UDG Spring Conference Birmingham 2016

Transcription

1 Use of Monitoring Equipment to Proactively Manage the Waste Water Network Presenters James Mason (RPS), Joanna Kelsey (Severn Trent Water) Contributors Andrew Bailey (RPS), John Hateley (RPS) Ruth Clarke (Innovyze), Richard Body (Innovyze), Phil Gelder (RAA) Summary During AMP5 Severn Trent Water have significantly invested in network monitoring to proactively identify blockages and prevent pollution incidents. For AMP6, Severn Trent are building on the success of this programme to implement a wider scale monitoring programme with a more automated approach to data analysis and proactive incident identification. This paper describes the strategy, why the business believes that investing in a Smart wastewater network is the right approach, and describes the approach to selecting monitor locations and equipment to maximise the benefits delivered and outcomes achieved from this investment. Introduction Imagine a world where we are able to predict and prevent failures so that all of our customers receive reliable water and waste water services. Seven Trent Smart Networks Vision Proactive management of the wastewater network is a key workstream within Severn Trent Water s broader strategy to develop a smart network and an approach that will optimise reactive responses to system issues and incidents through allowing the business to predict where problems may occur or are imminent. There are many components to a smart waste water network; this paper will focus on describing the overall strategy and the results of a trial study in 2015 to test monitor site and equipment selection methodologies. A unique facet of this project was an intention from the outset to involve staff from Severn Trent Network Control. It was considered that the final solution to network monitoring would sit within that department. By placing the analytics solution in Network Control a direct link between the monitor alarms and onsite crews could be achieved. As the majority of Network Control staff has a predominantly clean water background, and no experience of InfoWorks, it was important to ensure that the solution met the requirements of Network Control to handle the additional alarms efficiently. The trial catchment The trial catchment selected was West and Southwest Birmingham, within the Minworth (Birmingham) catchment. The primary focus of the trial catchment was the Aston and Handsworth Drainage Area as this has one of the highest rates of blockage incidents in the Severn Trent region. Overall, 20 Monitors were located in the drainage area, primarily targeted at identifying blockage formation. The other main areas of the interest were in the Bournbrook and Upper Rea Main Drainage Areas. These catchments were reported to have a relatively high number of flooding and pollution drivers, and were located adjacent to the Aston and Handsworth Drainage Area. 26 monitors were located in this area, targeted to assess potential pollution incidents (from dual manholes and misconnections) or hydraulic flooding incidents. 1

2 Walsall Town Centre Aston and Handsworth DA Birmingham City Centre Site selection Flooding Other Causes (FOC) risk sites An assessment was initially undertaken in InfoNet of key factors associated with FOC incidents, reported blockages, reported flooding incidents, repairs, CCTV survey data and demographic data from ONS Census. Each of these datasets was then applied to InfoNet objects for either the pipes or properties. A prioritised list of pipes was created based on a risk score determined for each pipe in the network based on the following variables: Number of properties draining to the pipe Pipe diameter Demographic information Property type Historic incidents 2

3 The aim of the scoring was to attempt to select twenty of the highest risks sites to monitor as part of the trial. Therefore the 60 highest scoring sites in the Aston and Handsworth Drainage areas were then manually interrogated to determine where repairs have been completed and no further incidents have been reported, or where the pipes are already being investigated as part of the Planned Works Programme (PWP) or External Flooding Initiative (EFI). The remaining sites were then further assessed against normal flow monitor site selection criteria such as access to the manhole chambers and anticipated hydraulic conditions, such as theoretical velocity checks, turbulence and general likelihood to deliver quality data. Following the prioritisation exercise, the highest risk sites tended to be where long runs of terraced properties were connected to a single sewer. The main factors which resulted in these kinds of sites being high scoring were the number of terrace properties attached to a single pipe and a history of repeat blockages. CSOs Figure 2. Example of a site with high risk score. Severn Trent has committed to installing EDM equipment at the majority of CSOs during AMP6, therefore it was important that some CSOs were included within the ICM Live pilot study. CSOs have been selected for monitoring where pollution incidents have been reported at their respective outfalls, as well as where the Sewage Management Plan Study has indicated overflow may be hydraulically deficient. Dual Manholes Dual Manhole sites were selected based on a history of pollution incidents at the outfall of the surface water system. Hydraulic Flooding Locations 10 monitors were installed at reported hydraulic flooding locations. The final locations were drawn from sites at which the frequency and severity of reported flooding was sufficient to consider them for the proportion of a capital scheme during AMP6. Monitor types benefits and drawbacks Two types of monitors were installed: The key feature of both of the monitors is the wireless communications technology. The monitors were configured to send the data wirelessly and automatically to the manufactures server, via mobile phone networks. The monitor data was also automatically forwarded to Severn Trent. The wireless communications were found to be reliable for all trial monitors, allowing the monitor data to be collected without regular visits from the contractor. 3

4 The monitors were configured to dial in daily, and at pre-configured threshold based on the pipe soffit and spill / flooding level allowing monitor data to be received by Severn Trent in near real time for as levels changed within the system, while maintaining practical battery usage. Depth only monitors were installed in chambers with relatively small pipes (less than 300mm diameter) where it was deemed possible that a flow monitor may increase blockage risk. Pressure / velocity device monitors were used where dry weather velocity data was more likely to be recorded, and the contribution to blockage risk minimized. The table below highlights the positive and negative aspects of each type of monitor that were confirmed during the project. Benefits Pressure & Velocity Velocity data Generally more consistent depth data Can provide observed flow data Ultrasonic May be installed without man entry Lower cost No sensor in flow, therefore no increase in blockage risk. Concerns Higher Cost Requires manhole entry Sensor in flow may contribute to blockage risk No velocity data More susceptible to noisy data Needs line of sight to invert Software implementation Innovyze software ICM Live was chosen to process the incoming monitor data, weather data and generate alerts / alarms automatically according to rules defined in the software. The rules used create the alerts / alarms were generic in logic allowing depth variables to be configured on a per site basis. For a typical CSO or FOC site a proactive alert would be issued if: i. Is the recorded depth > anticipated diurnal peak depth for the duration of 1 hour ii. Is the recorded rainfall in this 1 hour <2mm in depth iii. Is condition i true whilst ii is true? For a typical CSO site a reactive pollution alert would be issued if: i. Is the recorded depth > spill level for the duration of 1 hour ii. Is the recorded rainfall in this 1 hour <2mm in depth iii. Is condition i or ii true whilst ii is true? Figure 3 below shows an example of a location where there is a clear change in the dry weather depth trend prior to a blockage occurring. This is typical of many sites and has been a key indicator the existing manual screening process. 4

5 Blockage Partial restriction Figure 3. Depth record at known pollution location These rules are designed to look at the data over periods of time, rather than individual measurements, this reduces the risk of an alarm being raised based on anomalous data, as consecutive measurements above the threshold required to trigger the rule. The disadvantage of this is that response is delayed by an hour. These rules were created as an initial trial of what may be possible using available data and proved appropriate for the selected locations. These rules have been proved to be applicable to real world sites and will form the basis of a company-wide reactive system. The primary challenges faced during implementation were related to the configuration of the IT and telemetry systems to allow the software to run, and for the software to continue to run in real time. The main challenges included providing the ICM Live services permission to access network data stores, and to have sufficient server space dedicated to the simulation results. The key to overcoming these challenges was to involve IT as early as possibly in order for them to work with Innovyze ensuring that the services are configured, and that the appropriate databased have been configured to process the incoming data. The ICM Live system was fed by three sources of observed data: Three tipping bucket rain gauges installed as part of the ongoing Minworth Sewerage Management Plan Study Observed RADAR rainfall data supplied by Weatherquest based upon Met Office data Flow and depth monitor data from the installed monitors No forecast / nowcast rainfall was available as the initial phase of the project was to assess the opportunity to improve the reactive response to early identification of blockage related incidents before they manifest as a flooding or pollution incident. Future trials within Severn Trent are planned to look at using this data to help transition to a truly pro-active system. The observed radar data had a cell size of 4km x 4km at 15 minute intervals, and covered the entire Severn Trent region. This radar data was found not to record peak intensity at a high enough resolution to make reliable predictions using the hydraulic model, due to the spatial and temporal average nature of the data. The weather data was limited during the trial to this relatively low resolution to keep data volumes 5

6 manageable, and was fit for the purpose of determining whether rainfall was occurring in the catchment to allow definition between event based incidents and other cause incidents in DWF. ICM Live proved to be capable of processing and visualising the observed data from the various sources albeit there were several features of the software package which were not fully utilised during these initial stages of the trial. The software was operated by experienced hydraulic modellers and it is recognized that introduction of this software into Network Control would require significant extra staff training for the Control staff or the recruitment of network modellers into Network Control. Monitoring Outcomes During the trial, three potential incidents were identified and these were all related to blockages detected in the Aston and Handsworth Drainage Area. Two issues occurred on ex-section 24 sewers at locations with a reported history of blockages. These sewers were cleansed pro-actively to eliminate the short term risk of the accumulation of debris affecting the customer. Figure 4. Depth recorded by monitor at reported flooding location. The first detected incident was identified by a rise in level at a manhole to over the depth of the incoming and outgoing pipe soffit. Site inspection revealed the chamber to be surcharged to the depth indicated by the monitor and subsequent jetting cleared the issue. It is not known for certain when (or if) this restriction would have resulted in a flooding or customer complaint; however due to the severity of the possible impact (flooding) it was deemed that the sewer had to be cleansed. The trial also successfully detected a blockage downstream of a CSO. The blockage was cleared by Severn Trent operational staff and a full assessment of the impact of the spill carried out in accordance with Standard Operating Procedure including self-reporting of the incident to the EA. Due to the shallow depth to spill at this CSO (approximately 140mm) the first data showing an indication of a problem at this CSO was received as the CSO was beginning to spill. Whilst it was not possible to prevent spill, the impact was mitigated 6

7 Upstream Downstream First potential indication of blockage forming Pipe downstream of CSO blocked Increase in depth due to rainfall Figure 5. Depth at upstream monitor (Blue) and downstream monitor (Green) Considering the scale of the monitor requirements throughout Severn Trent within AMP6, the trial was relatively limited in scope, geographical area and time. However, valuable conclusions were drawn to inform the broader strategy for a relatively small investment. The primary benefits and outcomes of the trial were: To develop a methodology for screening monitor data, and using rainfall to limit alarm numbers To understand the necessary IT configuration to implement ICM Live. This has assisted in implementing the first phase of a real time analytics solution to process network monitor data To establish that ICM Live is best implemented to target specific catchments and would not be the most appropriate tool at this point for screening data from a large number of monitors across a whole water company region. The installation of a large number of monitors across the entire Severn Trent region is likely to create a significant number of alarms. For these to be efficiently managed by Network Control a system which integrates closes with their existing process of working was found to be required. To prove that the computing and model resources exist for Severn Trent to implement a predictive ICM Live system in the future Lessons learnt Do not underestimate the work required to configure the IT systems to handle the many streams of data for an ICM live system; Locating high risk assets by desk study requires comparison on multiple data sets; utilising incident hot spot information alone is not effective; ICM Live is powerful software with more functionality than is required for real time assessment of current system performance; 7

8 Increased network monitoring can provide the intelligence to reduce flooding and pollution incident frequency and severity; The use of a significantly aggregated rainfall product is not sufficient to use ICM Live to its fullest potential. Conclusions The trial project proved the concept that computer software can effectively screen sewer monitor data in real time to identify potential flooding or pollution incidents. The trial has proven that the wastewater network can be proactively managed using relatively simple data and rules, and that this can form the basis of a large scale smart networks programme. The trial has also demonstrated the potential, as well as the system requirements, for targeted implementation of complex analytics based on forecasts. 8