Use this URL to cite or link to this record in EThOS: https://ethos.bl.uk/OrderDetails.do?uin=uk.bl.ethos.739652
Title: A multistage stochastic modelling framework for the optimal operation of DER aggregators under multidimensional uncertainty using stochastic dual dynamic programming
Author: Fatouros, Panagiotis
ISNI:       0000 0004 7229 0780
Awarding Body: Imperial College London
Current Institution: Imperial College London
Date of Award: 2017
Availability of Full Text:
Access from EThOS:
Access from Institution:
Abstract:
The emerging paradigm shift towards the Smart Grid concept, has vigorously encouraged the broad deployment of distributed energy resources (DER), such as energy storage (ES) and flexible demand (FD) and renewable micro-generators, in the energy system. In deregulated power systems, the deployment of flexibility pertaining to ES and FD is associated with their efficient integration in the electricity market. However, significant participation barriers have triggered the introduction of distributed energy resources (DER) aggregators in electricity markets, which settle the necessary framework for the market realisation of their promising operational flexibility potential. The significant number and diversity of resources pertaining to the DER aggregator portfolio, combined with multiple stochastic components affecting its optimal operation demonstrate a high-dimensional stochastic problem. Existing literature focusing on the problem of the optimal operation of DER aggregators exhibits significant limitations, since two-stage stochastic formulations are adopted. In this context, this thesis proposes, analyses and evaluates a novel multistage stochastic model, where multidimensional stochasticity is efficiently considered. Suitable dimensionality reduction and decomposition techniques have been deployed to tackle the computational issues stemming from the high dimensionality of the problem. Stochastic Dual Dynamic programming (SDDP) is deployed to alleviate computational tractability problems. Autoregressive models (AR) are employed to articulate temporal and cross-variable dependencies among the stochastic variables. Two novel extensions of the traditional SDDP algorithm, where linear (i.e. AR) models are integrated in the solution process and enhance solution quality, are proposed. A simulation framework for the validation and assessment of the proposed extended SDDP models, which compares them against scenario tree formulations with different structural characteristics, is presented. Case studies demonstrate that the extended SDDP models achieve a better trade-off between solution efficiency and computational performance. Additionally, results highlight the value of strategic positioning of the DER aggregator portfolio, when limited renewable generation is available. Finally, the effect of strategic decision-making based on less accurate information is shown to be intensified when the aggregator manages a more flexible portfolio.
Supervisor: Strbac, Goran Sponsor: Not available
Qualification Name: Thesis (Ph.D.) Qualification Level: Doctoral
EThOS ID: uk.bl.ethos.739652  DOI:
Share: