Water heater pilot goes out

Hi, I was hoping for some advice. In the house I recently moved in to, I have a natural gas, domestic hot water heater used for the in-slab radiant heating in the garage (Bradford White MI403S8FBNH). It was supposedly a functional system when the previous owner moved out. Unfortunately I have not been able to get the hot water heater to stay lit. I would find the pilot light would mysteriously go out after 1-2 days of lighting it. I replaced the thermocouple (ensuring the new one was well immersed in the pilot flame), but the problem persisted (pilot stayed lit when i took my finger off the valve, but i'd find it extinguished after a couple of days). I was generally unhappy with the fitment of the burner, pilot light assembly and gas control valve - it appeared to have been all retrofitted with non-factory parts by the previous owner (scavenged from a different model of hot water heater i would guess). As a result, the inner burner chamber covers were modified to accommodate the main burner pipe being in a different location, etc. This is a sealed combustion chamber water heater, and the modifications to the inner burner door left it not well sealed. As a result, I didn't go too in depth diagnosing the problem at this point, but instead ordered all factory correct parts for this water heater. I've recently installed the factory correct parts (thermostat/gas control valve, burner and pilot assembly, and new inner burner covers with fresh gaskets). To my disappointment, I still have a problem with the pilot going out - only this time it's much sooner. I dug a little deeper this time, and here are my findings of the current state of affairs:

- The pilot flame keeps the thermocouple well warm enough to signal the gas valve that there is a flame present. After lighting, I get ~9mV from the thermocouple loaded by the valve magnet, and 30mv open-circuit. Both these numbers appears to be within specs. I haven't tried but I believe i could leave the water heater in 'pilot' mode indefinitely.

- When the main burner comes on, the pilot appears to go out (hard to tell in the torrent of main burner flames), and the combustion chamber just kind of fills with a rolling blue 'cloud' of flame - the thermocouple is no longer directly engulfed in a specific flame tip, but rather is surrounded by the less specific main burner flames. As a result, when the main burner comes on (and the pilot seems to go out), the thermocouple begins to cool off, and I can see the mV output of the thermocouple steadily fall. After a few minutes the thermocouple closed circuit voltage falls below 2mv, the gas control valve clicks off, and the gas supply is shut , extinguishing the main burner flame and pilot flame (exactly as the gas control valve is supposed to do). I initially had the pilot assembly screwed in to what i believe to be the factory location. I've since rebent the bracket a few times, playing with different locations to get the thermocouple to stay more consistently immersed in the main burner flame (to no avail). Herein lies my first question: Is the pilot supposed to go out when the main burner comes on? If it were to stay on, the thermocouple would stay immersed in the pilot flame, whose location which is much more fixed than the main burner flames, and the thermocouple would stay hot.

- I measured the incoming natural gas pressure with a crude u-tube manometer, and found it stayed at 7-8" water column (i measured 3 loading conditions: with the hot water heater off, with only the pilot lit, and with the main burner going). This appears to be withing specs. I thought maybe the gas pressure was too low (due to plumbing restriction or frozen condensate or something) and this was causing the pilot to go out when the main burner came on (again not sure if this is intended behavior), but this doesn't appear to be a factor.

- I've tried to clean the combustion air intake path ('screen-lock' perforated floor of the combustion chamber) with shop vac and compressed air.

- I've tried the other thermocouple i had (the replacement i purchased the first time around), but the problem persisted.

- I've tried the old burner and gas orifice that was in the original setup (threaded into my new factory correct burner 'L' pipe - they appeared to be the exact same parts). no change.

- As mentioned above, the main burner flame seems a little 'wandery' in that it just kind of rolls around the combustion chamber. I would have expected more of a high pressure 'jet' of flame emerging from the holes in the burner. I don't know for sure what it's supposed to look like though. The nature of the flame does not change if I remove the inner burner access cover (allowing uninhibited fresh air to come in through there rather than the 'screenlock' perforated floor). The behavior where the pilot appears to go out when the main burner comes on, also is consistent with/without the combustion chamber cover.

I'm kind of running out of things to try at this point, and would appreciate some insight! (Sorry for the wall of text)

 

Chimney draft seems to be the only thing you haven't checked. I'd visually check ALL the vent piping from the heater outlet to the top of the chimney for obstructions. Also, be sure that there is a continuous rise from the heater outlet to the chimney. Could also be a flue obstruction in the heater itself. You would need to remove the internal baffle to check.

 

Thanks Furd, you're right - I have not done a thorough check of the flue and vent duct work (other than a quick look externally). I'll try that tonight, and maybe also check the regulated gas pressure at the tap on the valve body

 

Furd, You were correct, the flue had a blockage of rust chips. As I worked the wavy metal insert out, a huge pile of rust chips were falling into a pile in the burner compartment. I was previously skeptical of this being the problem, as, the first time i ran the water heater, there was a pretty convincing amount of exhaust/water vapor rising out of the flue - I took that to be an indicator that the flue was clear, but I guess not. I wire brushed the remaining rust flakes off the wavy insert, and inside of the flue (burner removed), and vacuumed up the debris. I put everything back together and the main burner flame cleaned up considerably. The pilot also continues to burn now when the main burner comes on (keeping the thermocouple nice and hot - I think this should fix my original problem).

I think what was happening when the flue was blocked was, when the main burner came on, the combustion products were filling the combustion chamber and starving the pilot (and the main burner itself) of oxygen. Even with the burner access cover off, All the exhaust gasses were probably traveling past the pilot, in the direction of the access hole, again starving the pilot of o2, and achieving the same result.

I think using a domestic water heater in this way, for radiant heat, is pretty taxing on it. It's constantly having to warm a steady incoming stream of stone cold return water (unlike for hot tap water applications where the water heater only rarely has to heat a full tank of completely cold water). The flue is probably rarely ever able to get up to a high enough temperature that the exhaust goes above it's dew point temp. So there's constantly condensation forming in the flue. With natural gas this condensate is acidic - hence all the corrosion I found.