Thames Water is a British utility tasked with managing water and wastewater services to more than 15 million customers located in the catchment area of the Thames River. This includes Greater London and the cities of Oxford, Swindon and Crawley. The utility operates 350 sewage works, 32 sludge treatment centers and supplies nearly a billion gallons of drinking water daily.
This is a complex and extremely large business. While static management of the business using spreadsheets and management judgement was previously possible, increased complexity and the need to reduce operating costs led Thames Water to investigate the implementation of a decision support tool (DST). After evaluating several vendors, Thames Water chose Business Modelling Associates to develop a supply and demand tool. The decision was based on BMA's earlier work with other water and sewage utilities. Using River Logic's Modeling and Optimization Platform, BMA developed an integrated business model that allowed Thames Water to optimize their operations. In the course of this exercise, Thames Water were able to identify potential savings in operational costs of £10 million annually.
Business Challenges Facing Thames Water
Thames Water operate a large number of sewage treatment works along with numerous sludge treatment centers and anaerobic digester sites as well as two sludge-fired combined heating and power plants. The organization has, over the last seven years, been actively realigning its strategies by closing inefficient processes, reducing its environmental impact and focusing on energy efficiency.
During this process, Thames Water realized that its spreadsheet-based planning tools were inflexible, and its historical costing processes made it difficult to make informed operational and strategic decisions, particularly in response to unplanned outages and disruptions. A specific factor that added greatly to operating costs were excessive demurrage charges from sludge haulage companies as a result of disruptions caused by plant outages, plant upgrades, unplanned sludge movements and road closures. Other factors that added to complexity were the need to accommodate sludge trading between wastewater utilities, the need to optimize treatments costs, and the optimization of energy recovery and power generation.
Specific Business Needs
In order to improve overall operation and reduce costs, Thames Water identified several priorities including:
- Development of daily, weekly and annual planning tools: These include long-term planning of operations, a medium-term strategy that focused on optimizing the delivery of sludge to the plants and short-term interventions focused on sludge tanker routing to deal with unplanned incidents and outages.
- Reduction of waste: To reduce demurrage charges from transport contractors due to unplanned events, to run plants efficiently and to optimize power generation.
- Optimizing sludge trading: To establish, in real time, sludge costs so as to improve sludge trading decisions.
- Minimizing operation costs and increasing cost recoveries: To identify the most cost-effective ways of processing sludge, including the possibility of rerouting to more efficient plants. To maximize energy output of CHP plants.
Choosing Appropriate Decision Support Tools
Access to the right business intelligence was crucial if Thames Water were to optimize their operations. At the first level, the organization needs to know what has happened. The next level, that of predictive analytics, focuses on understanding the reasons for how and why certain results were achieved. The final step, known as Prescriptive Analytics is the use of Thames Water's data and information to manage and prescribe day-to-day operational decisions. It was at this level that Thames Water wanted to operate.
In making this decision, Thames Water approached three consultants with demonstrable expertise in large-scale data analytics. Eventually, they selected Business Modelling Associates, who had already developed similar tools for other water and sewage utilities.
Using a 5th generation programming language, River Logic's Enterprise Optimizer enabled BMA to create an analytical model that facilitated the optimization of Thames Water's complex operations. With its built-in modelling tools, Enterprise Optimizer enabled the creation of mathematical relationships that closely represented how Thames Water's business operated in the real world.
The first step was the development of a baseline model. This model was designed so that it included the following key parameters:
- An understanding of all options for sludge storage, transportation, treatment and disposal: This includes inter-site transport requirements, different treatment methods, intermediate storage, and the sale and disposal of sludge.
- Operational constraints: The ability for users to change and modify operational constraints.
- Financial aspects: Income and expenditure associated with each option.
- Provision for validation and calibration: To ensure that predicted outputs are valid.
Once the base model was completed and demonstrated its proof of concept, the next step was to develop a more detailed model that provided for additional functionality, user templates and provision to drive the model by site availability.
The final step was to develop the interactive daily model and to incorporate various refinements such as allowing the same core data to drive each model.
Scenario Testing and Validation
Having prepared the base model, it was necessary to verify that the model's outputs were realistic and valid. Several test scenarios were evaluated, including:
- Asset base optimization
- Implications of planned and unplanned outages
- Impact of delays in capital investments
- Modelling of logistics and treatment costs
Several scenarios were evaluated to ensure that the predicted outcomes were valid by comparing the model's output to known outcomes. These were able to demonstrate that the model was correct.
The next step was model calibration to ensure that predicted volumes, costs, and outputs were correct. This established that the model's calculated profit and loss accounts were correct to within a few percent. Close correlation with other variables, such as power generation and the usage of wastewater treatment chemicals, was also achieved.
Additionally, the model closely predicted total sludge output. However, it suggested that operating costs were 30 percent lower than current costs. This was found to be due to the model assuming 100 percent site availability and ignoring demurrage charges. This will be corrected in the new daily model.
Benefits of Decision Support Technology
Thames Water believe they have derived significant benefits from the implementation of decision support technology, and in particular, the adoption of prescriptive analytics. Apart from effectively modelling their sludge operations, Thames Water has been able to identify annual savings in operating costs of £10 million ($13 million). Specific benefits include:
- A better understanding of the cost and volume of return liquor
- The identification and separation of costs
- A comprehensive representation of the wastewater treatment system as a whole in place of the earlier, fragmented plant-by-plant understanding
- The optimization of inter-site treatment of sludge
- The identification of plant constraints
- Clarification of inter-dependencies between plant, equipment, transport and sites
- Good user interaction thanks to innovative dashboards and reporting systems
- The ability to quickly assess different scenarios and review the impacts of potential business decisions
- Potential annual operational savings that will reduce costs and improve overall efficiencies
The adoption of a prescriptive analytics-based, supply and demand decision support platform by Thames Water has brought about greater understanding of what is a very complex business. Apart from direct benefits of optimizing wastewater and sludge processing, operational savings have been identified. These benefits have led to a decision to procure an additional model for planning long-term sludge investments over the next 40 years and to consider the impact of deregulation of this market.