Energy & Environment
Showcase of Selected Outcomes
Demand Response (DR) planning and optimisation tools
In the context of EC projects Wattalyst and OPTi our team developed methodologies for designing incentive-based DR programs as well as for ADR contracts. On the latter, two different approaches were developed and evaluated in a prototype tool targeted to utility companies. The prototype has been developed with the collaboration of IBM India. In the Fixed versus Learning-based Incentives in ADR Contracts approach, each consumer is modelled based on the modification of the optimal consumption schedule in the case of participation in DR, which leads to estimation of the incentives necessary for the consumer to actually participate. For each consumer there is a minimum incentive value that triggers him to participate. On the contrary, in Net Benefit-based Incentives in ADR Contracts approach each consumer is assumed to act rationally by choosing the consumption schedule that maximises her total Net Benefit. The approaches are robust with regard to sources of supply (i.e. are agnostic in that sense) in order to take into account energy supply from Distributed Resources as well.Download video demo
Price-based energy demand flexibility tool
The Price Based Flexibility Profiling (PBFP) tool is part of the Nobel Grid suite of tools for Demand-side Management, and more specifically manual demand-response schemes. It produces a profile for each customer, reflecting real-time demand flexibility as a function of multiple parameters, such as environmental context/ conditions, energy retail prices at peak and off-peak periods and individual/group preferences. These profiles can be used for calculating the flexibility obtained by each endpoint for any price and outdoor temperature during the peak and off-peak periods, as well as, for calculating incentive-compatible rewards to participants based on economic theory.
Business modelling analysis for Smart Grid actors
AUEB has lead the work of Business Modelling analysis for emerging Smart Grids in the context of project NobelGrid. A novel methodology has been developed that is based on a use case and product driven value network analysis and business modelling canvas generation. The methodology will be applied to the projects' pilots and will help us realise and evaluate the market potential of the Smart Grid technologies and the resulting interactions among the market players, namely DSOs, Aggregators, Retailers and Prosumers. Furthermore, in the next phase of this work (in progress) AUEB develops Cost Benefit Analysis tools for both Smart Grid networks (in project NobelGrid) and District Heating and Cooling networks (in project OPTi). Once evaluated the tools will be publicly released for the energy community.
Download project report
Multi-parametric forecast models of the electricity demand
The efficient evolution of the electricity grid depends on the active engagement of the end users (citizens) by means of their participation in demand response programs. More specifically, the retailer may announce dynamic prices during the peak periods, targeting to incentivize its clients to decrease their consumption or shift their elastic loads to the off-peak periods. In this context, this project (group internal research project) focuses on the profiling of the electricity consumers, i.e., the formulation of their demand with respect to the price and the environmental temperature. The project firstly classified the end-users according to their demand shifting behaviour and then developed two relevant models: the former is designed to capture the demand shifts within different periods in a day, in response to the values of the dynamic prices and temperatures in all those periods, while the latter assumes that the consumption during each period depends only on the price and the temperature of this particular period.
In constructing customer profiles based on historical data, the above models are fitted to a publicly available energy consumption dataset derived form a dynamic-pricing program at the city of London, using least square linear regression. Our findings suggest that the clustering of the users according to their shifting history and the subsequent assignment of the suitable model to each one of them, improves the forecast accuracy and achieves a percentage error lower than 4% (referring to the aggregate demand of the whole retailer’s clientele). Additionally, an algorithm is designed which may be utilized by the retailer for meeting its obligation for a balanced portfolio. The algorithm takes as input the models’ forecasts and computes the values of the dynamic prices to be applied such that the collective reaction of the end-users leads to the desired level of aggregate consumption. The models may be extended to capture further parameters that affect the consumption behaviour of the users such as their income, the members of the households and their social position (working or retired). Additionally, they may be modified for providing forecasts with higher granularity than the two-slotted separation of the planning horizon (peak and off-peak periods) in in their current form, and distinguish between the working days and the weekends targeting to improve their forecast accuracy.
- Utility-based user demand modelling
- Price-based demand profiling
- Behavioural economics
- Design of incentive compatible contracts for Automated/Manual DR campaigns
- Business model evaluation
- Economic/Societal Cost Benefit Analysis
- Auction mechanisms and bidding strategies