Oh, boy, do I feel dumb. But hopefully someone will learn
something from my embarrassment that will help to keep
them safe.

We have a gas-fired water heater here. It has been our
practice for a while now to cut the thermostat on that to
minima if we are leaving the house for more than a day
at a time. Why pay and waste energy to heat and re-heat
water which no one will use?

On returning, I turn the t-stat back up to a level which
provides comfortably hot water. There isn't any degree
rating on the rotary knob, so I have figured out empirically
what the "hot enough" point is.

It never, ever would have occurred to me that I should install
some sort of a physical guard to prevent knob travel beyond
that point. And I could have absolutely bloody well sworn that
I had shown my wife what the set point was supposed to be.

Apparently not. We came home today from holiday travels,
and I went to wash my hands and said, "Ah, it's cold, the
water thermostat should be adjusted back up."

She was in the garage and said, "I'll get it!"

I was roadburnt, and didn't pay any attention to the possible
ramifications of having someone else do a job which I normally
would do.

A few hours later (during which there was lots of hot water use
for baths and whatnot) we were relaxing in the back of the house.
Suddenly, we looked at one another. "Did you hear that?" "Yep.
But no idea what it might be."

Reconnoitering the front room and kitchen, nothing. She looked
into the garage and said, "Ack! The water heater has exploded!"

I ran out and saw nothing of the kind (fortunately). What I did
see was a pressure relief pipe (which in this case discharges
straight onto the concrete floor) spewing steaming water.

Glanced at the rotary knob on the water heater thermostat. It
was ALL of the way over. Probably thirty angular degrees of travel
past the "hot enough" point. If there had been someone with the
wit to inscribe a burner setting from 1 to 10 on the rotary
surround, this setting would have been 11. Utterly maxed out.

Touched the pressure relief pipe just below the PR valve. Nearly
burned my finger on it.

Oh feathers. Cut the thermostat to zero. Turned on the utility
sink hot tap. Incredibly scaldingly hot water came out and kept
on coming out, in a huge cloud of steam. Sent the wife to run
other taps on full.

I should have had the presence of mind to go valve off the cold
water inflow, but there is no way to do that next to the heater --
that action requires a trip outside to shut off the master, and I
had my hands full dealing with lots of suddenly very wet stuff
which we had stored in the garage.

After a couple of minutes, the steam coming from the discharged
water was greatly lessened, and the pressure relief pipe felt much
cooler when touched, though still flowing visibly. I tripped the lever
on the relief valve and that reset it. Flow stopped. Crisis over.

The unit has now cooled down to garage ambient temperature,
and I am not going to turn it back up until we have carefully
thought through the return-to-service process.

Okay, lessons learnt up to this point:

(a) Always, ALWAYS document on the device what the proper
normal setting should be;

(b) Strongly consider mechanically excluding settings above
"normal" -- even settings which would not have blown the
relief would have been easily hot enough to scald at the tap;

(c) Take the time to arrange graceful failure modes even for
seldom activated systems -- routing the pressure relief pipe
to an exit hose had occurred to me before, but I had never
made it an action item;

(d) Brief people on how to set controls if they're not yet so
informed -- and people who find themselves unsure should
take time to ask;

(e) Be sure to test safety systems on a regular schedule --
I shudder to think what would have happened if that relief
valve had not tripped when it did. I have flipped the test lever
in the past when I thought of it, but will be formally rigorous
about it in future.

That's what I have learned. Here's what I am yet to learn,
and about which hopefully someone will be able to inform me:

(f) Of what gotchas should I be aware when attempting to
return the unit to service?

(g) What are the odds that the heater has been damaged
by the overheat/overpressure event? The relief valve tag
specifies release at 150psi/210F, which is pretty impressive.
What's more, this unit has a build tag dated 1985! Assuming
that it was installed at that time (it may be old stock) that is
impressive longevity. My experience has been that modern
water heaters seldom last more than a decade even without
adverse events of this type. And while I would have liked to
eke out another few years, if the heater is likely to have been
compromised, I don't want to risk having it split wide open
a few months later on.

(h) Even if the tank and heater assembly seem OK, should
I think about replacing the relief valve itself? It was fine before
this, with no drips or hisses. I do want to be quite sure that
the relief will open when it needs to, though. Note that there
is also an unknown in that a new-in-box valve might itself
be junk -- the current valve at least has been demonstrated
to work properly under field conditions.

Comments and feedback actively solicited. I'd prefer to see
them here in the newsgroup rather than mailed to me: this
in-box is utterly soaked with spam.

And before I close, let me take this opportunity to wish the
top of the season to you all!

C