Resilient and Dynamically Adaptive Water Distribution Networks for a Sustainable Future

Water distribution networks (WDNs) are an integral component for achieving resilient water resources and supply management.   In the UK the WDNs have been sectorised into “static” discrete District Meter Areas  (DMAs) mainly for the purpose of leakage management. By installing kept-shut boundary valves to create DMAs, the redundancy of connectivity and supply within large interconnected networks is severely reduced thereby affecting operational resilience, water quality, and energy losses.  Although DMAs have contributed towards a step change in managing leakage over several decades, DMAs have introduced operational restrictions that affect both customers and utilities.  New technologies and systems-based approaches are urgently needed to improve operational resilience, hydraulic pressure and assets utilisation within WDNs during a time of escalating environmental, regulatory and financial constraints. This work followed concepts in communication networks, where “software-defined networks” and physical routers significantly improve the quality of service through dynamic connectivity and network adaptability (e.g. adaptive path control for application-aware routing across networks). Similarly, we have been working on the next generation of intelligent water distribution networks that dynamically adapt their connectivity and hydraulic conditions in order to deliver a notable improvement of the 4Rs of resilience for networks (redundancy, reliability, resistance, response & recovery).  This is achieved by replacing the Kept-shut boundary and control valves with multi-function self-powered automatic control valves (multi-function network controllers, MFNCs). The MFNCs adaptively control the network connectivity and hydraulic conditions based on mathematical optimisation in order to switch between hydraulic states that are specific for pressure control, incident response, water quality control and leakage management. This project has been a long-term collaboration between an early adopter network operator (Bristol Water), a technology company with extensive experience in pressure control (Cla-Val) and a world leading research-led university (Imperial College London). A demonstrator (the “Field Lab”), operated by BW, was implemented in 2012 as a “playground” for the development and integration of modelling, optimisation methods, and control technologies. The “Field Lab” includes three dynamically adaptive DMAs, 7900 customer connections and 59km of mains. The design and implementation of dynamically adaptive hydraulic control in WDNs (aka DMA v2.0) require a greater knowledge of the system hydraulics and the application of advanced modelling and optimisation methods based on fundamental scientific principles and operational experience. The ultimate goal is to improve the resilience, pressure management and operational efficiency, thereby also increasing asset life. Computationally efficient hydraulic modelling methods tailored for the control of dynamically adaptive networks were developed by Imperial College London (two PhD studentships funded by Cla-Val and BW) together with the mathematical formulation of performance, resilience, observability and stability metrics. Robust optimisation methods were combined with the vast operational experience of BW to define the placement and operation of MFNCs. An extensive development of the control equipment to deliver the broad range of control options has been carried out in parallel between Cla-Val and Imperial College London (e.g. a dual directional position control). Using the combined knowhow and experience of a Utility, Product Manufacturer and a University working collaboratively has been instrumental to the success of this project. The dynamic adaptability of both connectivity and hydraulic conditions has been non-existent within water distribution networks so far. The most “sophisticated” pressure control methods, which are mainly applied in single-feed DMAs, rely upon a simplistic feedback control loop from a critical point. These so called flow modulation methods are solely used to reduce leakage by minimising the pressure at a critical network point. The project developed and implemented both analytical methods and control technologies to enable the concurrent design, operation and control of dynamically adaptive water distribution networks that automatically configure their connectivity and hydraulic conditions. This allows network operators to periodically switch between operational modes by combining changes to network connectivity and the generation of hydraulic conditions using robust mathematical optimisation and control methods.  This unique operational framework sequentially implements “control to optimise”, “control to discover” and “control under extreme events” states for water distribution networks, which take into account the hydraulic dynamics and model uncertainties. For example, under the “control to optimise” application state, specific hydraulic conditions within adaptively configured areas are generated to minimise average zone pressure (AZP), variations in zonal pressure (VZPTM) and the cumulative pressure induced stress (CPISTM), while maximising the resilience of a water distribution network. The integrated analytics and technology for dynamically adaptive networks is now mature and provides a significant step change in way water distribution networks are managed in the UK, and is  at the forefront of innovation in resilient and sustainable water distribution networks. What measurable benefits have been achieved?
  1. A significant improvement in the 4Rs of resilience (and a reduction in customer interruptions) for the “Field Lab” network during major incidents. For example, a burst occurred in a 900mm diameter main in September 2014, which was upstream of an inlet to the “Field Lab”. A population of 121,450 people over a period of 48 hours had an interruption of supply. Nearly 4,000 customer connections maintained the supply of water with the “Field Lab” because of the improved resilience (and redundancy) through dynamic network connectivity and hydraulic adaptability. Furthermore, there was no need to manually open “kept shut” boundary valves and no discolouration complaints were received due to flow reversals. The hydraulic control has gradually and continuously “conditioned” the network to flow reversals.
  2. A reduction in background leakage in excess of 10% of what was achieved with “traditional” methods of pressure control (lower energy losses in interconnected multi-feed networks).
  3. A notable 80% reduction in bursts (from 15-20 bursts/annum to 3-5 bursts/annum) due to the reduction in AZP, VZP and hydraulically “calming” the network. Both background leakage and bursts/annum have had their lowest annual values recorded for this area since the implementation of the “Field Lab” in 2012.
  4. The number of water quality discolouration complaints have dropped from around 15/annum to zero for the last four years (and since the implementation of the “Field Lab”). No additional mains flushing was done since the implementation of the “Field Lab” in 2012.