Eight-Year-Old Glycol System Producing Lukewarm Water
A school's 8-year-old glycol system has delivered lukewarm water since November. The site log shows: collector sensor 96 °C at noon, tank sensor 44 °C, pump running continuously and hot to the touch, flow meter 3.5 L/min against a 14 L/min design, loop pressure 12 psi against a 40 psi cold fill, a heat-exchanger approach of 6 °C where commissioning recorded 2 °C, a dark brown sight glass, and a glycol sample at pH 5.1 (new glycol is about 9.0). Pressing the Schrader valve on the expansion tank releases fluid. The pump makes a gravel-like noise. Records show only an annual 'check the pump' visit, with no glycol testing, no strainer cleaning and no flow or temperature logging since commissioning.
1. Diagnose the failure in a logical order. Which reading is the primary fault, and how do the others connect to it?
Read the worked answer
The primary fault is low flow: 3.5 L/min against 14 L/min design. With a quarter of the design flow, the collector cannot reject heat, so the collector sensor climbs to 96 °C while the tank lags at 44 °C, and the controller keeps the pump running because the differential never closes.
Work upstream from the flow meter. Confirm the meter, then inspect the pump strainer or filter, which is the most likely blockage given brown fluid. Check isolation valves, a failed check valve, and air in the loop. The pump noise suggests cavitation or bearing damage, both consistent with a loop running at 12 psi against a 40 psi fill.
The expansion tank reading ties in: fluid from the Schrader valve means the bladder has failed, so there is no gas cushion and the loop pressure swings as it heats. That causes cavitation, which damages the pump and aerates the fluid.
The pH 5.1 and brown color explain the sludge: acidic glycol attacks copper and brass and generates the particles that plug the strainer. The 6 °C exchanger approach indicates the load side is fouled too. This is one failure chain, not four unrelated faults, and the missing records are why it went undetected.
2. List the corrective work and the measurements you would use to prove the repair succeeded.
Read the worked answer
Isolate and drain the loop, then flush with a cleaning solution until the flush water runs clear. Replace the expansion tank with one sized to the loop volume and pre-charged to the cold fill pressure; replace the pump, the pump strainer and any seized valves; restore or add an air separator and automatic air vent at the high point. Refill with inhibited propylene glycol mixed to the freeze protection the site requires and verify the concentration with a refractometer.
On the load side, backflush or descale the heat exchanger and confirm the approach returns toward the commissioning value at design flow.
Verification measurements: 14 L/min at the flow meter; collector-to-tank differential within the design charging range (commonly 5–15 °C); stable loop pressure from cold to operating temperature; no gravel noise and pump amps within nameplate; pH and freeze point recorded and logged with the date; exchanger approach back near 2 °C.
Glycol is toxic. Collect and dispose of the old fluid through the site's chemical waste stream, and never dump it to a drain or the ground.
3. Design the maintenance program that would have prevented this, and explain why each item matters.
Read the worked answer
Annual items, performed at a comparable solar condition so results are trendable:
- Glycol pH and freeze point. A pH well below new, or a freeze point that has drifted, means the inhibitor package is spent and replacement is due. This alone would have caught the failure years earlier.
- Visual check of fluid color and clarity through the sight glass.
- Expansion tank pre-charge, checked with the loop at zero pressure, because a failed bladder is silent until the pump starts cavitating.
- Strainer cleaning.
- Pump amp draw and listening for noise.
- Flow rate and collector-to-tank differential recorded in a log.
- Sensor check against a calibrated thermometer, since controller behavior depends entirely on sensor accuracy.
- Relief valve lift test and a leak inspection of collectors, roof penetrations and pipe insulation.
Every 3–5 years, or whenever the tests fail, drain, flush and recharge the glycol. Also check the potable-side T&P, the mixing valve, and the controller clock and battery after power outages. The point of the log is to make slow trends visible: falling flow, rising exchanger approach and falling pH are all early warnings that cost far less than a pump and heat exchanger.