Yesterday we sprang a leak - today's coffee time challenge is..

[MikeSharp01]

Yesterday I got a call from the other half to say that there was water dripping from the loft hatch back at mill stone manor!

 

We have a procedure for leaks, only used once before, and she had already done everything to prevent any further loss so the situation was calm and the amount of water was quite small - damp carpet. Had we been out it would have been another matter!

 

Long - story short, I came home and got into the loft and saw that the water was coming from the pressure set pump's primary pressure switch, a pin hole in the diaphragm housing! I soon had it apart, ordered a new switch as a spare and machined up a new diaphragm housing, fitted it and tested it. So me thinks - job done!

 

Overnight I have been wondering how the control system for the pressure set could be adapted to detect small leaks. It already has protection, trips, if the pump runs too long in trying repressurise the system or if the pressure rises above 1.2 bar. Neither of these would protect against a small leak.

 

So the challenge is to come up with a way of detecting very low, constant, flows such as a dripping tap or our pin hole. Several ideas come to mind, very sensitive flow meter and / or analog pressure sensor watching the slow reduction in pressure come to mind but neither would be fool proof and might well be impractical. Any thoughts?

 

A bit of background if it helps: We have the worst water pressure anywhere and we can get about 2L/min out of the main tap if nobody else on the hill is running anything. So about 20 years ago we put an extra tank in the roof (total of 400L), a UVC, a 100L expansion vessel and a pressure set into the loft which pressurises the cold water to 1Bar which then feeds the UVC. This approach means that the Hot and cold water are always at the same (roughly) pressure so the thermostatic showers work really well and at full bore are very refreshing. The pump will happily deliver 70L / min against 1 Bar and the expansion vessel (on the cold side - but there is a small one on the hot side as well to allow for the expansion of the hot water as it expands) ensures that you get several minutes of flow to the showers before the pump cycles the pressure up with a hysteresis of about 0.4 Bar.

 

[SteamyTea]

I am just off for a coffee, don't think I will be thinking about your problems though

[ToughButterCup]

35 minutes ago, MikeSharp01 said:

[...]

So the challenge is to come up with a way of detecting very low, constant, flows such as a dripping tap or our pin hole.

[...]

 

I will be watching this one like a hawk. Had we such a system, I'd have sorted out our analogous problem long ago.

[ProDave]

My "out of the box" solution is a drip tray where you expect the leak and 2 electrodes to detect the presence of water.

[MikeSharp01]

7 minutes ago, ProDave said:

 

My "out of the box" solution is a drip tray where you expect the leak and 2 electrodes to detect the presence of water.

 

The UVC is bundled and is piped to overflow but on this occasion the hole caused a jet which sprayed the water over the boundary. Plus this would not spot a dripping tap.

[Jeremy Harris]

4 minutes ago, ProDave said:

My "out of the box" solution is a drip tray where you expect the leak and 2 electrodes to detect the presence of water.

 

 

Mine too.  I use these fairly cheap and easy to use sensors, that have gold plated electrodes so tend to give better long-term reliability: https://www.rapidonline.com/Electronic-Components/Liquid-sensor-66470

 

They are easy to wire directly to a small alarm beeper, like this: https://www.rapidonline.com/rvfm-12v-low-profile-buzzer-35-0041  and a suitable power supply to make a very effective alarm.

[SteamyTea]

Most cars have some sort of rain sensor that can stop and start the wiper (mine seems to stop them when I pull up at traffic lights).

I have no idea where the sensor is, so must be fairly small.

[Ferdinand]

1 minute ago, SteamyTea said:

Most cars have some sort of rain sensor that can stop and start the wiper (mine seems to stop them when I pull up at traffic lights).

I have no idea where the sensor is, so must be fairly small.

 

You seem to be buying excessively new or excessively complicated cars .

[SteamyTea]

Or a ten year old basic Ford C-Max, The Scenic had it too, and they are over 20 years old.

 

Is there some kind of tape that could be wrapped around areas of high risk.  Something with two parallel conductors and some dry salt.  Connect to a meter and when it gets damp, circuit is made and a buzzer goes off.

Pipe condensation may be a problem, but pipes should be lagged enough to stop that anyway.

Edited January 6, 2018 by SteamyTea

[Ferdinand]

8 minutes ago, SteamyTea said:

Or a ten year old basic Ford C-Max, The Scenic had it too, and they are over 20 years old.

 

Is there some kind of tape that could be wrapped around areas of high risk.  Something with two parallel conductors and some dry salt.  Connect to a meter and when it gets damp, circuit is made and a buzzer goes off.

 

 

Yep. Modern and complicated ;-).

 

For the substantive point, what about the normal aluminium sealing tape - though it may be a sod to get off again, or cannibalise ribbon connector or 2 or 4-core phone wire?

 

F 

[Jeremy Harris]

I think the car sensors are capacitive- there are no conductors on the outer surface of the glass that I can see, and I think they work by the change in dielectric constant when rain drops bridge the internal electrodes under de the black layer inside the glass.  They probably have low level AC across them and get loaded down when a water droplet is on the outer face of the glass.  I could be wrong, but that's how I guessed they worked when I first looked at one.

 

Those WS-1 sensors I linked to earlier are so cheap and easy to use that it's not really worth mucking around with home made solutions, IMHO.   £4 and it will switch up to half an amp directly for an alarm, it's potted in a sealed box and has gold plated contacts that won't corrode. 

 

When I tried to make a beeper wire for measuring the RWL in our borehole I found that one problem was that you need a material between the sense wires that doesn't retain water - it must be strongly hydrophobic - otherwise the beeper keeps going when the thing is out of the water, from the water film bridging the electrodes.  The plastic that these WS1 sensors are made from doesn't allow a water film to stay put, so gets around that problem.

[Ferdinand]

Maplin have a £4.99 one that seems networked.

 

https://www.maplin.co.uk/p/foxx-project-water-sensor-a76wa

[Jeremy Harris]

5 minutes ago, Ferdinand said:

Maplin have a £4.99 one that seems networked.

 

https://www.maplin.co.uk/p/foxx-project-water-sensor-a76wa

 

Yes the Foxx system is a modular system that allows interconnectivity between sensors.  There are a range of Foxx modules available, I believe.  Seems a complex way to make a simple water level alarm though, when all that's needed is a cheap and reliable sensor and a beeper or LED (or both) , really.

[SteamyTea]

I think Mike's problem is for tiny, tiny leaks, ones that would not show up on a normal detector (I had one given to me by the water company, not sure what happened to it).

I think they used sound, but where known to be unreliable and gave too many false positives.

[dpmiller]

@JSHarris car rain detectors are optical, using the water droplet to refract a light beam (visible orange light in some early units) back to a receiver. If the screen is dry the light beam just passes out through the glass.

 

here's a standalone using the same tech http://rainsensors.com/

 

[Jeremy Harris]

1 hour ago, dpmiller said:

@JSHarris car rain detectors are optical, using the water droplet to refract a light beam (visible orange light in some early units) back to a receiver. If the screen is dry the light beam just passes out through the glass.

 

here's a standalone using the same tech http://rainsensors.com/

 

 

 

Interesting.  I'll point a CCTV camera at mine to see if I can see anything.  There's no visible light, but if they're using short wavelength IR LEDs then they should show on any CCTV with a night vision capability.  As it happens, I park facing one of the cameras, so I'll take a look and sew if there's any IR with the ignition on.

 

I'd assumed they were capacitive based on observations of the electrode pattern on the first car I had with this system, years ago now.  There seemed to be rows of interlocking "fingers" sputtered on to the inner face of the glass, so my guess was that they were just looking for a change in the dielectric constant, in the same way as external electrode level sensors work.  I made level sensors for paramotor fuel tanks for a few years, as kits, and these used a capacitive sensor with the fuel as the dielectric.  They worked pretty well, and were a lot easier that tryign to crane your neck around to look at a mirror pointing at a translucent fuel tank.

Edited January 6, 2018 by JSHarris

[Nickfromwales]

23 hours ago, ProDave said:

My "out of the box" solution is a drip tray where you expect the leak and 2 electrodes to detect the presence of water.

Basically that's what I would do. I've stuck a lot of stuff in attics over the years, then I spoke to a very helpful chap iirc from Pumpsukltd and he mentioned making a large glass fibre drip tray with up stands for all the equipment to sit on / in. From there a Hepworth trap and an overflow run in 32mm waste pipe, so basically a big shower tray. 

[SteamyTea]

I made a large drip tray for our spa baths to drip into.  I am surprised that it is not done more often, almost as a standard.

But a pipe can leak anywhere, but usually and a joint or junction I would think.

A simple way to detect a very small leak would be useful, though it may not show where the problem actually is.

[Onoff]

I seriously debated a tray under my half sunken bath. Tbh though I followed @Nickfromwales tips to the letter for sealing the trap in etc so reckon I'm good.

 

 

If water does get in there it'll be straight through that pear drop shaped hole in the ply and be held within the floor under the slab by the DPM. 

[ProDave]

In the last house, the source of bath leaks was the incredably poor standard sealing arrangements around a bath tap. When showering water splashing around the taps would occasionally leak through and drip under the bath. They are now sealed to death and don't leak any more.

[Nickfromwales]

2 hours ago, ProDave said:

In the last house, the source of bath leaks was the incredably poor standard sealing arrangements around a bath tap. When showering water splashing around the taps would occasionally leak through and drip under the bath. They are now sealed to death and don't leak any more.

So many 'plumbers' don't bother with doing that, but most sinks have a terrible cast where the tap sits, and cheap taps have a terrible nasty rubber o-ring and nowt else. That usually covers less than 2/3 of the hole. It's fatal with a shower over a bath but also not helped along by the fact most people inadvertently buy a standard bath instead of a dedicated shower bath, the latter has flatter sides and slowed decks to guide the splashing water to the drain. They have a much better finish and mate the taps far better. 

 

[ProDave]

Another "issue" with the standard baths we fitted at the last house was the pre drilled tap holes seemed overly large.  The bath we have for the new house, does not have pre drilled holes, I will have to drill them, and I will make sure the holes are only just large enough giving a greater area for any sealing  washer to sit on.

[MikeSharp01]

Not sure how  rain detectors will hack it - here is what I have been thinking over the last 24 hours.

 

Firstly we need some parameters - how much of a leak am I prepared to tolerate before I want all the pressure in the system gone and fresh input water to the system choked off. I think I would not be too unhappy with 100ml on the carpet, or in the ceiling plasterboard - not so worried about dripping taps but any system would find both,  If I could get better than 100ml I would be happy. How quickly to I want it to react - lets say 10 minutes for the smallest leaks, limited to 100ml.

1. Very sensitive pressure sensor based on the the PVT rule - will need quite a bit of intelligence and other sensors. Total water volume of the system from Non returns before pump to the taps is about 230L including the UVC and 50% of the expansion vessels. If 1 drip is about 0.1ml the volume change is 1 part in 2,300,000 to detect that would need a 22bit ADC compensated for all kinds of stuff like the temperature of hot water cooling in the pipes after the hot tap has been opened, the whole system warming up as the house warms up / cools down (although heating up the house would mask a tiny leak cooling would go the wrong way and look like a tiny leak if not compensated for). 22bit ADCs are cheap enough but the signal processing before the ADC will need to be mega stable and the sampling / smoothing / filtering software will need to be thought about. However a few drips, say 20, over 5 minutes would be 1ml, reduces the required resolution of the ADC and in my ideal world would trip the leak detection.
2. Modify or use a standard water meter. I have noticed that the water company water meter is very sensitive but I am not sure how much so, will it sense the dripping tap volume over a period? If so then it might be possible to use a water meter. There is another problem in where to put the flow meter, I would need one in the HOT final and the Cold final as a single one before the pump won't sense the expansion vessel releasing its water. As far as flow meters go I found this article which seems to show that they are not that sensitive and points out that people could get around water meters by using very low flows. Based on our trip level above 100ml over 10 minutes = 600ml / Hr with is well below the sensitivity of water meters - well on the curve shown here anyway which tends to say that below 10l/hr they may not even register. This article is more interesting in that is shows, for somewhat older meters, that the mean start rate of flow is around 20lm so if your flow is below that you stand a good chance of not paying for water - which may explain why they have never fitted one at Millstone Manor. So unless I can find a much more sensitive flow meter this approach won't work.

 

I need to think about this some more - perhaps a hybrid system of some sort.

[Jeremy Harris]

What is it you EXACTLY want to be warned about?

 

Sensitive leak detection at the one drip every few seconds level is near-impossible on the water feed side, as you need to be able to differentiate between legitimate slow water use (like the final stage of a cistern fill cycle, where the flow rate may well be tiny, or someone leaving a tap dripping) and genuine leakage, which may well be of the same order of magnitude.  It's possible, but means having loads of sensitive flow sensors on every pipe and then subtracting the legitimate flow from the total flow to get the leakage flow.

 

On the other hand, if you just want a warning of a slow leak that could cause damage if it's not checked, then just put a tray underneath the probable leakage point, as a bund, fit a simple water detector and let the output from that trigger an alarm/water shut off.

 

Out in our treatment plant shed, I fitted a 2 x 1 upstand all around the perimeter of the 22mm OSB floor, sealed the whole floor with several coats of the sealant stuff used for tanking behind showers and then just fitted one of those WS-1 sensors, screwed to the wall so that the sensor is almost touching the floor (it has two moulded in plastic stand offs to space the sensing conductors just clear of the floor).  That's wired to a 12 V supply and a beeper, so if one of the pressure vessels or any of the pipework in there develops a leak, the beeper sounds.  I could pretty easily wire the alarm to the pump power, to cut it off in the event of a leak.  As a back up, I glued in an overflow drain so that if the water level on the floor exceeds about 1/2" the water just runs to a visible outside drain via a pipe.

 

I'm planning to do exactly the same where the water softener sits and where the internal 100 litre pressure vessel sits, but with both of those I'm just going to use a suitably sized plastic tray as the bund.

[Onoff]

Old school!

 

From the electronics kit I got for Christmas in 1973, I used Veroboard instead of blotting paper...it went green outside! I remember trying to convince my Mum all she needed to do was watch for the lamp to light to know it was raining: