IWRA Proceedings

Authors: Craig Mauelshagen, Stephen Herndlhofer, Richard Martin, Pieter Jonas, Niki Roach
1 INTRODUCTION The environmental, regulatory and financial challenges within the UK water industry are drivers for innovation and performance improvement. Companies are continually looking for solutions to better manage complex business environments. Predicting and managing the impact of risk and uncertainty across an integrated network of assets is one of the key challenges faced. Yorkshire Water (YW) identified the need to develop an integrated risk and cost forecasting capability to represent the complex interdependencies between internal and external factors, regulatory commitments and business constraints. In 2011 YW began to work with Business Modelling Associates (BMA) and Cranfield University to develop an enterprise-wide risk and cost modelling capability. Prior to this YW's risk modelling capability has been distributed across a range of specialised models focused on individual risks, such as fluvial flooding, or business areas, such as asset maintenance. In contrast, enterprise risk management has been dominated by qualitative risk assessments and little modelling. This is a situation common to most companies (Lindhe et al., 2010; Ray et al., 2007). YW's development of a strategic risk and cost modelling capability puts it at the forefront of a move towards a new generation of enterprise-wide risk models (Bisias et al., 2012; Lai, 2012). In this paper we present a case study of YW's strategic risk and cost model (Source-to-sea model) and discuss the learning and experience gained through its development and implementation.

2 METHODS/MATERIAL The model was built using a technology developed by Riverlogic called Enterprise Optimizer® (EO) which is built on Constraint-Oriented Reasoning™ (COR). COR is a 5th generation programming language that enables EO to quickly create high-value analytical solutions in complex problem domains. Problems are defined as constraints and can be expressed in intuitive ways such as graphically or through symbols and relationships. Two key feature of EO made it suitable for this project. First, EO allows model building without writing code or managing formulas. Unlike conventional modelling, COR automatically generates mathematical representations of all system constraints and their interactions. This avoids the need for specialised developers and allows users with operational, engineering or financial backgrounds to build models and apply their expertise directly. Second, EO can represent an asset base, business processes and company accounts in a single representation. This allows the quantification of the impact of risks (or investment options) on the balance sheet, shareholder value and the regulatory accounts. It also allows both operational and financial constraints to be applied within a single model.

3 RESULTS AND DISCUSSION The Source-to-sea models representation of YW's business comprises:
* Water Processes
- All raw & treated Water Processes represented - Representation of customers (circa 5 million) down to production management zone level (circa 45000 customers) - Representation of water network connectivity and water flow - All water treatment works represented individually
* Sewerage
- Representation of customers down to a sewage treatment works catchment area * Sewage Treatment - All sewage treatment works represented - Surface water and infiltration represented * Sludge - All sludge processes represented, including sludge tankering - All sludge import sites individually represented - Sludge treatment variations by site represented
* Financial
- Operational accounts for Raw Water, Treated Water, Sewage and Sludge - Wholesale/retail accounting separation represented - Regulatory accounting (including Regulatory Capital Value) - Quantification of financial risk (Value at Risk)
* Time
- 40 years represented in annual time steps

Since implementation the source-to-sea model has been used to quantify systemic risks and quantify the uncertainty around financial impacts to understand the group's 'value at risk' (Mauelshagen et al., 2014). Further, by representing risk, cost and performance in a single model, it has helped ensure that risk is considered along-side other factors in strategic decisions. Some of the key learning derived from project has been from the challenges overcome. The first challenge was that no one expert, team or department had all the necessary data and domain knowledge required. Considerable effort was spend identifying and locating expertise and data. A benefit of this effort was identification of previously unknown areas of data paucity. As a result a 'data services' team has been created within YW. The second challenge concerned model implementation. Because an enterprise wide approach to risk and cost modelling is so novel, it encountered misunderstanding and resistance. Key factors in overcoming this were substantial training and workshops and development of bespoke user interfaces to ensure model outputs were relevant and accessible. A significant amount of work also went into redesigning the strategic risk management processes to align with the new modelling capability. Currently running risk scenarios on the model is offered as a service to risk owners. While the ultimate vision is for all risk owners and risk champions to have access to the model, this approach will allow extended user acceptance testing and capability building before wider implementation.

4 CONCLUSION
1. The quality and availability of data were a key enabler in embedding business analytics into YW. However, where sufficient data were not available this was not a blocker and indeed benefit was derived by driving improvements to data capture and management.
2. Equally important to data were the cultural and process changes that enabled analytics to form part of business-as-usual processes.
3. Combining asset and financial data in a single view at any level, strategic through to operational, created a step change in the way YW managed their operations.
4. YW's business analytics capability is providing the ability to understand and predict the impact of current and future industry changes and challenges. As the project fully delivers YW will be able to model, predict and improve their decision-making to meet the challenges of the future. 1. Bisias D, Flood M, Lo AW, Valavanis S. A survey of systemic risk analytics. Annu Rev Financ Econ. 2012;4(1):255–96.
2. Lai FJKP. A Value at risk approach to the assessment and evaluation of economic risk and return associated with large open pit slope design. 2012 [cited 2014 Sep 23]; Available from: http://minerva-access.unimelb.edu.au/handle/11343/37205
3. Lindhe A, Rosén L, Hokstad P. Risk Evaluation and Decision Support for Drinking Water Systems [Internet]. Techneau; 2010 [cited 2013 May 1]. Available from: http://publications.lib.chalmers.se/publication/154930-risk-evaluation-and-decision-support-for-drinking-water-systems
4. Mauelshagen CW, Pollard SJT, Owen D, Herndlhofer S, Firth P, McKenna J, et al. Protecting asset value and driving performance with a dynamic, risk-based contingency fund. Environ Syst Decis. 2014 Sep;34(3):417–24.
5. Ray B, Apte C, McAuliffe K, Deleris L, Cope E. Harnessing uncertainty: The future of risk analytics [Internet]. Research report. Yorktown Heights, New York: IBM TJ Watson Research Center; 2007 [cited 2014 Oct 30]. Available from: http://domino.research.ibm.com/library/cyberdig.nsf/papers/B910FD442135744585257434005349F4/$File/rc24534.pdf