Residential Night Consumption - Iwa-waterloss.org

IWA Water Loss 2012
Manila, 26-29 Feb 2012
Residential Night Consumption –
Assessment, Choice of Scaling Units,
Calculation of Variability
Marco Fantozzi, MIYA marco.fantozzi@miya-water.com
Allan Lambert, Water Loss Research & Analysis
Ltd./ ILMSS Ltd

Overview
• Draft best practice terminology for components
of minimum night flow (MNF)
• Estimating Residential night consumption
– Seasonal variations of MNF
– Day of week, time of night, duration of MNF
– Variability and Influence of Sample Size
– Choice of units for Assessed Residential Night Use
– Component analysis approach
– Variability of Estimates using Binomial Distributions
• Case study
• Conclusions

Draft best practice terminology for
components of minimum night flow
Minimum
Night Flow
(MNF)
Night
Consumption
NC
Utility
Night
Leakage
Night
Use
NU
Customer
Night
Leakage
CNL
Bursts
B
Exceptional Night Use
ENU
Assessed Residential
Night Use ARNU
Assessed Non Res. Night
Use ANRNU
Inside Buildings CNLI
Outside Buildings CNLO
Unreported Bursts UB
Reported Bursts RB
(not yet repaired)
UNL Background On service conns BLS
Point of
Delivery
Leakage BL
On mains BLM
Source: Night Flow Analysis Team of IWA Water Loss Specialist Group

Residential night consumption
• Night consumption = Night use + customer night leakage
• traditionally derived from manual readings of limited
number of metered properties at night
– and, more recently, using smart metering studies
Minimum
Night Flow
(MNF)
Utility
• traditionally calculated as an average, then expressed in terms Night of:
– ‘per property per hour’ or ‘per service connection per hour’
Night
Consumption
NC
– ‘per person per hour’ or ‘% of average consumption’ (apartments)
depending on custom and practice in different countries
Leakage
• wide variations in choice of units and average figures used
– UK 1.7 lit/prop/hr, Canada 3 lit/prop/hr, Malaysia 5 lit/prop/hr, …….
– Germany and Austria, 0.4 to 0.8 litres/person/hour
– reasons for different figures and choice of units needs to be better
understood
Night
Use
NU
Customer
Night
Leakage
CNL
Bursts
B
As
Asse
UNL Background On
Leakage BL
Exc
Ins
Outs
Unr
Re
(

Residential night consumption
The increasing availability of Smart Metering technology now provides
excellent opportunities for improving our understanding of night
consumption.
However, the full benefit will not be realised without:
– a practical process for selecting when to measure MNFs to assess
Utility Night Leakage; this also defines times and durations of night
consumption investigations
– an improved understanding of night consumption components and
their variability
– a logical method for selecting appropriate units for night consumption,
depending upon local or national circumstances
This paper proposes some approaches for achieving these outcomes,
with particular reference to assessments of residential night consumption,
for potential application to a wide variety of international situations.

Estimating Assessed Residential
Night Consumption
• For most reliable estimates of assessed residential night
consumption, need sample measurements:
– at times of year when exceptional night use is minimal (seasonal
considerations)
– at hour of night and days of week when residential consumption is
typically lowest
– with number of samples large enough to obtain meaningful results
• For approximate estimates, an average value can be used
• For more detailed estimates, it is preferable to study and understand
the typical largest components of residential night consumption

Zone Inflow (litres / minute)
01/07/2008
29/07/2008
26/08/2008
23/09/2008
21/10/2008
18/11/2008
16/12/2008
13/01/2009
10/02/2009
10/03/2009
07/04/2009
05/05/2009
02/06/2009
30/06/2009
28/07/2009
25/08/2009
22/09/2009
20/10/2009
17/11/2009
15/12/2009
12/01/2010
09/02/2010
09/03/2010
06/04/2010
04/05/2010
01/06/2010
29/06/2010
Pressure (m)
Seasonal variations of MNF
• Minimum Night Flow
varies both
systematically and
randomly throughout
the year
1200
1000
800
600
400
Daily Average Flow (l/min) Minimum Night Flow (l/min) Daily Average Inlet Pressure (m)
50
45
40
35
30
25
20
15
Western
Australia
example
• Identify the seasonal
patterns of MNF in
your systems
200
0
10
5
0
• Select ‘windows of
opportunity’ when
MNFs are at lowest
seasonal values, for
residential night
consumption sample
measurements
Northern
Italy
example

Day of week, time of night, duration
of MNF
MNF Litres/Household/hour
MNF Litres/Person/hour
• MNF varies with
hour of night, and
from weekdays to
weekends/
religious days
• Identify the typical
hours of MNF for
your system; see
this simple table
Australian
example
(Queensland)
In this example, MNFs typically occur 02-03 hrs on weekdays and 03-04 weekends
• Smart metering of
residential night
consumption
confirms that
minimum occurs
later at night at
weekends
5.5
5.0
4.5
4.0
3.5
3.0
2.5
2.0
1.5
1.0
0.5
Weekday, Weekend and 7-Day Week Average Night Consumption,
All Zones, litres/household/hour
0.0
0 1 2 3 4
Time (hours)
Weekends,
all Zones
Weekdays,
all Zones
7-Day Week,
all Zones
Weekday, Weekend and 7-Day Week Average Night Consumption,
All Zones, litres/person/hour
1.50
1.40
1.30
1.20
1.10
1.00
0.90
0.80
0.70
0.60
0.50
0.40
0.30
0.20
0.10
0.00
0 1 2 3 4
Time (hours)
Weekends,
all Zones
Weekdays,
all Zones
All days, all
Zones
Sources: Gold Coast WaterSaver End Use Study, Griffith University, ILMSS Ltd/Wide Bay Water

'
% of people 'active'
% of people 'active;
Gold Coast End Use Study
End-use study on 137
residential properties with high
resolution smart metering
(0.014 litres every 10 seconds)
allows different types of night
consumption to be identified
using ‘Trace Wizard’
Study confirmed
that higher % of
people stay up
later using water
at weekends
compared to
weekday nights
% of people 'active', All Zones, Weekdays Only
25%
20%
15%
Personal Hygiene Total
Use of Faucets
Toilet Flushing
10%
Shower
5%
0%
0 1 2 3 4
Time (hours)
% of people 'active, All Zones, Weekends only
25%
20%
15%
10%
5%
0%
0 1 2 3 4
Time (hours)
Sources: Gold Coast WaterSaver
% of
End
people 'active',
Use
All
Study,
Zones, Weekdays
Griffith
Only
University, ILMSS Ltd/Wide Bay Water
25%
20%

% of service connections in
sample
Variability and Influence of Sample Size
• Record cumulative readings on customer meters around 1 hour apart, to
nearest litre, noting if flow is occurring on each occasion (if possible)
• Most meters may show zero consumption, as residential night consumption
is usually generated by a small % of customers
– the example below is for 128 residential meters with roof tanks in ASEAN region
– 60% of the average of 5.9 lit/conn/hour arises from only 4% of customers!
70%
60%
50%
40%
30%
20%
10%
0%
Average 5.9
lit/conn/hour
0 >0-1 >1-2 >2-4 >4-8 >8-16 >16-32 >32-64 >64-128
Litres/connection/hour
% of meters in
sample
% contribution
to average of
5.9 lit/conn/hr
So averages of small sample sets are HIGHLY VARIABLE
These are NOT ‘normal’ distributions, so use Binomial or
Multinomial statistical distributions for advanced analysis

What are the estimates being used for
• Using night flows for targeting active leakage
control interventions
– required level of accuracy of assessed night
consumption decreases as ratio of assessed night
consumption to MNF increases
• Using night flows to check water balance
calculations, or other detailed analysis
– higher standards of accuracy required
– other parameters (average zone pressure, Night-Day
Factor) also important

Component analysis approach
• Identify probable main components of residential
night consumption for your systems
• Identify night use components that are directly
related to number of residents (‘people-based’)
– use ‘% active’ x ‘volume per event’ approach
• Classify the remaining components as ‘propertybased’,
being broadly independent of the
number of residents
– mainly leakage inside and outside buildings

Basic component analysis approach to
assessed residential night consumption
People-based components of
residential night consumption
Average
volume
% of people
active at time of
MNF
Average
litres
Toilet flushing 5.0 3.0% 0.15
Use of faucets 1.0 3.0% 0.03
Shower/bath 75 0.1% 0.08
0.26
l/person/hr
Property -based components Average % of properties
of residential night volume active at time of
Average
consumption
litres MNF l/property/hr
Dishwasher 5 0.1% 0.01
Clothes washer 50 0.1% 0.05
Water softener 0 0.0% 0.00
Leakage inside buildings 0.25
Leakage outside buildings 0.25
0.56
By using a seasonal
low MNF measured
on weeknights during
the typical hour of
the lowest MNF, the
components that are
most variable and
difficult to estimate
are minimised
If there are 3.0 occupants per property, average assessed residential night consumption is:
• 3.0 x 0.26 litres/person/hour + 0.56 litres/property/hour = 1.4 litres/residential prop/hour
• 0.26 litres/person/hour + 0.56/3 litres/property/hour = 0.45 litres/person/hour

Choice of units for Assessed
Residential Night Use
• If customer night leakage is low, and the majority of residential night
consumption arises from ‘people-based’ components
(toilet flushing, use of faucets, showers etc)
then ‘litres/person/hour’ is a logical unit to use, provided the residential
population is reliably known
– Austria and Germany are examples of this
• Otherwise, litres/property hour or litres/service connection/hour is
more logical
– UK, with many unmetered or internally metered properties, is such an
example
• But using the component-based approach, why not calculate both
and use whichever is most appropriate to local circumstances

Variability of Estimates using
Binomial Distributions
• Using toilet flushing as an example
– If N is the number of people in the Zone (e.g. 3000)
– p is the average % of ‘active’ people that flush a toilet during the hour of
MNF (e.g. 3 %).
– average number ‘active’ on a single night is N x p = 3000 x 3% = 90
people
• Using the Binomial, the standard deviation of the number of
residents ‘active’ on a single night is easily and quickly calculated as
√ {N x p x (1-p)} = √ {3000 x 3% x 97%} = 9.3 people
– so the 95% confidence limits for the number of people flushing a toilet is
90 +/- 2 x 9.3 = 90 +/- 19
– if the average toilet flush volume is 5 litres, the component of toilet
flushing is 450 litres +/- 95 litres, or 0.15 +/- 0.032 litres/person/hour
• A more detailed explanation is provided in the paper

Summary
• A WLSG draft terminology for Minimum Night Flow components is now available
• Selection of MNFs for detailed analyses should be based on lowest seasonal, day of
week and time of Night Consumption over a fixed hour
• Review annual pattern of daily inflows to identify seasonal ‘windows of opportunity’
• Identify fixed hour of MNF during windows of opportunity; if weekend and weekday
MNFs have different characteristics, use only weekday MNFs
• Smart Metering provides opportunities for better understanding of Night Consumption
• Attempt to separate Customer Night Leakage from Night Use
• Calculate Assessed Residential Consumption using component analysis
• Choice of units: litres/person/hour is appropriate if personal night use is the dominant
component of Night Consumption; otherwise use litres/property/hour or
litres/connection/hour.

IWA Water Loss 2012
Manila, 26-29 Feb 2012
Residential Night Consumption –
Assessment, Choice of Scaling Units,
Calculation of Variability
THANKS !
Marco Fantozzi, MIYA marco.fantozzi@miya-water.com
Allan Lambert, Water Loss Research & Analysis
Ltd./ ILMSS Ltd