Pressure Transients in Water Mains: What a UK Review Actually Found
TL;DR: A pressure transient travels along a metal water main at around 1,000 metres per second and is over in a fraction of a second. A logger sampling every 15 minutes has roughly a one in a million chance of being awake when it passes. The interesting part is what the UK’s own review concluded about them, which is not what the surge-monitoring market implies, and where the money in pressure work actually sits.
Last updated: 10 September 2026
On this page
Key takeaways
- Wave speed is 300 to 500 m/s in plastic pipe and 1,000 to 1,200 m/s in metal. A one metre per second velocity change produces roughly 30 metres of head in plastic and 100 metres in metal.
- A pressure transient is a sub-second event. Interval logging does not slow it down, it simply misses it.
- Defra and the DWI commissioned a review of exactly this question. Its first conclusion was that there is “little evidence in the literature for pressure fluctuations sufficient to cause ingress”.
- The same review put low-pressure events capable of causing ingress at 0.23 per 1,000 population per year, about 12,000 a year across England and Wales.
- Leakage varies with pressure raised to an exponent that can sit anywhere between 0.5 and 2.5. Managing average pressure pays. Chasing every transient usually does not.
- A battery LoRaWAN pressure sensor is an interval instrument. Buy it for profiling and alarms, not for surge capture, and be suspicious of anyone who says otherwise.
What is a pressure transient, and how big does it get?
A pressure transient is the pressure wave produced when moving water is made to change speed. Close a valve, trip a pump, or start one, and the momentum of the column has to go somewhere. It converts into pressure, and the pressure transient travels away from the disturbance as a wave at the speed of sound in that particular pipe.
The size is given by the Joukowsky relationship, H = c × Δv / g, where H is the head change, c is the wave speed and Δv is the velocity change. The Defra and Drinking Water Inspectorate review of this subject, carried out by WRc and published in March 2008, sets out the numbers that matter: wave speed in plastic pipes is typically 300 to 500 m/s and in metal pipes 1,000 to 1,200 m/s.
Put a modest number through it. The report’s own worked example uses a velocity change of 1 m/s and arrives at about 30 metres of head in plastic pipe and 100 metres in metal pipe. A hundred metres of head is roughly ten bar arriving on top of whatever the main was already carrying, which is why a pressure transient damages fittings, joints and meters rather than politely announcing itself.
Direction matters as much as magnitude. Downstream of a stopping pump, upstream of a starting one, downstream of a closing valve or upstream of an opening one, that change is negative. The main does not get squeezed, it gets pulled. That is the version of a pressure transient people worry about, and it is the one the next two sections are about.
Why a 15-minute logger will never see one
Because the event and the sampling interval are separated by six orders of magnitude. A wave crossing a five kilometre trunk main at 1,000 m/s does the round trip in about ten seconds, and the sharp part of the pressure change at any one point lasts a fraction of a second. A logger that wakes up every fifteen minutes is asleep for 99.9999% of the pressure transient’s life.
This is not a resolution problem that averaging can rescue. It is a sampling problem. The instrument does not record a smaller version of the pressure transient, it records the pressure fifteen minutes before and fifteen minutes after, both of which look normal. A pressure transient leaves no partial trace to interpolate. The event leaves no trace in the data at all, which is why sites with a clean pressure history can still be breaking fittings.
Catching a pressure transient requires a purpose-built high-speed logger sampling many times per second, deployed deliberately. The WRc report’s own suggested method is exactly that: choose pumping stations with different lengths of uninterrupted trunk main downstream, place monitors at the station, on the trunk main and out in the network, and “record pressures with a high speed logger during pump switching”. Note what that is: a study, at known locations, around a known trigger. It is not a permanent network-wide deployment.
Does a transient really suck contamination into a main?
Less often than the surge-monitoring market implies, at least in the UK. The mechanism is real: the negative half of a pressure transient passing a leak can, in principle, draw the surrounding groundwater in through the same hole that normally leaks out. What the UK evidence base does not support is the idea that this happens routinely across a distribution network.
The Defra and DWI review reached six conclusions worth quoting almost verbatim, because they are unusually direct for a government-commissioned document:
- There is “little evidence in the literature for pressure fluctuations sufficient to cause ingress”.
- Surge effects are of limited size and penetration in distribution systems, although they can occur in unbranched trunk mains and in smaller mains close to them.
- To cause ingress, the high point must be approximately 15 m or more above the demand point.
- Low pressure fluctuations sufficient to cause ingress happen 0.23 times per year per 1,000 population served.
- Across England and Wales, that rate translates to roughly 12,000 events a year.
- Where surge is sufficient to cause ingress, the answer is engineering, not monitoring: slow valve closure and controlled pump switching.
Read those together and a sensible position on pressure transient risk falls out. A pressure transient large enough to matter is a trunk-main problem with a specific hydraulic signature, not a distributed threat you can monitor your way out of. Twelve thousand events a year across a population of sixty million is genuinely rare per site and genuinely non-zero nationally. It justifies targeted pressure transient study at pumping stations and trunk mains. It does not justify buying surge loggers for every district metered area.
Where the money in pressure work actually is
In the average, not the spike. Leakage responds to pressure far more strongly than intuition suggests, and reducing the pressure a network runs at reduces both the background leakage and the rate at which new bursts appear. That is a continuous effect you can measure with ordinary interval data.
The governing relationship comes from Allan Lambert’s paper “What do we know about pressure:leakage relationships in distribution systems?”, which sets out that leakage L varies with pressure P raised to an exponent N1, and that N1 “could vary between 0.50 and 2.50” depending on the type of leak present. Round holes in rigid pipe sit near the bottom of that range. Cracks that open as pressure rises sit near the top.
The practical consequence is worth sitting with. At the square-root end, a 10% pressure reduction buys about 5% less leakage. At the upper end of the range, the same 10% reduction buys well over 20%. You cannot know which end of the range a district sits at without measuring pressure and night flow together, over weeks, at more than one point. That is an interval-logging job, and it is where the returns are.
What an interval pressure sensor can and cannot do
It can profile, alarm and evidence. It cannot capture a pressure transient, and no amount of configuration will change that, because a pressure transient is faster than the sampling architecture. Being explicit about the line between those two is the whole point of this section, because it is the line most product pages blur.
Our reference instrument for this work is the Milesight EM500-PP pipe pressure sensor at £197.52 ex VAT. The published specification is a vented gauge measuring 0 to 1,600 kPa, which is 0 to 16 bar, with accuracy of ±0.5% of full scale, 1 kPa resolution, 150% full-scale overload tolerance and long-term stability of ±0.3% FS per year. It connects with a G ½ inch male process connection on a 1.5 m cable, and runs from a single 19,000 mAh lithium thionyl chloride cell.
The battery figure is the one that tells you what kind of instrument this is: ten years at a ten-minute reporting interval. That number exists because the device spends almost all of its life asleep. An instrument that stayed awake fast enough to see a pressure transient would not last ten years on a battery, and an instrument that lasts ten years on a battery cannot see one. Physics, not marketing.
What it does give you is worth having. Threshold alarms fire when pressure crosses a limit you set. On-board storage holds 1,000 entries with retransmission, so a gateway outage does not become a hole in the record. And 1 kPa resolution across 16 bar is more than enough to see a district’s daily pressure curve, its night minimum and the slow drift that says a pressure reducing valve is losing its setting.
Pressure transient logger or interval sensor: how to choose
Start from the question you are trying to answer, not the hardware. Two questions dominate this area and they need opposite instruments.
- “Are we damaging our own network when the pumps switch?” That is a study. Use a high-speed pressure transient logger, at the pumping station and on the trunk main, around known switching events, for days rather than years. Then fix it in the hydraulics with slower valve closure and controlled pump starts.
- “What pressure are we actually running at, everywhere, all the time?” That is monitoring. Use battery interval sensors at enough points to cover the district, leave them for a year, and use the data to set pressure reducing valves and to find the districts where the leakage exponent is high.
Most utilities and large private networks need the second far more often than the first, and the second is the one that survives a business case. Our utilities and water networks solution is built around it, and it sits directly alongside the leakage work described in our guide to non-revenue water.
One practical note on hardware. If you already have a 4-20mA pressure transducer on site, you do not need a new sensor at all, only something to read it. The retrofit path is covered in our piece on 4-20mA current loops, and a LoRaWAN IO controller at the panel end is usually cheaper than replacing an instrument that is working perfectly well.
Frequently asked questions
How fast does a pressure transient logger need to sample?
Many times per second, because the sharp part of a pressure transient lasts a fraction of one. There is no single correct figure, and it depends on the pipe length and wave speed you are studying, but anything reporting at minute intervals is measuring a different phenomenon entirely.
Can a LoRaWAN sensor be set to a fast enough interval?
No, and not because of the sensor. LoRaWAN is a low duty cycle network by design, with airtime limits in the EU868 band that make continuous high-rate reporting impossible. Anyone offering pressure transient capture over LoRaWAN is either buffering on the device or describing something other than a pressure transient.
Is water hammer a real risk on a private network or building?
Damage risk, yes. Fast-acting solenoid valves are a classic cause, and the fix is a slower closure or a properly sized arrestor. Contamination risk is a different question and the UK review points at trunk mains with a high point 15 m or more above the demand point, not at ordinary building services.
What pressure range should we specify?
Cover the working range with room above it, and remember that accuracy quoted as a percentage of full scale gets worse in absolute terms as you increase the range. A 16 bar instrument at ±0.5% FS is ±8 kPa everywhere on the scale, including at the 2 bar you actually care about.
Where should the first sensors go?
The critical point of the district, meaning the location with the lowest pressure at peak demand, and immediately downstream of any pressure reducing valve. Those two positions answer most of the questions a pressure programme is asked in its first year.
Where this leaves you
A pressure transient is a genuine engineering phenomenon with a well-understood equation behind it, and it is also the thing that most pressure monitoring is quietly sold against while being incapable of detecting it. The UK’s own review found the ingress risk to be real but rare and concentrated in trunk mains, and recommended fixing it in the hydraulics rather than watching it.
Meanwhile the unglamorous work, knowing what pressure every district actually runs at and moving the average down, has a peer-reviewed exponent attached to it and pays for itself. Tell us the pipe, the working range and how many points you want to cover, and we will quote the sensors and the gateway inside one working day. Start with the utilities and water networks solution, or read the water pressure monitoring overview if you want the shorter version.
Next step
Get a priced kit list for your site
Answer three quick questions and tell us where to send it. An engineer replies within one working day with the parts, the prices and the lead time.
Rather talk it through? Call 023 9223 3611
Received. An engineer has it.
You will hear back within one working day with the parts, the prices and the lead time. A confirmation is on its way to your inbox.
If it is urgent, call 023 9223 3611.
