Tracing a Fault Across a Shared System

Why this matters

Recognizing that several units share one cause is half the job. The other half is physically finding where in the shared distribution the fault sits, usually with no accurate drawings, in a building that has been altered by three contractors over thirty years. This is a craft: mapping an undocumented system, isolating segments, and proving pieces innocent one at a time until the fault has nowhere left to hide. Guess instead of trace and you either replace the wrong shared component, expensive and disruptive to every unit, or you shut the whole building down chasing a fault a bisection would have cornered in an hour.

Build the map before you touch anything

You cannot trace a system you have not drawn. Sketch it, even roughly: the source (the main, the meter, the mechanical room, the service feed), the branch points, and which units hang off which branch. You do not need it to scale, you need the topology, what connects to what and in what order. Fill it by walking the runs, not by trusting labels, and mark every point where you can isolate or measure. That sketch is the board you play the rest of the diagnosis on, and thirty minutes making it saves hours of random poking.

Isolate the shared from the local

The first cut is the most valuable one: prove whether the fault is in the shared part of the system or in a unit's own dedicated part. Measure at the boundary where the shared supply hands off to a single unit. If the supply arriving at that boundary is already wrong, the fault is upstream in the shared system and no amount of work inside the unit will fix it. If the supply at the boundary is normal and the fault only appears past it, the shared system is innocent for that unit and the problem is local. Make this cut early, because it decides whether you are working on one unit's equipment or on infrastructure that touches everyone.

Bisect to corner the fault

Once you know the fault is in the shared distribution, do not walk the whole run end to end. Cut it in half. Pick a midpoint you can measure or isolate, and test there. The fault is on one side of that midpoint; the other side is now cleared. Move to the middle of the bad half and cut again. Each test roughly halves the search, so a long run that would take many sequential checks collapses to a handful. Bisection is the difference between tracing a system in an afternoon and tracing it in a day.

• Measure where you can (pressure, flow, voltage, temperature at a branch point) to see which side of the cut carries the fault.
• Isolate where you cannot measure (close a valve, open a disconnect, cap a branch) to remove half the system and see if the symptom changes.

Use the units as instruments

The units hanging off the system are free sensors. A unit downstream of a suspected fault point should show the symptom; a unit upstream of it, or on a different branch, should not. If a unit that shares the branch has not complained, check it anyway, a milder version of the same symptom that has not yet reached the tenant's threshold to call is still a data point that confirms the branch. The distribution of who has the symptom and who does not, laid over your map, points straight at the common segment feeding exactly the affected units and no others.

Prove the segment innocent, do not assume it

The discipline that separates a trace from a guess is clearing each segment on evidence, not on hunch. A segment is innocent only when a measurement or an isolation test says so, not because it looks fine or because you would rather the fault be somewhere cheaper. Write down what each test cleared. The trap in shared-system work is the untested assumption, "the main is probably fine," that turns out to be exactly where the fault lived, after you have replaced two branches that were never the problem.

Flag the disruption before you cause it

A shared-system repair often means access to common areas or a shutdown that hits units which never complained. Tell whoever manages the building before you isolate a segment that will interrupt service to bystanders. Surprising three tenants with no water to corner one tenant's fault turns a good diagnosis into a complaint, so stage the disruptive cuts and warn the affected units first.

References

• Trade-standard practice for shared mechanical, electrical, and plumbing distribution tracing
• Local building code guidance on common-area and shared-system access
• See related: The Shared System: Multiple Tenants, One Cause; Two Units Share a System and Only One Complains (decision tree)