Outcomes, Insights, and Best Practices from IIC Testbeds: Intelligent Urban Water Supply Testbed
equipment built from various manufacturers
over a long span of time, interoperability is a
big challenge and standardization is clearly
important. Learning from this testbed in this
area provides valuable input as WPG works
in a number of standards in the water supply
industrial vertical in China, as one of leading
contributors.
remote monitoring capability, for example,
manual on-site inspections can be reduced
with substantial savings in human resources.
To be successful in analytics, enough data
must be accumulated so that reliable models
can be built to establish the operational
norms from which anomalies can be
detected, faults diagnosed and finally
predicted. Root cause analysis and
recommended repair procedures are also
included as a part of the work flow.
Predictive maintenance is still at a very early
state because large amounts of data must be
collected to enable reliable model building.
The testbed is exploring building machine
learning models to identify machine running
states and anomalies.
Because we cannot afford to wait for
standards to be implemented and replace all
of the equipment before implementing IIoT
systems, it is essential to be able to adapt to
the existing equipment that has already
been installed. The testbed provides a
flexible data adaptation layer to transform
data so analytics can be performed in a
normalized format.
Progress has been made in monitoring
equipment energy efficiency, currently on a
pump-by-pump basis. The next step is to
compare it to their peers’ normalized usage
patterns when a sufficient number of pumps
are connected. The goal is to increase energy
savings by 30% from the current baseline.
T ESTBED O UTCOMES
From the use case perspective, the testbed
plan is divided into 5 stages. After
establishing the system and connectivity to
the equipment, the first stage seeks to
monitor the operational state of the
equipment, then to identify anomalies and
perform diagnostics. In the second stage, the
testbed
moves
toward
predictive
maintenance. The third phase involves water
quality monitoring and analytics. The fourth
stage focuses on stored water quality
improvement and balancing the water
demand and supply. The fifth stage looks at
the overall business model.
The testbed has connected to a number of
pumps equipped with water quality
monitoring capability. It is able to monitor
quality measurement and receive an alert if
the quality is degraded. At this time, the
testbed does not yet have instrumentation
broad enough to enable model-building to
infer water quality across a network.
Based on feedback from the customers, the
testbed has been working in a number of
areas that are not explicitly outlined in the
testbed proposal however are found to be
valuable to the customers. These include
features to:
Across
industry
sectors,
predictive
maintenance has been considered the key
IIoT application. However, visibility into the
operation states of the remote equipment,
e.g. if and how well they are working,
already offers value to the operators. With
- 32 -
November 2017