Repeat Conflict Flashes After Rain at a Rural Intersection
A signal on a rural 55 mph arterial has flashed 12 times in eight weeks, always during or within a day of rain. The MMU event log shows seven POWER FAIL events (some only 0.2 s), three CONFLICT events (phase 2 green with phase 4 green, 130 ms), and two WATCHDOG events. Maintenance has been resetting the MMU and returning the signal to service each time. The intersection uses span wire, a pole-mounted cabinet, and loop detectors whose lead-in cable shares a conduit with the 120 V signal conductors for the final 60 ft. One minor-street loop reads 0.6 megohms to ground when wet and more than 100 megohms when dry. All of the conflict events occurred between 5 and 7 a.m., when irrigation runs and dew is heavy.
1. What is the likely root cause, and what would you do before resetting the MMU again?
Read the worked answer
The pattern points at moisture, not at random electronics. Every event follows rain, the wet loop reading collapses to 0.6 megohms, and the timing matches dew and irrigation.
Short power fail events are suspicious. A genuine utility outage lasts far longer than 0.2 s. Sub-second power fail entries usually mean a loose neutral or service connection, a failing surge arrestor, or an MMU power supply on the edge. Check the service termination torque, the arrestor, and the MMU supply before treating these as utility outages.
The conflicts. A wet loop or a wet splice sharing a conduit with 120 V conductors can couple current into a load switch input or a signal circuit and produce a false indication, which the MMU correctly sees as a conflict. Inspect for water in the conduit and cabinet, check the door gasket and cabinet heater, megger every loop circuit wet and dry, and inspect all splices. Gel-filled or potted splices are required; dry-twist splices in a wet handhole are the usual failure.
Do not just reset. Pull and archive the full event log first; resetting destroys the evidence you need to identify the conflicting channels.
2. What immediate traffic control action is appropriate while the cause is investigated, and what governs that decision?
Read the worked answer
A signal in flash is a stop condition. Leaving a failed signal in yellow on the 55 mph arterial and red on the minor street is not adequate, because a yellow indication does not require the arterial driver to stop. The defensible action is all-red flash with STOP signs on every approach, or flaggers if the stop condition cannot be installed safely, until the signal is restored.
Site-specific factors to weigh:
- 55 mph approach speeds require large sign sizes, long advance placement, and visibility for stopping sight distance.
- The minor street volume is low, so an all-way stop is workable, but the arterial drivers will not expect it, so add warning signs in advance.
- The number of repeated failures means the signal is not reliable; a portable temporary signal may be worth deploying if repair extends past a day.
Document it. Record the time in flash, the traffic control installed, notification to the responsible engineer, and who authorized it. A reliability record with 12 events in eight weeks also justifies a scheduled replacement of the affected components rather than another reset.
3. What preventive and record-keeping measures would you require so this does not recur?
Read the worked answer
Physical corrections:
- Separate loop lead-in from the 120 V signal conductors; dedicated conduit or at minimum a divided raceway. This addresses the coupling path directly.
- Replace wet splices with potted or gel-filled splices, and re-seal the handholes.
- Correct cabinet moisture: verify the gasket, desiccant or heater, and drainage at the pad.
- Replace the surge arrestor and check service conductor terminations.
Testing program: annual megger test of every loop circuit with results recorded wet and dry, plus an annual cabinet inspection before the rainy season.
Records: track each flash event as a reliability metric with cause, downtime, and corrective action. Twelve events in eight weeks is a pattern that justifies component replacement, not repeated resets.
Verification: after the work, watch the event log through at least one significant rainstorm before closing the work order, and confirm the conflict and power fail counters stay at zero.