level 1 gave you the map: the power path from the intake to the scanner, the heat path from the gantry or the cold head compressor to the outdoor condenser, and a habit of looking outside the scanner before opening it. That habit finds the weak link in many calls. The calls that come to Level 2 are the ones where the habit is not enough: the trip happens twice a week and never while you watch, two links fail together, or three parties each insist that their part is fine.

Level 2 work has three steps. First, prove which link failed, with records rather than opinions. Second, follow the fault inside that link to the block or the part, as far as your training and authorisation allow. Third, hand what remains to the owner of that link in a form that leaves nothing to argue about. This lesson is about the first and the third step; the rest of the course is about the second, link by link.

Five sources of evidence and their blind spots

The scanner's error log is the record you know best, and it is the least reliable about the site. It reports what the scanner felt: a power fail message, a water temperature warning, a compressor stop, an inverter fault in its own generator. It rarely knows why. On many systems the first site-related entry is followed by a cascade of secondary errors as subsystems lose power or overheat one after another; the loudest message at the end of the cascade is often the least useful.

The UPS event log records changes of mode: to battery, to static bypass, back to inverter, overloads, input voltage or frequency out of tolerance, battery test results, fan and temperature alarms. Its weaknesses are a limited buffer that overwrites the oldest events, a clock that nobody set, and a difference between what the front panel stores and what a network card or monitoring software stores. Some warnings light an indicator without writing anything to the log at all. Save the log in every form the UPS offers before anyone resets an alarm.

The chiller controller keeps an alarm history: high and low pressure, flow, antifreeze or low water temperature, compressor overload, fan faults, sensor faults. Some controllers keep only active alarms and lose the history when power is removed; others separate real trips from pre-alarms that only warn. Learn which kind is on your site, and photograph the alarm screen before the facility resets it.

A power quality analyser gives what no device log gives: the voltage and current waveforms at the scanner input, with dips, swells, transients, harmonics and imbalance stamped with their time. It sees only what happens where it is connected and while it records, so placement and duration decide its value. Room and water loggers give temperature and humidity trends, often the only witness to what happened at night.

Add a sixth source that is not a device: people. The radiographer who filmed a blinking UPS indicator on a phone, the guard who heard the chiller stop, the facility electrician who switched a feeder last Tuesday. Ask early, write it down with the time, and treat it as a lead to confirm with a record.

One timeline from many clocks

Records only become evidence when they share a time base. Choose one reference clock, such as a mobile network time or the hospital's time server, and note the offset of every device against it on the day you collect the logs: scanner, UPS, chiller, analyser, loggers. Offsets of minutes are common; after a long outage a device without a working backup battery may be years off. Write the offsets on the timeline itself, so that the next reader does not have to repeat your work.

Then build a table: time, source, event, value. Put the events of all devices in one column of time. Mark the resolution of each source, because a UPS that records to the second and a chiller that records to the minute cannot prove which came first when they are thirty seconds apart. When the order matters and the resolution is too coarse, the analyser or a dedicated logger becomes the referee.

Read the table with three rules. A cause comes before its effect, after allowing for the clock uncertainty. A real cause repeats: one coincidence proves little, three events with the same order begin to prove something. And a cause usually has a dose: the trip is more frequent on hot afternoons, during heavy sequences or when a large motor starts. When all three point the same way, you have a hypothesis worth a test; when they disagree, keep recording.

Trends beat snapshots

A single reading taken when the fault is absent tells you only that the fault is absent. Infrastructure faults usually develop over days or seasons, and trends show them long before a trip. The useful trends are few: supply and return temperatures and delta T on each loop, loop pressures, chiller condensing pressure against outdoor temperature, compressor running time, UPS load and its peaks, battery block voltages and internal resistance, room temperature and humidity.

Compare like with like. A condensing pressure means nothing without the outdoor temperature at the same moment; a UPS load means nothing without knowing whether the scanner was idle or running its heaviest protocol. Note the conditions with every reading, and compare with readings taken under similar conditions.

Learn the shape of the change. A step change, where a reading moves at once and then stays, points at an event: a valve turned, a setting changed, a pump replaced, a phase reconnected. Look for what happened on that day. A slow drift points at wear or fouling: a strainer filling, a condenser coil clogging, a pump impeller wearing, a battery string ageing. A drift that speeds up is a failure approaching, and the time to act is before it arrives.

The interfaces between building and scanner

Between the infrastructure and the scanner there are signals as well as power and water. A chiller sends a general alarm through a dry contact to a panel or to the scanner; a UPS sends status signals, and some imaging systems start a controlled shutdown when the UPS reports battery low; an emergency-off loop runs through several cabinets; flow and temperature interlocks stop a compressor or an amplifier. Each of these paths has relays, terminals, cables and sometimes a small power supply of its own.

When the two sides disagree, suspect the path between them. A chiller that runs normally while the scanner reports its alarm, or a scanner that shuts down for battery low while the UPS shows full batteries, often has a fault in the interface: a stuck or burnt relay, a loose terminal, a contact that was rewired during another job. Read the interface drawings in the installation documents, check the signal at each end with the systems in a safe state, and repair it within the scope agreed with its owner. Never bridge an interface signal to silence an alarm; a silenced alarm will not warn anyone when the real fault comes.

Owners, authorisation and the evidence pack

Level 1 asked you to know who owns each link. At Level 2 you also need to know what you may do on each one. As a rule, a scanner service engineer may read displays and logs, measure from safe test points with the right instruments, inspect, clean filters and coils where the owner allows it, and reset alarms by the documented procedure. Opening a refrigerant circuit, switching building switchgear, operating a UPS maintenance bypass or changing protection settings of a UPS or a chiller belong to people trained and authorised for that equipment. Entering a service menu with a password you were given informally is not authorisation.

When the evidence points outside your scope, write the evidence pack. Keep it short and factual: the symptom and its pattern, the timeline with offsets, the readings with their conditions, what you have ruled out and how, the link you believe failed and the reason, and the question you need the owner to answer. Attach photos, log files and analyser extracts. Send it in writing to the owner and copy the people who decide on the scanner's use.

A good pack turns a dispute into a job. The chiller contractor who receives 'primary supply temperature rose above the site limit eleven minutes before each compressor trip, three times this week, condensing pressure high for the outdoor temperature' does not argue; he brings coil cleaner.

• Save every log before any reset, in every format the device offers.
• Note clock offsets against one reference on the day of collection.
• Record conditions with every reading: load, protocol, outdoor and room temperature.
• Look for steps and drifts in trends, not only for limits crossed.
• Check interface signals at both ends before blaming either system.
• Stay inside your authorisation; send the rest as a written evidence pack.

On-site checklist

• Collect scanner, UPS, chiller and logger records before resets.
• Write a single timeline with sources, resolution and clock offsets.
• Ask staff what they saw and heard, with times.
• Trend delta T, pressures, UPS load, battery data and room climate.
• Check interface relays and contacts at both ends.
• Send a written evidence pack to the owner of the failed link.

Check yourself