impact of aerosols in photovoltaic energy production · context introduction results summary and...

20
IMPACT OF AEROSOLS IN PHOTOVOLTAIC ENERGY PRODUCTION C. Guti´ errez 1 , S. Somot 2 , P. Nabat 2 , M. Mallet 2 , M.O. Molina 1 , M.A. Gaertner 1 , O. Perpi˜ an 3 1 Facultad de Ambientales y Bioqu´ ımica Universidad de Castilla-La Mancha, Toledo, Spain 2 Centre National de Recherches M´ et´ eorologiques (CNRM) et´ eo France, Toulouse, France 3 Electrical Engineering Department ETSIDI-UPM, Madrid, Spain Climate Change impacts in the Mediterranean Region. 16-18 october 2017

Upload: phamminh

Post on 08-Nov-2018

212 views

Category:

Documents


0 download

TRANSCRIPT

IMPACT OF AEROSOLS INPHOTOVOLTAIC ENERGY

PRODUCTION

C. Gutierrez1, S. Somot2, P. Nabat2, M. Mallet2, M.O. Molina1,M.A. Gaertner1, O. Perpinan3

1Facultad de Ambientales y BioquımicaUniversidad de Castilla-La Mancha, Toledo, Spain

2Centre National de Recherches Meteorologiques (CNRM)Meteo France, Toulouse, France

3Electrical Engineering DepartmentETSIDI-UPM, Madrid, Spain

Climate Change impacts in the Mediterranean Region.16-18 october 2017

INTRODUCTION

SCOPE : Increase of renewable energies installed capacity:

I High penetration challenges: variability.

MOTIVATION : Photovoltaic technology

I 1 variability: CLOUDS + AEROSOLS

Aerosols can affect the photovoltaic energy production in theshort-term and can lead to misleading bankability studies for some

projects in the long-term.

I 2 CGMs and RCMs discrepancy in shortwave solar radiation, SW,scenarios over Europe (Bartok et al.).

OBJECTIVE: Assess the impact of aerosols on PV productionover the Euro-Mediterranean region and its role in projected SWscenarios.

INTRODUCTION

APPROACH: A photovoltaic system transform solar irradiationinto electricity.

2 models are needed for the assessment/forecast of the PV production:

I Climate model for SW + PV production model.

2 different analysis:

I simulations 2003-2009

I simulations RCP 4.5 scenario

INTRODUCTION

APPROACH: A photovoltaic system transform solar irradiationinto electricity.

2 models are needed for the assessment/forecast of the PV production:

I Climate model for SW + PV production model.

2 different analysis:

I simulations 2003-2009

I simulations RCP 4.5 scenario

INTRODUCTION

APPROACH: A photovoltaic system transform solar irradiationinto electricity.

2 models are needed for the assessment/forecast of the PV production:

I Climate model for SW + PV production model.

2 different analysis:

I simulations 2003-2009

I simulations RCP 4.5 scenario

2003-2009

2 simulations AER and NO.CNRM-RCSM4

I AER includes realistic interannual monthly AOD climatology: Nabat et al.2015a, Climate Dynamics

Figura: Nabat et al. 2015a, Climate Dynamics

1. Surface shortwave (SW) comparison with satellite product.I CM-SAF satellite data. SARAH dataset.

2. PVoutput by tracking system

Annual mean of SW. Relative differences

10°W

0°10°E 20°E

30°E

40°E

50°E

25°N

30°N

35°N

40°N

45°N

50°N

55°N CAER.SAT

10°W

0°10°E 20°E

30°E

40°E

50°E

25°N

30°N

35°N

40°N

45°N

50°N

55°N CNO.SAT

−0.1

0.0

0.1

0.2

0.3

0.4

0.5

AER-SAT

NO-SAT

Relative difference in PV productionAER-NO

I Most of the domain around 5 %

I Higher values in northern europe, above 10 %

Seasonal relative differences.

FIXED

TWO

I Differences in the seasonal spatial pattern.I Winter has higher relative values.I Summer has lower relative values but around 5 % for fixed panels in most

of the domain.

PV real data: SEVILLA

AOD

valu

e

0.0

0.5

1.0

0.10 0.15 0.20

●●

●●

●●

●●

caercnosat

I Difference between AER and real data is generally less than 0.5 kWh.

I In some months, AER is closer to real data than CM-SAF dataset.

I AER is better than NO specially in months with high AOD.

SCENARIOS: MED-CORDEX

2 RCMs: CNRM-ALADIN5.2ENEA-PROTHEUS.

I Same GCM: CNRM-CM5

I 50km x 50 km

I RCP 4.5I Diferent aerosol scheme:

ENEA-PROTHEUS:no aerosolCNRM-ALADIN5.2:aerosoldataset (Szopa et al., 2012) 2Dvariability/time-evolving bydecade. 5 different species.

AOD anomaly JJA(2076/2100)-(1981/2005)

JJA anomaly. 1981/2005

SW CLT

I ENEA-PROTHEUS does not reproduce historical brightening.

I Increase in RSDS for both models.

I SW increase not directly linked with CLT decrease forCNRM-ALADIN.

JJA anomaly. 1981/2005

SW vs. CLT SW vs. AOD

I Linear regression between SW and CLT show higher scores forENEA-PROTHEUS ( 0.72 ALADIN5.2, 0.88 PROTHEUS).

JJA anomaly. 1981/2005

SW vs. CLT SW vs. AOD

I Linear regression between SW and CLT show higher scores forENEA-PROTHEUS ( 0.72 ALADIN5.2, 0.88 PROTHEUS).

I The linear model for AOD shows a r2 coefficient of 0.89.

I SW(clt, aod): model r2 is 0.97

JJA anomaly. 1981/2005

SW vs. CLT SW vs. AOD

I Linear regression between SW and CLT show higher scores forENEA-PROTHEUS ( 0.72 ALADIN5.2, 0.88 PROTHEUS).

I The linear model for AOD shows a r2 coefficient of 0.89.

I SW(clt, aod): model r2 is 0.97

JJA mean anomaly (2076/2100)-(1981/2005)

SW

W/m^2

CLT

%

I Different spatial patterns. SW in ENEA-PROTHEUS correlates better with

CLT.

I With some exceptions, the increase of SW is clear over the domain.

JJA mean anomaly (2076/2100)-(1981/2005)

SW

W/m^2

CLT

%

I Spatial correlation values:I CNRM: SW vs. AOD = - 0.55 / SW vs CLT = - 0.70I ENEA: SW vs. CLT= - 0.86

CONTEXTINTRODUCTION

RESULTSSUMMARY AND FUTURE

Summary

I Aerosols have an impact on PV production that is non negligiblein most places of the studied domain for annual and seasonal terms.

I There is a clear improvement on the representation of SW withthe aerosol-included simulations.

I Modelled PV production is close to real data for the periodanalysed.

I Scenarios show the impact of aerosols in SW could be as importantas clouds.

CONTEXTINTRODUCTION

RESULTSSUMMARY AND FUTURE

Next steps

I More real PV data for evaluation ?

I PV production analysis for scenario rcp4.5 with different trackingtypes.

I Use more RCMs ? (FPS MED-CORDEX aerosols)

CONTEXTINTRODUCTION

RESULTSSUMMARY AND FUTURE

Thank you for your attention

[email protected]