Northumbrian Water Group

Industrial energy systems integrating on-site renewable generation are subject to increased operational variability and uncertainty, particularly under high renewable penetration and carbon-constrained operation. Conventional deterministic forecasting approaches are insufficient for capturing uncertainty in renewable generation, grid import behaviour, and associated carbon intensity signals. Reliable probabilistic forecasting is therefore required to support next-generation industrial energy management systems. This paper presents a probabilistic multi-horizon forecasting framework based on a multi-output quantile Long Short-Term Memory (LSTM) architecture. The model generates probabilistic forecasts of photovoltaic (PV) generation, grid electricity import, and grid carbon intensity. Prediction interval reliability is improved using conformal prediction, enabling statistically consistent uncertainty calibration without distributional assumptions. The framework is evaluated using high-resolution industrial energy datasets under winter and summer operating conditions. Results demonstrate robust forecasting performance across multiple horizons, achieving median forecast errors below approximately 0.7 in winter and approximately 1.1 in summer. Calibrated prediction intervals achieve empirical coverage levels of approximately 0.80-0.83 in winter and 0.71-0.74 in summer. The results demonstrate the suitability of probabilistic deep learning methods for uncertainty-aware forecasting in industrial energy planning and management.

Energy Storage Systems (ESSs) play a vital role in storing carbon-neutral yet variable Renewable Energy (RE) for later use, facilitating demand response and reducing grid reliance. Although ESS integration with RE systems has been widely studied, limited research examines its role in the decarbonisation of industrial and critical infrastructures such as Water Distributi on Networks (WDNs). This paper presents an optimisation framework to evaluate the impact of Battery Energy Storage Systems (BESS) on the performance of PPA-based solar PV systems in such contexts. A Mixed-Integer Linear Programming (MILP) model was developed to minimise operational costs and carbon emissions under dynamic grid pricing and time-varying carbon intensity. Two scenarios were analysed: PPA-PV with grid interaction only, and PPA-PV integrated with a 3 MW/6 MWh BESS. Results show that BESS integration reduced grid imports by 8.2% in summer and 10.3% in winter, while lowering carbon emissions by 35.6% and 27.3%, respectively. Operational costs declined by 1.6% in summer and 5.9% winter. The findings demonstrate the effectiveness of Storage in improving renewable utilisation, reducing grid dependency, and accelerating industrial decarbonisation under PPA-based RE systems.

Water Distribution Networks (WDNs), with their substantial energy demands, present a significant opportunity for integrating Renewable Energy Systems (RESs) for decarbonising operations. In recent years, AI-based datadriven Energy Management approaches have become popular in optimising energy use, however, remain underexplored in the context of WDNs. Additionally, existing work in Energy Management fails to incorporate third-party ownership models for RESs such as Power Purchase Agreements (PPAs). This study presents an AIbased Smart Industrial Energy Planning (SIEP) framework that, for the first time, applies integrated forecasting and optimisation to WDNs while explicitly accounting for Third-Party Financed solar PV systems through Power Purchase Agreements (PPAs). The SIEP tool employs Multivariate Long Short-Term Memory Recurrent Neural Networks (LSTM-RNNs) for accurate day-ahead forecasting of Energy Demand and PV Generation. Additionally, a novel Data-Correction method involving statistical estimation and time-series forecasting is proposed for WDNs to improve the quality of Energy Demand data using the site's Water Outflow data, allowing the correction of erroneous measurements of energy consumption. Furthermore, a Mixed-Integer Linear Programming (MILP) optimisation model is developed to strategically schedule grid-electricity imports, achieving a balance between carbon emissions-reduction and energy cost-savings. The results demonstrate that, over an average summer week, a self-financed PV system can reduce up to 22% carbon emissions and up to 21% energy costs, while a third-party financed PV system can achieve up to 21% carbon and 12% cost reductions. The findings highlight the potential of leveraging AI-based Energy Optimisation for effectively decarbonising WDNs and other industries.

Large-scale construction activity associated with the enlargement of Abberton Reservoir in Essex, England was calculated, mapped and subsequently modelled to examine the effect of disturbance on four Special Protection Area designated dabbling waterfowl species, Anas strepera (Gadwall), A. clypeata (Shoveler), A. crecca (Teal) and A. penelope (Wigeon). The distribution of each species was compared with levels of construction disturbance and environmental variables using Hurdle Model analysis and spatial referencing. Numbers of all four species varied throughout the study period with significant increases observed across the reservoir during the four year construction period. Findings show that the most important environmental variable was shallow water with increases in this area of habitat as a result of planned enhancement measures being of particular benefit to Gadwall, Shoveler and Teal. Numbers of Wigeon were especially variable across the site during construction and were displaced from the Main Section during the most disturbing shoreline works, behaviour we attribute to the loss of suitable grazing habitat during the construction process. While results show some disturbance responses, maintenance of site integrity for all four species is attributed to the overall size of the reservoir complex (4.75 km(2)) and the phased construction programme which reduced the extent and impact of disturbance. Research presented here provides evidence for a sensitive and science based approach to better deliver conservation and development requirements. The monitoring prior to (2006-2009) and during the construction phase (2010-2013) has enabled the much needed, but often lacking, evidence-based reporting on construction disturbance effects and associated mitigation measures.

Sustainable drainage systems and ‘blue–green’ infrastructure provide a range of environmental, economic and social benefits in addition to managing water quantity and quality. Recognition of the multi-functionality of these systems and their multiple benefits could lead to joint efforts to deliver infrastructure that meets the strategic objectives of both public and private organisations. This paper reports on the evaluation of the multiple benefits of the Killingworth and Longbenton surface water management scheme, a partnership project in north-east England jointly funded by Northumbrian Water, the Environment Agency and North Tyneside Council. Using complementary assessment tools, the evaluation quantified and monetised six key benefits, assessed two qualitative benefits, illustrated the spatial distribution of five non-flood benefits and highlighted locations with the greatest opportunity for multi-beneficial intervention. It was found that the scheme generates significant flood damage reduction benefits, improves water quality, habitat size, carbon dioxide sequestration and amenity, and reduces noise pollution. Use of the tools for multiple benefit evaluation shows promise as an aid to facilitate partnership working towards implementation of multi-functional sustainable drainage systems and blue–green infrastructure.