| Author: | Alabi, Tobi Michael |
| Title: | A combined robust optimal planning and novel data-driven energy management for zero-carbon building multi-energy system |
| Advisors: | Lu, Lin Vivien (BEEE) |
| Degree: | Ph.D. |
| Year: | 2023 |
| Subject: | Buildings -- Energy conservation Sustainable buildings Buildings -- Energy consumption -- Data processing Hong Kong Polytechnic University -- Dissertations |
| Department: | Department of Building Environment and Energy Engineering |
| Pages: | xxviii, 222 pages : color illustrations, maps |
| Language: | English |
| Abstract: | Carbon neutrality is an ambitious goal that has been promulgated to be achieved on or before 2060. As a result, most of the current energy policies focus more on carbon emission reduction, efficiency and high penetration of renewable energy. However, to overcome the challenges associated with operating energy equipment separately, which include an increase in operating cost; energy loss; low efficiency; lack of optimal coordination and scheduling, and renewable energy intermittency. The concept of a Multi-Energy System (MES) that deals with the co-planning and operation of energy infrastructure in one-fold have been identified as a succinct approach to minimising the economic implications, reducing environmental hazards, and increasing the efficiency of an integrated energy infrastructure holistically. Nevertheless, whilst a significant number of resources have been invested in achieving low-carbon energy communities and energy infrastructure, limited research has been conducted towards the realisation of zero-carbon (ZC) MES. Based on this background, this thesis presents a comprehensive and systematic study of the robust planning and data-driven energy management approach for MES with zero-carbon potential. The influence of investment restriction, uncertainty quantification, and energy storage degradation are investigated in the planning decision. Novel energy management strategies are proposed considering realistic uncertainty quantification, electric-vehicle integration, energy flexibility, and carbon capture system integration. Furthermore, a novel data-driven method is developed for energy demand prediction using a deep learning method, and smart energy system control is developed using a deep reinforcement learning algorithm. In Chapter 3, an improved mathematical model is proposed for the optimal configuration of ZC-MES integrated with financial constraints. To achieve this aim, a mean-max approach is proposed for the selection of representative hourly data from the pool of available datasets. Next, mixed-integer linear programming (MILP) is formulated to describe the whole system, with the inclusion of new replacement cost formulation and financial constraints. The proposed approach is then verified using a developed residential district area in Hong Kong that was modelled using TRNSYS software as a case study under different scenarios. The simulation results show that the optimal investment cost obtained by the conventional approach can be further reduced by 2.90% through rational equipment configuration without violating the applicable constraints, while a financing rate of 5% and 10% on the additional funds under breakable constraint are of economic benefits to the investors. Hereupon, the proposed method provides a useful reference resource for energy planners, decision-makers, and academic researchers, on the feasibility of zero-emission in the energy sector and other related fields. To quantitatively and accurately analyse the impact of multi-energy demand and renewable energy uncertainty on MES capacity planning, a stochastic planning and operation (ZC-MES) taking the uncertainties of individual energy demand and environmental conditions into consideration is proposed in Chapter 4. The comprehensive mathematical model is developed as an optimisation problem using Monte Carlo scenario generation, fast-forward scenario reduction approach, chance-constrained programming, duality theory, and big-M linearisation approach. Furthermore, benders decomposition is applied to split the large-scale optimisation problem into an investment master problem (MP) and operation subproblem (SP), which are then solved iteratively. The obtained results indicate that considering all the energy demand uncertainties as an individual entity in the model, the selected capacities of PV, AC, and HWS increase by 60%, 21%, and 10.6%, respectively. At the same time, WT, WSHP, BES, and EC reduce by 14%, 15%, 11%, and 1.09%, respectively. Also, the optimal operation cost for the proposed model is only 5-10% of others compared to other scenarios. However, its annualised investment cost is 3-12% higher, but the overall economic implication is optimal. The underlying reason for this optimal result is the combination of energy storage temporal arbitrage and energy output shifting technique implemented by the algorithm to maintain an optimal interrelationship balance and to favour the optimal sizing of the chosen technologies. Additionally, a modelling strategy for ZCMES incorporating energy storage ageing influence and integrated demand response (IDR) is proposed in Chapter 5. An integrated clustering-scenario generation and reduction approach (IC-SGRA) is first developed to quantify the datasets uncertainties while selecting a representative day for the model. Secondly, the model is formulated as a multi-objective optimisation problem to evaluate the influence of decision-maker preference concerning investment cost and operation cost on optimal planning, and then a weighting sum method is adopted to solve the problem. Finally, a Markowitz portfolio risk theory approach is adopted to mitigate the risk associated with uncertainties during decision-making, and then an illustrative case study is used to analyse the proposed model. The simulation results reveal that the energy storage is overdesigned when ageing effects are not considered. The proposed approach can reduce investment and operation costs by 10.86% and 80.66%, respectively. In comparison, the overall expenditure is reduced by 23.09%. Furthermore, the effect of BES lifetime and IDR load factors are also examined on ZCMES optimal planning. It was affirmed that IDR is a promising strategy to encourage adopting zero-carbon policies flexibly and economically while choosing BES with high lifetime and tolerable capacity loss contribute to optimal planning. In the ZCMES operation schedule, an improved day-ahead optimal scheduling for a Virtual Power Plant (VPP) to enable optimal dispatch of Zero-Carbon Multi-Energy System (ZCMES) and EV multi-flexible potentials and the influence of uncertainties under various scenarios is developed in Chapter 6. The Latin Hypercube Sampling (LHS) method is used to quantify the uncertainties, while the multi-energy demand randomness is further analysed using a robust approach, and a robust-stochastic optimisation approach is developed to solve the mathematical problem. The obtained simulation results indicate that for a specified EV flexibility (e.g. vehicle-to-electricity use V2EU, vehicle-to-heat use V2HU, or vehicle-to-cooling use V2CU). The proposed multi-flexible approach provides both economic and technical benefits by reducing the overall cost by 8.5%, having a high EV flexibility ratio (0.54), and reducing the technical stress on energy storage by smoothing the state of charge (SOC) level through discharge behaviour regulation. Finally, it was observed that the multi-flexible performance is influenced by the magnitude of multi-energy demand and generation, charging station standards, EV parameters, and robust control parameters. Thus, such a great feasibility analysis of EV multi-flexible potential provides valuable policy and planning reference tools for the energy stakeholders towards energy carbon neutrality in urban areas. Furthermore, a deep learning approach and optimisation model for the optimal day-ahead scheduling of ZCMES virtual power plants are investigated in Chapter 7. Technically, a carbon capture system (CCS) is introduced to harness the carbon emission associated with some equipment, consideration of electric vehicle multi-flexible potentials, followed by a clean energy marketer (CEM) strategy to ensure system reliability sustainably. An integrated recurrent unit-bidirectional long-short term memory (GRU-BiLSTM) is developed for day-ahead multivariable time-series prediction. This is followed by an autoencoder (AE) for scenario generation and scene reduction using the fast-forward reduction algorithm. A robust-stochastic modelling approach is then applied for optimal decision-making. As a case study, the proposed model is verified using accurate historical multi-energy data of a district in Arizona, the United States. Remarkably, it was observed that the CEM trading period restriction influenced the scheduling behaviour of ZCMES and the charging pattern of EVs. However, the integration of EV flexibility reduces dependency on the external grid and optimises the power consumption of CCS using part of cogeneration electrical output instead of total reliance on the external grid. Thus, the proposed model strengthens carbon-neutral feasibility in urban centres and serves as a reference tool for sustainable energy policymakers. Finally, a deep reinforcement learning agent (DRL), integrated with an automated hyperparameter selection feature, is developed for the real-time scheduling of a multi-energy system (MES) coupled with post-carbon capture systems (PCCS) and direct-air capture systems (DACS) is developed in Chapter 8. Various possible configurations are evaluated using real-time multi-energy data of a district in Arizona, the United States, and CDRT parameters from manufacturers' catalogues and pilot project documentation. The results show that the proposed DRL agent can meet the prosumers' multi-energy demand and schedule the CDRT energy demand economically without specified constraints violation and outperform rule-based scheduling by 23.65%. In summary, the main novelty and originality of this thesis can be briefly summarised as follows: First, a robust optimal planning model is developed for the feasibility of ZCMES, the model incorporates investment constraints, uncertainty influence on capacity sizing, and energy storage degradation effect on capacity sizing. Second, an improved co-optimization model that considers energy storage ageing effect and IDR flexibility approach on the planning and operation decision making is developed. The model guaranteed realistic optimal configuration and provide excellent strategy for the adoption ZCMES via IDR. Third, a data-driven energy management approach for the real-time and smart control of ZCMES is developed to optimally schedule ZCMES considering EV multi-flexible approach, and clean energy trading. Lastly, a deep reinforcement learning is developed for autonomous control of ZCMES components in real-time with minimal latency. The model is applied to ZCMES configuration with carbon capture technology integration. The above study on the planning and energy management strategy for MES can help researchers and policymakers to evaluate the technical and economic viability of achieving zero-carbon emissions in the energy sector. Also, the systemic research methodology and framework provide significant guidance for relative stakeholders to develop a smart energy management strategy for MES considering transport sector integration and energy flexibility to accelerate the progress of carbon neutrality in urban centres. |
| Rights: | All rights reserved |
| Access: | open access |
Copyright Undertaking
As a bona fide Library user, I declare that:
- I will abide by the rules and legal ordinances governing copyright regarding the use of the Database.
- I will use the Database for the purpose of my research or private study only and not for circulation or further reproduction or any other purpose.
- I agree to indemnify and hold the University harmless from and against any loss, damage, cost, liability or expenses arising from copyright infringement or unauthorized usage.
By downloading any item(s) listed above, you acknowledge that you have read and understood the copyright undertaking as stated above, and agree to be bound by all of its terms.
Please use this identifier to cite or link to this item:
https://theses.lib.polyu.edu.hk/handle/200/12567