How much incoming water supply does a building need?
What you’ll learn
By the end of this article, you will understand:
how to estimate incoming water demand at feasibility stage
the difference between a directly mains-fed system and a stored-and-boosted system
how occupancy and sanitary fittings can be used to estimate cold-water storage
how loading units can be used to estimate diversified demand
why the chosen water-supply arrangement can significantly change the instantaneous demand placed on the external network
what information is worth carrying forward into the next design stage
The article deliberately stays at feasibility level: the aim is not to complete the final hydraulic design, but to establish a sensible basis for storage, incoming flow and system arrangement.
When we looked at gas, the question was relatively straightforward: establish the connected gas load and arrive at an incoming gas demand.
Water is slightly different because before we can decide what the incoming supply needs to provide, we need to understand how the building is going to be supplied.
At its simplest, we might have a building supplied directly from the incoming mains, or we might have an incoming mains supply filling a cold-water storage tank, with water then distributed around the building from that tank, normally using a booster set (although this can also be achieved through gravity feeding the outlets). I have worked on food production facilities whereby the term ‘towns water’ is used for mains water.
There are plenty of variations on those arrangements. Some outlets may remain directly mains-fed while the remainder of the building is tank-fed. Storage may be at high or low level. There may be several pressure zones in a tall building. There may be separate potable and non-potable systems. Potable water is water that is suitable for human consumption — in other words, water of drinking-water quality. In UK regulations you may also see this described as wholesome water.
Non-potable water is water that is not intended to be used for drinking or other uses requiring potable-quality water. It may, for example, be used for WC flushing, irrigation or certain process applications.
Where potable and non-potable systems are provided within the same building, they must be kept properly separated so that there is no risk of the non-potable system contaminating the potable water supply. This will be covered later in backflow prevention article.
We are not going into all of those arrangements here.
What I want to establish is much simpler:
How much water does the building need, and what demand does that place on the incoming water supply?
I tend to approach that using the same three questions:
1. What are we trying to establish?
2. Why do we need to know it?
3. How do we calculate it?
WHAT are we trying to establish?
At feasibility stage, I normally want to arrive at one of two answers.
If the building is going to have stored cold water, I want to know approximately how much usable water storage is required and how quickly the incoming main needs to refill it.
If the building is going to be supplied directly from the mains, I want to know the likely peak diversified flow rate required from the incoming supply.
Those are two quite different demands.
A building might momentarily require 2 or 3 litres per second internally when a number of appliances are being used at the same time, but that does not necessarily mean that the water authority must provide 2 or 3 litres per second directly from its network.
If we put a storage tank between the network and the building, the tank can deal with the short-term peaks while the water main replenishes it much more slowly.
That distinction becomes important when we are making a new water application or investigating whether an existing connection can be reused.
The exact information requested by a water undertaker varies, but eventually somebody is going to have to put some engineering behind the application. We may be asked for a peak flow rate, anticipated demand, proposed connection size, storage volume, fitting schedule or some combination of these.
Writing “office building” on an application form is not a design calculation.
We need to know what the building is likely to do.
What guidance are we using?
For current UK design work, the main framework is the BS EN 806 series, together with BS 8558, which provides complementary UK guidance. BS EN 806-3 contains a simplified method for sizing water distribution pipework. The old BS 6700 has been withdrawn, although some of the traditional UK calculation material survives through BS 8558 and continues to be familiar to engineers who have been doing this for a while.
CIBSE Guide G, Public health and plumbing engineering, is also extremely useful when looking at water usage, storage, diversity and the practical design of systems.
There is one thing worth saying here because it can catch people out.
Do not mix loading-unit systems.
A loading unit is not a universal physical unit such as a litre per second. Different design methods can assign different numerical loading units to the same appliance and then use a corresponding table or curve to convert those loading units into flow.
If you use the appliance values from one method, use the diversity or conversion method that belongs to it. Do not take loading units from one table and then read the flow from a completely different system.
“Do not mix loading-unit systems. Use the appliance values and diversity relationship from the same method.”
For the worked example later in this article I am using the traditional UK loading-unit approach that I have used on projects, rather than attempting to reproduce the full BS EN 806-3 calculation method.
Under this approach, the connected sanitary appliances are assigned loading units and the total loading-unit value is converted to a probable design flow rate using the relationship historically given in BS 6700. That design flow can then be used to size the distribution pipework against sensible hydraulic limits for pressure drop and water velocity.
For the quick-selection table below, I have taken the corresponding flow rates and checked the copper pipe sizes against established hydraulic pipe-sizing data. The selected pipe size is limited to approximately 500 Pa/m maximum pressure loss, with a maximum water velocity of about 1.0 m/s up to 54 mm pipework and about 1.5 m/s for 67 mm and above.
That produces the following practical sizing table:
Maximum loading units (LU) | Copper pipe size (mm) | Approx. design flow (l/s) | Approx. velocity (m/s) |
— | 15 | 0.090 | 0.62 |
9 | 22 | 0.269 | 0.84 |
28 | 28 | 0.545 | 1.01 |
55 | 35 | 0.840 | 1.00 |
95 | 42 | 1.230 | 1.00 |
200 | 54 | 2.100 | 1.00 |
600 | 67 | 4.880 | 1.50 |
900 | 76 | 6.300 | 1.50 |
2,500 | 108 | 13.000 | 1.50 |
4,300 | 133 | 19.900 | 1.50 |
7,000 | 159 | 28.300 | 1.50 |
The important point is that this is not intended to be a replacement for the underlying standards or hydraulic calculation. It is a practical quick-selection method based on the traditional UK loading-unit relationship and normal pipe-sizing criteria. Once a system becomes more complex, has long index runs, limited available pressure, unusual fittings or several pressure zones, the complete system should still be checked hydraulically.
WHY do we need to know the incoming water demand?
The obvious reason is that we need to establish whether the water supply is adequate.
On a new development, that allows the water undertaker to assess the proposed connection against its network.
On an existing site, perhaps with a capped supply or a building being refurbished or extended, we need to establish whether the existing connection has enough capacity for the proposed use.
But there is another reason.
The way in which we supply the building can change the peak demand imposed on the external network quite dramatically.
A directly mains-fed building places its diversified peak demand onto the water network as it occurs. A stored-water system does not.
The building takes water from its tank and the network simply replenishes that tank over a period of time.
The tank is effectively a buffer between the building demand and the infrastructure outside.
It does not reduce the amount of water the building eventually consumes. What it can reduce considerably is the instantaneous flow rate demanded from the network.
“A storage tank does not reduce how much water a building uses. It changes when the water is demanded from the network.”
Do we need mains water or stored and boosted water?
Sometimes, at feasibility stage, we simply do not know.
We may not yet have pressure and flow test information. We may know that there is a water main outside the site, but that tells us very little about what pressure and flow will be available at the building under design conditions.
Ofwat's guaranteed minimum service standard is only seven metres static head, or approximately 0.7 bar, in the communication pipe (the water authorities pipe into our building). That is a minimum service standard, not a sensible design assumption for a commercial building. Once we allow for the height of the building, pipework resistance, valves, meters and the minimum pressure required at the outlets, the available pressure can disappear fairly quickly. (Ofwat)
For that reason, if the pressure and flow have not yet been established and I am looking at anything beyond a straightforward low-rise building, I would normally make an allowance at feasibility stage for cold-water storage and boosting.
If later testing demonstrates that the mains can supply the building directly at the required flow and residual pressure, we can reconsider it.
That is a much safer direction to work in than designing the building on the assumption that the main will be adequate and discovering later that we need space, power and drainage for a tank and booster set that nobody allowed for.
There is also a regulatory point worth knowing. Under the Water Supply (Water Fittings) Regulations in England and Wales, installation of a pump or booster drawing more than 12 litres per minute ( 0.20 l/s) requires notification to the water undertaker. (Water Regs UK Limited)
Again, this is why I prefer to understand the supply strategy early rather than leave it until somebody starts sizing pipes.
HOW do we calculate it?
I am going to use a fictional office building so that we can follow the calculation from very little information through to a reasonable feasibility estimate.
Assume we have:
Gross internal floor area (GIFA) : 5,000 m²
Eventually we know that the building contains:
36 WCs
42 wash-hand basins
15 urinals
3 sinks
3 showers
We are going to look at three calculations.
First, an occupancy-based cold-water storage calculation.
Second, an appliance-based cold-water storage calculation.
Then a loading-unit calculation for a directly mains-fed building.
The useful thing is not simply getting three numbers. It is seeing whether the different approaches tell a coherent engineering story.
HOW — Method 1: cold-water storage from occupancy
At the earliest stage of a project, we may have little more than a floor area and a building use. That does not mean we cannot do anything. We can work forwards.
For an office, Approved Document B currently gives a floor-space factor of 6 m² per person for calculating occupant numbers where a more appropriate known occupancy is unavailable. It is primarily fire-safety guidance, so if we have an actual design occupancy we should of course use that instead, but it gives us a perfectly useful early-stage starting point. (GOV.UK)
For our 5,000 m² office:
5,000 m² ÷ 6 m²/person = 833.3 people
Round that up:
Design occupancy = 834 people
CIBSE Guide G gives indicative water storage figures for different types of buildings. For an office without canteen facilities, a figure of 40 litres per person is given; with canteen facilities the figure is higher. The important point is not to regard these figures as some immutable law of nature. CIBSE itself makes clear that cold-water storage needs to be assessed on a project-by-project basis. (StudyLib)
Using 40 litres per person:
834 people × 40 litres/person = 33,360 litres
That gives us the underlying daily design quantity from which we can establish an appropriate stored volume.
Now we need to make an engineering decision about how much of that demand we actually want to store.
For this worked example, we are going to assume an 8-hour equivalent storage provision. As a simple preliminary-design allowance, this is taken as one-third of the calculated daily demand:
33,360 × 8/24 = 11,120 litres usable storage
This is an engineering design assumption for the worked example, not a statement that the building uses water uniformly over a 24-hour period.
I prefer the word usable here because it helps distinguish the amount of water we actually need from the physical size of the tank.
A 12,000-litre tank does not necessarily contain 12,000 litres of usable water in service. We need to allow for the float-valve shut-off level, warning and overflow arrangements, low-water level, any unusable volume below the effective outlet and the manufacturer's actual tank geometry.
At feasibility stage, a useful rule of thumb might be to assume that the usable water volume is approximately 80% of the nominal tank capacity.
That is not something I would blindly use to order a tank. At detailed design I would check the manufacturer's actual usable capacity.
But for feasibility:
11,120 ÷ 0.80 = 13,900 litres nominal
So our first approach tells us:
Usable cold-water storage ≈ 11.1 m³
Nominal tank capacity ≈ 13.9 m³
Call it about 14 m³ nominal at this stage.
That is already enough to start talking sensibly about plantroom space, floor loading and booster-set provision.
HOW — Method 2: cold-water storage from appliances
As the project design phase progresses, we may end up knowing the sanitary fitting schedule.
That gives us another way of looking at the building.
For our example we have:
Appliance | Quantity | Storage allowance used for example | Calculated storage |
WC | 36 | 180 L | 6,480 L |
Wash-hand basin | 42 | 90 L | 3,780 L |
Urinal | 15 | 110 L | 1,650 L |
Sink | 3 | 135 L | 405 L |
Shower | 3 | 185 L | 555 L |
Total | 12,870 L |
The individual allowances here are the fitting-based values adopted for this worked example. On a real project I would always identify the particular design reference being used rather than create a generic spreadsheet labelled simply water calculation and allow it to survive for twenty years with nobody remembering where the numbers came from.
The result is:
12,870 litres usable storage
Again applying our preliminary 80% usable-to-nominal assumption:
12,870 ÷ 0.80 = 16,088 litres nominal
So, say:
Nominal tank capacity ≈ 16 m³
Now this becomes interesting.
Using occupancy, we arrived at approximately:
14 m³ nominal
Using the actual fitting schedule, we arrived at approximately:
16 m³ nominal
They are not identical, and I would not expect them to be.
They are completely different ways of estimating the same underlying building demand.
What matters to me at feasibility stage is that one method has not given me a 5 m³ tank while the other has given me a 30 m³ tank.
Both calculations are telling me that this building probably wants a tank somewhere in the region of 14–16 m³ nominal capacity.
That gives me considerably more confidence in the answer.
I would probably carry something in that range through the next stage and refine it once the occupancy, actual fittings, client resilience requirement, water-company information and tank arrangement become clearer.
This is what early-stage engineering often looks like.
You do not always have perfect information.
What you want is two or three sensible routes to an answer that broadly agree with each other.
“Early-stage engineering is not about having perfect information. It is about finding two or three sensible routes to an answer that broadly agree.”
How much incoming mains water is required to fill the tank?
Knowing the tank volume does not finish the calculation.
The water authority is interested in what we are taking from its network.
So we need to establish how quickly the tank needs to refill.
Let us use a convenient 13,000 litres usable volume, which is approximately the result of our appliance calculation.
We now need to establish the flow rate required to refill the tank. BS 8558 states that, where other information is unavailable, a maximum filling time of four hours may be assumed for a cold-water storage cistern.
For this example we will therefore assume that the tank is required to refill within four hours.
The calculation is simply:
13,000 litres ÷ 4 hours = 3,250 litres/hour
Then convert litres per hour to litres per second:
3,250 ÷ 3,600 = 0.903 l/s
Call it:
0.90 l/s
That is the important number as far as the external network is concerned.
The building itself may be drawing water from the tank at considerably greater instantaneous rates, but if we can replenish the storage over four hours, our nominal incoming refill requirement is only about:
0.90 litres per second
There is a qualification here.
That calculation is the basic tank recovery rate. If the project requires the tank to recover fully within four hours while the building remains heavily occupied and continues using water, then I would look at the demand and recovery profile more carefully rather than assume that dividing the tank volume by four is the end of the exercise.
But for a feasibility calculation, 0.90 l/s gives us a useful network demand.
So our stored-water solution is currently looking like:
Usable storage: approximately 11–13 m³
Nominal tank capacity: approximately 14–16 m³
Incoming refill rate: approximately 0.90 l/s
HOW — What if the building is directly mains-fed?
Now let us remove the tank.
The water main itself has to respond to the building's instantaneous demand.
We therefore need a way of estimating how many appliances are likely to operate simultaneously.
Simply adding every tap and appliance at its full flow rate would produce an enormous and unrealistic number because 36 WCs, 42 basins, 15 urinals, three sinks and three showers are not all going to operate at exactly the same moment.
This is where loading units and diversity come in.
The loading-unit method gives each type of outlet a numerical value that reflects not only its flow but also how frequently and for how long it is likely to operate.
We total those loading units and then use the corresponding diversity relationship to convert them into a probable simultaneous flow.
As a cross-check, the 2004 edition of CIBSE Guide G: Public Health Engineering, Table 2.2, gave indicative 24-hour cold-water storage allowances based on the number and type of fittings installed. Typical values included 180 litres per WC, 90 litres per basin and 110 litres per urinal, with ranges of 90–180 litres per sink and 140–230 litres per shower. For this example, I have used the midpoint of those ranges for sinks and showers.
For this example, using the traditional UK loading-unit values adopted for the calculation:
Appliance | Qty | Cold LU each | Cold LU | Hot LU each | Hot LU |
WC cistern | 36 | 2 | 72 | – | – |
Wash-hand basin | 42 | 1.5 | 63 | 1.5 | 63 |
Urinal | 15 | 0* | 0 | – | – |
Sink | 3 | 3 | 9 | 3 | 9 |
Shower | 3 | 3 | 9 | 3 | 9 |
Total | 153 LU | 81 LU |
*The urinal demand is ignored in this particular worked example because of the flushing arrangement assumed. That should not simply be copied to every project; directly supplied flushing valves or different urinal systems need to be considered on their own merits. For the wash-hand basins I have retained the 1.5 LU value used in my traditional project calculations. Higher loading-unit values may be adopted where a pronounced peak period of use is expected, so this should be considered against the actual building use rather than copied automatically.
The cold-water distribution demand is therefore:
153 LU
The hot-water system also ultimately receives its water from the incoming cold-water supply.
In the approach I have traditionally used at feasibility stage, I therefore include the downstream hot-water demand when considering the total duty through the building's cold-water incomer:
153 LU cold + 81 LU hot = 234 LU
Using the corresponding loading-unit-to-flow relationship for the same calculation method gives approximately:
234 LU ≈ 2.30 l/s
That is our directly supplied incoming water demand.
And it is important that we use the conversion curve or table belonging to the same loading-unit method that produced the 234 LU. As I mentioned earlier, do not take these traditional UK loading-unit values and then apply an unrelated BS EN 806-3 table to them. That is mixing calculation methodologies.
At detailed design I would then take the required flow and available pressure and carry out the hydraulic calculation properly, working from the incoming connection through the meter, valves, backflow protection, pipework, elevation and fittings to establish whether sufficient residual pressure remains at the most hydraulically remote outlet. We will discuss this in another article.
That is a separate calculation.
For now we simply want the demand.
And that demand is approximately:
2.30 l/s
So what has the tank actually done?
We can now put the two arrangements alongside each other.
For a directly mains-fed building:
Incoming peak demand ≈ 2.30 L/s
For our stored-water arrangement:
Incoming tank refill demand ≈ 0.90 L/s
The relationship is:
2.30 ÷ 0.90 = 2.56
So in this example the directly mains-fed arrangement places approximately 2.5 times the instantaneous flow demand on the water network compared with filling the cold-water storage tank over four hours.
That is quite a difference.
And this is the point of the exercise.
The tank has not magically reduced how much water the occupants use.
It has changed when we need the water from the network.
The building can take relatively large short-term flows from its own stored supply while the water main replenishes that storage at a much lower and steadier rate.
That can make a substantial difference when assessing an existing connection or discussing a new supply with the water undertaker.
What would I actually carry at feasibility stage?
For this example I would probably leave the calculation with the following design information:
Occupancy: approximately 834 people
Nominal cold-water storage allowance: approximately 14–16 m³
Usable stored volume: approximately 11–13 m³
Tank refill requirement: approximately 0.90 l/s based on four-hour recovery
Direct mains demand: approximately 2.30 l/s using the adopted loading-unit method
Then I would record the assumptions.
I would state where the occupancy came from, what storage allowance I used, what resilience assumption I made, where the appliance quantities came from, which loading-unit methodology was used, what tank usable-volume ratio I assumed and what recovery period I adopted.
That way, when better information arrives, somebody can actually update the design rather than reverse-engineering a spreadsheet full of unexplained numbers.
There may also be other loads to add.
A commercial kitchen, process equipment, irrigation, specialist equipment, water treatment plant, humidification or other fixed demands should not disappear simply because they are not shown on the sanitary fitting schedule.
This is why I regard these calculations as an engineering framework rather than a recipe.
Start with what you know.
If all you have is a floor area, estimate the occupancy.
If you know the occupancy, you can start looking at likely water storage.
Once you know the appliances, check the answer from the fitting schedule.
If the building might be directly mains-fed, convert the appliance demand into a diversified peak flow using a recognised and internally consistent loading-unit method.
Then compare the answers.
For our 5,000 m² office, two different approaches suggest a cold-water tank of roughly 14–16 m³ nominal capacity, while the difference between stored and directly mains-fed supply changes the external network demand from about 0.90 l/s to 2.30 l/s.
That is enough information to have a meaningful conversation with the water undertaker, assess an existing supply and make sensible space and plant allowances in the building.
We have not sized a single domestic water pipe yet.
And at feasibility stage, we do not need to.
We have answered the question we actually started with:
How much incoming water supply does the building need?
What you’ve learned
By working through the example, we have established that there is more than one sensible route to estimating water demand.
For the 5,000 m² office example:
the occupancy-based method suggested approximately 14 m³ nominal cold-water storage
the fitting-based method suggested approximately 16 m³ nominal cold-water storage
a stored system required an incoming refill rate of approximately 0.90 l/s
the same building supplied directly from the mains produced a diversified incoming demand of approximately 2.30 l/s
The important lesson is not just the numbers.
A storage tank does not reduce how much water the building uses. It changes when the water is demanded from the external network. That distinction can have a major effect when assessing an existing supply, making a water application or deciding whether storage and boosting should be allowed for at feasibility stage. Pasted markdown
The wider design lesson is equally important:
Start with what you know, state your assumptions, use internally consistent calculation methods, and compare more than one route to the answer where you can.
Best wishes - Kevin
Comments