Frequency fluctuations and stability of power grids with a

Acknowledgements

Abstract

Frequency fluctuations and stability of power grids with a large renewable penetration ratio

María Martínez-Barbeito, Damià Gomila, Pere Colet

IFISC (CSIC-UIB) Palma de Mallorca – Spain

maria@ifisc.uib-csic.es

http://ifisc.uib-csic.es@ifisc_mallorca www.facebook.com/ifisc

To combine the numerical model with observations we use 10-minute total demandand generation data from Red Eléctrica de España ( https://demanda.ree.es/visiona/home )

Model Data assimilation

● Conventional generation nodes [2, 3]:

Future scenarios

Current transition towards a more sustainable energy system

QR code

Demand distributed according to population of each node

+ fast fluctuations(Ornstein-Uhlenbeck noise)

𝑃!" = 𝑃!#$$ + 𝜖𝜉!%&

Generation distributed by technology in each generation

node

𝜆! , 𝑃!'()

■ To avoid large frequency fluctuations that threaten grid stability as windpenetration increases, we consider additional control capability in conventionalpower plants.

■ We plot the amountof secondary controlneeded to reduce theprobability of frequencydeviations larger than±0.2 Hz to that of thecurrent case.

● Consumer nodes:

● Renewable generation: modelled as a negative load

Grid frequencyis a good indicator of the

generation-demand balance.

Motivation

[1] Ulbig, A., Borsche, T. S., & Andersson, G. (2014). IFACProceedings Volumes, 47(3), 7290-7297.[2] Saadat, H. (1999). Power system analysis (Vol. 2). McGraw-hill.[3] Filatrella, G., Nielsen, A. H., & Pedersen, N. F. (2008). TheEuropean Physical Journal B, 61(4), 485-491.

High voltage grid of Gran Canaria

Frequency data: https://osf.io/by5hu/

We propose a dynamical modelto study power grid stability inscenarios of high renewablepenetration. We consider thehigh voltage grid as a network ofsubstations and power plantsinteracting via transmission lines.In particular, we present GranCanaria as a case study.

GenerationDemand

50 Hz

49.5 50.5

loadgeneration transmitted power

Primary control

Secondary control

dispatch

Model validation

■ The model reproduces the statistics of realdata, and it captures the frequency peaksassociated to renewable generation variability.

■ In the rank distribution, we see that the modelworks best for the smaller frequency deviations.

■ The slope in the power spectrum indicates thatfrequency fluctuations are correlated, and theOrnstein-Uhlenbeck noise is a good choice tomodel fast power variations.

References

We acknowledge Agencia Estatal de Investigación (AEI,Spain), and Fondo Europeo de Desarrollo Regional (FEDER,EU) under grant PACSS (RTI2018-093732-B-C22) and theMaria de Maeztu program for Units of Excellence in R&D(MDM-2017-0711).

■ Intermittent and unpredictable nature of renewableenergy sources adds fluctuations on the generation side■ Reduced conventional control capacity■ Loss of rotational inertia increases frequency fluctuations

Integrating a high share of renewable generationin the power grid is a challenging task [1].

■ We plot the rank distribution of frequencyfluctuations for

The legend displays the fraction of thetotal generated energy provided by wind.

■ As expected, we observe that higherwind penetration leads to largerfrequency fluctuations.

■ We also see that for the same installedwind power, there is roughly twice as muchwind generation in summer than in winter.This is because trade winds prevail all yearin Gran Canaria, but they are most intenseduring spring and summer.

wind x1 wind x4wind x2

● current installed wind power● 2 and 4 times the current installed wind power

● Large fluctuations are associated withdeterministic events which the model cannotpredict unless they are recorded in the 10-minute data (model input).