Impact of different time series aggregation methods on optimal energy system design

Kotzur, Leander; Markewitz, Peter; Robinius, Martin; Stolten, Detlef

doi:10.1016/j.renene.2017.10.017

Cited by 242 publications

(177 citation statements)

References 44 publications

Supporting

Mentioning

158

Contrasting

Unclassified

Order By: Relevance

“…The building optimization is based on a typical Mixed-Integer Linear Program (MILP) with the objective of minimizing the annual energy cost of a single building as proposed in the vast literature. The operation and design of the supply system is modeled with the object-oriented system modeling framework FINE [65,71,72]. The binary variables are considered to sufficiently incorporate the economy of scale of the technologies.…”

Section: Optimizing Structure Scale and Operationmentioning

confidence: 99%

Bottom-up energy supply optimization of a national building stock

Kotzur¹,

Markewitz²,

Robinius³

et al. 2020

Energy and Buildings

Self Cite

View full text Add to dashboard Cite

The installation and operation distributed energy resources (DER) and the electrification of the heat supply significantly changes the interaction of the residential building stock with the grid infrastructure. Evaluating the mass deployment of DER at the national level would require analyzing millions of individual buildings, entailing significant computational burden.To overcome this, this work proposes a novel bottom-up model that consists of an aggregation algorithm to create a spatially distributed set of typical residential buildings from census data. Each typical building is then optimized with a Mixed-Integer Linear Program to derive its cost optimal technology adoption and operation, determining its changing grid load in future scenarios.The model is validated for Germany, with 200 typical buildings considered to sufficiently represent the diversity of the residential building stock. In a future scenario for 2050, photovoltaic and heat pumps are predicted to be the most economically and ecologically robust supply solutions for the different building types. Nevertheless, their electricity generation and demand temporally do not match, resulting in a doubling of the peak electricity grid load in the rural areas during the winter. The urban areas can compensate this with efficient co-generation units, which are not cost-efficient in the rural areas.

show abstract

Section: Optimizing Structure Scale and Operationmentioning

confidence: 99%

Bottom-up energy supply optimization of a national building stock

Kotzur¹,

Markewitz²,

Robinius³

et al. 2020

Energy and Buildings

Self Cite

View full text Add to dashboard Cite

show abstract

“…Reference [9] also compares different techniques in the context of different local energy systems (averaging, k-means, k-medoids, hierarchical) for obtaining representative days, 3days or weeks. It finds that medoids perform better than centroids but recommends overall the use of hierarchical clustering due to the reproducibility of the results.…”

Section: State Of the Art And Contributionmentioning

confidence: 99%

Clustering methods assessment for investment in zero emission neighborhoods’ energy system

Pinel¹

2020

International Journal of Electrical Power & Energy Systems

View full text Add to dashboard Cite

This paper investigates the use of clustering in the context of designing the energy system of Zero Emission Neighborhoods (ZEN). ZENs are neighborhoods who aim to have net zero emissions during their lifetime. While previous work has used and studied clustering for designing the energy system of neighborhoods, no article dealt with neighborhoods such as ZEN, which have high requirements for the solar irradiance time series, include a CO2 factor time series and have a zero emission balance limiting the possibilities. To this end several methods are used and their results compared. The results are on the one hand the performances of the clustering itself and on the other hand, the performances of each method in the optimization model where the data is used. Various aspects related to the clustering methods are tested. The different aspects studied are: the goal (clustering to obtain days or hours), the algorithm (k-means or k-medoids), the normalization method (based on the standard deviation or range of values) and the use of heuristic. The results highlight that k-means offers better results than k-medoids and that k-means was systematically underestimating the objective value while k-medoids was constantly overestimating it. When the choice between clustering days and hours is possible, it appears that clustering days offers the best precision and solving time. The choice depends on the formulation used for the optimization model and the need to model seasonal storage. The choice of the normalization method has the least impact, but the range of values method show some advantages in terms of solving time. When a good representation of the solar irradiance time series is needed, a higher number of days or using hours is necessary. The choice depends on what solving time is acceptable.

show abstract

“…The need for long samples poses a computational challenge since accurately modelling systems with significant shares of VRE generation also requires high temporal resolution. Previous studies indicate that models with low resolution fail to capture VRE output fluctuations and underestimate required flexible and dispatchable generation capacity (De Jonghe et al, 2011;Poncelet et al, 2016;Collins et al, 2017;Kotzur et al, 2018). For many realistic PSMs, it is computationally unfeasible to solve the optimisation problem using both a long sample of data and a high temporal resolution (Pfenninger, 2017).…”

Section: The Computational Cost Of Weather and Climate Variabilitymentioning

confidence: 99%

“…days or weeks) obtained by clustering the full dataset. Numerous studies explore the efficacy of such approaches in reproducing model outputs at reduced computational expense (de Sisternes & Webster, 2013;Pfenninger, 2017;Nahmmacher et al, 2016;Kotzur et al, 2018;Härtel et al, 2017;Poncelet et al, 2017). They arrive at a number of common conclusions.…”

Section: Established Data Reduction Approachesmentioning

confidence: 99%