Water discolouration is becoming
an increasingly difficult problem for water companies to manage. In 2013 alone,
2 million UK customers were affected by water discolouration. Rising customer
expectations, tighter regulatory demands and ageing water systems means that
predicting, managing and reducing water discolouration is a key priority for
the improvement of water quality provision to consumers in the UK.
How does water discolouration occur?
The formation of water discolouration
is a complex process and is not completely understood. Water discolouration is
believed to be primarily caused by the mobilisation of discolouration material.
This is thought to be caused by hydraulic changes detaching and transporting
the discolouration material through the network, resulting in discoloured water
at the tap. Trunk mains (the motorways in a water transport network) are
thought to be at especially high risk as their large size acts as a reservoir,
allowing discolouration material to build up.How is water discolouration currently managed?
At present, water discolouration is
managed in a reactive way – once a sufficient number of customers have notified
their water provider of discoloured water in their area, the water company
responds by cleaning parts of the water network to try and alleviate the issue.
Trunk mains cleaning programs, where trunk mains are periodically cleaned in an
attempt to reduce discolouration material build-up, are also a current
management tool for water discolouration. However, these programs are expensive
and difficult to implement without disrupting supply and so to reduce costs for
both the water company and consumers, this only occurs infrequently. Therefore,
there is a strong need to develop new methods that predictive where water
discolouration might occur and provide the intelligence required to plan
preventative maintenance.
How can water discolouration be managed better?
STREAM researcher Gregory Meyers has
developed a new, data-driven method of forecasting water discolouration. The
developed system can detect the mobilisation of discolouration material and
estimate if sufficient turbidity will be generated to exceed a preselected
threshold and approximate how long the material will take to reach a downstream
meter and taps. This new method could therefore be used as an early warning
system, which will allow water companies to deal with water discolouration
proactively rather than reactively. In addition, the method is cheaper than
traditional periodic trunk mains cleaning programs.
The method designed by STREAM
researcher Gregory Meyers is a data-driven modelling approach. In order to
predict water discolouration, three processes need to be considered by the
model. The model must detect if sufficient hydraulic force capable of
mobilising discolouration material has occurred. Secondly, the model must be
able to assess the resultant turbidity. Finally, it must be able to estimate
where mobilisation of discolouration material has occurred and how long it will
take to reach downstream turbidity meters. Meyers compared two different
modelling approaches and three different machine-learning methods using flow,
turbidity and hydraulic data collected from a trunk main network. The best
performing model was able to reliably predict turbidity up to 5 hours ahead of
its detection at downstream meters.
In order to further validate this
method, the method needs to be tested on multiple water systems. Once
validated, the method can be used to improve management processes regarding
water discolouration. This will reduce the amount of water discolouration
experienced by customers throughout the UK.
For full reference:
Meyers, G., Kapelan, Z., Keedwell, E., 2017. Short-term forecasting of turbidity in trunk main networks. Water Research 124, 67–76.
