Shelley, Samuel Robert

Dr Shelley worked toward his PhD as part of the EPSRC Centre for Doctoral Training in Metamaterials at the University of Exeter, graduating in 2018 with his thesis The Control of Fluid Flow Using Metamaterial Concepts. Working as a postdoctoral researcher for the University he then investigated the interaction of turbulence with acoustic surface waves. In 2020 he joined nuron as a data scientist under a KTP project with the University of Sheffield before accepting the lead scientist role where he is responsible for the development of measurements and the delivery of data science algorithms.

Titles from this speaker

Distributed Optical Measurements Enable Smart CSO Management

Abstract

Wastewater intermittently enters our rivers, lakes and oceans via storm overflows at an alarming volume and frequency. This is a problem we face all around the world, for example in the UK only 14% of rivers meet good environmental standards. Significant investment is being made to tackle this issue, as a result of public and or regulatory pressure, and because we all know that we need to do better. However, to really make an impact and improve performance, wastewater collection system operators need to understand what is happening within their networks. Not just at sewer outflows or every 1km or so where a discrete monitor is placed, usually only providing a low frequency measurement of either flow or depth. If operators know what is happening throughout their system in real time they can learn to predict and optimize in real time. Creating a continually accurate digital twin of existing wastewater infrastructure makes this possible. Such 'Digital twins' enabled by the ability to rapidly monitor what is currently occurring in the system and predict what will happen. Having both capabilities allows for the smart allocation of resources and minimizes the risk of undesirable events. To produce and maintain an accurate data-driven digital twin requires dense (spatial and temporal) data about the network are required to drive and enable the smart networks. Such data is not currently available to sewer network operators but if available could unlock substantial savings and performance improvements. In this presentation we will cover the two main options for increasing the density of in-sewer monitoring to the level required for a digital twin to provide reliable information: 1. discrete sensors, and 2. distributed sensors. A comparison between these two types of sensors will be provided to highlight the benefits of the distributed sensor approach. Results from a case study utilizing a distributed sensor to monitor the conditions in a sewer will also then be discussed. Telemetered discrete sensors as mentioned above are able to monitor at a single location only, generally access chambers, providing information about limited parameters at that location. In gravity fed sewer networks this leaves the rest of the network unmonitored. Most discrete sensors measure only depth whilst a few measure flow. A reasonably low-cost option to purchase if only a few sensors are required, although the need for regular maintenance visits confirm that is actually a false economy. Discrete sensors can become very expensive when deployed for scale solutions. Equally, the communication and data management infrastructure required to operate the network of sensors quickly becomes unwieldly as each sensor is a separate object. For example, each sensor would need its own power supply, which would need replacing periodically, and would transmit data back independently requiring a separate correlation process to reveal the full picture. Even so, as the data points are isolated and local, in a gravity fed sewer the majority of the monitored network will still be invisible, thus requiring inference from nearby sensors which will then reduce the overall accuracy of the systems as a whole. Distributed sensors in contrast are continuous and in pipe, and therefore able to monitor the entirety of the network they cover. This means that the need for inference in the system is minimized as the sensor is providing information everywhere. A single data acquisition hub is able to monitor over 100km. Equally the acquisition infrastructure is located at a single point above ground. This means power only needs to be provided at one location rather than hundreds or thousands of separate ones. Data is then also transmitted from that single point making it easy to correlate results and gain insight from the acquired data. As the power requirements are easier to meet, the data is able to be transmitted continuously allowing for real time monitoring. Such systems can also provide other benefits. For example, the distributed measurement system developed by nuron has a dual purpose to provide fibre networking for broadband and 5G connectivity. Distributed fibre optic sensors operate by sending laser pulses down the length of the contained fibre. These pulses interact with natural imperfections in the fibre to produce backscatter whose properties such as amplitude, frequency and phase are affected by the temperature and strain experienced by the fibre. By measuring the properties of the backscatter, the conditions around the fibre can be measured and, by time of flight, the location of each measurement point can be found. This potentially provides a map of the flow conditions experienced within the entire collection network. One of the most useful outcomes of a distributed sensing system is to continuously validate and drive a digital twin such as that described above. For example, in this presentation we demonstrate the detection of high pipe occupancy and a blockage, both key to prevention of CSO spill events. Figure 1 shows the measured flow rate in a real sewer environment over the course of a single day. Normal nighttime behaviour of the network can first be observed consisting of almost no flow. As people begin to wake up at 6.30 this transitions to normal daytime flows. A heavy rainfall event occurs at about 15:00 leading to a sharp increase in flow rate which could lead to a CSO spill event. Figure 2. shows a web interface used to alert users to the presence of a blockage in the system which could also lead to CSO spill event. To simulate blockage situations a sensor was installed into the ICAIR test facility operated by the University of Sheffield. This allows us to conduct experiments that wouldn't be possible to do in active sewer environment. Such a system can be used to produce the required high quality dense data that a smart city needs to develop a digital twin. As the wastewater network is not an isolated system, other data sources, such as rainfall, are integrated into the algorithms to produce accurate predictions of the behaviour and needs of the wastewater network. By using historic data, the predictions of the system can be validated to ensure their accuracy. Equally, underlying trends of the system can be identified and accounted for. Machine learning and AI are used to interpret the large data volumes such a system produces to provide meaningful and actionable insights that customers require.

This paper was presented at the WEF Collection Systems Conference in Detroit, Michigan, April 19-22.

SpeakerShelley, Samuel Robert

Presentation time

10:45:00

11:15:00

Session time

8:30:00

11:15:00

Session number13

Session locationHuntington Place, Detroit, Michigan

TopicCombined Sewer Overflow, Flow Monitoring, Sensors

Author(s)