New user?

gravity flow water system [closed]

Design and construction of a gravity flow water system for a community of 3000 people in Kyrgyzstan. There is already a reservoir tank above the village so this project involves constructing a spring catchment and then pipeline to the reservoir tank.

The engineers are designing a gravity-flow water supply from a spring to an existing reservoir tank above a village of about 3000 people. Previously, a borehole and pump was used to supply the tank, but this stopped working nearly 20 years ago, so people currently drink from the river. The engineers have assessed that the distribution network from the reservoir to the tap stands in the village is still in useable condition, so in this project they plan to construct just a spring catchment and a pipeline from the spring to the tank. The pipe will discharge freely into the tank. They have estimated the length of pipeline required as 6km, but until I arrived they had not surveyed the height differences of the land or measured the flow from the spring. Without calculation, they decided to use 90mm diameter HDP pipe (which they are now in process of procuring through a tender). I have now helped them measure the flow of the spring as 6 litre/s and have started to survey the pipeline route with them. So far we have surveyed 1.6km starting at the spring, and there is a drop of 150m in height from the spring so far. Based on this topography and pipe diameter, I have calculated the ‘natural flow’ of the pipeline as about 13 litre/s (using pp47 of ‘A Handbook of Gravity Flow Water Systems’ by Thomas D. Jordan). This is more than the flow from the spring, so it implies that the pipeline will not flow full and therefore will not be pressurised. Jordan states that this is not a problem for pipeline sections without any tap stands situated on the section. However, Engineering in Emergencies pp365 states that a pipe not flowing full creates problems of air-locks, water hammer and variable flow. As I understand it, there are likely to be 2 options (assuming the rest of the topography survey shows a similar gradient the rest of the way to the reservoir - this has to be checked): (a) Continue with the design of 90mm diameter pipes, knowing that the pipe will not flow full. (b) Re-design the pipeline with a smaller diameter pipe that allows the pipeline to flow full. This may need the addition of break pressure tanks to the design. If anyone with experience of gravity flow systems is able to comment on whether or not a pipe not flowing full is likely to cause problems, and therefore which option (a) or (b) is more sensible, I would really appreciate the advice.

Here is additional information since my first email:

PROFILE - Kyzyltoo We have now surveyed the whole pipeline and overall there is about a 240m drop over 5.9km from spring to tank. I am ... (more)

edit retag flag offensive reopen merge delete

Closed for the following reason the question is answered, right answer was accepted by KnowledgePointAdmin
close date 2014-01-27 15:20:59.483360

6 Answers

Sort by » oldest newest most liked

Looks as though you are using the outside diameter of the pipe instead of the inside (72.9mm for 90mm pipe See Appendix 16 of Engineering in emergencies). The pipe will give about 4.5 -5.0 l/s over that gradient., which will probably be ample for that population, especially if they continue to use river water for washing etc. Maybe you could consider a twin pipe if the next size up is not available (at least for part of the line if you need the full flow all the time). Max pressure on the pipe would be 15 bar if there is a stop valve at the lower end. If no valve you will need to allow for overflow to pass safely around the tank (and not wash out its foundations!) It will flow full over those lengths which lie below the grade line but air pockets will form at the high spots. I suggest tapping the pipe at the high spots and putting in bleed valves if air valves are not available. This would require occasional opening of all those valves to let the air out (tedious task). Is the pipe being buried below the frost line? Freezing could be a problem in those parts of the pipe which remain full at zero flow. Regards, Alan Reed

edit flag offensive delete publish link more

As I wrote the section in “Engineering in Emergencies” that made the comment concerning partially full pipes creating problems with air locks I had better respond! The problem is that air will be entrained in the flow so if there are sections where the pipe flows full then air may well collect in high spots creating air locks. But in the case described there should be no problem, assuming the pipeline is designed and built to avoid air pockets, because it would be simplest to fit a gate valve at the beginning of the pipeline to regulate the flow and adjust to ensure the pipe flows full. Given that the pipes have already been ordered then it gives the opportunity to adjust the flow to the optimum and upgrade to higher flows at times when the spring yield is greater. That is a question: the flow might have been estimated at this time of the year but how does it vary throughout the year? If it is found that the pipe cannot take the maximum flow when the yield is highest, then an overflow will need to be designed into the spring box/capture to avoid seepage and erosion of the protected spring.

If there is a drop in height of 150m in 1.6 km then it will be essential to include break pressure tanks in the system to avoid bursting the pipes. Do we know the pressure rating of the pipes that have been ordered? It is not just the pipe pressure rating but that of the joints as well. If the flow has been calculated on the basis of a 150m pressure drop then think again because this will exceed the pressure rating of the HDPE pipe. The system will need to be designed in stages to avoid the high pressures. This will also mean that the flow will be less than originally calculated and, perhaps, match the yield more closely.

I hope this is useful Jan Davies

edit flag offensive delete publish link more

The problems that are listed of air locks etc are principally problems that would affect consumers drawing directly from the pipeline. In this case as I understand the problem the pipe from the spring will run into a tank from which the consumers will draw water. If the spring flows at a steady rate there will be an even flow of water down the line which will find its own level. The only problem may be if there were any high spots in the line rather than a continuous fall. These would be where airlocks may occur and inhibit flow or set up a pulsing flow.

To get the pipe to run full is a relatively simple matter of artificially increasing the head-loss in the pipe which can be easily done by introducing chokes such as a partially closed valve to increase the head-loss and cause water to back up behind it.

If the section nearest the tank could be arranged to drop below the tank before rising to enter the tank then a section will be created that will always run full. The pipe velocity by my calcs is about 0.9 m/s which is only slightly below the normal range.

I trust that the spring will be protected in some way to prevent pollution.

Regards, Steve Oxtoby

edit flag offensive delete publish link more

The 150m fall over 1600m gives a hydraulic gradient of 0.0938 (1 in 10.67) which as you indicate is far steeper than required, so natural flow is likely for at least parts of the pipe. I calculate that for 6 litres/sec a 90mm EXTERNAL diameter polyethylene pipe (assuming 71.6mm internal diameter) would only need a hydraulic gradient of 0.032 (that is a total head-loss of 51.2m rather than 150m).

If you really need to use this pipe then at the lower end of the pipe you could add head-loss to use up the extra 98.8m of head, to ensure that the pipe flows full. Putting a partly closed valve on the pipe, or inserting a section of small diameter pipe that gave a high head-loss, or inserting a small diameter orifice in the pipe are ways of adding head-loss. However if you have stand-posts served by the pipe between the spring and the tank you will need to control the dynamic head above ground (i.e. pressure above ground under flow conditions) so that it is not excessive at the taps (e.g. less than 20m head) so you are probably going to have to use one or more break-pressure tank(s).

You need to bear in mind that without a break-pressure tank, if flow stops (for example because the tank becomes full and a float valve at inlet closes), that the pipe work will be subjected to the full static head (i.e. 150m head at the lower end) - this is another reason for providing one or more break-pressure tanks.

A 50mm internal diameter pipe (63mm outside diameter) over the whole length would give too great a head-loss for the desired flow rate. If you really want to get the right flow for the 150m head-loss then you could use the 71.6mm diameter for part of the length and a 50mm diameter for the remainder. However, this assumes that there are no stand posts on this supply pipe so pressure in it does not also need controlling.

The whole system will need a proper hydraulic design carried out when the topography is known and a suitable site for the break pressure tank(s) can be located.


Brian Skinner

edit flag offensive delete publish link more

I designed gravity systems in Rwanda for several years so interesting to see someone else wrestling with similar issues. My suggestions: • Priority must be to establish the head difference between the spring and the tank – without this, one is truly stuck! • Next given the spring flow and the head difference less a safety margin (the book referred to gives figures), check the pipe size required to deliver the maximum spring flow • Now check velocity in pipe, o if it is lower than the minimum figure in the book, you may get deposits, so will have to ensure there are washouts etc, see below – but do not reduce the diameter unless you are willing to lose some of the flow – usually a bad idea o if it is higher than the maximum figure in the book, you will get pipe damage, so increase the diameter • Air locking is indeed a big issue on low pressure schemes but is not insurmountable! Some strategies we adopted in Rwanda o Have as steep and as constant a downhill gradient as possible immediately below the spring and keep on going until you are below the level of the tank – this may mean a slightly longer pipeline but worth every penny! o Once you are down to that level, try to lay the pipe at a constant fall and then a constant rise of say 5%, with highpoints always below the level of the tank and low points at or near the maximum pressure rating of the pipe o At each high point have a wash out valve and an air valve – the air valve can be as crude as a small hole in an otherwise dead end pipe sticking up from the main line. The wash out valve will allow you to fill the pipe without sucking in air (if you use the washouts at the low points you will suck in air and create more problems than you solve) o At each low point have a wash out valve – I repeat not for filling the pipe – but essential to allow a rapid emptying and therefore cleaning out of sediment in the pipe as well as repairs. o Air locking can occur not only at physical/planned high points but also in accidental high points created by trying to lay pipes horizontal or not paying enough attention to laying them at a constant gradient – it is not rocket science but needs discipline and although the pipe line may appear longer as it will tend to sweep in curves, the actual difference is usually minimal – I can suggest ways to do this if necessary. They can also occur where there is a relative high point in between the pipe line and the hydraulic gradient – so worth plotting out and avoiding • If the pipe can be laid as suggested above, the only place where you might get major problems with half empty pipes are in the first section between the spring and the point where pipe falls ... (more)

edit flag offensive delete publish link more

I'm not an expert on this but it seems that the 90 mm pipe is too big for the spring and far too big for the population to be served and would be a waste of money, unless the plan is to use water for irrigation. If the scheme goes ahead with 90 mm pipe it's always possible to put a gate valve on the discharge end to regulate the flow (or use a float valve in the tank) and have the pipe under pressure. If this is done, or smaller diameter pipe is used, then break-pressure tank(s) may be required as suggested, but that's not technically very complex although there would be some maintenance to plan. If open flow conditions are chosen it could be possible to deal with potential problems by placing some gate valves along the line to regulate flow and reduce likelihood of hammer and maybe some automatic purges to deal with airlocks at high points.

Another good reference would be the gravity flow chapter in the ACF manual 'Water, sanitation and hygiene for populations at risk', which is available on disk.


John Adams

edit flag offensive delete publish link more